{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.9998639733387744, "eval_steps": 1.0, "global_step": 14702, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "grad_norm": 47.324086306621034, "learning_rate": 2.2624434389140274e-08, "loss": 1.5816, "step": 1 }, { "epoch": 0.0, "grad_norm": 46.55241795899525, "learning_rate": 4.524886877828055e-08, "loss": 1.5823, "step": 2 }, { "epoch": 0.0, "grad_norm": 49.13895279550911, "learning_rate": 6.787330316742082e-08, "loss": 1.5408, "step": 3 }, { "epoch": 0.0, "grad_norm": 45.15076674235958, "learning_rate": 9.04977375565611e-08, "loss": 1.4732, "step": 4 }, { "epoch": 0.0, "grad_norm": 49.90336245215782, "learning_rate": 1.1312217194570136e-07, "loss": 1.4564, "step": 5 }, { "epoch": 0.0, "grad_norm": 44.83436542714078, "learning_rate": 1.3574660633484163e-07, "loss": 1.5114, "step": 6 }, { "epoch": 0.0, "grad_norm": 45.865964499669275, "learning_rate": 1.583710407239819e-07, "loss": 1.6831, "step": 7 }, { "epoch": 0.0, "grad_norm": 44.347514723877744, "learning_rate": 1.809954751131222e-07, "loss": 1.5625, "step": 8 }, { "epoch": 0.0, "grad_norm": 45.83313773425252, "learning_rate": 2.0361990950226246e-07, "loss": 1.5317, "step": 9 }, { "epoch": 0.0, "grad_norm": 41.45860560211038, "learning_rate": 2.2624434389140273e-07, "loss": 1.5189, "step": 10 }, { "epoch": 0.0, "grad_norm": 42.688663468191805, "learning_rate": 2.4886877828054297e-07, "loss": 1.4978, "step": 11 }, { "epoch": 0.0, "grad_norm": 40.78012648920121, "learning_rate": 2.7149321266968326e-07, "loss": 1.4846, "step": 12 }, { "epoch": 0.0, "grad_norm": 41.60739726170051, "learning_rate": 2.9411764705882356e-07, "loss": 1.5908, "step": 13 }, { "epoch": 0.0, "grad_norm": 43.45241133693418, "learning_rate": 3.167420814479638e-07, "loss": 1.4583, "step": 14 }, { "epoch": 0.0, "grad_norm": 45.4984372095751, "learning_rate": 3.393665158371041e-07, "loss": 1.6334, "step": 15 }, { "epoch": 0.0, "grad_norm": 41.60364787946703, "learning_rate": 3.619909502262444e-07, "loss": 1.452, "step": 16 }, { "epoch": 0.0, "grad_norm": 39.592622664320274, "learning_rate": 3.846153846153847e-07, "loss": 1.5667, "step": 17 }, { "epoch": 0.0, "grad_norm": 41.5719924341463, "learning_rate": 4.072398190045249e-07, "loss": 1.5943, "step": 18 }, { "epoch": 0.0, "grad_norm": 39.85639252580765, "learning_rate": 4.298642533936652e-07, "loss": 1.4992, "step": 19 }, { "epoch": 0.0, "grad_norm": 37.89783825675412, "learning_rate": 4.5248868778280546e-07, "loss": 1.4051, "step": 20 }, { "epoch": 0.0, "grad_norm": 33.555674518958746, "learning_rate": 4.751131221719457e-07, "loss": 1.3337, "step": 21 }, { "epoch": 0.0, "grad_norm": 29.557897923439626, "learning_rate": 4.977375565610859e-07, "loss": 1.4861, "step": 22 }, { "epoch": 0.0, "grad_norm": 28.049966085087643, "learning_rate": 5.203619909502263e-07, "loss": 1.392, "step": 23 }, { "epoch": 0.0, "grad_norm": 28.450726909765585, "learning_rate": 5.429864253393665e-07, "loss": 1.3398, "step": 24 }, { "epoch": 0.0, "grad_norm": 27.17760006850618, "learning_rate": 5.656108597285068e-07, "loss": 1.3233, "step": 25 }, { "epoch": 0.0, "grad_norm": 29.94609817035504, "learning_rate": 5.882352941176471e-07, "loss": 1.496, "step": 26 }, { "epoch": 0.0, "grad_norm": 25.099922019613906, "learning_rate": 6.108597285067874e-07, "loss": 1.2783, "step": 27 }, { "epoch": 0.0, "grad_norm": 19.902377368792834, "learning_rate": 6.334841628959276e-07, "loss": 1.2601, "step": 28 }, { "epoch": 0.0, "grad_norm": 17.740497386863442, "learning_rate": 6.56108597285068e-07, "loss": 1.2398, "step": 29 }, { "epoch": 0.0, "grad_norm": 16.540755999769434, "learning_rate": 6.787330316742082e-07, "loss": 1.0646, "step": 30 }, { "epoch": 0.0, "grad_norm": 17.797706005619172, "learning_rate": 7.013574660633485e-07, "loss": 1.1898, "step": 31 }, { "epoch": 0.0, "grad_norm": 16.71336993358581, "learning_rate": 7.239819004524888e-07, "loss": 1.1811, "step": 32 }, { "epoch": 0.0, "grad_norm": 15.755817054284199, "learning_rate": 7.46606334841629e-07, "loss": 1.1472, "step": 33 }, { "epoch": 0.0, "grad_norm": 14.307559200505587, "learning_rate": 7.692307692307694e-07, "loss": 1.2365, "step": 34 }, { "epoch": 0.0, "grad_norm": 13.162217628169689, "learning_rate": 7.918552036199095e-07, "loss": 1.1582, "step": 35 }, { "epoch": 0.0, "grad_norm": 15.175861273017157, "learning_rate": 8.144796380090498e-07, "loss": 1.0578, "step": 36 }, { "epoch": 0.01, "grad_norm": 13.611671510782166, "learning_rate": 8.3710407239819e-07, "loss": 1.0618, "step": 37 }, { "epoch": 0.01, "grad_norm": 11.263794135489798, "learning_rate": 8.597285067873304e-07, "loss": 1.1027, "step": 38 }, { "epoch": 0.01, "grad_norm": 12.06017805672213, "learning_rate": 8.823529411764707e-07, "loss": 1.0898, "step": 39 }, { "epoch": 0.01, "grad_norm": 13.02973373693591, "learning_rate": 9.049773755656109e-07, "loss": 1.1553, "step": 40 }, { "epoch": 0.01, "grad_norm": 15.627880018949988, "learning_rate": 9.276018099547512e-07, "loss": 1.0333, "step": 41 }, { "epoch": 0.01, "grad_norm": 14.56903556430717, "learning_rate": 9.502262443438914e-07, "loss": 0.9733, "step": 42 }, { "epoch": 0.01, "grad_norm": 11.218788463930702, "learning_rate": 9.728506787330317e-07, "loss": 1.0478, "step": 43 }, { "epoch": 0.01, "grad_norm": 9.174858574331619, "learning_rate": 9.954751131221719e-07, "loss": 0.9326, "step": 44 }, { "epoch": 0.01, "grad_norm": 11.473450399932556, "learning_rate": 1.0180995475113123e-06, "loss": 0.9595, "step": 45 }, { "epoch": 0.01, "grad_norm": 12.496532138592856, "learning_rate": 1.0407239819004527e-06, "loss": 0.8796, "step": 46 }, { "epoch": 0.01, "grad_norm": 10.948400278224577, "learning_rate": 1.0633484162895929e-06, "loss": 1.0361, "step": 47 }, { "epoch": 0.01, "grad_norm": 9.810414417895084, "learning_rate": 1.085972850678733e-06, "loss": 0.949, "step": 48 }, { "epoch": 0.01, "grad_norm": 10.427447348702707, "learning_rate": 1.1085972850678732e-06, "loss": 0.9804, "step": 49 }, { "epoch": 0.01, "grad_norm": 16.10481674138479, "learning_rate": 1.1312217194570136e-06, "loss": 0.9345, "step": 50 }, { "epoch": 0.01, "grad_norm": 13.864301964288783, "learning_rate": 1.153846153846154e-06, "loss": 0.8113, "step": 51 }, { "epoch": 0.01, "grad_norm": 10.942810263929978, "learning_rate": 1.1764705882352942e-06, "loss": 0.8108, "step": 52 }, { "epoch": 0.01, "grad_norm": 11.625878490193923, "learning_rate": 1.1990950226244346e-06, "loss": 0.8921, "step": 53 }, { "epoch": 0.01, "grad_norm": 8.018411115938568, "learning_rate": 1.2217194570135748e-06, "loss": 0.8493, "step": 54 }, { "epoch": 0.01, "grad_norm": 8.439552239774411, "learning_rate": 1.244343891402715e-06, "loss": 0.9298, "step": 55 }, { "epoch": 0.01, "grad_norm": 7.652302205122982, "learning_rate": 1.2669683257918552e-06, "loss": 0.8978, "step": 56 }, { "epoch": 0.01, "grad_norm": 7.744335253674733, "learning_rate": 1.2895927601809958e-06, "loss": 0.8151, "step": 57 }, { "epoch": 0.01, "grad_norm": 6.91407272961568, "learning_rate": 1.312217194570136e-06, "loss": 0.8813, "step": 58 }, { "epoch": 0.01, "grad_norm": 6.164028922350728, "learning_rate": 1.3348416289592762e-06, "loss": 0.7766, "step": 59 }, { "epoch": 0.01, "grad_norm": 6.0763494523346075, "learning_rate": 1.3574660633484164e-06, "loss": 0.7626, "step": 60 }, { "epoch": 0.01, "grad_norm": 5.821241933822436, "learning_rate": 1.3800904977375568e-06, "loss": 0.8693, "step": 61 }, { "epoch": 0.01, "grad_norm": 6.098963089757372, "learning_rate": 1.402714932126697e-06, "loss": 0.9117, "step": 62 }, { "epoch": 0.01, "grad_norm": 7.486550675384216, "learning_rate": 1.4253393665158371e-06, "loss": 0.8778, "step": 63 }, { "epoch": 0.01, "grad_norm": 7.628281051641215, "learning_rate": 1.4479638009049775e-06, "loss": 0.7032, "step": 64 }, { "epoch": 0.01, "grad_norm": 7.166089376289424, "learning_rate": 1.4705882352941177e-06, "loss": 0.8802, "step": 65 }, { "epoch": 0.01, "grad_norm": 5.537778495645337, "learning_rate": 1.493212669683258e-06, "loss": 0.7473, "step": 66 }, { "epoch": 0.01, "grad_norm": 6.446435892440292, "learning_rate": 1.5158371040723981e-06, "loss": 0.8916, "step": 67 }, { "epoch": 0.01, "grad_norm": 6.7515210205527945, "learning_rate": 1.5384615384615387e-06, "loss": 0.7532, "step": 68 }, { "epoch": 0.01, "grad_norm": 6.397752826014561, "learning_rate": 1.561085972850679e-06, "loss": 0.8038, "step": 69 }, { "epoch": 0.01, "grad_norm": 6.410458846809738, "learning_rate": 1.583710407239819e-06, "loss": 0.7311, "step": 70 }, { "epoch": 0.01, "grad_norm": 5.944989756627878, "learning_rate": 1.6063348416289595e-06, "loss": 0.7307, "step": 71 }, { "epoch": 0.01, "grad_norm": 6.834158735141168, "learning_rate": 1.6289592760180997e-06, "loss": 0.764, "step": 72 }, { "epoch": 0.01, "grad_norm": 5.3338567864969155, "learning_rate": 1.6515837104072399e-06, "loss": 0.6867, "step": 73 }, { "epoch": 0.01, "grad_norm": 5.291423480136725, "learning_rate": 1.67420814479638e-06, "loss": 0.8161, "step": 74 }, { "epoch": 0.01, "grad_norm": 6.767243902494978, "learning_rate": 1.6968325791855207e-06, "loss": 0.7883, "step": 75 }, { "epoch": 0.01, "grad_norm": 6.79534224852462, "learning_rate": 1.7194570135746609e-06, "loss": 0.6856, "step": 76 }, { "epoch": 0.01, "grad_norm": 4.6697021545763, "learning_rate": 1.742081447963801e-06, "loss": 0.7719, "step": 77 }, { "epoch": 0.01, "grad_norm": 4.732938758687531, "learning_rate": 1.7647058823529414e-06, "loss": 0.7128, "step": 78 }, { "epoch": 0.01, "grad_norm": 6.988881014617995, "learning_rate": 1.7873303167420816e-06, "loss": 0.7028, "step": 79 }, { "epoch": 0.01, "grad_norm": 7.161983351757184, "learning_rate": 1.8099547511312218e-06, "loss": 0.7224, "step": 80 }, { "epoch": 0.01, "grad_norm": 5.524640554306544, "learning_rate": 1.832579185520362e-06, "loss": 0.6801, "step": 81 }, { "epoch": 0.01, "grad_norm": 5.830791500495103, "learning_rate": 1.8552036199095024e-06, "loss": 0.7416, "step": 82 }, { "epoch": 0.01, "grad_norm": 5.294140231237935, "learning_rate": 1.8778280542986426e-06, "loss": 0.6492, "step": 83 }, { "epoch": 0.01, "grad_norm": 8.009330539734682, "learning_rate": 1.9004524886877828e-06, "loss": 0.8042, "step": 84 }, { "epoch": 0.01, "grad_norm": 7.575644345714258, "learning_rate": 1.9230769230769234e-06, "loss": 0.7082, "step": 85 }, { "epoch": 0.01, "grad_norm": 6.459333839847727, "learning_rate": 1.9457013574660634e-06, "loss": 0.7182, "step": 86 }, { "epoch": 0.01, "grad_norm": 5.80715570330473, "learning_rate": 1.9683257918552038e-06, "loss": 0.6314, "step": 87 }, { "epoch": 0.01, "grad_norm": 5.447084308804807, "learning_rate": 1.9909502262443437e-06, "loss": 0.8296, "step": 88 }, { "epoch": 0.01, "grad_norm": 5.262205045988835, "learning_rate": 2.0135746606334846e-06, "loss": 0.6603, "step": 89 }, { "epoch": 0.01, "grad_norm": 5.846945680006292, "learning_rate": 2.0361990950226245e-06, "loss": 0.7179, "step": 90 }, { "epoch": 0.01, "grad_norm": 5.172276581137405, "learning_rate": 2.058823529411765e-06, "loss": 0.6702, "step": 91 }, { "epoch": 0.01, "grad_norm": 5.006354666298998, "learning_rate": 2.0814479638009053e-06, "loss": 0.7271, "step": 92 }, { "epoch": 0.01, "grad_norm": 4.924292404388525, "learning_rate": 2.1040723981900453e-06, "loss": 0.7608, "step": 93 }, { "epoch": 0.01, "grad_norm": 5.790612318345288, "learning_rate": 2.1266968325791857e-06, "loss": 0.7752, "step": 94 }, { "epoch": 0.01, "grad_norm": 5.965318445264805, "learning_rate": 2.149321266968326e-06, "loss": 0.7647, "step": 95 }, { "epoch": 0.01, "grad_norm": 6.010675440272277, "learning_rate": 2.171945701357466e-06, "loss": 0.7228, "step": 96 }, { "epoch": 0.01, "grad_norm": 5.16174287982009, "learning_rate": 2.1945701357466065e-06, "loss": 0.6611, "step": 97 }, { "epoch": 0.01, "grad_norm": 5.9438947745821435, "learning_rate": 2.2171945701357465e-06, "loss": 0.6668, "step": 98 }, { "epoch": 0.01, "grad_norm": 5.102385364399148, "learning_rate": 2.2398190045248873e-06, "loss": 0.6678, "step": 99 }, { "epoch": 0.01, "grad_norm": 5.65456241531793, "learning_rate": 2.2624434389140273e-06, "loss": 0.7047, "step": 100 }, { "epoch": 0.01, "grad_norm": 5.850744611788548, "learning_rate": 2.2850678733031677e-06, "loss": 0.6983, "step": 101 }, { "epoch": 0.01, "grad_norm": 5.769974039671421, "learning_rate": 2.307692307692308e-06, "loss": 0.7315, "step": 102 }, { "epoch": 0.01, "grad_norm": 6.051256504043777, "learning_rate": 2.330316742081448e-06, "loss": 0.6675, "step": 103 }, { "epoch": 0.01, "grad_norm": 5.416544548296377, "learning_rate": 2.3529411764705885e-06, "loss": 0.6852, "step": 104 }, { "epoch": 0.01, "grad_norm": 7.18084158882696, "learning_rate": 2.3755656108597284e-06, "loss": 0.6705, "step": 105 }, { "epoch": 0.01, "grad_norm": 5.130120842619215, "learning_rate": 2.3981900452488693e-06, "loss": 0.6827, "step": 106 }, { "epoch": 0.01, "grad_norm": 6.019887714583096, "learning_rate": 2.4208144796380092e-06, "loss": 0.7535, "step": 107 }, { "epoch": 0.01, "grad_norm": 5.933745513990921, "learning_rate": 2.4434389140271496e-06, "loss": 0.6801, "step": 108 }, { "epoch": 0.01, "grad_norm": 5.351594396678103, "learning_rate": 2.46606334841629e-06, "loss": 0.723, "step": 109 }, { "epoch": 0.01, "grad_norm": 5.027409944173364, "learning_rate": 2.48868778280543e-06, "loss": 0.6353, "step": 110 }, { "epoch": 0.02, "grad_norm": 5.296079690592357, "learning_rate": 2.5113122171945704e-06, "loss": 0.7517, "step": 111 }, { "epoch": 0.02, "grad_norm": 5.211046813190376, "learning_rate": 2.5339366515837104e-06, "loss": 0.7573, "step": 112 }, { "epoch": 0.02, "grad_norm": 5.643720086511115, "learning_rate": 2.5565610859728508e-06, "loss": 0.5656, "step": 113 }, { "epoch": 0.02, "grad_norm": 4.437590229082709, "learning_rate": 2.5791855203619916e-06, "loss": 0.6163, "step": 114 }, { "epoch": 0.02, "grad_norm": 4.880547970865982, "learning_rate": 2.6018099547511316e-06, "loss": 0.6876, "step": 115 }, { "epoch": 0.02, "grad_norm": 4.965789057982377, "learning_rate": 2.624434389140272e-06, "loss": 0.6646, "step": 116 }, { "epoch": 0.02, "grad_norm": 6.109575339514163, "learning_rate": 2.647058823529412e-06, "loss": 0.7097, "step": 117 }, { "epoch": 0.02, "grad_norm": 5.163770524534186, "learning_rate": 2.6696832579185524e-06, "loss": 0.5827, "step": 118 }, { "epoch": 0.02, "grad_norm": 4.684590410516369, "learning_rate": 2.6923076923076923e-06, "loss": 0.5878, "step": 119 }, { "epoch": 0.02, "grad_norm": 5.357062670189611, "learning_rate": 2.7149321266968327e-06, "loss": 0.6379, "step": 120 }, { "epoch": 0.02, "grad_norm": 5.501204324799209, "learning_rate": 2.737556561085973e-06, "loss": 0.7096, "step": 121 }, { "epoch": 0.02, "grad_norm": 5.667376649225802, "learning_rate": 2.7601809954751135e-06, "loss": 0.6436, "step": 122 }, { "epoch": 0.02, "grad_norm": 6.2898383760149255, "learning_rate": 2.782805429864254e-06, "loss": 0.6664, "step": 123 }, { "epoch": 0.02, "grad_norm": 5.632240377569418, "learning_rate": 2.805429864253394e-06, "loss": 0.639, "step": 124 }, { "epoch": 0.02, "grad_norm": 4.489296366020518, "learning_rate": 2.8280542986425343e-06, "loss": 0.6329, "step": 125 }, { "epoch": 0.02, "grad_norm": 5.243772137295284, "learning_rate": 2.8506787330316743e-06, "loss": 0.6102, "step": 126 }, { "epoch": 0.02, "grad_norm": 4.231034008924022, "learning_rate": 2.8733031674208147e-06, "loss": 0.6501, "step": 127 }, { "epoch": 0.02, "grad_norm": 5.492055497905278, "learning_rate": 2.895927601809955e-06, "loss": 0.6471, "step": 128 }, { "epoch": 0.02, "grad_norm": 4.458290693407378, "learning_rate": 2.9185520361990955e-06, "loss": 0.541, "step": 129 }, { "epoch": 0.02, "grad_norm": 4.793731343488034, "learning_rate": 2.9411764705882355e-06, "loss": 0.6343, "step": 130 }, { "epoch": 0.02, "grad_norm": 4.871186999217534, "learning_rate": 2.963800904977376e-06, "loss": 0.6353, "step": 131 }, { "epoch": 0.02, "grad_norm": 4.125123506609178, "learning_rate": 2.986425339366516e-06, "loss": 0.6555, "step": 132 }, { "epoch": 0.02, "grad_norm": 5.000864247890614, "learning_rate": 3.0090497737556562e-06, "loss": 0.6381, "step": 133 }, { "epoch": 0.02, "grad_norm": 5.411697960785994, "learning_rate": 3.0316742081447962e-06, "loss": 0.7221, "step": 134 }, { "epoch": 0.02, "grad_norm": 6.125010330780797, "learning_rate": 3.054298642533937e-06, "loss": 0.7271, "step": 135 }, { "epoch": 0.02, "grad_norm": 4.990452474692772, "learning_rate": 3.0769230769230774e-06, "loss": 0.5793, "step": 136 }, { "epoch": 0.02, "grad_norm": 4.930065595340568, "learning_rate": 3.0995475113122174e-06, "loss": 0.6289, "step": 137 }, { "epoch": 0.02, "grad_norm": 5.313732197536749, "learning_rate": 3.122171945701358e-06, "loss": 0.5748, "step": 138 }, { "epoch": 0.02, "grad_norm": 5.306015232745663, "learning_rate": 3.1447963800904978e-06, "loss": 0.6026, "step": 139 }, { "epoch": 0.02, "grad_norm": 5.983303724819943, "learning_rate": 3.167420814479638e-06, "loss": 0.6488, "step": 140 }, { "epoch": 0.02, "grad_norm": 4.940847623079017, "learning_rate": 3.190045248868778e-06, "loss": 0.7011, "step": 141 }, { "epoch": 0.02, "grad_norm": 4.795527290390542, "learning_rate": 3.212669683257919e-06, "loss": 0.6674, "step": 142 }, { "epoch": 0.02, "grad_norm": 5.53820352359425, "learning_rate": 3.2352941176470594e-06, "loss": 0.6312, "step": 143 }, { "epoch": 0.02, "grad_norm": 5.624603465348181, "learning_rate": 3.2579185520361994e-06, "loss": 0.5864, "step": 144 }, { "epoch": 0.02, "grad_norm": 4.6446256166785815, "learning_rate": 3.2805429864253398e-06, "loss": 0.6339, "step": 145 }, { "epoch": 0.02, "grad_norm": 4.970591232334842, "learning_rate": 3.3031674208144797e-06, "loss": 0.6411, "step": 146 }, { "epoch": 0.02, "grad_norm": 4.687374988871516, "learning_rate": 3.32579185520362e-06, "loss": 0.6959, "step": 147 }, { "epoch": 0.02, "grad_norm": 4.49425673592456, "learning_rate": 3.34841628959276e-06, "loss": 0.669, "step": 148 }, { "epoch": 0.02, "grad_norm": 4.620513767898124, "learning_rate": 3.371040723981901e-06, "loss": 0.5703, "step": 149 }, { "epoch": 0.02, "grad_norm": 4.522125905598737, "learning_rate": 3.3936651583710413e-06, "loss": 0.6375, "step": 150 }, { "epoch": 0.02, "grad_norm": 4.7696520126131645, "learning_rate": 3.4162895927601813e-06, "loss": 0.6178, "step": 151 }, { "epoch": 0.02, "grad_norm": 4.45962863690474, "learning_rate": 3.4389140271493217e-06, "loss": 0.5984, "step": 152 }, { "epoch": 0.02, "grad_norm": 4.572796520749408, "learning_rate": 3.4615384615384617e-06, "loss": 0.6874, "step": 153 }, { "epoch": 0.02, "grad_norm": 4.390585223297342, "learning_rate": 3.484162895927602e-06, "loss": 0.6562, "step": 154 }, { "epoch": 0.02, "grad_norm": 4.116227149244516, "learning_rate": 3.506787330316742e-06, "loss": 0.5704, "step": 155 }, { "epoch": 0.02, "grad_norm": 4.961590266098592, "learning_rate": 3.529411764705883e-06, "loss": 0.6752, "step": 156 }, { "epoch": 0.02, "grad_norm": 4.256875003880694, "learning_rate": 3.552036199095023e-06, "loss": 0.628, "step": 157 }, { "epoch": 0.02, "grad_norm": 6.702355331888742, "learning_rate": 3.5746606334841633e-06, "loss": 0.5917, "step": 158 }, { "epoch": 0.02, "grad_norm": 4.272391416292173, "learning_rate": 3.5972850678733032e-06, "loss": 0.6089, "step": 159 }, { "epoch": 0.02, "grad_norm": 4.633713926731377, "learning_rate": 3.6199095022624436e-06, "loss": 0.6267, "step": 160 }, { "epoch": 0.02, "grad_norm": 5.175796106300143, "learning_rate": 3.642533936651584e-06, "loss": 0.6574, "step": 161 }, { "epoch": 0.02, "grad_norm": 5.0072674144621425, "learning_rate": 3.665158371040724e-06, "loss": 0.6438, "step": 162 }, { "epoch": 0.02, "grad_norm": 4.187272731461751, "learning_rate": 3.687782805429865e-06, "loss": 0.6008, "step": 163 }, { "epoch": 0.02, "grad_norm": 4.9227493687004, "learning_rate": 3.710407239819005e-06, "loss": 0.5809, "step": 164 }, { "epoch": 0.02, "grad_norm": 4.906786983254411, "learning_rate": 3.7330316742081452e-06, "loss": 0.6538, "step": 165 }, { "epoch": 0.02, "grad_norm": 4.588266663351423, "learning_rate": 3.755656108597285e-06, "loss": 0.6323, "step": 166 }, { "epoch": 0.02, "grad_norm": 5.488280362206947, "learning_rate": 3.7782805429864256e-06, "loss": 0.6334, "step": 167 }, { "epoch": 0.02, "grad_norm": 4.849895508432989, "learning_rate": 3.8009049773755656e-06, "loss": 0.603, "step": 168 }, { "epoch": 0.02, "grad_norm": 4.658650694319783, "learning_rate": 3.8235294117647055e-06, "loss": 0.6027, "step": 169 }, { "epoch": 0.02, "grad_norm": 4.809442256076622, "learning_rate": 3.846153846153847e-06, "loss": 0.5709, "step": 170 }, { "epoch": 0.02, "grad_norm": 5.504781646277077, "learning_rate": 3.868778280542987e-06, "loss": 0.6588, "step": 171 }, { "epoch": 0.02, "grad_norm": 4.779234836524345, "learning_rate": 3.891402714932127e-06, "loss": 0.6251, "step": 172 }, { "epoch": 0.02, "grad_norm": 5.157215712063347, "learning_rate": 3.914027149321267e-06, "loss": 0.6081, "step": 173 }, { "epoch": 0.02, "grad_norm": 4.452015851797737, "learning_rate": 3.9366515837104075e-06, "loss": 0.6197, "step": 174 }, { "epoch": 0.02, "grad_norm": 4.460589761431551, "learning_rate": 3.959276018099548e-06, "loss": 0.5652, "step": 175 }, { "epoch": 0.02, "grad_norm": 5.73531680702189, "learning_rate": 3.9819004524886875e-06, "loss": 0.6016, "step": 176 }, { "epoch": 0.02, "grad_norm": 5.4304143917830405, "learning_rate": 4.004524886877829e-06, "loss": 0.7197, "step": 177 }, { "epoch": 0.02, "grad_norm": 5.089019407911785, "learning_rate": 4.027149321266969e-06, "loss": 0.6007, "step": 178 }, { "epoch": 0.02, "grad_norm": 5.59800261806494, "learning_rate": 4.049773755656109e-06, "loss": 0.5145, "step": 179 }, { "epoch": 0.02, "grad_norm": 5.693589549795068, "learning_rate": 4.072398190045249e-06, "loss": 0.6264, "step": 180 }, { "epoch": 0.02, "grad_norm": 5.407924131839285, "learning_rate": 4.0950226244343895e-06, "loss": 0.5818, "step": 181 }, { "epoch": 0.02, "grad_norm": 4.724598015982189, "learning_rate": 4.11764705882353e-06, "loss": 0.5239, "step": 182 }, { "epoch": 0.02, "grad_norm": 4.717749408782815, "learning_rate": 4.1402714932126695e-06, "loss": 0.5379, "step": 183 }, { "epoch": 0.03, "grad_norm": 4.583757125478642, "learning_rate": 4.162895927601811e-06, "loss": 0.5606, "step": 184 }, { "epoch": 0.03, "grad_norm": 5.173519691867786, "learning_rate": 4.185520361990951e-06, "loss": 0.5966, "step": 185 }, { "epoch": 0.03, "grad_norm": 5.221904938559157, "learning_rate": 4.208144796380091e-06, "loss": 0.6387, "step": 186 }, { "epoch": 0.03, "grad_norm": 4.541599429288671, "learning_rate": 4.230769230769231e-06, "loss": 0.5527, "step": 187 }, { "epoch": 0.03, "grad_norm": 4.667769806137969, "learning_rate": 4.2533936651583714e-06, "loss": 0.5874, "step": 188 }, { "epoch": 0.03, "grad_norm": 5.311565786848941, "learning_rate": 4.276018099547512e-06, "loss": 0.7047, "step": 189 }, { "epoch": 0.03, "grad_norm": 4.874907465142788, "learning_rate": 4.298642533936652e-06, "loss": 0.6195, "step": 190 }, { "epoch": 0.03, "grad_norm": 4.473952448270892, "learning_rate": 4.321266968325793e-06, "loss": 0.6708, "step": 191 }, { "epoch": 0.03, "grad_norm": 4.467141358015583, "learning_rate": 4.343891402714932e-06, "loss": 0.6047, "step": 192 }, { "epoch": 0.03, "grad_norm": 5.2686132960475245, "learning_rate": 4.366515837104073e-06, "loss": 0.6328, "step": 193 }, { "epoch": 0.03, "grad_norm": 4.626293540962446, "learning_rate": 4.389140271493213e-06, "loss": 0.6284, "step": 194 }, { "epoch": 0.03, "grad_norm": 4.582429057524721, "learning_rate": 4.411764705882353e-06, "loss": 0.6396, "step": 195 }, { "epoch": 0.03, "grad_norm": 5.648289579781185, "learning_rate": 4.434389140271493e-06, "loss": 0.6515, "step": 196 }, { "epoch": 0.03, "grad_norm": 4.495533438702332, "learning_rate": 4.457013574660634e-06, "loss": 0.6094, "step": 197 }, { "epoch": 0.03, "grad_norm": 4.954121242845797, "learning_rate": 4.479638009049775e-06, "loss": 0.5781, "step": 198 }, { "epoch": 0.03, "grad_norm": 4.204465477240456, "learning_rate": 4.502262443438914e-06, "loss": 0.586, "step": 199 }, { "epoch": 0.03, "grad_norm": 4.993322786300454, "learning_rate": 4.5248868778280546e-06, "loss": 0.6251, "step": 200 }, { "epoch": 0.03, "grad_norm": 4.127682586174253, "learning_rate": 4.547511312217195e-06, "loss": 0.5356, "step": 201 }, { "epoch": 0.03, "grad_norm": 4.5285529588991364, "learning_rate": 4.570135746606335e-06, "loss": 0.515, "step": 202 }, { "epoch": 0.03, "grad_norm": 4.744936477204903, "learning_rate": 4.592760180995475e-06, "loss": 0.6117, "step": 203 }, { "epoch": 0.03, "grad_norm": 3.9012871587992723, "learning_rate": 4.615384615384616e-06, "loss": 0.586, "step": 204 }, { "epoch": 0.03, "grad_norm": 4.611456870027148, "learning_rate": 4.6380090497737566e-06, "loss": 0.6019, "step": 205 }, { "epoch": 0.03, "grad_norm": 4.320652802100966, "learning_rate": 4.660633484162896e-06, "loss": 0.5293, "step": 206 }, { "epoch": 0.03, "grad_norm": 4.5688016555479924, "learning_rate": 4.6832579185520365e-06, "loss": 0.6136, "step": 207 }, { "epoch": 0.03, "grad_norm": 4.950863166808455, "learning_rate": 4.705882352941177e-06, "loss": 0.6006, "step": 208 }, { "epoch": 0.03, "grad_norm": 5.363584428988444, "learning_rate": 4.728506787330317e-06, "loss": 0.6301, "step": 209 }, { "epoch": 0.03, "grad_norm": 4.991673125772768, "learning_rate": 4.751131221719457e-06, "loss": 0.5809, "step": 210 }, { "epoch": 0.03, "grad_norm": 4.53383919402537, "learning_rate": 4.773755656108598e-06, "loss": 0.588, "step": 211 }, { "epoch": 0.03, "grad_norm": 4.591130926570458, "learning_rate": 4.7963800904977385e-06, "loss": 0.5489, "step": 212 }, { "epoch": 0.03, "grad_norm": 4.323494536790714, "learning_rate": 4.819004524886878e-06, "loss": 0.6226, "step": 213 }, { "epoch": 0.03, "grad_norm": 4.403162804664311, "learning_rate": 4.8416289592760185e-06, "loss": 0.5748, "step": 214 }, { "epoch": 0.03, "grad_norm": 4.392539219264291, "learning_rate": 4.864253393665159e-06, "loss": 0.5848, "step": 215 }, { "epoch": 0.03, "grad_norm": 4.840397234228632, "learning_rate": 4.886877828054299e-06, "loss": 0.5771, "step": 216 }, { "epoch": 0.03, "grad_norm": 4.664124372742642, "learning_rate": 4.909502262443439e-06, "loss": 0.6055, "step": 217 }, { "epoch": 0.03, "grad_norm": 5.516984149180152, "learning_rate": 4.93212669683258e-06, "loss": 0.6222, "step": 218 }, { "epoch": 0.03, "grad_norm": 5.067549214799464, "learning_rate": 4.95475113122172e-06, "loss": 0.5812, "step": 219 }, { "epoch": 0.03, "grad_norm": 4.952233316706347, "learning_rate": 4.97737556561086e-06, "loss": 0.5841, "step": 220 }, { "epoch": 0.03, "grad_norm": 4.84707944398091, "learning_rate": 5e-06, "loss": 0.6202, "step": 221 }, { "epoch": 0.03, "grad_norm": 4.7331871922830615, "learning_rate": 5.022624434389141e-06, "loss": 0.5927, "step": 222 }, { "epoch": 0.03, "grad_norm": 4.125132992957289, "learning_rate": 5.045248868778281e-06, "loss": 0.5743, "step": 223 }, { "epoch": 0.03, "grad_norm": 4.832511863559642, "learning_rate": 5.067873303167421e-06, "loss": 0.5291, "step": 224 }, { "epoch": 0.03, "grad_norm": 4.3756688803172095, "learning_rate": 5.090497737556561e-06, "loss": 0.5433, "step": 225 }, { "epoch": 0.03, "grad_norm": 5.305790144628681, "learning_rate": 5.1131221719457016e-06, "loss": 0.6039, "step": 226 }, { "epoch": 0.03, "grad_norm": 4.377369075264761, "learning_rate": 5.135746606334842e-06, "loss": 0.5285, "step": 227 }, { "epoch": 0.03, "grad_norm": 5.449460486285878, "learning_rate": 5.158371040723983e-06, "loss": 0.6129, "step": 228 }, { "epoch": 0.03, "grad_norm": 4.56008922775789, "learning_rate": 5.180995475113123e-06, "loss": 0.609, "step": 229 }, { "epoch": 0.03, "grad_norm": 4.965163458545332, "learning_rate": 5.203619909502263e-06, "loss": 0.6233, "step": 230 }, { "epoch": 0.03, "grad_norm": 4.449054036739927, "learning_rate": 5.2262443438914036e-06, "loss": 0.5563, "step": 231 }, { "epoch": 0.03, "grad_norm": 4.5269479375018955, "learning_rate": 5.248868778280544e-06, "loss": 0.6806, "step": 232 }, { "epoch": 0.03, "grad_norm": 4.76519453579965, "learning_rate": 5.2714932126696835e-06, "loss": 0.6311, "step": 233 }, { "epoch": 0.03, "grad_norm": 4.504696846685045, "learning_rate": 5.294117647058824e-06, "loss": 0.5887, "step": 234 }, { "epoch": 0.03, "grad_norm": 4.193977973008326, "learning_rate": 5.316742081447964e-06, "loss": 0.5909, "step": 235 }, { "epoch": 0.03, "grad_norm": 4.3106929260284295, "learning_rate": 5.339366515837105e-06, "loss": 0.5172, "step": 236 }, { "epoch": 0.03, "grad_norm": 4.780001793323971, "learning_rate": 5.361990950226244e-06, "loss": 0.4837, "step": 237 }, { "epoch": 0.03, "grad_norm": 4.334902506332388, "learning_rate": 5.384615384615385e-06, "loss": 0.5039, "step": 238 }, { "epoch": 0.03, "grad_norm": 4.223117131606598, "learning_rate": 5.407239819004525e-06, "loss": 0.6184, "step": 239 }, { "epoch": 0.03, "grad_norm": 4.666276977515013, "learning_rate": 5.4298642533936655e-06, "loss": 0.6063, "step": 240 }, { "epoch": 0.03, "grad_norm": 4.6084217162928285, "learning_rate": 5.452488687782805e-06, "loss": 0.643, "step": 241 }, { "epoch": 0.03, "grad_norm": 5.035783517894395, "learning_rate": 5.475113122171946e-06, "loss": 0.615, "step": 242 }, { "epoch": 0.03, "grad_norm": 4.693414300344119, "learning_rate": 5.497737556561087e-06, "loss": 0.5687, "step": 243 }, { "epoch": 0.03, "grad_norm": 4.477118423069263, "learning_rate": 5.520361990950227e-06, "loss": 0.506, "step": 244 }, { "epoch": 0.03, "grad_norm": 4.915829014078916, "learning_rate": 5.5429864253393675e-06, "loss": 0.6216, "step": 245 }, { "epoch": 0.03, "grad_norm": 4.727089451369661, "learning_rate": 5.565610859728508e-06, "loss": 0.5768, "step": 246 }, { "epoch": 0.03, "grad_norm": 3.8374705094863284, "learning_rate": 5.588235294117647e-06, "loss": 0.5353, "step": 247 }, { "epoch": 0.03, "grad_norm": 4.042693162925187, "learning_rate": 5.610859728506788e-06, "loss": 0.4587, "step": 248 }, { "epoch": 0.03, "grad_norm": 5.303765054717771, "learning_rate": 5.633484162895928e-06, "loss": 0.5969, "step": 249 }, { "epoch": 0.03, "grad_norm": 5.041866209999289, "learning_rate": 5.656108597285069e-06, "loss": 0.5782, "step": 250 }, { "epoch": 0.03, "grad_norm": 5.314399616659564, "learning_rate": 5.678733031674208e-06, "loss": 0.6363, "step": 251 }, { "epoch": 0.03, "grad_norm": 4.258796549478307, "learning_rate": 5.7013574660633486e-06, "loss": 0.5276, "step": 252 }, { "epoch": 0.03, "grad_norm": 6.068149510032359, "learning_rate": 5.723981900452489e-06, "loss": 0.5316, "step": 253 }, { "epoch": 0.03, "grad_norm": 4.826302008184719, "learning_rate": 5.746606334841629e-06, "loss": 0.5973, "step": 254 }, { "epoch": 0.03, "grad_norm": 4.464112754418021, "learning_rate": 5.769230769230769e-06, "loss": 0.5936, "step": 255 }, { "epoch": 0.03, "grad_norm": 4.463730426742132, "learning_rate": 5.79185520361991e-06, "loss": 0.5483, "step": 256 }, { "epoch": 0.03, "grad_norm": 4.415747986470693, "learning_rate": 5.8144796380090506e-06, "loss": 0.5474, "step": 257 }, { "epoch": 0.04, "grad_norm": 4.311835526859164, "learning_rate": 5.837104072398191e-06, "loss": 0.5738, "step": 258 }, { "epoch": 0.04, "grad_norm": 4.341182104293865, "learning_rate": 5.859728506787331e-06, "loss": 0.5039, "step": 259 }, { "epoch": 0.04, "grad_norm": 4.24943794990233, "learning_rate": 5.882352941176471e-06, "loss": 0.4754, "step": 260 }, { "epoch": 0.04, "grad_norm": 4.323475757436853, "learning_rate": 5.904977375565611e-06, "loss": 0.6001, "step": 261 }, { "epoch": 0.04, "grad_norm": 4.549078559900586, "learning_rate": 5.927601809954752e-06, "loss": 0.5231, "step": 262 }, { "epoch": 0.04, "grad_norm": 5.29008315918479, "learning_rate": 5.950226244343892e-06, "loss": 0.5431, "step": 263 }, { "epoch": 0.04, "grad_norm": 4.551434145441224, "learning_rate": 5.972850678733032e-06, "loss": 0.5188, "step": 264 }, { "epoch": 0.04, "grad_norm": 4.014840367224011, "learning_rate": 5.995475113122172e-06, "loss": 0.536, "step": 265 }, { "epoch": 0.04, "grad_norm": 4.671533531025163, "learning_rate": 6.0180995475113125e-06, "loss": 0.4998, "step": 266 }, { "epoch": 0.04, "grad_norm": 4.689789233763135, "learning_rate": 6.040723981900453e-06, "loss": 0.5625, "step": 267 }, { "epoch": 0.04, "grad_norm": 4.92951754286318, "learning_rate": 6.0633484162895924e-06, "loss": 0.5102, "step": 268 }, { "epoch": 0.04, "grad_norm": 4.608479107843144, "learning_rate": 6.085972850678733e-06, "loss": 0.6335, "step": 269 }, { "epoch": 0.04, "grad_norm": 4.665817552173435, "learning_rate": 6.108597285067874e-06, "loss": 0.5802, "step": 270 }, { "epoch": 0.04, "grad_norm": 4.599385421126586, "learning_rate": 6.1312217194570145e-06, "loss": 0.5896, "step": 271 }, { "epoch": 0.04, "grad_norm": 4.565272527280676, "learning_rate": 6.153846153846155e-06, "loss": 0.5371, "step": 272 }, { "epoch": 0.04, "grad_norm": 4.951314251195843, "learning_rate": 6.176470588235295e-06, "loss": 0.5488, "step": 273 }, { "epoch": 0.04, "grad_norm": 4.753506503008498, "learning_rate": 6.199095022624435e-06, "loss": 0.5399, "step": 274 }, { "epoch": 0.04, "grad_norm": 4.504779533642551, "learning_rate": 6.221719457013575e-06, "loss": 0.4895, "step": 275 }, { "epoch": 0.04, "grad_norm": 4.3417829396284695, "learning_rate": 6.244343891402716e-06, "loss": 0.5055, "step": 276 }, { "epoch": 0.04, "grad_norm": 4.447677953155331, "learning_rate": 6.266968325791856e-06, "loss": 0.5391, "step": 277 }, { "epoch": 0.04, "grad_norm": 4.784073885209587, "learning_rate": 6.2895927601809956e-06, "loss": 0.4711, "step": 278 }, { "epoch": 0.04, "grad_norm": 5.32513529620868, "learning_rate": 6.312217194570136e-06, "loss": 0.609, "step": 279 }, { "epoch": 0.04, "grad_norm": 4.69330444457034, "learning_rate": 6.334841628959276e-06, "loss": 0.5509, "step": 280 }, { "epoch": 0.04, "grad_norm": 4.498934242355806, "learning_rate": 6.357466063348417e-06, "loss": 0.587, "step": 281 }, { "epoch": 0.04, "grad_norm": 4.534903554225795, "learning_rate": 6.380090497737556e-06, "loss": 0.4532, "step": 282 }, { "epoch": 0.04, "grad_norm": 3.8669024656007998, "learning_rate": 6.402714932126697e-06, "loss": 0.5319, "step": 283 }, { "epoch": 0.04, "grad_norm": 3.820843527967093, "learning_rate": 6.425339366515838e-06, "loss": 0.5421, "step": 284 }, { "epoch": 0.04, "grad_norm": 4.322121676924435, "learning_rate": 6.447963800904978e-06, "loss": 0.5307, "step": 285 }, { "epoch": 0.04, "grad_norm": 3.962125741742938, "learning_rate": 6.470588235294119e-06, "loss": 0.5509, "step": 286 }, { "epoch": 0.04, "grad_norm": 4.247009534852939, "learning_rate": 6.493212669683258e-06, "loss": 0.5919, "step": 287 }, { "epoch": 0.04, "grad_norm": 4.329772680596671, "learning_rate": 6.515837104072399e-06, "loss": 0.5274, "step": 288 }, { "epoch": 0.04, "grad_norm": 4.149350457328093, "learning_rate": 6.538461538461539e-06, "loss": 0.5278, "step": 289 }, { "epoch": 0.04, "grad_norm": 4.2137271929822235, "learning_rate": 6.5610859728506795e-06, "loss": 0.6136, "step": 290 }, { "epoch": 0.04, "grad_norm": 4.295935175893815, "learning_rate": 6.583710407239819e-06, "loss": 0.5302, "step": 291 }, { "epoch": 0.04, "grad_norm": 4.796289383983139, "learning_rate": 6.6063348416289595e-06, "loss": 0.5851, "step": 292 }, { "epoch": 0.04, "grad_norm": 4.113245341170916, "learning_rate": 6.6289592760181e-06, "loss": 0.5213, "step": 293 }, { "epoch": 0.04, "grad_norm": 4.08163602906666, "learning_rate": 6.65158371040724e-06, "loss": 0.4859, "step": 294 }, { "epoch": 0.04, "grad_norm": 4.532482620996132, "learning_rate": 6.67420814479638e-06, "loss": 0.5838, "step": 295 }, { "epoch": 0.04, "grad_norm": 5.851215696416483, "learning_rate": 6.69683257918552e-06, "loss": 0.5262, "step": 296 }, { "epoch": 0.04, "grad_norm": 4.84589571886412, "learning_rate": 6.7194570135746615e-06, "loss": 0.4793, "step": 297 }, { "epoch": 0.04, "grad_norm": 4.318899080625823, "learning_rate": 6.742081447963802e-06, "loss": 0.5945, "step": 298 }, { "epoch": 0.04, "grad_norm": 4.440142965362735, "learning_rate": 6.764705882352942e-06, "loss": 0.5617, "step": 299 }, { "epoch": 0.04, "grad_norm": 4.84829205613685, "learning_rate": 6.787330316742083e-06, "loss": 0.5636, "step": 300 }, { "epoch": 0.04, "grad_norm": 4.207782313921567, "learning_rate": 6.809954751131222e-06, "loss": 0.5563, "step": 301 }, { "epoch": 0.04, "grad_norm": 4.6592367226448355, "learning_rate": 6.832579185520363e-06, "loss": 0.55, "step": 302 }, { "epoch": 0.04, "grad_norm": 4.797367006653022, "learning_rate": 6.855203619909503e-06, "loss": 0.6299, "step": 303 }, { "epoch": 0.04, "grad_norm": 4.433443078815136, "learning_rate": 6.8778280542986434e-06, "loss": 0.588, "step": 304 }, { "epoch": 0.04, "grad_norm": 4.885788915405099, "learning_rate": 6.900452488687783e-06, "loss": 0.4871, "step": 305 }, { "epoch": 0.04, "grad_norm": 4.579871402784918, "learning_rate": 6.923076923076923e-06, "loss": 0.6587, "step": 306 }, { "epoch": 0.04, "grad_norm": 4.354344702967108, "learning_rate": 6.945701357466064e-06, "loss": 0.5218, "step": 307 }, { "epoch": 0.04, "grad_norm": 5.0225127911094125, "learning_rate": 6.968325791855204e-06, "loss": 0.6233, "step": 308 }, { "epoch": 0.04, "grad_norm": 4.223500557389156, "learning_rate": 6.990950226244344e-06, "loss": 0.5905, "step": 309 }, { "epoch": 0.04, "grad_norm": 4.208123175547726, "learning_rate": 7.013574660633484e-06, "loss": 0.5089, "step": 310 }, { "epoch": 0.04, "grad_norm": 3.769843414442755, "learning_rate": 7.036199095022625e-06, "loss": 0.4271, "step": 311 }, { "epoch": 0.04, "grad_norm": 4.396811671621064, "learning_rate": 7.058823529411766e-06, "loss": 0.5694, "step": 312 }, { "epoch": 0.04, "grad_norm": 4.596009984873259, "learning_rate": 7.081447963800906e-06, "loss": 0.4803, "step": 313 }, { "epoch": 0.04, "grad_norm": 3.97438124154271, "learning_rate": 7.104072398190046e-06, "loss": 0.5036, "step": 314 }, { "epoch": 0.04, "grad_norm": 5.173346956659891, "learning_rate": 7.126696832579186e-06, "loss": 0.6478, "step": 315 }, { "epoch": 0.04, "grad_norm": 5.417444679211975, "learning_rate": 7.1493212669683265e-06, "loss": 0.5444, "step": 316 }, { "epoch": 0.04, "grad_norm": 4.696128268217875, "learning_rate": 7.171945701357467e-06, "loss": 0.6359, "step": 317 }, { "epoch": 0.04, "grad_norm": 4.5613550864692405, "learning_rate": 7.1945701357466065e-06, "loss": 0.5483, "step": 318 }, { "epoch": 0.04, "grad_norm": 4.32709109590873, "learning_rate": 7.217194570135747e-06, "loss": 0.5488, "step": 319 }, { "epoch": 0.04, "grad_norm": 4.438429576213771, "learning_rate": 7.239819004524887e-06, "loss": 0.5856, "step": 320 }, { "epoch": 0.04, "grad_norm": 4.361050008722124, "learning_rate": 7.262443438914028e-06, "loss": 0.5852, "step": 321 }, { "epoch": 0.04, "grad_norm": 3.8690044451046286, "learning_rate": 7.285067873303168e-06, "loss": 0.4792, "step": 322 }, { "epoch": 0.04, "grad_norm": 4.366360436607657, "learning_rate": 7.307692307692308e-06, "loss": 0.5985, "step": 323 }, { "epoch": 0.04, "grad_norm": 4.543156462356783, "learning_rate": 7.330316742081448e-06, "loss": 0.5698, "step": 324 }, { "epoch": 0.04, "grad_norm": 4.004362120969393, "learning_rate": 7.352941176470589e-06, "loss": 0.5533, "step": 325 }, { "epoch": 0.04, "grad_norm": 4.810032411347552, "learning_rate": 7.37556561085973e-06, "loss": 0.5619, "step": 326 }, { "epoch": 0.04, "grad_norm": 4.523937987753119, "learning_rate": 7.39819004524887e-06, "loss": 0.5901, "step": 327 }, { "epoch": 0.04, "grad_norm": 4.9012309676822845, "learning_rate": 7.42081447963801e-06, "loss": 0.6451, "step": 328 }, { "epoch": 0.04, "grad_norm": 4.487979890130035, "learning_rate": 7.44343891402715e-06, "loss": 0.5427, "step": 329 }, { "epoch": 0.04, "grad_norm": 4.297561861211119, "learning_rate": 7.4660633484162904e-06, "loss": 0.4672, "step": 330 }, { "epoch": 0.05, "grad_norm": 4.227667921604703, "learning_rate": 7.488687782805431e-06, "loss": 0.5162, "step": 331 }, { "epoch": 0.05, "grad_norm": 4.1053040668753, "learning_rate": 7.51131221719457e-06, "loss": 0.5316, "step": 332 }, { "epoch": 0.05, "grad_norm": 4.1551252729054005, "learning_rate": 7.533936651583711e-06, "loss": 0.635, "step": 333 }, { "epoch": 0.05, "grad_norm": 4.329868394087096, "learning_rate": 7.556561085972851e-06, "loss": 0.6031, "step": 334 }, { "epoch": 0.05, "grad_norm": 3.8991176594716803, "learning_rate": 7.579185520361992e-06, "loss": 0.4362, "step": 335 }, { "epoch": 0.05, "grad_norm": 4.99892314065285, "learning_rate": 7.601809954751131e-06, "loss": 0.5797, "step": 336 }, { "epoch": 0.05, "grad_norm": 4.986733237369014, "learning_rate": 7.6244343891402715e-06, "loss": 0.5269, "step": 337 }, { "epoch": 0.05, "grad_norm": 4.34726219901186, "learning_rate": 7.647058823529411e-06, "loss": 0.5781, "step": 338 }, { "epoch": 0.05, "grad_norm": 4.425787455104366, "learning_rate": 7.669683257918553e-06, "loss": 0.5308, "step": 339 }, { "epoch": 0.05, "grad_norm": 4.535073711274931, "learning_rate": 7.692307692307694e-06, "loss": 0.5241, "step": 340 }, { "epoch": 0.05, "grad_norm": 4.106795410215829, "learning_rate": 7.714932126696834e-06, "loss": 0.4888, "step": 341 }, { "epoch": 0.05, "grad_norm": 4.795360637516257, "learning_rate": 7.737556561085974e-06, "loss": 0.5246, "step": 342 }, { "epoch": 0.05, "grad_norm": 4.460905726302721, "learning_rate": 7.760180995475115e-06, "loss": 0.5129, "step": 343 }, { "epoch": 0.05, "grad_norm": 4.48806670521333, "learning_rate": 7.782805429864253e-06, "loss": 0.4968, "step": 344 }, { "epoch": 0.05, "grad_norm": 4.56233259188479, "learning_rate": 7.805429864253394e-06, "loss": 0.5017, "step": 345 }, { "epoch": 0.05, "grad_norm": 4.875514745311065, "learning_rate": 7.828054298642534e-06, "loss": 0.6065, "step": 346 }, { "epoch": 0.05, "grad_norm": 4.171160689484168, "learning_rate": 7.850678733031675e-06, "loss": 0.5073, "step": 347 }, { "epoch": 0.05, "grad_norm": 4.717833811669195, "learning_rate": 7.873303167420815e-06, "loss": 0.55, "step": 348 }, { "epoch": 0.05, "grad_norm": 4.454537371871018, "learning_rate": 7.895927601809955e-06, "loss": 0.4854, "step": 349 }, { "epoch": 0.05, "grad_norm": 4.081477846318132, "learning_rate": 7.918552036199096e-06, "loss": 0.4853, "step": 350 }, { "epoch": 0.05, "grad_norm": 4.24445651802358, "learning_rate": 7.941176470588236e-06, "loss": 0.5629, "step": 351 }, { "epoch": 0.05, "grad_norm": 5.280784594194673, "learning_rate": 7.963800904977375e-06, "loss": 0.5297, "step": 352 }, { "epoch": 0.05, "grad_norm": 4.498063724493121, "learning_rate": 7.986425339366517e-06, "loss": 0.5065, "step": 353 }, { "epoch": 0.05, "grad_norm": 4.285815352669205, "learning_rate": 8.009049773755657e-06, "loss": 0.4847, "step": 354 }, { "epoch": 0.05, "grad_norm": 4.560942940010623, "learning_rate": 8.031674208144798e-06, "loss": 0.6003, "step": 355 }, { "epoch": 0.05, "grad_norm": 4.450579009490884, "learning_rate": 8.054298642533938e-06, "loss": 0.5015, "step": 356 }, { "epoch": 0.05, "grad_norm": 4.524567495670942, "learning_rate": 8.076923076923077e-06, "loss": 0.6448, "step": 357 }, { "epoch": 0.05, "grad_norm": 4.473722450598201, "learning_rate": 8.099547511312217e-06, "loss": 0.535, "step": 358 }, { "epoch": 0.05, "grad_norm": 4.2010587947453155, "learning_rate": 8.122171945701358e-06, "loss": 0.4983, "step": 359 }, { "epoch": 0.05, "grad_norm": 3.8116289693742322, "learning_rate": 8.144796380090498e-06, "loss": 0.5254, "step": 360 }, { "epoch": 0.05, "grad_norm": 3.9045041589322476, "learning_rate": 8.167420814479639e-06, "loss": 0.4808, "step": 361 }, { "epoch": 0.05, "grad_norm": 4.061961815678199, "learning_rate": 8.190045248868779e-06, "loss": 0.5125, "step": 362 }, { "epoch": 0.05, "grad_norm": 4.280785467920682, "learning_rate": 8.21266968325792e-06, "loss": 0.5309, "step": 363 }, { "epoch": 0.05, "grad_norm": 4.3734446547709815, "learning_rate": 8.23529411764706e-06, "loss": 0.5742, "step": 364 }, { "epoch": 0.05, "grad_norm": 4.31852756894498, "learning_rate": 8.257918552036199e-06, "loss": 0.5275, "step": 365 }, { "epoch": 0.05, "grad_norm": 4.489514364184108, "learning_rate": 8.280542986425339e-06, "loss": 0.5, "step": 366 }, { "epoch": 0.05, "grad_norm": 3.9295178991154884, "learning_rate": 8.303167420814481e-06, "loss": 0.4676, "step": 367 }, { "epoch": 0.05, "grad_norm": 4.359505505349747, "learning_rate": 8.325791855203621e-06, "loss": 0.5035, "step": 368 }, { "epoch": 0.05, "grad_norm": 4.833644355388557, "learning_rate": 8.348416289592762e-06, "loss": 0.4912, "step": 369 }, { "epoch": 0.05, "grad_norm": 4.438964399938822, "learning_rate": 8.371040723981902e-06, "loss": 0.4597, "step": 370 }, { "epoch": 0.05, "grad_norm": 5.80457154417493, "learning_rate": 8.393665158371041e-06, "loss": 0.567, "step": 371 }, { "epoch": 0.05, "grad_norm": 4.718867965229304, "learning_rate": 8.416289592760181e-06, "loss": 0.4958, "step": 372 }, { "epoch": 0.05, "grad_norm": 4.486969924073737, "learning_rate": 8.438914027149322e-06, "loss": 0.4962, "step": 373 }, { "epoch": 0.05, "grad_norm": 4.417199547472419, "learning_rate": 8.461538461538462e-06, "loss": 0.4876, "step": 374 }, { "epoch": 0.05, "grad_norm": 4.258334229000609, "learning_rate": 8.484162895927603e-06, "loss": 0.5355, "step": 375 }, { "epoch": 0.05, "grad_norm": 4.764837893642473, "learning_rate": 8.506787330316743e-06, "loss": 0.5205, "step": 376 }, { "epoch": 0.05, "grad_norm": 4.5931115611330116, "learning_rate": 8.529411764705883e-06, "loss": 0.565, "step": 377 }, { "epoch": 0.05, "grad_norm": 4.746871752346259, "learning_rate": 8.552036199095024e-06, "loss": 0.506, "step": 378 }, { "epoch": 0.05, "grad_norm": 4.292182820209345, "learning_rate": 8.574660633484162e-06, "loss": 0.5001, "step": 379 }, { "epoch": 0.05, "grad_norm": 4.35858427009413, "learning_rate": 8.597285067873304e-06, "loss": 0.5344, "step": 380 }, { "epoch": 0.05, "grad_norm": 4.382170855811785, "learning_rate": 8.619909502262445e-06, "loss": 0.5531, "step": 381 }, { "epoch": 0.05, "grad_norm": 4.585181518957607, "learning_rate": 8.642533936651585e-06, "loss": 0.5091, "step": 382 }, { "epoch": 0.05, "grad_norm": 5.061603866556143, "learning_rate": 8.665158371040726e-06, "loss": 0.5786, "step": 383 }, { "epoch": 0.05, "grad_norm": 4.236640697773424, "learning_rate": 8.687782805429864e-06, "loss": 0.5075, "step": 384 }, { "epoch": 0.05, "grad_norm": 4.175799572477735, "learning_rate": 8.710407239819005e-06, "loss": 0.491, "step": 385 }, { "epoch": 0.05, "grad_norm": 4.4687039128280635, "learning_rate": 8.733031674208145e-06, "loss": 0.5272, "step": 386 }, { "epoch": 0.05, "grad_norm": 3.8797908248373596, "learning_rate": 8.755656108597286e-06, "loss": 0.4955, "step": 387 }, { "epoch": 0.05, "grad_norm": 4.2696451106036015, "learning_rate": 8.778280542986426e-06, "loss": 0.5357, "step": 388 }, { "epoch": 0.05, "grad_norm": 4.631023691363582, "learning_rate": 8.800904977375566e-06, "loss": 0.5293, "step": 389 }, { "epoch": 0.05, "grad_norm": 4.117055986178787, "learning_rate": 8.823529411764707e-06, "loss": 0.5586, "step": 390 }, { "epoch": 0.05, "grad_norm": 4.2636358879292615, "learning_rate": 8.846153846153847e-06, "loss": 0.562, "step": 391 }, { "epoch": 0.05, "grad_norm": 4.378325838282376, "learning_rate": 8.868778280542986e-06, "loss": 0.4811, "step": 392 }, { "epoch": 0.05, "grad_norm": 3.8669465509001304, "learning_rate": 8.891402714932126e-06, "loss": 0.4526, "step": 393 }, { "epoch": 0.05, "grad_norm": 4.523310142256112, "learning_rate": 8.914027149321268e-06, "loss": 0.5512, "step": 394 }, { "epoch": 0.05, "grad_norm": 4.373198768829935, "learning_rate": 8.936651583710409e-06, "loss": 0.5556, "step": 395 }, { "epoch": 0.05, "grad_norm": 3.989984831879709, "learning_rate": 8.95927601809955e-06, "loss": 0.5158, "step": 396 }, { "epoch": 0.05, "grad_norm": 4.125234161881922, "learning_rate": 8.98190045248869e-06, "loss": 0.5332, "step": 397 }, { "epoch": 0.05, "grad_norm": 4.272050043238035, "learning_rate": 9.004524886877828e-06, "loss": 0.5286, "step": 398 }, { "epoch": 0.05, "grad_norm": 3.7107861706221965, "learning_rate": 9.027149321266969e-06, "loss": 0.4275, "step": 399 }, { "epoch": 0.05, "grad_norm": 3.9828442814965705, "learning_rate": 9.049773755656109e-06, "loss": 0.5361, "step": 400 }, { "epoch": 0.05, "grad_norm": 4.3433483104721065, "learning_rate": 9.07239819004525e-06, "loss": 0.5396, "step": 401 }, { "epoch": 0.05, "grad_norm": 3.614866946507279, "learning_rate": 9.09502262443439e-06, "loss": 0.4441, "step": 402 }, { "epoch": 0.05, "grad_norm": 4.0508349413825275, "learning_rate": 9.11764705882353e-06, "loss": 0.4902, "step": 403 }, { "epoch": 0.05, "grad_norm": 3.933629542614728, "learning_rate": 9.14027149321267e-06, "loss": 0.4889, "step": 404 }, { "epoch": 0.06, "grad_norm": 4.246770072280458, "learning_rate": 9.162895927601811e-06, "loss": 0.4951, "step": 405 }, { "epoch": 0.06, "grad_norm": 4.5478995346196935, "learning_rate": 9.18552036199095e-06, "loss": 0.4655, "step": 406 }, { "epoch": 0.06, "grad_norm": 3.900124560884611, "learning_rate": 9.20814479638009e-06, "loss": 0.4841, "step": 407 }, { "epoch": 0.06, "grad_norm": 4.065356585807601, "learning_rate": 9.230769230769232e-06, "loss": 0.5168, "step": 408 }, { "epoch": 0.06, "grad_norm": 4.0360517494819845, "learning_rate": 9.253393665158373e-06, "loss": 0.4705, "step": 409 }, { "epoch": 0.06, "grad_norm": 4.129132364601867, "learning_rate": 9.276018099547513e-06, "loss": 0.5408, "step": 410 }, { "epoch": 0.06, "grad_norm": 4.062276159893152, "learning_rate": 9.298642533936652e-06, "loss": 0.5367, "step": 411 }, { "epoch": 0.06, "grad_norm": 4.091941693164324, "learning_rate": 9.321266968325792e-06, "loss": 0.5283, "step": 412 }, { "epoch": 0.06, "grad_norm": 4.083379197240744, "learning_rate": 9.343891402714933e-06, "loss": 0.4964, "step": 413 }, { "epoch": 0.06, "grad_norm": 4.125925884924888, "learning_rate": 9.366515837104073e-06, "loss": 0.4981, "step": 414 }, { "epoch": 0.06, "grad_norm": 4.061565292133338, "learning_rate": 9.389140271493213e-06, "loss": 0.5292, "step": 415 }, { "epoch": 0.06, "grad_norm": 3.935570167659488, "learning_rate": 9.411764705882354e-06, "loss": 0.4672, "step": 416 }, { "epoch": 0.06, "grad_norm": 4.015519185932199, "learning_rate": 9.434389140271494e-06, "loss": 0.4883, "step": 417 }, { "epoch": 0.06, "grad_norm": 4.1038909999626805, "learning_rate": 9.457013574660635e-06, "loss": 0.5024, "step": 418 }, { "epoch": 0.06, "grad_norm": 3.8912641502677197, "learning_rate": 9.479638009049773e-06, "loss": 0.4756, "step": 419 }, { "epoch": 0.06, "grad_norm": 4.456666506644494, "learning_rate": 9.502262443438914e-06, "loss": 0.5367, "step": 420 }, { "epoch": 0.06, "grad_norm": 4.030377461676069, "learning_rate": 9.524886877828054e-06, "loss": 0.4929, "step": 421 }, { "epoch": 0.06, "grad_norm": 4.295450835918703, "learning_rate": 9.547511312217196e-06, "loss": 0.4729, "step": 422 }, { "epoch": 0.06, "grad_norm": 4.799435769761408, "learning_rate": 9.570135746606337e-06, "loss": 0.6111, "step": 423 }, { "epoch": 0.06, "grad_norm": 4.466120524019092, "learning_rate": 9.592760180995477e-06, "loss": 0.5428, "step": 424 }, { "epoch": 0.06, "grad_norm": 4.263897824618048, "learning_rate": 9.615384615384616e-06, "loss": 0.5037, "step": 425 }, { "epoch": 0.06, "grad_norm": 4.163864328816326, "learning_rate": 9.638009049773756e-06, "loss": 0.4594, "step": 426 }, { "epoch": 0.06, "grad_norm": 4.175261212796942, "learning_rate": 9.660633484162897e-06, "loss": 0.5335, "step": 427 }, { "epoch": 0.06, "grad_norm": 4.5642415237492315, "learning_rate": 9.683257918552037e-06, "loss": 0.5528, "step": 428 }, { "epoch": 0.06, "grad_norm": 4.290362037402862, "learning_rate": 9.705882352941177e-06, "loss": 0.5095, "step": 429 }, { "epoch": 0.06, "grad_norm": 4.43578139470838, "learning_rate": 9.728506787330318e-06, "loss": 0.5577, "step": 430 }, { "epoch": 0.06, "grad_norm": 4.302834026610889, "learning_rate": 9.751131221719458e-06, "loss": 0.5217, "step": 431 }, { "epoch": 0.06, "grad_norm": 3.746240980832705, "learning_rate": 9.773755656108599e-06, "loss": 0.4286, "step": 432 }, { "epoch": 0.06, "grad_norm": 4.737683963994727, "learning_rate": 9.796380090497737e-06, "loss": 0.538, "step": 433 }, { "epoch": 0.06, "grad_norm": 3.7134187372750715, "learning_rate": 9.819004524886878e-06, "loss": 0.4257, "step": 434 }, { "epoch": 0.06, "grad_norm": 4.169826987108958, "learning_rate": 9.841628959276018e-06, "loss": 0.4939, "step": 435 }, { "epoch": 0.06, "grad_norm": 4.012214557549502, "learning_rate": 9.86425339366516e-06, "loss": 0.514, "step": 436 }, { "epoch": 0.06, "grad_norm": 3.890701316274265, "learning_rate": 9.8868778280543e-06, "loss": 0.4949, "step": 437 }, { "epoch": 0.06, "grad_norm": 3.7092421147311345, "learning_rate": 9.90950226244344e-06, "loss": 0.4688, "step": 438 }, { "epoch": 0.06, "grad_norm": 4.134060216178577, "learning_rate": 9.93212669683258e-06, "loss": 0.49, "step": 439 }, { "epoch": 0.06, "grad_norm": 4.138555452917669, "learning_rate": 9.95475113122172e-06, "loss": 0.5186, "step": 440 }, { "epoch": 0.06, "grad_norm": 4.239755058882701, "learning_rate": 9.97737556561086e-06, "loss": 0.5248, "step": 441 }, { "epoch": 0.06, "grad_norm": 3.5906645345369888, "learning_rate": 1e-05, "loss": 0.4277, "step": 442 }, { "epoch": 0.06, "grad_norm": 5.137724362903945, "learning_rate": 9.99999987866092e-06, "loss": 0.5142, "step": 443 }, { "epoch": 0.06, "grad_norm": 4.539470682330847, "learning_rate": 9.999999514643683e-06, "loss": 0.4486, "step": 444 }, { "epoch": 0.06, "grad_norm": 5.02051033290526, "learning_rate": 9.999998907948308e-06, "loss": 0.4871, "step": 445 }, { "epoch": 0.06, "grad_norm": 4.925685240568301, "learning_rate": 9.999998058574822e-06, "loss": 0.5042, "step": 446 }, { "epoch": 0.06, "grad_norm": 4.882793261258588, "learning_rate": 9.999996966523272e-06, "loss": 0.5325, "step": 447 }, { "epoch": 0.06, "grad_norm": 4.455540032917785, "learning_rate": 9.999995631793704e-06, "loss": 0.503, "step": 448 }, { "epoch": 0.06, "grad_norm": 4.814631615234224, "learning_rate": 9.999994054386188e-06, "loss": 0.5122, "step": 449 }, { "epoch": 0.06, "grad_norm": 4.24333021805767, "learning_rate": 9.999992234300798e-06, "loss": 0.5005, "step": 450 }, { "epoch": 0.06, "grad_norm": 3.563085867482712, "learning_rate": 9.999990171537621e-06, "loss": 0.4011, "step": 451 }, { "epoch": 0.06, "grad_norm": 3.8263809401959468, "learning_rate": 9.999987866096762e-06, "loss": 0.4387, "step": 452 }, { "epoch": 0.06, "grad_norm": 3.892276853997083, "learning_rate": 9.99998531797833e-06, "loss": 0.4165, "step": 453 }, { "epoch": 0.06, "grad_norm": 3.7791664499357376, "learning_rate": 9.999982527182446e-06, "loss": 0.4926, "step": 454 }, { "epoch": 0.06, "grad_norm": 3.7897281894916865, "learning_rate": 9.99997949370925e-06, "loss": 0.4161, "step": 455 }, { "epoch": 0.06, "grad_norm": 4.48452072507302, "learning_rate": 9.999976217558886e-06, "loss": 0.5854, "step": 456 }, { "epoch": 0.06, "grad_norm": 3.9029547118654597, "learning_rate": 9.999972698731516e-06, "loss": 0.436, "step": 457 }, { "epoch": 0.06, "grad_norm": 3.8753256372243756, "learning_rate": 9.99996893722731e-06, "loss": 0.4952, "step": 458 }, { "epoch": 0.06, "grad_norm": 3.781639218291935, "learning_rate": 9.999964933046447e-06, "loss": 0.5236, "step": 459 }, { "epoch": 0.06, "grad_norm": 4.443021096223136, "learning_rate": 9.999960686189125e-06, "loss": 0.4909, "step": 460 }, { "epoch": 0.06, "grad_norm": 4.0242718676758695, "learning_rate": 9.999956196655548e-06, "loss": 0.5023, "step": 461 }, { "epoch": 0.06, "grad_norm": 4.672182316403683, "learning_rate": 9.999951464445938e-06, "loss": 0.5088, "step": 462 }, { "epoch": 0.06, "grad_norm": 5.9757116370347125, "learning_rate": 9.99994648956052e-06, "loss": 0.5808, "step": 463 }, { "epoch": 0.06, "grad_norm": 3.6196168095826127, "learning_rate": 9.999941271999536e-06, "loss": 0.4307, "step": 464 }, { "epoch": 0.06, "grad_norm": 4.164086270917012, "learning_rate": 9.999935811763242e-06, "loss": 0.4764, "step": 465 }, { "epoch": 0.06, "grad_norm": 4.255748216570893, "learning_rate": 9.999930108851902e-06, "loss": 0.4403, "step": 466 }, { "epoch": 0.06, "grad_norm": 4.149961280092839, "learning_rate": 9.99992416326579e-06, "loss": 0.5124, "step": 467 }, { "epoch": 0.06, "grad_norm": 3.8482584296744093, "learning_rate": 9.9999179750052e-06, "loss": 0.49, "step": 468 }, { "epoch": 0.06, "grad_norm": 3.8296109895038066, "learning_rate": 9.999911544070427e-06, "loss": 0.4867, "step": 469 }, { "epoch": 0.06, "grad_norm": 3.848225819302578, "learning_rate": 9.999904870461785e-06, "loss": 0.453, "step": 470 }, { "epoch": 0.06, "grad_norm": 3.8812965337776415, "learning_rate": 9.9998979541796e-06, "loss": 0.51, "step": 471 }, { "epoch": 0.06, "grad_norm": 3.509133626458995, "learning_rate": 9.999890795224206e-06, "loss": 0.4727, "step": 472 }, { "epoch": 0.06, "grad_norm": 4.532018953374427, "learning_rate": 9.999883393595949e-06, "loss": 0.5028, "step": 473 }, { "epoch": 0.06, "grad_norm": 4.1755388094206, "learning_rate": 9.99987574929519e-06, "loss": 0.4849, "step": 474 }, { "epoch": 0.06, "grad_norm": 4.342241247830162, "learning_rate": 9.999867862322299e-06, "loss": 0.405, "step": 475 }, { "epoch": 0.06, "grad_norm": 3.841375582239115, "learning_rate": 9.99985973267766e-06, "loss": 0.4124, "step": 476 }, { "epoch": 0.06, "grad_norm": 4.351344513423413, "learning_rate": 9.999851360361666e-06, "loss": 0.5139, "step": 477 }, { "epoch": 0.07, "grad_norm": 4.22735232581215, "learning_rate": 9.999842745374726e-06, "loss": 0.5135, "step": 478 }, { "epoch": 0.07, "grad_norm": 3.8317792767719876, "learning_rate": 9.999833887717257e-06, "loss": 0.4654, "step": 479 }, { "epoch": 0.07, "grad_norm": 3.882614688212824, "learning_rate": 9.999824787389686e-06, "loss": 0.4994, "step": 480 }, { "epoch": 0.07, "grad_norm": 3.6451690398816243, "learning_rate": 9.999815444392458e-06, "loss": 0.5125, "step": 481 }, { "epoch": 0.07, "grad_norm": 3.753405819108101, "learning_rate": 9.999805858726026e-06, "loss": 0.4659, "step": 482 }, { "epoch": 0.07, "grad_norm": 3.6887999895773462, "learning_rate": 9.999796030390855e-06, "loss": 0.4545, "step": 483 }, { "epoch": 0.07, "grad_norm": 4.09844656163741, "learning_rate": 9.999785959387422e-06, "loss": 0.5575, "step": 484 }, { "epoch": 0.07, "grad_norm": 3.5396947572368673, "learning_rate": 9.999775645716215e-06, "loss": 0.4579, "step": 485 }, { "epoch": 0.07, "grad_norm": 4.076131687633431, "learning_rate": 9.999765089377735e-06, "loss": 0.4535, "step": 486 }, { "epoch": 0.07, "grad_norm": 3.947601533370477, "learning_rate": 9.999754290372496e-06, "loss": 0.4498, "step": 487 }, { "epoch": 0.07, "grad_norm": 4.007417544347977, "learning_rate": 9.99974324870102e-06, "loss": 0.5041, "step": 488 }, { "epoch": 0.07, "grad_norm": 3.8995413007100073, "learning_rate": 9.999731964363843e-06, "loss": 0.4614, "step": 489 }, { "epoch": 0.07, "grad_norm": 3.79904985911, "learning_rate": 9.999720437361513e-06, "loss": 0.4855, "step": 490 }, { "epoch": 0.07, "grad_norm": 4.253773799537596, "learning_rate": 9.99970866769459e-06, "loss": 0.4435, "step": 491 }, { "epoch": 0.07, "grad_norm": 3.926491160745131, "learning_rate": 9.999696655363646e-06, "loss": 0.4541, "step": 492 }, { "epoch": 0.07, "grad_norm": 4.2287934729785235, "learning_rate": 9.999684400369263e-06, "loss": 0.5564, "step": 493 }, { "epoch": 0.07, "grad_norm": 4.379296666453499, "learning_rate": 9.999671902712035e-06, "loss": 0.5195, "step": 494 }, { "epoch": 0.07, "grad_norm": 3.587809635448618, "learning_rate": 9.99965916239257e-06, "loss": 0.4594, "step": 495 }, { "epoch": 0.07, "grad_norm": 3.771105468970707, "learning_rate": 9.999646179411486e-06, "loss": 0.481, "step": 496 }, { "epoch": 0.07, "grad_norm": 3.950693377623531, "learning_rate": 9.999632953769413e-06, "loss": 0.5124, "step": 497 }, { "epoch": 0.07, "grad_norm": 3.916250554965569, "learning_rate": 9.999619485466991e-06, "loss": 0.4815, "step": 498 }, { "epoch": 0.07, "grad_norm": 3.8347724423229783, "learning_rate": 9.999605774504877e-06, "loss": 0.3951, "step": 499 }, { "epoch": 0.07, "grad_norm": 4.037156051124367, "learning_rate": 9.999591820883736e-06, "loss": 0.5126, "step": 500 }, { "epoch": 0.07, "grad_norm": 4.127237373898493, "learning_rate": 9.999577624604244e-06, "loss": 0.4964, "step": 501 }, { "epoch": 0.07, "grad_norm": 3.8751807754609686, "learning_rate": 9.99956318566709e-06, "loss": 0.4872, "step": 502 }, { "epoch": 0.07, "grad_norm": 4.086511488974735, "learning_rate": 9.999548504072974e-06, "loss": 0.4842, "step": 503 }, { "epoch": 0.07, "grad_norm": 4.322069064968943, "learning_rate": 9.999533579822611e-06, "loss": 0.5294, "step": 504 }, { "epoch": 0.07, "grad_norm": 4.242402105127522, "learning_rate": 9.999518412916723e-06, "loss": 0.5004, "step": 505 }, { "epoch": 0.07, "grad_norm": 3.651864103240707, "learning_rate": 9.999503003356047e-06, "loss": 0.4292, "step": 506 }, { "epoch": 0.07, "grad_norm": 4.181855185563273, "learning_rate": 9.999487351141333e-06, "loss": 0.5523, "step": 507 }, { "epoch": 0.07, "grad_norm": 4.658082436740373, "learning_rate": 9.999471456273337e-06, "loss": 0.5056, "step": 508 }, { "epoch": 0.07, "grad_norm": 4.3269049084366, "learning_rate": 9.999455318752833e-06, "loss": 0.4957, "step": 509 }, { "epoch": 0.07, "grad_norm": 3.7573559007478425, "learning_rate": 9.999438938580604e-06, "loss": 0.4598, "step": 510 }, { "epoch": 0.07, "grad_norm": 3.608159223432518, "learning_rate": 9.999422315757443e-06, "loss": 0.4187, "step": 511 }, { "epoch": 0.07, "grad_norm": 4.141417851585135, "learning_rate": 9.999405450284161e-06, "loss": 0.4947, "step": 512 }, { "epoch": 0.07, "grad_norm": 4.489402672407533, "learning_rate": 9.999388342161571e-06, "loss": 0.4703, "step": 513 }, { "epoch": 0.07, "grad_norm": 4.027577560126352, "learning_rate": 9.999370991390507e-06, "loss": 0.535, "step": 514 }, { "epoch": 0.07, "grad_norm": 4.308871213695158, "learning_rate": 9.99935339797181e-06, "loss": 0.5162, "step": 515 }, { "epoch": 0.07, "grad_norm": 4.031732761813992, "learning_rate": 9.999335561906336e-06, "loss": 0.4784, "step": 516 }, { "epoch": 0.07, "grad_norm": 4.214849584512343, "learning_rate": 9.999317483194948e-06, "loss": 0.5057, "step": 517 }, { "epoch": 0.07, "grad_norm": 3.77222345467758, "learning_rate": 9.999299161838524e-06, "loss": 0.4629, "step": 518 }, { "epoch": 0.07, "grad_norm": 3.6943668863338437, "learning_rate": 9.999280597837954e-06, "loss": 0.4058, "step": 519 }, { "epoch": 0.07, "grad_norm": 4.226765299329125, "learning_rate": 9.99926179119414e-06, "loss": 0.4477, "step": 520 }, { "epoch": 0.07, "grad_norm": 3.8214781901559958, "learning_rate": 9.999242741907992e-06, "loss": 0.4668, "step": 521 }, { "epoch": 0.07, "grad_norm": 4.071593989792488, "learning_rate": 9.999223449980434e-06, "loss": 0.4754, "step": 522 }, { "epoch": 0.07, "grad_norm": 3.950445558887973, "learning_rate": 9.999203915412408e-06, "loss": 0.5088, "step": 523 }, { "epoch": 0.07, "grad_norm": 4.5151638297204055, "learning_rate": 9.999184138204859e-06, "loss": 0.5198, "step": 524 }, { "epoch": 0.07, "grad_norm": 4.210869658943398, "learning_rate": 9.999164118358743e-06, "loss": 0.4788, "step": 525 }, { "epoch": 0.07, "grad_norm": 4.299127829264426, "learning_rate": 9.999143855875038e-06, "loss": 0.5753, "step": 526 }, { "epoch": 0.07, "grad_norm": 4.284540692145413, "learning_rate": 9.999123350754722e-06, "loss": 0.5305, "step": 527 }, { "epoch": 0.07, "grad_norm": 3.8779870772494545, "learning_rate": 9.999102602998794e-06, "loss": 0.4566, "step": 528 }, { "epoch": 0.07, "grad_norm": 3.9778523698899297, "learning_rate": 9.99908161260826e-06, "loss": 0.4708, "step": 529 }, { "epoch": 0.07, "grad_norm": 4.498298141893578, "learning_rate": 9.999060379584139e-06, "loss": 0.5276, "step": 530 }, { "epoch": 0.07, "grad_norm": 4.1366355559078505, "learning_rate": 9.99903890392746e-06, "loss": 0.5245, "step": 531 }, { "epoch": 0.07, "grad_norm": 4.165988225753585, "learning_rate": 9.999017185639266e-06, "loss": 0.43, "step": 532 }, { "epoch": 0.07, "grad_norm": 3.636109536896207, "learning_rate": 9.998995224720614e-06, "loss": 0.4189, "step": 533 }, { "epoch": 0.07, "grad_norm": 3.810994667745163, "learning_rate": 9.998973021172564e-06, "loss": 0.4441, "step": 534 }, { "epoch": 0.07, "grad_norm": 4.447889296094373, "learning_rate": 9.9989505749962e-06, "loss": 0.4842, "step": 535 }, { "epoch": 0.07, "grad_norm": 3.4970834783485016, "learning_rate": 9.998927886192607e-06, "loss": 0.4615, "step": 536 }, { "epoch": 0.07, "grad_norm": 4.328587359199753, "learning_rate": 9.99890495476289e-06, "loss": 0.5025, "step": 537 }, { "epoch": 0.07, "grad_norm": 4.196277261888811, "learning_rate": 9.998881780708156e-06, "loss": 0.5694, "step": 538 }, { "epoch": 0.07, "grad_norm": 3.7187197869163593, "learning_rate": 9.998858364029536e-06, "loss": 0.4629, "step": 539 }, { "epoch": 0.07, "grad_norm": 4.130776293878367, "learning_rate": 9.998834704728164e-06, "loss": 0.4622, "step": 540 }, { "epoch": 0.07, "grad_norm": 3.6440908819343707, "learning_rate": 9.998810802805189e-06, "loss": 0.5451, "step": 541 }, { "epoch": 0.07, "grad_norm": 3.6939719491146965, "learning_rate": 9.99878665826177e-06, "loss": 0.4665, "step": 542 }, { "epoch": 0.07, "grad_norm": 3.7604152602957144, "learning_rate": 9.998762271099079e-06, "loss": 0.51, "step": 543 }, { "epoch": 0.07, "grad_norm": 3.903824418184381, "learning_rate": 9.9987376413183e-06, "loss": 0.4802, "step": 544 }, { "epoch": 0.07, "grad_norm": 3.7258302791912454, "learning_rate": 9.998712768920627e-06, "loss": 0.4604, "step": 545 }, { "epoch": 0.07, "grad_norm": 3.7173692174278923, "learning_rate": 9.998687653907271e-06, "loss": 0.4647, "step": 546 }, { "epoch": 0.07, "grad_norm": 3.77259934326476, "learning_rate": 9.998662296279447e-06, "loss": 0.4895, "step": 547 }, { "epoch": 0.07, "grad_norm": 4.378792806066716, "learning_rate": 9.998636696038389e-06, "loss": 0.4708, "step": 548 }, { "epoch": 0.07, "grad_norm": 3.6339630188806304, "learning_rate": 9.998610853185336e-06, "loss": 0.4589, "step": 549 }, { "epoch": 0.07, "grad_norm": 3.6962007055309396, "learning_rate": 9.998584767721545e-06, "loss": 0.4405, "step": 550 }, { "epoch": 0.07, "grad_norm": 3.736461755779509, "learning_rate": 9.99855843964828e-06, "loss": 0.5223, "step": 551 }, { "epoch": 0.08, "grad_norm": 3.9078836598202553, "learning_rate": 9.998531868966822e-06, "loss": 0.4701, "step": 552 }, { "epoch": 0.08, "grad_norm": 4.178005012250725, "learning_rate": 9.998505055678458e-06, "loss": 0.4723, "step": 553 }, { "epoch": 0.08, "grad_norm": 3.988783480305429, "learning_rate": 9.99847799978449e-06, "loss": 0.4325, "step": 554 }, { "epoch": 0.08, "grad_norm": 4.5135517325552765, "learning_rate": 9.99845070128623e-06, "loss": 0.388, "step": 555 }, { "epoch": 0.08, "grad_norm": 4.111968680172977, "learning_rate": 9.998423160185006e-06, "loss": 0.4818, "step": 556 }, { "epoch": 0.08, "grad_norm": 4.183180006522272, "learning_rate": 9.998395376482152e-06, "loss": 0.5187, "step": 557 }, { "epoch": 0.08, "grad_norm": 3.802773183308847, "learning_rate": 9.99836735017902e-06, "loss": 0.4653, "step": 558 }, { "epoch": 0.08, "grad_norm": 4.415024573479558, "learning_rate": 9.998339081276966e-06, "loss": 0.468, "step": 559 }, { "epoch": 0.08, "grad_norm": 3.9388361023121434, "learning_rate": 9.998310569777364e-06, "loss": 0.3933, "step": 560 }, { "epoch": 0.08, "grad_norm": 3.521669914349715, "learning_rate": 9.998281815681597e-06, "loss": 0.4042, "step": 561 }, { "epoch": 0.08, "grad_norm": 4.1621612598897535, "learning_rate": 9.998252818991062e-06, "loss": 0.5166, "step": 562 }, { "epoch": 0.08, "grad_norm": 3.8546044512804083, "learning_rate": 9.998223579707162e-06, "loss": 0.4659, "step": 563 }, { "epoch": 0.08, "grad_norm": 4.338222310756935, "learning_rate": 9.998194097831323e-06, "loss": 0.4981, "step": 564 }, { "epoch": 0.08, "grad_norm": 3.7490454747881783, "learning_rate": 9.998164373364974e-06, "loss": 0.4366, "step": 565 }, { "epoch": 0.08, "grad_norm": 4.042683715732491, "learning_rate": 9.998134406309555e-06, "loss": 0.4755, "step": 566 }, { "epoch": 0.08, "grad_norm": 4.140921372878578, "learning_rate": 9.99810419666652e-06, "loss": 0.4486, "step": 567 }, { "epoch": 0.08, "grad_norm": 4.641791809779396, "learning_rate": 9.998073744437338e-06, "loss": 0.5257, "step": 568 }, { "epoch": 0.08, "grad_norm": 3.770645408142371, "learning_rate": 9.998043049623487e-06, "loss": 0.4768, "step": 569 }, { "epoch": 0.08, "grad_norm": 3.5603099308465453, "learning_rate": 9.998012112226456e-06, "loss": 0.4464, "step": 570 }, { "epoch": 0.08, "grad_norm": 3.6614880833936407, "learning_rate": 9.997980932247747e-06, "loss": 0.4738, "step": 571 }, { "epoch": 0.08, "grad_norm": 3.46305082003212, "learning_rate": 9.997949509688871e-06, "loss": 0.4493, "step": 572 }, { "epoch": 0.08, "grad_norm": 5.480045297007061, "learning_rate": 9.997917844551355e-06, "loss": 0.515, "step": 573 }, { "epoch": 0.08, "grad_norm": 4.177410070531614, "learning_rate": 9.997885936836738e-06, "loss": 0.4815, "step": 574 }, { "epoch": 0.08, "grad_norm": 3.8784400562091483, "learning_rate": 9.997853786546563e-06, "loss": 0.4509, "step": 575 }, { "epoch": 0.08, "grad_norm": 3.7270699311908206, "learning_rate": 9.997821393682396e-06, "loss": 0.4142, "step": 576 }, { "epoch": 0.08, "grad_norm": 3.8323667433158635, "learning_rate": 9.997788758245808e-06, "loss": 0.4328, "step": 577 }, { "epoch": 0.08, "grad_norm": 4.174880729462535, "learning_rate": 9.99775588023838e-06, "loss": 0.5396, "step": 578 }, { "epoch": 0.08, "grad_norm": 4.075801840551763, "learning_rate": 9.99772275966171e-06, "loss": 0.5225, "step": 579 }, { "epoch": 0.08, "grad_norm": 5.409259930052314, "learning_rate": 9.997689396517408e-06, "loss": 0.6094, "step": 580 }, { "epoch": 0.08, "grad_norm": 4.1225319309606405, "learning_rate": 9.997655790807088e-06, "loss": 0.4492, "step": 581 }, { "epoch": 0.08, "grad_norm": 4.376329995519756, "learning_rate": 9.997621942532383e-06, "loss": 0.5059, "step": 582 }, { "epoch": 0.08, "grad_norm": 4.4453057028562695, "learning_rate": 9.997587851694939e-06, "loss": 0.5902, "step": 583 }, { "epoch": 0.08, "grad_norm": 4.187961913845949, "learning_rate": 9.997553518296406e-06, "loss": 0.4571, "step": 584 }, { "epoch": 0.08, "grad_norm": 4.307484700966834, "learning_rate": 9.997518942338455e-06, "loss": 0.4408, "step": 585 }, { "epoch": 0.08, "grad_norm": 4.544041648059544, "learning_rate": 9.997484123822761e-06, "loss": 0.4985, "step": 586 }, { "epoch": 0.08, "grad_norm": 3.4498617661883966, "learning_rate": 9.997449062751012e-06, "loss": 0.4, "step": 587 }, { "epoch": 0.08, "grad_norm": 3.9053665350031608, "learning_rate": 9.997413759124916e-06, "loss": 0.407, "step": 588 }, { "epoch": 0.08, "grad_norm": 3.7464909444563492, "learning_rate": 9.99737821294618e-06, "loss": 0.4267, "step": 589 }, { "epoch": 0.08, "grad_norm": 4.398236018561587, "learning_rate": 9.997342424216534e-06, "loss": 0.5282, "step": 590 }, { "epoch": 0.08, "grad_norm": 4.096434609210357, "learning_rate": 9.997306392937713e-06, "loss": 0.4478, "step": 591 }, { "epoch": 0.08, "grad_norm": 3.972496781429311, "learning_rate": 9.997270119111467e-06, "loss": 0.5371, "step": 592 }, { "epoch": 0.08, "grad_norm": 4.189989376589614, "learning_rate": 9.997233602739553e-06, "loss": 0.535, "step": 593 }, { "epoch": 0.08, "grad_norm": 3.8479556075593777, "learning_rate": 9.997196843823747e-06, "loss": 0.5057, "step": 594 }, { "epoch": 0.08, "grad_norm": 3.764284446291677, "learning_rate": 9.997159842365832e-06, "loss": 0.4245, "step": 595 }, { "epoch": 0.08, "grad_norm": 3.822795711124626, "learning_rate": 9.997122598367604e-06, "loss": 0.4296, "step": 596 }, { "epoch": 0.08, "grad_norm": 3.3313448908941568, "learning_rate": 9.99708511183087e-06, "loss": 0.4246, "step": 597 }, { "epoch": 0.08, "grad_norm": 3.748864285846869, "learning_rate": 9.997047382757453e-06, "loss": 0.5359, "step": 598 }, { "epoch": 0.08, "grad_norm": 4.268101924858505, "learning_rate": 9.997009411149177e-06, "loss": 0.5262, "step": 599 }, { "epoch": 0.08, "grad_norm": 3.7388749003550257, "learning_rate": 9.996971197007892e-06, "loss": 0.4346, "step": 600 }, { "epoch": 0.08, "grad_norm": 3.752834939091205, "learning_rate": 9.996932740335448e-06, "loss": 0.4487, "step": 601 }, { "epoch": 0.08, "grad_norm": 4.551308222477313, "learning_rate": 9.996894041133715e-06, "loss": 0.4702, "step": 602 }, { "epoch": 0.08, "grad_norm": 3.6472671302648196, "learning_rate": 9.996855099404567e-06, "loss": 0.4597, "step": 603 }, { "epoch": 0.08, "grad_norm": 3.516856938769006, "learning_rate": 9.9968159151499e-06, "loss": 0.3393, "step": 604 }, { "epoch": 0.08, "grad_norm": 3.4248624280607833, "learning_rate": 9.996776488371611e-06, "loss": 0.3854, "step": 605 }, { "epoch": 0.08, "grad_norm": 3.4359268653983803, "learning_rate": 9.996736819071616e-06, "loss": 0.4394, "step": 606 }, { "epoch": 0.08, "grad_norm": 3.818163609274083, "learning_rate": 9.99669690725184e-06, "loss": 0.4481, "step": 607 }, { "epoch": 0.08, "grad_norm": 3.912057076741779, "learning_rate": 9.996656752914217e-06, "loss": 0.4377, "step": 608 }, { "epoch": 0.08, "grad_norm": 4.058320137962852, "learning_rate": 9.996616356060702e-06, "loss": 0.4773, "step": 609 }, { "epoch": 0.08, "grad_norm": 4.08217901701637, "learning_rate": 9.99657571669325e-06, "loss": 0.4341, "step": 610 }, { "epoch": 0.08, "grad_norm": 4.046808115051464, "learning_rate": 9.996534834813838e-06, "loss": 0.5139, "step": 611 }, { "epoch": 0.08, "grad_norm": 3.971304276889696, "learning_rate": 9.996493710424447e-06, "loss": 0.431, "step": 612 }, { "epoch": 0.08, "grad_norm": 4.44827033427837, "learning_rate": 9.996452343527073e-06, "loss": 0.507, "step": 613 }, { "epoch": 0.08, "grad_norm": 4.658515341323335, "learning_rate": 9.996410734123725e-06, "loss": 0.5809, "step": 614 }, { "epoch": 0.08, "grad_norm": 4.264836071245019, "learning_rate": 9.996368882216423e-06, "loss": 0.4307, "step": 615 }, { "epoch": 0.08, "grad_norm": 4.099604534111894, "learning_rate": 9.996326787807198e-06, "loss": 0.4378, "step": 616 }, { "epoch": 0.08, "grad_norm": 4.101610395898905, "learning_rate": 9.996284450898093e-06, "loss": 0.4318, "step": 617 }, { "epoch": 0.08, "grad_norm": 3.729014586412663, "learning_rate": 9.99624187149116e-06, "loss": 0.4307, "step": 618 }, { "epoch": 0.08, "grad_norm": 3.6475556048812017, "learning_rate": 9.996199049588471e-06, "loss": 0.3968, "step": 619 }, { "epoch": 0.08, "grad_norm": 4.087762326631072, "learning_rate": 9.996155985192101e-06, "loss": 0.4817, "step": 620 }, { "epoch": 0.08, "grad_norm": 4.125607906132819, "learning_rate": 9.99611267830414e-06, "loss": 0.4856, "step": 621 }, { "epoch": 0.08, "grad_norm": 4.0604641304426785, "learning_rate": 9.996069128926691e-06, "loss": 0.4937, "step": 622 }, { "epoch": 0.08, "grad_norm": 3.914453764891072, "learning_rate": 9.996025337061868e-06, "loss": 0.4246, "step": 623 }, { "epoch": 0.08, "grad_norm": 4.183340275524261, "learning_rate": 9.995981302711795e-06, "loss": 0.4825, "step": 624 }, { "epoch": 0.09, "grad_norm": 3.913152794370873, "learning_rate": 9.99593702587861e-06, "loss": 0.5265, "step": 625 }, { "epoch": 0.09, "grad_norm": 3.6739430471665715, "learning_rate": 9.995892506564461e-06, "loss": 0.4581, "step": 626 }, { "epoch": 0.09, "grad_norm": 4.000276912378069, "learning_rate": 9.995847744771514e-06, "loss": 0.451, "step": 627 }, { "epoch": 0.09, "grad_norm": 3.6919016686691437, "learning_rate": 9.995802740501933e-06, "loss": 0.5176, "step": 628 }, { "epoch": 0.09, "grad_norm": 4.2005750643292865, "learning_rate": 9.99575749375791e-06, "loss": 0.4545, "step": 629 }, { "epoch": 0.09, "grad_norm": 3.5553165075851654, "learning_rate": 9.995712004541634e-06, "loss": 0.4524, "step": 630 }, { "epoch": 0.09, "grad_norm": 3.460400346639752, "learning_rate": 9.99566627285532e-06, "loss": 0.5212, "step": 631 }, { "epoch": 0.09, "grad_norm": 3.71626468505182, "learning_rate": 9.995620298701183e-06, "loss": 0.4492, "step": 632 }, { "epoch": 0.09, "grad_norm": 3.42875349277366, "learning_rate": 9.995574082081454e-06, "loss": 0.4034, "step": 633 }, { "epoch": 0.09, "grad_norm": 4.187160315259293, "learning_rate": 9.99552762299838e-06, "loss": 0.4623, "step": 634 }, { "epoch": 0.09, "grad_norm": 4.010534826251181, "learning_rate": 9.995480921454212e-06, "loss": 0.4604, "step": 635 }, { "epoch": 0.09, "grad_norm": 3.934643391754611, "learning_rate": 9.99543397745122e-06, "loss": 0.4864, "step": 636 }, { "epoch": 0.09, "grad_norm": 3.7960499596327337, "learning_rate": 9.99538679099168e-06, "loss": 0.4346, "step": 637 }, { "epoch": 0.09, "grad_norm": 3.5328112031614065, "learning_rate": 9.995339362077883e-06, "loss": 0.4543, "step": 638 }, { "epoch": 0.09, "grad_norm": 3.60286521042545, "learning_rate": 9.995291690712132e-06, "loss": 0.4779, "step": 639 }, { "epoch": 0.09, "grad_norm": 4.070434464944307, "learning_rate": 9.995243776896739e-06, "loss": 0.4742, "step": 640 }, { "epoch": 0.09, "grad_norm": 4.038570914605018, "learning_rate": 9.99519562063403e-06, "loss": 0.4481, "step": 641 }, { "epoch": 0.09, "grad_norm": 3.5516633864892575, "learning_rate": 9.995147221926343e-06, "loss": 0.4278, "step": 642 }, { "epoch": 0.09, "grad_norm": 4.075570493208812, "learning_rate": 9.995098580776028e-06, "loss": 0.495, "step": 643 }, { "epoch": 0.09, "grad_norm": 3.9152935313184174, "learning_rate": 9.995049697185442e-06, "loss": 0.4187, "step": 644 }, { "epoch": 0.09, "grad_norm": 6.0601143125810575, "learning_rate": 9.995000571156962e-06, "loss": 0.4907, "step": 645 }, { "epoch": 0.09, "grad_norm": 3.9772321680935026, "learning_rate": 9.99495120269297e-06, "loss": 0.4817, "step": 646 }, { "epoch": 0.09, "grad_norm": 3.392430064059209, "learning_rate": 9.994901591795863e-06, "loss": 0.3914, "step": 647 }, { "epoch": 0.09, "grad_norm": 3.9810671865753253, "learning_rate": 9.994851738468047e-06, "loss": 0.4571, "step": 648 }, { "epoch": 0.09, "grad_norm": 3.5778310292291997, "learning_rate": 9.994801642711945e-06, "loss": 0.4249, "step": 649 }, { "epoch": 0.09, "grad_norm": 3.9916210544230903, "learning_rate": 9.994751304529987e-06, "loss": 0.4927, "step": 650 }, { "epoch": 0.09, "grad_norm": 3.9506093809108416, "learning_rate": 9.994700723924612e-06, "loss": 0.505, "step": 651 }, { "epoch": 0.09, "grad_norm": 3.7971838255019716, "learning_rate": 9.994649900898283e-06, "loss": 0.4268, "step": 652 }, { "epoch": 0.09, "grad_norm": 3.418366931911641, "learning_rate": 9.99459883545346e-06, "loss": 0.4437, "step": 653 }, { "epoch": 0.09, "grad_norm": 3.69588589550883, "learning_rate": 9.994547527592626e-06, "loss": 0.4753, "step": 654 }, { "epoch": 0.09, "grad_norm": 3.691128924255343, "learning_rate": 9.994495977318267e-06, "loss": 0.4203, "step": 655 }, { "epoch": 0.09, "grad_norm": 4.672302244527737, "learning_rate": 9.994444184632888e-06, "loss": 0.437, "step": 656 }, { "epoch": 0.09, "grad_norm": 4.0680205816039505, "learning_rate": 9.994392149539003e-06, "loss": 0.4509, "step": 657 }, { "epoch": 0.09, "grad_norm": 3.81406019468143, "learning_rate": 9.994339872039135e-06, "loss": 0.4895, "step": 658 }, { "epoch": 0.09, "grad_norm": 4.393285714248007, "learning_rate": 9.994287352135826e-06, "loss": 0.516, "step": 659 }, { "epoch": 0.09, "grad_norm": 3.7607980171846482, "learning_rate": 9.99423458983162e-06, "loss": 0.4161, "step": 660 }, { "epoch": 0.09, "grad_norm": 4.399028529901324, "learning_rate": 9.994181585129079e-06, "loss": 0.466, "step": 661 }, { "epoch": 0.09, "grad_norm": 3.728795440037079, "learning_rate": 9.994128338030778e-06, "loss": 0.4888, "step": 662 }, { "epoch": 0.09, "grad_norm": 3.5397066421002115, "learning_rate": 9.994074848539297e-06, "loss": 0.4508, "step": 663 }, { "epoch": 0.09, "grad_norm": 4.173566959930643, "learning_rate": 9.994021116657239e-06, "loss": 0.5457, "step": 664 }, { "epoch": 0.09, "grad_norm": 3.711658593035554, "learning_rate": 9.993967142387207e-06, "loss": 0.421, "step": 665 }, { "epoch": 0.09, "grad_norm": 3.819673713371357, "learning_rate": 9.993912925731819e-06, "loss": 0.4702, "step": 666 }, { "epoch": 0.09, "grad_norm": 4.127672973212655, "learning_rate": 9.993858466693712e-06, "loss": 0.4636, "step": 667 }, { "epoch": 0.09, "grad_norm": 3.945174108587737, "learning_rate": 9.993803765275525e-06, "loss": 0.4313, "step": 668 }, { "epoch": 0.09, "grad_norm": 3.845374084033308, "learning_rate": 9.993748821479913e-06, "loss": 0.4677, "step": 669 }, { "epoch": 0.09, "grad_norm": 3.760679560538588, "learning_rate": 9.993693635309545e-06, "loss": 0.5376, "step": 670 }, { "epoch": 0.09, "grad_norm": 4.748079488116355, "learning_rate": 9.9936382067671e-06, "loss": 0.4941, "step": 671 }, { "epoch": 0.09, "grad_norm": 3.592883128958613, "learning_rate": 9.993582535855265e-06, "loss": 0.4317, "step": 672 }, { "epoch": 0.09, "grad_norm": 3.4557037750506416, "learning_rate": 9.993526622576744e-06, "loss": 0.3733, "step": 673 }, { "epoch": 0.09, "grad_norm": 3.7223883822111303, "learning_rate": 9.99347046693425e-06, "loss": 0.4827, "step": 674 }, { "epoch": 0.09, "grad_norm": 3.4993647782019432, "learning_rate": 9.993414068930511e-06, "loss": 0.4717, "step": 675 }, { "epoch": 0.09, "grad_norm": 3.302972508966449, "learning_rate": 9.99335742856826e-06, "loss": 0.4128, "step": 676 }, { "epoch": 0.09, "grad_norm": 3.851670079507773, "learning_rate": 9.99330054585025e-06, "loss": 0.4811, "step": 677 }, { "epoch": 0.09, "grad_norm": 3.9097350681821954, "learning_rate": 9.99324342077924e-06, "loss": 0.4739, "step": 678 }, { "epoch": 0.09, "grad_norm": 3.877344101394417, "learning_rate": 9.993186053358001e-06, "loss": 0.4894, "step": 679 }, { "epoch": 0.09, "grad_norm": 4.258242316343126, "learning_rate": 9.993128443589322e-06, "loss": 0.5199, "step": 680 }, { "epoch": 0.09, "grad_norm": 3.87681449979008, "learning_rate": 9.993070591475996e-06, "loss": 0.4452, "step": 681 }, { "epoch": 0.09, "grad_norm": 3.9150448620656975, "learning_rate": 9.993012497020831e-06, "loss": 0.4984, "step": 682 }, { "epoch": 0.09, "grad_norm": 3.837686063711313, "learning_rate": 9.992954160226646e-06, "loss": 0.4379, "step": 683 }, { "epoch": 0.09, "grad_norm": 3.7719792006290063, "learning_rate": 9.992895581096274e-06, "loss": 0.4044, "step": 684 }, { "epoch": 0.09, "grad_norm": 3.3410354716862525, "learning_rate": 9.992836759632558e-06, "loss": 0.4122, "step": 685 }, { "epoch": 0.09, "grad_norm": 3.9735288323064846, "learning_rate": 9.992777695838351e-06, "loss": 0.4687, "step": 686 }, { "epoch": 0.09, "grad_norm": 3.7732542021001616, "learning_rate": 9.992718389716521e-06, "loss": 0.4055, "step": 687 }, { "epoch": 0.09, "grad_norm": 3.676873178765638, "learning_rate": 9.992658841269949e-06, "loss": 0.4173, "step": 688 }, { "epoch": 0.09, "grad_norm": 3.777328968483565, "learning_rate": 9.992599050501521e-06, "loss": 0.4461, "step": 689 }, { "epoch": 0.09, "grad_norm": 7.375840242029813, "learning_rate": 9.99253901741414e-06, "loss": 0.5065, "step": 690 }, { "epoch": 0.09, "grad_norm": 4.049434091394213, "learning_rate": 9.992478742010722e-06, "loss": 0.4827, "step": 691 }, { "epoch": 0.09, "grad_norm": 3.815076487472994, "learning_rate": 9.992418224294191e-06, "loss": 0.408, "step": 692 }, { "epoch": 0.09, "grad_norm": 3.7215514007247417, "learning_rate": 9.992357464267483e-06, "loss": 0.4469, "step": 693 }, { "epoch": 0.09, "grad_norm": 4.442078129026385, "learning_rate": 9.992296461933551e-06, "loss": 0.455, "step": 694 }, { "epoch": 0.09, "grad_norm": 4.476812999395245, "learning_rate": 9.992235217295352e-06, "loss": 0.3574, "step": 695 }, { "epoch": 0.09, "grad_norm": 3.4321575054172695, "learning_rate": 9.992173730355858e-06, "loss": 0.3525, "step": 696 }, { "epoch": 0.09, "grad_norm": 3.79868790685427, "learning_rate": 9.992112001118058e-06, "loss": 0.4123, "step": 697 }, { "epoch": 0.09, "grad_norm": 4.724403923344181, "learning_rate": 9.992050029584943e-06, "loss": 0.4812, "step": 698 }, { "epoch": 0.1, "grad_norm": 3.685702755132042, "learning_rate": 9.991987815759523e-06, "loss": 0.4392, "step": 699 }, { "epoch": 0.1, "grad_norm": 5.083626629028651, "learning_rate": 9.991925359644818e-06, "loss": 0.5263, "step": 700 }, { "epoch": 0.1, "grad_norm": 3.9108763518405567, "learning_rate": 9.991862661243859e-06, "loss": 0.4687, "step": 701 }, { "epoch": 0.1, "grad_norm": 4.2320559810397755, "learning_rate": 9.991799720559687e-06, "loss": 0.479, "step": 702 }, { "epoch": 0.1, "grad_norm": 3.660168904655609, "learning_rate": 9.99173653759536e-06, "loss": 0.4234, "step": 703 }, { "epoch": 0.1, "grad_norm": 3.877394768025337, "learning_rate": 9.991673112353944e-06, "loss": 0.4821, "step": 704 }, { "epoch": 0.1, "grad_norm": 3.5990519778068695, "learning_rate": 9.991609444838518e-06, "loss": 0.4247, "step": 705 }, { "epoch": 0.1, "grad_norm": 3.8667297484580048, "learning_rate": 9.991545535052168e-06, "loss": 0.4594, "step": 706 }, { "epoch": 0.1, "grad_norm": 3.8285057502303825, "learning_rate": 9.991481382998001e-06, "loss": 0.5021, "step": 707 }, { "epoch": 0.1, "grad_norm": 4.311172764500985, "learning_rate": 9.991416988679127e-06, "loss": 0.4711, "step": 708 }, { "epoch": 0.1, "grad_norm": 4.33732834926924, "learning_rate": 9.991352352098673e-06, "loss": 0.4347, "step": 709 }, { "epoch": 0.1, "grad_norm": 3.6016900225496817, "learning_rate": 9.991287473259778e-06, "loss": 0.4481, "step": 710 }, { "epoch": 0.1, "grad_norm": 3.8036375398206634, "learning_rate": 9.991222352165588e-06, "loss": 0.4418, "step": 711 }, { "epoch": 0.1, "grad_norm": 3.8811273048073733, "learning_rate": 9.991156988819264e-06, "loss": 0.4325, "step": 712 }, { "epoch": 0.1, "grad_norm": 3.681157984238525, "learning_rate": 9.991091383223979e-06, "loss": 0.4331, "step": 713 }, { "epoch": 0.1, "grad_norm": 3.9788961955233866, "learning_rate": 9.991025535382919e-06, "loss": 0.5042, "step": 714 }, { "epoch": 0.1, "grad_norm": 4.2259938281142375, "learning_rate": 9.990959445299277e-06, "loss": 0.4829, "step": 715 }, { "epoch": 0.1, "grad_norm": 3.8690853276517037, "learning_rate": 9.990893112976264e-06, "loss": 0.477, "step": 716 }, { "epoch": 0.1, "grad_norm": 4.078451021264796, "learning_rate": 9.990826538417095e-06, "loss": 0.4523, "step": 717 }, { "epoch": 0.1, "grad_norm": 3.4596753646532785, "learning_rate": 9.990759721625005e-06, "loss": 0.4591, "step": 718 }, { "epoch": 0.1, "grad_norm": 3.3912449460144662, "learning_rate": 9.990692662603236e-06, "loss": 0.4202, "step": 719 }, { "epoch": 0.1, "grad_norm": 3.73921571556323, "learning_rate": 9.990625361355042e-06, "loss": 0.4466, "step": 720 }, { "epoch": 0.1, "grad_norm": 3.6131432937474854, "learning_rate": 9.99055781788369e-06, "loss": 0.4487, "step": 721 }, { "epoch": 0.1, "grad_norm": 3.2585820216137305, "learning_rate": 9.99049003219246e-06, "loss": 0.3647, "step": 722 }, { "epoch": 0.1, "grad_norm": 3.9843107319411373, "learning_rate": 9.990422004284639e-06, "loss": 0.4534, "step": 723 }, { "epoch": 0.1, "grad_norm": 3.9435401893352693, "learning_rate": 9.99035373416353e-06, "loss": 0.44, "step": 724 }, { "epoch": 0.1, "grad_norm": 3.8022671450385817, "learning_rate": 9.990285221832447e-06, "loss": 0.4791, "step": 725 }, { "epoch": 0.1, "grad_norm": 4.2114961241532205, "learning_rate": 9.990216467294714e-06, "loss": 0.4579, "step": 726 }, { "epoch": 0.1, "grad_norm": 4.935065084058531, "learning_rate": 9.99014747055367e-06, "loss": 0.5342, "step": 727 }, { "epoch": 0.1, "grad_norm": 4.083128138283087, "learning_rate": 9.990078231612663e-06, "loss": 0.4108, "step": 728 }, { "epoch": 0.1, "grad_norm": 4.138499361707397, "learning_rate": 9.990008750475053e-06, "loss": 0.4772, "step": 729 }, { "epoch": 0.1, "grad_norm": 3.5088964544514614, "learning_rate": 9.989939027144213e-06, "loss": 0.399, "step": 730 }, { "epoch": 0.1, "grad_norm": 4.694041309826441, "learning_rate": 9.989869061623527e-06, "loss": 0.5379, "step": 731 }, { "epoch": 0.1, "grad_norm": 4.264073216575001, "learning_rate": 9.989798853916388e-06, "loss": 0.4783, "step": 732 }, { "epoch": 0.1, "grad_norm": 3.5826479350296205, "learning_rate": 9.989728404026208e-06, "loss": 0.5197, "step": 733 }, { "epoch": 0.1, "grad_norm": 3.417301838381401, "learning_rate": 9.989657711956406e-06, "loss": 0.4254, "step": 734 }, { "epoch": 0.1, "grad_norm": 3.798774328447033, "learning_rate": 9.989586777710409e-06, "loss": 0.4265, "step": 735 }, { "epoch": 0.1, "grad_norm": 3.913007478860528, "learning_rate": 9.989515601291663e-06, "loss": 0.4974, "step": 736 }, { "epoch": 0.1, "grad_norm": 3.6875165257678315, "learning_rate": 9.989444182703623e-06, "loss": 0.4242, "step": 737 }, { "epoch": 0.1, "grad_norm": 3.572320014978045, "learning_rate": 9.989372521949754e-06, "loss": 0.4516, "step": 738 }, { "epoch": 0.1, "grad_norm": 3.7445727481417843, "learning_rate": 9.989300619033532e-06, "loss": 0.4445, "step": 739 }, { "epoch": 0.1, "grad_norm": 3.3851288817735288, "learning_rate": 9.989228473958452e-06, "loss": 0.3786, "step": 740 }, { "epoch": 0.1, "grad_norm": 3.5709662799659085, "learning_rate": 9.989156086728014e-06, "loss": 0.4286, "step": 741 }, { "epoch": 0.1, "grad_norm": 3.778351111153784, "learning_rate": 9.989083457345727e-06, "loss": 0.4352, "step": 742 }, { "epoch": 0.1, "grad_norm": 3.685216319556074, "learning_rate": 9.989010585815121e-06, "loss": 0.4972, "step": 743 }, { "epoch": 0.1, "grad_norm": 3.777178133832618, "learning_rate": 9.98893747213973e-06, "loss": 0.4344, "step": 744 }, { "epoch": 0.1, "grad_norm": 3.8687530557799623, "learning_rate": 9.988864116323106e-06, "loss": 0.4731, "step": 745 }, { "epoch": 0.1, "grad_norm": 4.085141653211006, "learning_rate": 9.988790518368806e-06, "loss": 0.4521, "step": 746 }, { "epoch": 0.1, "grad_norm": 4.1145828248194745, "learning_rate": 9.988716678280403e-06, "loss": 0.4501, "step": 747 }, { "epoch": 0.1, "grad_norm": 3.7193444909260913, "learning_rate": 9.988642596061482e-06, "loss": 0.5015, "step": 748 }, { "epoch": 0.1, "grad_norm": 4.917313075285732, "learning_rate": 9.988568271715637e-06, "loss": 0.4606, "step": 749 }, { "epoch": 0.1, "grad_norm": 3.5126891746301854, "learning_rate": 9.988493705246478e-06, "loss": 0.3585, "step": 750 }, { "epoch": 0.1, "grad_norm": 3.5419650264366442, "learning_rate": 9.98841889665762e-06, "loss": 0.3714, "step": 751 }, { "epoch": 0.1, "grad_norm": 3.3703627217257055, "learning_rate": 9.988343845952697e-06, "loss": 0.3927, "step": 752 }, { "epoch": 0.1, "grad_norm": 3.4501143112668413, "learning_rate": 9.98826855313535e-06, "loss": 0.4079, "step": 753 }, { "epoch": 0.1, "grad_norm": 3.684980108789023, "learning_rate": 9.988193018209236e-06, "loss": 0.4183, "step": 754 }, { "epoch": 0.1, "grad_norm": 4.283584472856652, "learning_rate": 9.988117241178019e-06, "loss": 0.4893, "step": 755 }, { "epoch": 0.1, "grad_norm": 4.1689276698709605, "learning_rate": 9.988041222045378e-06, "loss": 0.4408, "step": 756 }, { "epoch": 0.1, "grad_norm": 4.238580465936238, "learning_rate": 9.987964960815e-06, "loss": 0.4429, "step": 757 }, { "epoch": 0.1, "grad_norm": 3.5894013525341437, "learning_rate": 9.98788845749059e-06, "loss": 0.4158, "step": 758 }, { "epoch": 0.1, "grad_norm": 3.673336558920734, "learning_rate": 9.987811712075856e-06, "loss": 0.4367, "step": 759 }, { "epoch": 0.1, "grad_norm": 3.6262102686533346, "learning_rate": 9.98773472457453e-06, "loss": 0.427, "step": 760 }, { "epoch": 0.1, "grad_norm": 4.196042907509368, "learning_rate": 9.987657494990344e-06, "loss": 0.4706, "step": 761 }, { "epoch": 0.1, "grad_norm": 4.673696888706038, "learning_rate": 9.987580023327046e-06, "loss": 0.446, "step": 762 }, { "epoch": 0.1, "grad_norm": 3.8227312344165583, "learning_rate": 9.9875023095884e-06, "loss": 0.5045, "step": 763 }, { "epoch": 0.1, "grad_norm": 3.638975237754077, "learning_rate": 9.987424353778172e-06, "loss": 0.5065, "step": 764 }, { "epoch": 0.1, "grad_norm": 3.6679058798109443, "learning_rate": 9.987346155900152e-06, "loss": 0.3963, "step": 765 }, { "epoch": 0.1, "grad_norm": 3.8570756172792624, "learning_rate": 9.98726771595813e-06, "loss": 0.4627, "step": 766 }, { "epoch": 0.1, "grad_norm": 3.5110486686090616, "learning_rate": 9.987189033955918e-06, "loss": 0.429, "step": 767 }, { "epoch": 0.1, "grad_norm": 3.8010077215494062, "learning_rate": 9.987110109897331e-06, "loss": 0.5489, "step": 768 }, { "epoch": 0.1, "grad_norm": 3.6954294843676876, "learning_rate": 9.9870309437862e-06, "loss": 0.4554, "step": 769 }, { "epoch": 0.1, "grad_norm": 3.5308495521496233, "learning_rate": 9.98695153562637e-06, "loss": 0.4402, "step": 770 }, { "epoch": 0.1, "grad_norm": 3.726641526885968, "learning_rate": 9.986871885421693e-06, "loss": 0.442, "step": 771 }, { "epoch": 0.11, "grad_norm": 3.959428228668203, "learning_rate": 9.986791993176035e-06, "loss": 0.493, "step": 772 }, { "epoch": 0.11, "grad_norm": 3.8096944171933314, "learning_rate": 9.986711858893274e-06, "loss": 0.4286, "step": 773 }, { "epoch": 0.11, "grad_norm": 3.608793882455618, "learning_rate": 9.986631482577301e-06, "loss": 0.4582, "step": 774 }, { "epoch": 0.11, "grad_norm": 3.6629003959858655, "learning_rate": 9.986550864232014e-06, "loss": 0.4343, "step": 775 }, { "epoch": 0.11, "grad_norm": 3.893838330078622, "learning_rate": 9.986470003861328e-06, "loss": 0.4328, "step": 776 }, { "epoch": 0.11, "grad_norm": 3.974281313458577, "learning_rate": 9.986388901469167e-06, "loss": 0.4229, "step": 777 }, { "epoch": 0.11, "grad_norm": 4.099052063884827, "learning_rate": 9.986307557059467e-06, "loss": 0.5141, "step": 778 }, { "epoch": 0.11, "grad_norm": 3.9045886793257436, "learning_rate": 9.986225970636177e-06, "loss": 0.4066, "step": 779 }, { "epoch": 0.11, "grad_norm": 3.3244847222072345, "learning_rate": 9.986144142203253e-06, "loss": 0.3775, "step": 780 }, { "epoch": 0.11, "grad_norm": 3.757298910100173, "learning_rate": 9.986062071764674e-06, "loss": 0.4374, "step": 781 }, { "epoch": 0.11, "grad_norm": 4.241532348780391, "learning_rate": 9.985979759324418e-06, "loss": 0.4311, "step": 782 }, { "epoch": 0.11, "grad_norm": 3.5667612243330034, "learning_rate": 9.985897204886481e-06, "loss": 0.4955, "step": 783 }, { "epoch": 0.11, "grad_norm": 3.4959197471170746, "learning_rate": 9.985814408454871e-06, "loss": 0.5121, "step": 784 }, { "epoch": 0.11, "grad_norm": 3.493931471825948, "learning_rate": 9.985731370033604e-06, "loss": 0.4138, "step": 785 }, { "epoch": 0.11, "grad_norm": 3.8465923646054114, "learning_rate": 9.985648089626712e-06, "loss": 0.4321, "step": 786 }, { "epoch": 0.11, "grad_norm": 3.601071087176339, "learning_rate": 9.985564567238237e-06, "loss": 0.3604, "step": 787 }, { "epoch": 0.11, "grad_norm": 3.4323796853201394, "learning_rate": 9.985480802872236e-06, "loss": 0.3989, "step": 788 }, { "epoch": 0.11, "grad_norm": 3.831090589289812, "learning_rate": 9.985396796532768e-06, "loss": 0.4984, "step": 789 }, { "epoch": 0.11, "grad_norm": 3.6875062718599683, "learning_rate": 9.985312548223915e-06, "loss": 0.4594, "step": 790 }, { "epoch": 0.11, "grad_norm": 3.31948354183924, "learning_rate": 9.985228057949765e-06, "loss": 0.4021, "step": 791 }, { "epoch": 0.11, "grad_norm": 3.5722201109357488, "learning_rate": 9.985143325714419e-06, "loss": 0.4129, "step": 792 }, { "epoch": 0.11, "grad_norm": 4.26927885324235, "learning_rate": 9.985058351521987e-06, "loss": 0.463, "step": 793 }, { "epoch": 0.11, "grad_norm": 4.348880226337345, "learning_rate": 9.984973135376598e-06, "loss": 0.4676, "step": 794 }, { "epoch": 0.11, "grad_norm": 3.9371664851630723, "learning_rate": 9.984887677282384e-06, "loss": 0.4671, "step": 795 }, { "epoch": 0.11, "grad_norm": 3.571262523493251, "learning_rate": 9.984801977243494e-06, "loss": 0.3868, "step": 796 }, { "epoch": 0.11, "grad_norm": 4.203468810602752, "learning_rate": 9.984716035264089e-06, "loss": 0.4183, "step": 797 }, { "epoch": 0.11, "grad_norm": 3.4349035050037324, "learning_rate": 9.984629851348338e-06, "loss": 0.4087, "step": 798 }, { "epoch": 0.11, "grad_norm": 3.392596917880574, "learning_rate": 9.984543425500426e-06, "loss": 0.4077, "step": 799 }, { "epoch": 0.11, "grad_norm": 4.199614171491465, "learning_rate": 9.984456757724545e-06, "loss": 0.4126, "step": 800 }, { "epoch": 0.11, "grad_norm": 3.643039255571014, "learning_rate": 9.984369848024905e-06, "loss": 0.4364, "step": 801 }, { "epoch": 0.11, "grad_norm": 3.583455450046676, "learning_rate": 9.98428269640572e-06, "loss": 0.4346, "step": 802 }, { "epoch": 0.11, "grad_norm": 3.367025362228926, "learning_rate": 9.984195302871224e-06, "loss": 0.4096, "step": 803 }, { "epoch": 0.11, "grad_norm": 3.8915808756163752, "learning_rate": 9.984107667425658e-06, "loss": 0.4487, "step": 804 }, { "epoch": 0.11, "grad_norm": 3.92520972796167, "learning_rate": 9.984019790073271e-06, "loss": 0.4979, "step": 805 }, { "epoch": 0.11, "grad_norm": 3.3705437873806114, "learning_rate": 9.983931670818335e-06, "loss": 0.3826, "step": 806 }, { "epoch": 0.11, "grad_norm": 5.58009240327504, "learning_rate": 9.983843309665122e-06, "loss": 0.4525, "step": 807 }, { "epoch": 0.11, "grad_norm": 3.4891774195199803, "learning_rate": 9.983754706617922e-06, "loss": 0.3832, "step": 808 }, { "epoch": 0.11, "grad_norm": 3.490984777874846, "learning_rate": 9.983665861681039e-06, "loss": 0.4333, "step": 809 }, { "epoch": 0.11, "grad_norm": 3.4796172396993055, "learning_rate": 9.983576774858776e-06, "loss": 0.4105, "step": 810 }, { "epoch": 0.11, "grad_norm": 3.623877598117047, "learning_rate": 9.983487446155467e-06, "loss": 0.4714, "step": 811 }, { "epoch": 0.11, "grad_norm": 3.9157425782240907, "learning_rate": 9.983397875575442e-06, "loss": 0.452, "step": 812 }, { "epoch": 0.11, "grad_norm": 3.7550483440487703, "learning_rate": 9.98330806312305e-06, "loss": 0.4144, "step": 813 }, { "epoch": 0.11, "grad_norm": 4.074309321641612, "learning_rate": 9.983218008802648e-06, "loss": 0.4393, "step": 814 }, { "epoch": 0.11, "grad_norm": 3.94343976971396, "learning_rate": 9.98312771261861e-06, "loss": 0.4627, "step": 815 }, { "epoch": 0.11, "grad_norm": 4.81536971936767, "learning_rate": 9.983037174575317e-06, "loss": 0.482, "step": 816 }, { "epoch": 0.11, "grad_norm": 3.559378879001055, "learning_rate": 9.982946394677165e-06, "loss": 0.4354, "step": 817 }, { "epoch": 0.11, "grad_norm": 3.5273401128559367, "learning_rate": 9.982855372928557e-06, "loss": 0.3963, "step": 818 }, { "epoch": 0.11, "grad_norm": 3.677765740988804, "learning_rate": 9.982764109333912e-06, "loss": 0.413, "step": 819 }, { "epoch": 0.11, "grad_norm": 3.8265036932324827, "learning_rate": 9.982672603897661e-06, "loss": 0.3936, "step": 820 }, { "epoch": 0.11, "grad_norm": 3.725080439851199, "learning_rate": 9.982580856624242e-06, "loss": 0.4016, "step": 821 }, { "epoch": 0.11, "grad_norm": 3.466893552766595, "learning_rate": 9.982488867518112e-06, "loss": 0.4332, "step": 822 }, { "epoch": 0.11, "grad_norm": 3.905178033324297, "learning_rate": 9.982396636583734e-06, "loss": 0.4573, "step": 823 }, { "epoch": 0.11, "grad_norm": 3.7192892404410105, "learning_rate": 9.982304163825584e-06, "loss": 0.3654, "step": 824 }, { "epoch": 0.11, "grad_norm": 3.5917448384665027, "learning_rate": 9.982211449248151e-06, "loss": 0.3848, "step": 825 }, { "epoch": 0.11, "grad_norm": 3.413832195205937, "learning_rate": 9.982118492855934e-06, "loss": 0.4179, "step": 826 }, { "epoch": 0.11, "grad_norm": 3.5818484360549787, "learning_rate": 9.982025294653445e-06, "loss": 0.3788, "step": 827 }, { "epoch": 0.11, "grad_norm": 3.621148857340954, "learning_rate": 9.98193185464521e-06, "loss": 0.4777, "step": 828 }, { "epoch": 0.11, "grad_norm": 3.7236659638226492, "learning_rate": 9.98183817283576e-06, "loss": 0.4141, "step": 829 }, { "epoch": 0.11, "grad_norm": 5.082482522313879, "learning_rate": 9.981744249229643e-06, "loss": 0.4343, "step": 830 }, { "epoch": 0.11, "grad_norm": 3.724205285966483, "learning_rate": 9.98165008383142e-06, "loss": 0.3627, "step": 831 }, { "epoch": 0.11, "grad_norm": 3.5855671806569394, "learning_rate": 9.98155567664566e-06, "loss": 0.4254, "step": 832 }, { "epoch": 0.11, "grad_norm": 3.564208577987073, "learning_rate": 9.981461027676945e-06, "loss": 0.4192, "step": 833 }, { "epoch": 0.11, "grad_norm": 3.5669653548199576, "learning_rate": 9.981366136929868e-06, "loss": 0.4176, "step": 834 }, { "epoch": 0.11, "grad_norm": 4.650139945461489, "learning_rate": 9.981271004409036e-06, "loss": 0.472, "step": 835 }, { "epoch": 0.11, "grad_norm": 3.9054689199671624, "learning_rate": 9.981175630119065e-06, "loss": 0.4346, "step": 836 }, { "epoch": 0.11, "grad_norm": 3.946291450996487, "learning_rate": 9.981080014064584e-06, "loss": 0.4487, "step": 837 }, { "epoch": 0.11, "grad_norm": 5.101364913559284, "learning_rate": 9.980984156250236e-06, "loss": 0.5002, "step": 838 }, { "epoch": 0.11, "grad_norm": 3.3611800356569654, "learning_rate": 9.980888056680672e-06, "loss": 0.4188, "step": 839 }, { "epoch": 0.11, "grad_norm": 3.525915886245597, "learning_rate": 9.980791715360556e-06, "loss": 0.4462, "step": 840 }, { "epoch": 0.11, "grad_norm": 3.7695690425272574, "learning_rate": 9.980695132294565e-06, "loss": 0.4617, "step": 841 }, { "epoch": 0.11, "grad_norm": 3.374492245942483, "learning_rate": 9.980598307487383e-06, "loss": 0.3978, "step": 842 }, { "epoch": 0.11, "grad_norm": 4.13573204316627, "learning_rate": 9.980501240943716e-06, "loss": 0.4697, "step": 843 }, { "epoch": 0.11, "grad_norm": 3.9453998713440837, "learning_rate": 9.98040393266827e-06, "loss": 0.4326, "step": 844 }, { "epoch": 0.11, "grad_norm": 3.362898831932386, "learning_rate": 9.98030638266577e-06, "loss": 0.3885, "step": 845 }, { "epoch": 0.12, "grad_norm": 3.99155697086253, "learning_rate": 9.98020859094095e-06, "loss": 0.4512, "step": 846 }, { "epoch": 0.12, "grad_norm": 3.1553985740062775, "learning_rate": 9.980110557498556e-06, "loss": 0.3845, "step": 847 }, { "epoch": 0.12, "grad_norm": 3.396360435171731, "learning_rate": 9.980012282343348e-06, "loss": 0.3626, "step": 848 }, { "epoch": 0.12, "grad_norm": 3.6191917682157246, "learning_rate": 9.979913765480093e-06, "loss": 0.4226, "step": 849 }, { "epoch": 0.12, "grad_norm": 3.3257239567099224, "learning_rate": 9.979815006913576e-06, "loss": 0.4223, "step": 850 }, { "epoch": 0.12, "grad_norm": 3.4765561548130903, "learning_rate": 9.979716006648588e-06, "loss": 0.4208, "step": 851 }, { "epoch": 0.12, "grad_norm": 3.6000723019447665, "learning_rate": 9.979616764689932e-06, "loss": 0.4169, "step": 852 }, { "epoch": 0.12, "grad_norm": 3.591548420629017, "learning_rate": 9.97951728104243e-06, "loss": 0.441, "step": 853 }, { "epoch": 0.12, "grad_norm": 3.733091335741796, "learning_rate": 9.979417555710908e-06, "loss": 0.4714, "step": 854 }, { "epoch": 0.12, "grad_norm": 3.5407060952680025, "learning_rate": 9.979317588700203e-06, "loss": 0.4504, "step": 855 }, { "epoch": 0.12, "grad_norm": 3.8263889714651964, "learning_rate": 9.979217380015173e-06, "loss": 0.468, "step": 856 }, { "epoch": 0.12, "grad_norm": 3.6133149312142114, "learning_rate": 9.979116929660677e-06, "loss": 0.4284, "step": 857 }, { "epoch": 0.12, "grad_norm": 4.293519791213985, "learning_rate": 9.979016237641593e-06, "loss": 0.4211, "step": 858 }, { "epoch": 0.12, "grad_norm": 3.984186575176655, "learning_rate": 9.978915303962808e-06, "loss": 0.4453, "step": 859 }, { "epoch": 0.12, "grad_norm": 3.7304933332919195, "learning_rate": 9.978814128629219e-06, "loss": 0.4221, "step": 860 }, { "epoch": 0.12, "grad_norm": 4.049131102319004, "learning_rate": 9.978712711645738e-06, "loss": 0.4371, "step": 861 }, { "epoch": 0.12, "grad_norm": 4.706791905804332, "learning_rate": 9.978611053017286e-06, "loss": 0.4396, "step": 862 }, { "epoch": 0.12, "grad_norm": 4.2876972094227925, "learning_rate": 9.978509152748798e-06, "loss": 0.4473, "step": 863 }, { "epoch": 0.12, "grad_norm": 3.642419962129459, "learning_rate": 9.978407010845222e-06, "loss": 0.4432, "step": 864 }, { "epoch": 0.12, "grad_norm": 3.614610041872827, "learning_rate": 9.978304627311513e-06, "loss": 0.4491, "step": 865 }, { "epoch": 0.12, "grad_norm": 3.619767879031412, "learning_rate": 9.978202002152639e-06, "loss": 0.4528, "step": 866 }, { "epoch": 0.12, "grad_norm": 3.5949601309585035, "learning_rate": 9.978099135373584e-06, "loss": 0.4104, "step": 867 }, { "epoch": 0.12, "grad_norm": 3.511080844286839, "learning_rate": 9.977996026979338e-06, "loss": 0.4916, "step": 868 }, { "epoch": 0.12, "grad_norm": 4.429117274409899, "learning_rate": 9.977892676974908e-06, "loss": 0.4704, "step": 869 }, { "epoch": 0.12, "grad_norm": 3.934657505976631, "learning_rate": 9.977789085365308e-06, "loss": 0.4869, "step": 870 }, { "epoch": 0.12, "grad_norm": 3.923025924006177, "learning_rate": 9.977685252155569e-06, "loss": 0.4746, "step": 871 }, { "epoch": 0.12, "grad_norm": 3.497867966075496, "learning_rate": 9.977581177350726e-06, "loss": 0.4032, "step": 872 }, { "epoch": 0.12, "grad_norm": 3.480162714090871, "learning_rate": 9.977476860955834e-06, "loss": 0.4893, "step": 873 }, { "epoch": 0.12, "grad_norm": 3.593438384123019, "learning_rate": 9.977372302975954e-06, "loss": 0.4104, "step": 874 }, { "epoch": 0.12, "grad_norm": 3.697003171759777, "learning_rate": 9.977267503416163e-06, "loss": 0.4446, "step": 875 }, { "epoch": 0.12, "grad_norm": 3.557581692349847, "learning_rate": 9.977162462281544e-06, "loss": 0.4898, "step": 876 }, { "epoch": 0.12, "grad_norm": 3.4165205885802, "learning_rate": 9.977057179577199e-06, "loss": 0.3479, "step": 877 }, { "epoch": 0.12, "grad_norm": 3.673897224972876, "learning_rate": 9.976951655308236e-06, "loss": 0.4796, "step": 878 }, { "epoch": 0.12, "grad_norm": 3.334408649092537, "learning_rate": 9.976845889479778e-06, "loss": 0.3552, "step": 879 }, { "epoch": 0.12, "grad_norm": 4.025145247564824, "learning_rate": 9.976739882096956e-06, "loss": 0.4284, "step": 880 }, { "epoch": 0.12, "grad_norm": 3.9755975723189905, "learning_rate": 9.976633633164918e-06, "loss": 0.4904, "step": 881 }, { "epoch": 0.12, "grad_norm": 4.000112214348706, "learning_rate": 9.976527142688818e-06, "loss": 0.4817, "step": 882 }, { "epoch": 0.12, "grad_norm": 4.048943265080465, "learning_rate": 9.976420410673826e-06, "loss": 0.4072, "step": 883 }, { "epoch": 0.12, "grad_norm": 4.059771759965085, "learning_rate": 9.976313437125124e-06, "loss": 0.441, "step": 884 }, { "epoch": 0.12, "grad_norm": 3.7104112085074514, "learning_rate": 9.9762062220479e-06, "loss": 0.4275, "step": 885 }, { "epoch": 0.12, "grad_norm": 3.8064792833852894, "learning_rate": 9.976098765447361e-06, "loss": 0.4473, "step": 886 }, { "epoch": 0.12, "grad_norm": 3.6497871565989763, "learning_rate": 9.975991067328722e-06, "loss": 0.4182, "step": 887 }, { "epoch": 0.12, "grad_norm": 3.910563473578966, "learning_rate": 9.97588312769721e-06, "loss": 0.3682, "step": 888 }, { "epoch": 0.12, "grad_norm": 4.736760347089691, "learning_rate": 9.975774946558064e-06, "loss": 0.5077, "step": 889 }, { "epoch": 0.12, "grad_norm": 4.335399107141557, "learning_rate": 9.975666523916534e-06, "loss": 0.4451, "step": 890 }, { "epoch": 0.12, "grad_norm": 5.210713410897175, "learning_rate": 9.975557859777882e-06, "loss": 0.4314, "step": 891 }, { "epoch": 0.12, "grad_norm": 3.333919694098228, "learning_rate": 9.975448954147383e-06, "loss": 0.4633, "step": 892 }, { "epoch": 0.12, "grad_norm": 3.2781030787961605, "learning_rate": 9.975339807030323e-06, "loss": 0.3557, "step": 893 }, { "epoch": 0.12, "grad_norm": 3.9564521896118667, "learning_rate": 9.975230418431998e-06, "loss": 0.3791, "step": 894 }, { "epoch": 0.12, "grad_norm": 3.539918866250391, "learning_rate": 9.97512078835772e-06, "loss": 0.3835, "step": 895 }, { "epoch": 0.12, "grad_norm": 3.1926419704327227, "learning_rate": 9.975010916812808e-06, "loss": 0.3931, "step": 896 }, { "epoch": 0.12, "grad_norm": 3.732044061059728, "learning_rate": 9.974900803802595e-06, "loss": 0.3826, "step": 897 }, { "epoch": 0.12, "grad_norm": 3.9711207593511935, "learning_rate": 9.974790449332424e-06, "loss": 0.4824, "step": 898 }, { "epoch": 0.12, "grad_norm": 3.8279510442709364, "learning_rate": 9.974679853407653e-06, "loss": 0.4088, "step": 899 }, { "epoch": 0.12, "grad_norm": 5.734331647371014, "learning_rate": 9.97456901603365e-06, "loss": 0.4148, "step": 900 }, { "epoch": 0.12, "grad_norm": 3.5793358816230687, "learning_rate": 9.974457937215795e-06, "loss": 0.4347, "step": 901 }, { "epoch": 0.12, "grad_norm": 3.306032063963768, "learning_rate": 9.974346616959476e-06, "loss": 0.3218, "step": 902 }, { "epoch": 0.12, "grad_norm": 3.5363827482568997, "learning_rate": 9.9742350552701e-06, "loss": 0.4171, "step": 903 }, { "epoch": 0.12, "grad_norm": 4.336884617542349, "learning_rate": 9.974123252153078e-06, "loss": 0.501, "step": 904 }, { "epoch": 0.12, "grad_norm": 3.586129428624226, "learning_rate": 9.97401120761384e-06, "loss": 0.3733, "step": 905 }, { "epoch": 0.12, "grad_norm": 3.569179830570361, "learning_rate": 9.973898921657823e-06, "loss": 0.3873, "step": 906 }, { "epoch": 0.12, "grad_norm": 3.7033365685667015, "learning_rate": 9.973786394290475e-06, "loss": 0.4364, "step": 907 }, { "epoch": 0.12, "grad_norm": 3.647325447184198, "learning_rate": 9.973673625517259e-06, "loss": 0.4325, "step": 908 }, { "epoch": 0.12, "grad_norm": 3.781637895900407, "learning_rate": 9.973560615343649e-06, "loss": 0.4377, "step": 909 }, { "epoch": 0.12, "grad_norm": 3.439122077593651, "learning_rate": 9.973447363775128e-06, "loss": 0.433, "step": 910 }, { "epoch": 0.12, "grad_norm": 3.477923612758594, "learning_rate": 9.973333870817197e-06, "loss": 0.3762, "step": 911 }, { "epoch": 0.12, "grad_norm": 3.804711122651126, "learning_rate": 9.973220136475359e-06, "loss": 0.3737, "step": 912 }, { "epoch": 0.12, "grad_norm": 3.555070830680235, "learning_rate": 9.973106160755136e-06, "loss": 0.4617, "step": 913 }, { "epoch": 0.12, "grad_norm": 3.802386771119473, "learning_rate": 9.972991943662062e-06, "loss": 0.4222, "step": 914 }, { "epoch": 0.12, "grad_norm": 3.7914744975951558, "learning_rate": 9.97287748520168e-06, "loss": 0.4433, "step": 915 }, { "epoch": 0.12, "grad_norm": 3.3718587381302205, "learning_rate": 9.972762785379543e-06, "loss": 0.4518, "step": 916 }, { "epoch": 0.12, "grad_norm": 3.3672512825912904, "learning_rate": 9.97264784420122e-06, "loss": 0.4059, "step": 917 }, { "epoch": 0.12, "grad_norm": 4.3466853906737075, "learning_rate": 9.972532661672288e-06, "loss": 0.4373, "step": 918 }, { "epoch": 0.13, "grad_norm": 3.965511883046672, "learning_rate": 9.97241723779834e-06, "loss": 0.4635, "step": 919 }, { "epoch": 0.13, "grad_norm": 3.6508130018830873, "learning_rate": 9.972301572584977e-06, "loss": 0.3858, "step": 920 }, { "epoch": 0.13, "grad_norm": 3.6901730176537084, "learning_rate": 9.972185666037812e-06, "loss": 0.4036, "step": 921 }, { "epoch": 0.13, "grad_norm": 4.478265680075037, "learning_rate": 9.972069518162472e-06, "loss": 0.4564, "step": 922 }, { "epoch": 0.13, "grad_norm": 4.786857209247731, "learning_rate": 9.971953128964593e-06, "loss": 0.3848, "step": 923 }, { "epoch": 0.13, "grad_norm": 3.752284819779497, "learning_rate": 9.971836498449824e-06, "loss": 0.3962, "step": 924 }, { "epoch": 0.13, "grad_norm": 3.8970493933299424, "learning_rate": 9.971719626623827e-06, "loss": 0.3994, "step": 925 }, { "epoch": 0.13, "grad_norm": 3.752868955765207, "learning_rate": 9.971602513492276e-06, "loss": 0.4562, "step": 926 }, { "epoch": 0.13, "grad_norm": 3.685661363213032, "learning_rate": 9.971485159060851e-06, "loss": 0.3741, "step": 927 }, { "epoch": 0.13, "grad_norm": 3.4318588385589512, "learning_rate": 9.97136756333525e-06, "loss": 0.3743, "step": 928 }, { "epoch": 0.13, "grad_norm": 3.3192379418653086, "learning_rate": 9.97124972632118e-06, "loss": 0.4115, "step": 929 }, { "epoch": 0.13, "grad_norm": 3.356108367137899, "learning_rate": 9.971131648024362e-06, "loss": 0.3863, "step": 930 }, { "epoch": 0.13, "grad_norm": 3.871682075418105, "learning_rate": 9.971013328450526e-06, "loss": 0.4802, "step": 931 }, { "epoch": 0.13, "grad_norm": 3.483042379841636, "learning_rate": 9.970894767605412e-06, "loss": 0.4011, "step": 932 }, { "epoch": 0.13, "grad_norm": 4.101610440002108, "learning_rate": 9.970775965494778e-06, "loss": 0.3929, "step": 933 }, { "epoch": 0.13, "grad_norm": 3.8707378820476976, "learning_rate": 9.97065692212439e-06, "loss": 0.4021, "step": 934 }, { "epoch": 0.13, "grad_norm": 3.87812304773653, "learning_rate": 9.970537637500024e-06, "loss": 0.4063, "step": 935 }, { "epoch": 0.13, "grad_norm": 3.641058185959528, "learning_rate": 9.970418111627471e-06, "loss": 0.3816, "step": 936 }, { "epoch": 0.13, "grad_norm": 4.147227407333795, "learning_rate": 9.970298344512533e-06, "loss": 0.509, "step": 937 }, { "epoch": 0.13, "grad_norm": 3.7849581932856364, "learning_rate": 9.970178336161018e-06, "loss": 0.481, "step": 938 }, { "epoch": 0.13, "grad_norm": 3.3854310765994353, "learning_rate": 9.970058086578758e-06, "loss": 0.3993, "step": 939 }, { "epoch": 0.13, "grad_norm": 3.5356227379163485, "learning_rate": 9.969937595771584e-06, "loss": 0.424, "step": 940 }, { "epoch": 0.13, "grad_norm": 3.7546532395016046, "learning_rate": 9.969816863745345e-06, "loss": 0.3958, "step": 941 }, { "epoch": 0.13, "grad_norm": 4.014826312509291, "learning_rate": 9.969695890505904e-06, "loss": 0.4497, "step": 942 }, { "epoch": 0.13, "grad_norm": 3.274811606261925, "learning_rate": 9.969574676059129e-06, "loss": 0.4043, "step": 943 }, { "epoch": 0.13, "grad_norm": 3.4919613318964573, "learning_rate": 9.969453220410903e-06, "loss": 0.386, "step": 944 }, { "epoch": 0.13, "grad_norm": 4.01207374834551, "learning_rate": 9.969331523567126e-06, "loss": 0.4501, "step": 945 }, { "epoch": 0.13, "grad_norm": 3.7879664953570145, "learning_rate": 9.969209585533697e-06, "loss": 0.4455, "step": 946 }, { "epoch": 0.13, "grad_norm": 3.4891821580323943, "learning_rate": 9.96908740631654e-06, "loss": 0.3814, "step": 947 }, { "epoch": 0.13, "grad_norm": 3.3367790297528352, "learning_rate": 9.968964985921584e-06, "loss": 0.3584, "step": 948 }, { "epoch": 0.13, "grad_norm": 4.039464598541204, "learning_rate": 9.968842324354769e-06, "loss": 0.4051, "step": 949 }, { "epoch": 0.13, "grad_norm": 3.6514307221496387, "learning_rate": 9.968719421622049e-06, "loss": 0.4057, "step": 950 }, { "epoch": 0.13, "grad_norm": 3.6893314355308426, "learning_rate": 9.968596277729392e-06, "loss": 0.4538, "step": 951 }, { "epoch": 0.13, "grad_norm": 3.749158973773489, "learning_rate": 9.96847289268277e-06, "loss": 0.4558, "step": 952 }, { "epoch": 0.13, "grad_norm": 3.634099573552168, "learning_rate": 9.968349266488176e-06, "loss": 0.4391, "step": 953 }, { "epoch": 0.13, "grad_norm": 3.6501901407423967, "learning_rate": 9.968225399151607e-06, "loss": 0.473, "step": 954 }, { "epoch": 0.13, "grad_norm": 4.003862240929404, "learning_rate": 9.968101290679078e-06, "loss": 0.4744, "step": 955 }, { "epoch": 0.13, "grad_norm": 3.170656810385927, "learning_rate": 9.96797694107661e-06, "loss": 0.3597, "step": 956 }, { "epoch": 0.13, "grad_norm": 3.2126804803114957, "learning_rate": 9.967852350350239e-06, "loss": 0.3713, "step": 957 }, { "epoch": 0.13, "grad_norm": 3.9244444022439353, "learning_rate": 9.967727518506014e-06, "loss": 0.4029, "step": 958 }, { "epoch": 0.13, "grad_norm": 3.9250407600026835, "learning_rate": 9.96760244554999e-06, "loss": 0.4671, "step": 959 }, { "epoch": 0.13, "grad_norm": 3.726891974534164, "learning_rate": 9.96747713148824e-06, "loss": 0.4533, "step": 960 }, { "epoch": 0.13, "grad_norm": 3.063861765143171, "learning_rate": 9.96735157632685e-06, "loss": 0.3493, "step": 961 }, { "epoch": 0.13, "grad_norm": 3.6743024061074103, "learning_rate": 9.967225780071908e-06, "loss": 0.4598, "step": 962 }, { "epoch": 0.13, "grad_norm": 3.825091738655482, "learning_rate": 9.96709974272952e-06, "loss": 0.4573, "step": 963 }, { "epoch": 0.13, "grad_norm": 3.33183979124593, "learning_rate": 9.966973464305805e-06, "loss": 0.4203, "step": 964 }, { "epoch": 0.13, "grad_norm": 3.3783634817049606, "learning_rate": 9.966846944806892e-06, "loss": 0.4571, "step": 965 }, { "epoch": 0.13, "grad_norm": 3.802062654231774, "learning_rate": 9.966720184238923e-06, "loss": 0.4198, "step": 966 }, { "epoch": 0.13, "grad_norm": 3.6537116937291936, "learning_rate": 9.966593182608048e-06, "loss": 0.4289, "step": 967 }, { "epoch": 0.13, "grad_norm": 3.38801367824599, "learning_rate": 9.966465939920431e-06, "loss": 0.4613, "step": 968 }, { "epoch": 0.13, "grad_norm": 3.6495516711937817, "learning_rate": 9.96633845618225e-06, "loss": 0.3735, "step": 969 }, { "epoch": 0.13, "grad_norm": 3.7521678504824085, "learning_rate": 9.966210731399692e-06, "loss": 0.4631, "step": 970 }, { "epoch": 0.13, "grad_norm": 3.660324605328734, "learning_rate": 9.966082765578956e-06, "loss": 0.4542, "step": 971 }, { "epoch": 0.13, "grad_norm": 3.460149039893484, "learning_rate": 9.965954558726249e-06, "loss": 0.4072, "step": 972 }, { "epoch": 0.13, "grad_norm": 4.180120715966375, "learning_rate": 9.9658261108478e-06, "loss": 0.4744, "step": 973 }, { "epoch": 0.13, "grad_norm": 4.560377129649893, "learning_rate": 9.96569742194984e-06, "loss": 0.4337, "step": 974 }, { "epoch": 0.13, "grad_norm": 4.083409344908946, "learning_rate": 9.965568492038616e-06, "loss": 0.447, "step": 975 }, { "epoch": 0.13, "grad_norm": 3.5184550207982794, "learning_rate": 9.965439321120383e-06, "loss": 0.3577, "step": 976 }, { "epoch": 0.13, "grad_norm": 3.7624505765210574, "learning_rate": 9.965309909201414e-06, "loss": 0.3992, "step": 977 }, { "epoch": 0.13, "grad_norm": 3.461227958137638, "learning_rate": 9.965180256287988e-06, "loss": 0.4183, "step": 978 }, { "epoch": 0.13, "grad_norm": 3.485345489172927, "learning_rate": 9.9650503623864e-06, "loss": 0.3497, "step": 979 }, { "epoch": 0.13, "grad_norm": 3.697789984025576, "learning_rate": 9.96492022750295e-06, "loss": 0.4307, "step": 980 }, { "epoch": 0.13, "grad_norm": 3.1995733086402627, "learning_rate": 9.964789851643959e-06, "loss": 0.3195, "step": 981 }, { "epoch": 0.13, "grad_norm": 4.022680910534052, "learning_rate": 9.964659234815752e-06, "loss": 0.4245, "step": 982 }, { "epoch": 0.13, "grad_norm": 3.697613338060207, "learning_rate": 9.964528377024669e-06, "loss": 0.3851, "step": 983 }, { "epoch": 0.13, "grad_norm": 3.1691627322432394, "learning_rate": 9.964397278277063e-06, "loss": 0.3425, "step": 984 }, { "epoch": 0.13, "grad_norm": 3.612845789581019, "learning_rate": 9.964265938579294e-06, "loss": 0.4342, "step": 985 }, { "epoch": 0.13, "grad_norm": 4.341699351898425, "learning_rate": 9.96413435793774e-06, "loss": 0.486, "step": 986 }, { "epoch": 0.13, "grad_norm": 4.03922168321354, "learning_rate": 9.964002536358784e-06, "loss": 0.3875, "step": 987 }, { "epoch": 0.13, "grad_norm": 4.553653083372239, "learning_rate": 9.963870473848827e-06, "loss": 0.4923, "step": 988 }, { "epoch": 0.13, "grad_norm": 4.2548522220126115, "learning_rate": 9.963738170414277e-06, "loss": 0.4451, "step": 989 }, { "epoch": 0.13, "grad_norm": 3.8858785162974194, "learning_rate": 9.963605626061556e-06, "loss": 0.4243, "step": 990 }, { "epoch": 0.13, "grad_norm": 3.7920558989134476, "learning_rate": 9.963472840797096e-06, "loss": 0.5219, "step": 991 }, { "epoch": 0.13, "grad_norm": 3.90493193201016, "learning_rate": 9.963339814627344e-06, "loss": 0.4535, "step": 992 }, { "epoch": 0.14, "grad_norm": 3.781331806117128, "learning_rate": 9.963206547558756e-06, "loss": 0.4292, "step": 993 }, { "epoch": 0.14, "grad_norm": 3.5640552877353304, "learning_rate": 9.963073039597798e-06, "loss": 0.4129, "step": 994 }, { "epoch": 0.14, "grad_norm": 3.881371688686478, "learning_rate": 9.962939290750953e-06, "loss": 0.4283, "step": 995 }, { "epoch": 0.14, "grad_norm": 3.5636639000844514, "learning_rate": 9.962805301024709e-06, "loss": 0.3643, "step": 996 }, { "epoch": 0.14, "grad_norm": 7.156204408383229, "learning_rate": 9.962671070425573e-06, "loss": 0.4514, "step": 997 }, { "epoch": 0.14, "grad_norm": 4.5438271908073125, "learning_rate": 9.962536598960058e-06, "loss": 0.4483, "step": 998 }, { "epoch": 0.14, "grad_norm": 3.5236990981557894, "learning_rate": 9.96240188663469e-06, "loss": 0.4301, "step": 999 }, { "epoch": 0.14, "grad_norm": 3.940066693249747, "learning_rate": 9.962266933456008e-06, "loss": 0.409, "step": 1000 }, { "epoch": 0.14, "grad_norm": 3.4440928126585164, "learning_rate": 9.962131739430563e-06, "loss": 0.4234, "step": 1001 }, { "epoch": 0.14, "grad_norm": 3.8670481818666453, "learning_rate": 9.961996304564916e-06, "loss": 0.4154, "step": 1002 }, { "epoch": 0.14, "grad_norm": 3.7641912490658727, "learning_rate": 9.96186062886564e-06, "loss": 0.4053, "step": 1003 }, { "epoch": 0.14, "grad_norm": 3.5765729041011003, "learning_rate": 9.961724712339322e-06, "loss": 0.4668, "step": 1004 }, { "epoch": 0.14, "grad_norm": 3.331495491074906, "learning_rate": 9.961588554992557e-06, "loss": 0.3953, "step": 1005 }, { "epoch": 0.14, "grad_norm": 3.3104408027822725, "learning_rate": 9.961452156831951e-06, "loss": 0.3729, "step": 1006 }, { "epoch": 0.14, "grad_norm": 3.160251042425179, "learning_rate": 9.961315517864131e-06, "loss": 0.3656, "step": 1007 }, { "epoch": 0.14, "grad_norm": 3.9212722947957235, "learning_rate": 9.961178638095723e-06, "loss": 0.3716, "step": 1008 }, { "epoch": 0.14, "grad_norm": 3.4066826105318073, "learning_rate": 9.961041517533372e-06, "loss": 0.414, "step": 1009 }, { "epoch": 0.14, "grad_norm": 4.1752580624411735, "learning_rate": 9.960904156183735e-06, "loss": 0.4438, "step": 1010 }, { "epoch": 0.14, "grad_norm": 3.4832878760588586, "learning_rate": 9.960766554053477e-06, "loss": 0.3847, "step": 1011 }, { "epoch": 0.14, "grad_norm": 4.058238321001869, "learning_rate": 9.960628711149276e-06, "loss": 0.4156, "step": 1012 }, { "epoch": 0.14, "grad_norm": 3.581098253055751, "learning_rate": 9.960490627477823e-06, "loss": 0.387, "step": 1013 }, { "epoch": 0.14, "grad_norm": 3.3508335890339653, "learning_rate": 9.960352303045821e-06, "loss": 0.4309, "step": 1014 }, { "epoch": 0.14, "grad_norm": 3.1360416367872395, "learning_rate": 9.960213737859984e-06, "loss": 0.359, "step": 1015 }, { "epoch": 0.14, "grad_norm": 3.5555155420202986, "learning_rate": 9.960074931927038e-06, "loss": 0.4051, "step": 1016 }, { "epoch": 0.14, "grad_norm": 3.649645747248781, "learning_rate": 9.959935885253715e-06, "loss": 0.3828, "step": 1017 }, { "epoch": 0.14, "grad_norm": 3.728192871645846, "learning_rate": 9.959796597846769e-06, "loss": 0.3977, "step": 1018 }, { "epoch": 0.14, "grad_norm": 3.71527614985473, "learning_rate": 9.95965706971296e-06, "loss": 0.393, "step": 1019 }, { "epoch": 0.14, "grad_norm": 4.197867378428486, "learning_rate": 9.959517300859055e-06, "loss": 0.4025, "step": 1020 }, { "epoch": 0.14, "grad_norm": 3.0293074045549173, "learning_rate": 9.959377291291846e-06, "loss": 0.3545, "step": 1021 }, { "epoch": 0.14, "grad_norm": 3.4950761738586777, "learning_rate": 9.95923704101812e-06, "loss": 0.3916, "step": 1022 }, { "epoch": 0.14, "grad_norm": 3.480785482635199, "learning_rate": 9.959096550044688e-06, "loss": 0.3598, "step": 1023 }, { "epoch": 0.14, "grad_norm": 3.585592259094522, "learning_rate": 9.958955818378372e-06, "loss": 0.4317, "step": 1024 }, { "epoch": 0.14, "grad_norm": 3.358784111661446, "learning_rate": 9.958814846025996e-06, "loss": 0.3922, "step": 1025 }, { "epoch": 0.14, "grad_norm": 3.6211125614431228, "learning_rate": 9.95867363299441e-06, "loss": 0.4177, "step": 1026 }, { "epoch": 0.14, "grad_norm": 3.623163583643859, "learning_rate": 9.958532179290458e-06, "loss": 0.3814, "step": 1027 }, { "epoch": 0.14, "grad_norm": 3.3223236419918587, "learning_rate": 9.958390484921015e-06, "loss": 0.4225, "step": 1028 }, { "epoch": 0.14, "grad_norm": 3.392412392181483, "learning_rate": 9.958248549892953e-06, "loss": 0.3305, "step": 1029 }, { "epoch": 0.14, "grad_norm": 3.2264757355673437, "learning_rate": 9.958106374213161e-06, "loss": 0.3739, "step": 1030 }, { "epoch": 0.14, "grad_norm": 3.781127575228496, "learning_rate": 9.957963957888542e-06, "loss": 0.3576, "step": 1031 }, { "epoch": 0.14, "grad_norm": 3.8100992366581816, "learning_rate": 9.957821300926007e-06, "loss": 0.4552, "step": 1032 }, { "epoch": 0.14, "grad_norm": 3.775749686649478, "learning_rate": 9.957678403332477e-06, "loss": 0.3838, "step": 1033 }, { "epoch": 0.14, "grad_norm": 3.716358990362804, "learning_rate": 9.957535265114895e-06, "loss": 0.3683, "step": 1034 }, { "epoch": 0.14, "grad_norm": 3.5191569691111293, "learning_rate": 9.957391886280201e-06, "loss": 0.3852, "step": 1035 }, { "epoch": 0.14, "grad_norm": 3.6225849167186848, "learning_rate": 9.957248266835358e-06, "loss": 0.3642, "step": 1036 }, { "epoch": 0.14, "grad_norm": 3.0956299663821527, "learning_rate": 9.957104406787335e-06, "loss": 0.3583, "step": 1037 }, { "epoch": 0.14, "grad_norm": 4.27119778392982, "learning_rate": 9.956960306143113e-06, "loss": 0.4252, "step": 1038 }, { "epoch": 0.14, "grad_norm": 4.494643766355487, "learning_rate": 9.956815964909691e-06, "loss": 0.4298, "step": 1039 }, { "epoch": 0.14, "grad_norm": 3.4689970881619105, "learning_rate": 9.95667138309407e-06, "loss": 0.4133, "step": 1040 }, { "epoch": 0.14, "grad_norm": 3.230993587065826, "learning_rate": 9.956526560703268e-06, "loss": 0.3933, "step": 1041 }, { "epoch": 0.14, "grad_norm": 3.2401655615761458, "learning_rate": 9.956381497744317e-06, "loss": 0.3219, "step": 1042 }, { "epoch": 0.14, "grad_norm": 3.255267778345084, "learning_rate": 9.956236194224253e-06, "loss": 0.3572, "step": 1043 }, { "epoch": 0.14, "grad_norm": 3.6003794612943683, "learning_rate": 9.956090650150132e-06, "loss": 0.376, "step": 1044 }, { "epoch": 0.14, "grad_norm": 6.430654911330745, "learning_rate": 9.955944865529019e-06, "loss": 0.354, "step": 1045 }, { "epoch": 0.14, "grad_norm": 3.4126816333649663, "learning_rate": 9.955798840367985e-06, "loss": 0.3567, "step": 1046 }, { "epoch": 0.14, "grad_norm": 3.25139836577505, "learning_rate": 9.955652574674122e-06, "loss": 0.3973, "step": 1047 }, { "epoch": 0.14, "grad_norm": 3.61859306479588, "learning_rate": 9.955506068454527e-06, "loss": 0.3991, "step": 1048 }, { "epoch": 0.14, "grad_norm": 4.011914347909685, "learning_rate": 9.955359321716311e-06, "loss": 0.4378, "step": 1049 }, { "epoch": 0.14, "grad_norm": 3.6780501736091207, "learning_rate": 9.955212334466597e-06, "loss": 0.3618, "step": 1050 }, { "epoch": 0.14, "grad_norm": 3.7558733299085327, "learning_rate": 9.955065106712517e-06, "loss": 0.4007, "step": 1051 }, { "epoch": 0.14, "grad_norm": 4.2539607581446415, "learning_rate": 9.954917638461221e-06, "loss": 0.431, "step": 1052 }, { "epoch": 0.14, "grad_norm": 4.105232796515715, "learning_rate": 9.954769929719863e-06, "loss": 0.3864, "step": 1053 }, { "epoch": 0.14, "grad_norm": 3.7608835923829655, "learning_rate": 9.954621980495614e-06, "loss": 0.4653, "step": 1054 }, { "epoch": 0.14, "grad_norm": 3.84367335071056, "learning_rate": 9.954473790795654e-06, "loss": 0.4203, "step": 1055 }, { "epoch": 0.14, "grad_norm": 3.1777133605164427, "learning_rate": 9.954325360627173e-06, "loss": 0.3758, "step": 1056 }, { "epoch": 0.14, "grad_norm": 3.812402766788447, "learning_rate": 9.954176689997379e-06, "loss": 0.4191, "step": 1057 }, { "epoch": 0.14, "grad_norm": 3.6256116216717134, "learning_rate": 9.954027778913486e-06, "loss": 0.4242, "step": 1058 }, { "epoch": 0.14, "grad_norm": 3.2903142191961767, "learning_rate": 9.953878627382723e-06, "loss": 0.3734, "step": 1059 }, { "epoch": 0.14, "grad_norm": 3.343848579313104, "learning_rate": 9.953729235412327e-06, "loss": 0.3969, "step": 1060 }, { "epoch": 0.14, "grad_norm": 4.009459571994383, "learning_rate": 9.95357960300955e-06, "loss": 0.3615, "step": 1061 }, { "epoch": 0.14, "grad_norm": 3.4574799133615857, "learning_rate": 9.953429730181653e-06, "loss": 0.4044, "step": 1062 }, { "epoch": 0.14, "grad_norm": 3.139434337891451, "learning_rate": 9.953279616935914e-06, "loss": 0.3517, "step": 1063 }, { "epoch": 0.14, "grad_norm": 3.7959780788751156, "learning_rate": 9.953129263279616e-06, "loss": 0.4399, "step": 1064 }, { "epoch": 0.14, "grad_norm": 3.5236698383197624, "learning_rate": 9.952978669220056e-06, "loss": 0.4384, "step": 1065 }, { "epoch": 0.15, "grad_norm": 3.929271901049085, "learning_rate": 9.952827834764545e-06, "loss": 0.4333, "step": 1066 }, { "epoch": 0.15, "grad_norm": 3.75538058990571, "learning_rate": 9.952676759920401e-06, "loss": 0.447, "step": 1067 }, { "epoch": 0.15, "grad_norm": 3.706935746040335, "learning_rate": 9.95252544469496e-06, "loss": 0.381, "step": 1068 }, { "epoch": 0.15, "grad_norm": 3.52602486720364, "learning_rate": 9.952373889095566e-06, "loss": 0.3966, "step": 1069 }, { "epoch": 0.15, "grad_norm": 3.286671180249141, "learning_rate": 9.952222093129571e-06, "loss": 0.3494, "step": 1070 }, { "epoch": 0.15, "grad_norm": 3.461796652449324, "learning_rate": 9.952070056804346e-06, "loss": 0.3906, "step": 1071 }, { "epoch": 0.15, "grad_norm": 3.554996386437928, "learning_rate": 9.951917780127268e-06, "loss": 0.3644, "step": 1072 }, { "epoch": 0.15, "grad_norm": 3.473397151101057, "learning_rate": 9.951765263105731e-06, "loss": 0.35, "step": 1073 }, { "epoch": 0.15, "grad_norm": 3.1009899051286824, "learning_rate": 9.951612505747134e-06, "loss": 0.367, "step": 1074 }, { "epoch": 0.15, "grad_norm": 3.1643083745704015, "learning_rate": 9.951459508058892e-06, "loss": 0.3792, "step": 1075 }, { "epoch": 0.15, "grad_norm": 3.1570436837608975, "learning_rate": 9.951306270048432e-06, "loss": 0.3958, "step": 1076 }, { "epoch": 0.15, "grad_norm": 3.2155956437798774, "learning_rate": 9.951152791723193e-06, "loss": 0.4332, "step": 1077 }, { "epoch": 0.15, "grad_norm": 3.93547773097145, "learning_rate": 9.950999073090621e-06, "loss": 0.45, "step": 1078 }, { "epoch": 0.15, "grad_norm": 3.5908526439787245, "learning_rate": 9.950845114158177e-06, "loss": 0.3575, "step": 1079 }, { "epoch": 0.15, "grad_norm": 3.674515480182691, "learning_rate": 9.950690914933336e-06, "loss": 0.3409, "step": 1080 }, { "epoch": 0.15, "grad_norm": 3.918025718065109, "learning_rate": 9.95053647542358e-06, "loss": 0.3506, "step": 1081 }, { "epoch": 0.15, "grad_norm": 3.6262593009502875, "learning_rate": 9.950381795636406e-06, "loss": 0.3524, "step": 1082 }, { "epoch": 0.15, "grad_norm": 4.241810374382643, "learning_rate": 9.950226875579319e-06, "loss": 0.4456, "step": 1083 }, { "epoch": 0.15, "grad_norm": 4.099083331973089, "learning_rate": 9.95007171525984e-06, "loss": 0.4746, "step": 1084 }, { "epoch": 0.15, "grad_norm": 3.9122307713641646, "learning_rate": 9.949916314685505e-06, "loss": 0.3907, "step": 1085 }, { "epoch": 0.15, "grad_norm": 3.632645388810754, "learning_rate": 9.949760673863846e-06, "loss": 0.3909, "step": 1086 }, { "epoch": 0.15, "grad_norm": 3.526036626177289, "learning_rate": 9.949604792802425e-06, "loss": 0.3303, "step": 1087 }, { "epoch": 0.15, "grad_norm": 3.7009039601983567, "learning_rate": 9.949448671508804e-06, "loss": 0.3764, "step": 1088 }, { "epoch": 0.15, "grad_norm": 4.020670744671599, "learning_rate": 9.949292309990563e-06, "loss": 0.3918, "step": 1089 }, { "epoch": 0.15, "grad_norm": 3.5134631408006443, "learning_rate": 9.949135708255288e-06, "loss": 0.3634, "step": 1090 }, { "epoch": 0.15, "grad_norm": 3.5054925310015754, "learning_rate": 9.948978866310584e-06, "loss": 0.344, "step": 1091 }, { "epoch": 0.15, "grad_norm": 3.9006421167343204, "learning_rate": 9.94882178416406e-06, "loss": 0.4022, "step": 1092 }, { "epoch": 0.15, "grad_norm": 3.913935591574752, "learning_rate": 9.94866446182334e-06, "loss": 0.4127, "step": 1093 }, { "epoch": 0.15, "grad_norm": 3.2056212367027785, "learning_rate": 9.948506899296064e-06, "loss": 0.3485, "step": 1094 }, { "epoch": 0.15, "grad_norm": 4.230556869156749, "learning_rate": 9.948349096589874e-06, "loss": 0.4271, "step": 1095 }, { "epoch": 0.15, "grad_norm": 3.5588245867250956, "learning_rate": 9.94819105371243e-06, "loss": 0.4176, "step": 1096 }, { "epoch": 0.15, "grad_norm": 3.7428661694568355, "learning_rate": 9.948032770671405e-06, "loss": 0.4849, "step": 1097 }, { "epoch": 0.15, "grad_norm": 3.776330472552753, "learning_rate": 9.947874247474483e-06, "loss": 0.3856, "step": 1098 }, { "epoch": 0.15, "grad_norm": 3.8823734161965415, "learning_rate": 9.947715484129352e-06, "loss": 0.4364, "step": 1099 }, { "epoch": 0.15, "grad_norm": 4.0679998887348185, "learning_rate": 9.947556480643724e-06, "loss": 0.3688, "step": 1100 }, { "epoch": 0.15, "grad_norm": 3.809744854514048, "learning_rate": 9.94739723702531e-06, "loss": 0.4038, "step": 1101 }, { "epoch": 0.15, "grad_norm": 3.165115375293726, "learning_rate": 9.947237753281845e-06, "loss": 0.408, "step": 1102 }, { "epoch": 0.15, "grad_norm": 3.271009551758034, "learning_rate": 9.947078029421066e-06, "loss": 0.3906, "step": 1103 }, { "epoch": 0.15, "grad_norm": 3.127837799338739, "learning_rate": 9.946918065450726e-06, "loss": 0.3794, "step": 1104 }, { "epoch": 0.15, "grad_norm": 3.75539924743593, "learning_rate": 9.94675786137859e-06, "loss": 0.3341, "step": 1105 }, { "epoch": 0.15, "grad_norm": 3.233122610852758, "learning_rate": 9.946597417212433e-06, "loss": 0.3531, "step": 1106 }, { "epoch": 0.15, "grad_norm": 3.77023416744597, "learning_rate": 9.946436732960042e-06, "loss": 0.3954, "step": 1107 }, { "epoch": 0.15, "grad_norm": 4.166535044988802, "learning_rate": 9.946275808629215e-06, "loss": 0.5038, "step": 1108 }, { "epoch": 0.15, "grad_norm": 3.5524132552889474, "learning_rate": 9.946114644227765e-06, "loss": 0.3965, "step": 1109 }, { "epoch": 0.15, "grad_norm": 3.763848413734846, "learning_rate": 9.945953239763513e-06, "loss": 0.4023, "step": 1110 }, { "epoch": 0.15, "grad_norm": 3.6589052025254043, "learning_rate": 9.945791595244293e-06, "loss": 0.4283, "step": 1111 }, { "epoch": 0.15, "grad_norm": 4.127144755512334, "learning_rate": 9.945629710677949e-06, "loss": 0.4406, "step": 1112 }, { "epoch": 0.15, "grad_norm": 3.267765552685049, "learning_rate": 9.94546758607234e-06, "loss": 0.3564, "step": 1113 }, { "epoch": 0.15, "grad_norm": 3.8779146531095643, "learning_rate": 9.945305221435336e-06, "loss": 0.4544, "step": 1114 }, { "epoch": 0.15, "grad_norm": 4.155901899508083, "learning_rate": 9.945142616774812e-06, "loss": 0.4144, "step": 1115 }, { "epoch": 0.15, "grad_norm": 3.7298970549980597, "learning_rate": 9.944979772098666e-06, "loss": 0.4047, "step": 1116 }, { "epoch": 0.15, "grad_norm": 3.6417691747167824, "learning_rate": 9.9448166874148e-06, "loss": 0.3985, "step": 1117 }, { "epoch": 0.15, "grad_norm": 3.7800509109469784, "learning_rate": 9.944653362731127e-06, "loss": 0.3748, "step": 1118 }, { "epoch": 0.15, "grad_norm": 3.286099567248507, "learning_rate": 9.944489798055579e-06, "loss": 0.3895, "step": 1119 }, { "epoch": 0.15, "grad_norm": 3.0431403679918363, "learning_rate": 9.944325993396089e-06, "loss": 0.3622, "step": 1120 }, { "epoch": 0.15, "grad_norm": 4.026004672089731, "learning_rate": 9.944161948760612e-06, "loss": 0.4799, "step": 1121 }, { "epoch": 0.15, "grad_norm": 3.722256282950157, "learning_rate": 9.943997664157108e-06, "loss": 0.3671, "step": 1122 }, { "epoch": 0.15, "grad_norm": 3.347174208373489, "learning_rate": 9.94383313959355e-06, "loss": 0.4079, "step": 1123 }, { "epoch": 0.15, "grad_norm": 3.7321533621045333, "learning_rate": 9.943668375077926e-06, "loss": 0.39, "step": 1124 }, { "epoch": 0.15, "grad_norm": 3.1372239239892976, "learning_rate": 9.943503370618228e-06, "loss": 0.3475, "step": 1125 }, { "epoch": 0.15, "grad_norm": 3.4664818762276455, "learning_rate": 9.94333812622247e-06, "loss": 0.4011, "step": 1126 }, { "epoch": 0.15, "grad_norm": 3.8498502667705785, "learning_rate": 9.943172641898669e-06, "loss": 0.429, "step": 1127 }, { "epoch": 0.15, "grad_norm": 4.172065930359833, "learning_rate": 9.943006917654859e-06, "loss": 0.4949, "step": 1128 }, { "epoch": 0.15, "grad_norm": 4.0242213866131085, "learning_rate": 9.94284095349908e-06, "loss": 0.3744, "step": 1129 }, { "epoch": 0.15, "grad_norm": 3.4361906021486437, "learning_rate": 9.942674749439391e-06, "loss": 0.3801, "step": 1130 }, { "epoch": 0.15, "grad_norm": 3.3126510029523444, "learning_rate": 9.942508305483857e-06, "loss": 0.3959, "step": 1131 }, { "epoch": 0.15, "grad_norm": 3.7163518565120177, "learning_rate": 9.942341621640558e-06, "loss": 0.3423, "step": 1132 }, { "epoch": 0.15, "grad_norm": 3.5679456274536068, "learning_rate": 9.942174697917581e-06, "loss": 0.442, "step": 1133 }, { "epoch": 0.15, "grad_norm": 3.758875120783958, "learning_rate": 9.942007534323032e-06, "loss": 0.4547, "step": 1134 }, { "epoch": 0.15, "grad_norm": 3.955889760305799, "learning_rate": 9.94184013086502e-06, "loss": 0.4312, "step": 1135 }, { "epoch": 0.15, "grad_norm": 4.06171163704184, "learning_rate": 9.941672487551672e-06, "loss": 0.4706, "step": 1136 }, { "epoch": 0.15, "grad_norm": 3.708333325152216, "learning_rate": 9.941504604391126e-06, "loss": 0.411, "step": 1137 }, { "epoch": 0.15, "grad_norm": 3.3554970015736463, "learning_rate": 9.941336481391529e-06, "loss": 0.3695, "step": 1138 }, { "epoch": 0.15, "grad_norm": 3.0221303706695872, "learning_rate": 9.94116811856104e-06, "loss": 0.3089, "step": 1139 }, { "epoch": 0.16, "grad_norm": 3.2354302127793613, "learning_rate": 9.940999515907832e-06, "loss": 0.3447, "step": 1140 }, { "epoch": 0.16, "grad_norm": 4.065984287593194, "learning_rate": 9.940830673440089e-06, "loss": 0.3542, "step": 1141 }, { "epoch": 0.16, "grad_norm": 3.7679432379715996, "learning_rate": 9.940661591166003e-06, "loss": 0.3917, "step": 1142 }, { "epoch": 0.16, "grad_norm": 3.8202769674731885, "learning_rate": 9.940492269093784e-06, "loss": 0.4097, "step": 1143 }, { "epoch": 0.16, "grad_norm": 3.859071437068066, "learning_rate": 9.940322707231647e-06, "loss": 0.3419, "step": 1144 }, { "epoch": 0.16, "grad_norm": 4.096745650412949, "learning_rate": 9.940152905587824e-06, "loss": 0.4367, "step": 1145 }, { "epoch": 0.16, "grad_norm": 3.39746317226797, "learning_rate": 9.939982864170554e-06, "loss": 0.3492, "step": 1146 }, { "epoch": 0.16, "grad_norm": 3.712596788737134, "learning_rate": 9.939812582988094e-06, "loss": 0.4636, "step": 1147 }, { "epoch": 0.16, "grad_norm": 3.782005602471775, "learning_rate": 9.939642062048704e-06, "loss": 0.4121, "step": 1148 }, { "epoch": 0.16, "grad_norm": 3.946887068990616, "learning_rate": 9.939471301360664e-06, "loss": 0.3959, "step": 1149 }, { "epoch": 0.16, "grad_norm": 3.587381548081942, "learning_rate": 9.939300300932263e-06, "loss": 0.3335, "step": 1150 }, { "epoch": 0.16, "grad_norm": 3.575671618420908, "learning_rate": 9.939129060771795e-06, "loss": 0.3815, "step": 1151 }, { "epoch": 0.16, "grad_norm": 3.286885552928116, "learning_rate": 9.938957580887575e-06, "loss": 0.4084, "step": 1152 }, { "epoch": 0.16, "grad_norm": 3.2383064283619913, "learning_rate": 9.938785861287927e-06, "loss": 0.349, "step": 1153 }, { "epoch": 0.16, "grad_norm": 3.8476959097021783, "learning_rate": 9.938613901981183e-06, "loss": 0.4, "step": 1154 }, { "epoch": 0.16, "grad_norm": 3.3253761416441794, "learning_rate": 9.938441702975689e-06, "loss": 0.3546, "step": 1155 }, { "epoch": 0.16, "grad_norm": 3.598824058934064, "learning_rate": 9.938269264279805e-06, "loss": 0.362, "step": 1156 }, { "epoch": 0.16, "grad_norm": 3.7712958114358934, "learning_rate": 9.9380965859019e-06, "loss": 0.3893, "step": 1157 }, { "epoch": 0.16, "grad_norm": 3.271642445729462, "learning_rate": 9.937923667850354e-06, "loss": 0.3686, "step": 1158 }, { "epoch": 0.16, "grad_norm": 3.615394312356984, "learning_rate": 9.93775051013356e-06, "loss": 0.3689, "step": 1159 }, { "epoch": 0.16, "grad_norm": 3.07282585480812, "learning_rate": 9.937577112759921e-06, "loss": 0.363, "step": 1160 }, { "epoch": 0.16, "grad_norm": 4.4988755578061905, "learning_rate": 9.937403475737856e-06, "loss": 0.3948, "step": 1161 }, { "epoch": 0.16, "grad_norm": 4.1335194919224, "learning_rate": 9.937229599075791e-06, "loss": 0.4366, "step": 1162 }, { "epoch": 0.16, "grad_norm": 4.033354764844315, "learning_rate": 9.937055482782165e-06, "loss": 0.3298, "step": 1163 }, { "epoch": 0.16, "grad_norm": 3.8267760184757664, "learning_rate": 9.93688112686543e-06, "loss": 0.4145, "step": 1164 }, { "epoch": 0.16, "grad_norm": 3.5276350715342444, "learning_rate": 9.936706531334046e-06, "loss": 0.4077, "step": 1165 }, { "epoch": 0.16, "grad_norm": 3.5664432881431214, "learning_rate": 9.936531696196488e-06, "loss": 0.4155, "step": 1166 }, { "epoch": 0.16, "grad_norm": 3.325537967294773, "learning_rate": 9.936356621461243e-06, "loss": 0.3357, "step": 1167 }, { "epoch": 0.16, "grad_norm": 3.3697654251215607, "learning_rate": 9.936181307136808e-06, "loss": 0.4231, "step": 1168 }, { "epoch": 0.16, "grad_norm": 3.3885738952111626, "learning_rate": 9.936005753231692e-06, "loss": 0.388, "step": 1169 }, { "epoch": 0.16, "grad_norm": 3.505221678227167, "learning_rate": 9.935829959754415e-06, "loss": 0.3918, "step": 1170 }, { "epoch": 0.16, "grad_norm": 3.422829743530292, "learning_rate": 9.93565392671351e-06, "loss": 0.3708, "step": 1171 }, { "epoch": 0.16, "grad_norm": 3.714970739393046, "learning_rate": 9.935477654117518e-06, "loss": 0.4405, "step": 1172 }, { "epoch": 0.16, "grad_norm": 3.512464857953762, "learning_rate": 9.935301141975e-06, "loss": 0.3645, "step": 1173 }, { "epoch": 0.16, "grad_norm": 3.56050127646893, "learning_rate": 9.935124390294516e-06, "loss": 0.396, "step": 1174 }, { "epoch": 0.16, "grad_norm": 4.029878710499042, "learning_rate": 9.934947399084654e-06, "loss": 0.4436, "step": 1175 }, { "epoch": 0.16, "grad_norm": 3.3872466464295385, "learning_rate": 9.934770168353995e-06, "loss": 0.3348, "step": 1176 }, { "epoch": 0.16, "grad_norm": 3.76285084545201, "learning_rate": 9.934592698111148e-06, "loss": 0.389, "step": 1177 }, { "epoch": 0.16, "grad_norm": 3.212844753356167, "learning_rate": 9.934414988364722e-06, "loss": 0.365, "step": 1178 }, { "epoch": 0.16, "grad_norm": 3.44046854056895, "learning_rate": 9.934237039123344e-06, "loss": 0.3795, "step": 1179 }, { "epoch": 0.16, "grad_norm": 3.953661252646465, "learning_rate": 9.934058850395652e-06, "loss": 0.448, "step": 1180 }, { "epoch": 0.16, "grad_norm": 3.9962966651230087, "learning_rate": 9.933880422190293e-06, "loss": 0.3394, "step": 1181 }, { "epoch": 0.16, "grad_norm": 3.6145466390890286, "learning_rate": 9.933701754515928e-06, "loss": 0.3893, "step": 1182 }, { "epoch": 0.16, "grad_norm": 3.7374998937281427, "learning_rate": 9.933522847381228e-06, "loss": 0.3928, "step": 1183 }, { "epoch": 0.16, "grad_norm": 3.8520386057047378, "learning_rate": 9.933343700794875e-06, "loss": 0.4833, "step": 1184 }, { "epoch": 0.16, "grad_norm": 4.205526618427121, "learning_rate": 9.93316431476557e-06, "loss": 0.4628, "step": 1185 }, { "epoch": 0.16, "grad_norm": 3.6509835551551757, "learning_rate": 9.932984689302012e-06, "loss": 0.4, "step": 1186 }, { "epoch": 0.16, "grad_norm": 3.795360957567366, "learning_rate": 9.932804824412922e-06, "loss": 0.3878, "step": 1187 }, { "epoch": 0.16, "grad_norm": 3.7556383095615122, "learning_rate": 9.932624720107031e-06, "loss": 0.4381, "step": 1188 }, { "epoch": 0.16, "grad_norm": 2.989883212437547, "learning_rate": 9.93244437639308e-06, "loss": 0.3229, "step": 1189 }, { "epoch": 0.16, "grad_norm": 3.65599076891545, "learning_rate": 9.932263793279823e-06, "loss": 0.3653, "step": 1190 }, { "epoch": 0.16, "grad_norm": 4.241512409432362, "learning_rate": 9.932082970776023e-06, "loss": 0.3497, "step": 1191 }, { "epoch": 0.16, "grad_norm": 3.5704308638221076, "learning_rate": 9.931901908890457e-06, "loss": 0.3914, "step": 1192 }, { "epoch": 0.16, "grad_norm": 3.3588114579084585, "learning_rate": 9.931720607631912e-06, "loss": 0.3814, "step": 1193 }, { "epoch": 0.16, "grad_norm": 3.9990625084620173, "learning_rate": 9.931539067009191e-06, "loss": 0.4862, "step": 1194 }, { "epoch": 0.16, "grad_norm": 3.5946360168074505, "learning_rate": 9.9313572870311e-06, "loss": 0.4263, "step": 1195 }, { "epoch": 0.16, "grad_norm": 4.373983919756624, "learning_rate": 9.931175267706466e-06, "loss": 0.436, "step": 1196 }, { "epoch": 0.16, "grad_norm": 3.5564304533699267, "learning_rate": 9.930993009044123e-06, "loss": 0.4248, "step": 1197 }, { "epoch": 0.16, "grad_norm": 4.129406798296237, "learning_rate": 9.930810511052913e-06, "loss": 0.4397, "step": 1198 }, { "epoch": 0.16, "grad_norm": 3.5937455647815613, "learning_rate": 9.930627773741699e-06, "loss": 0.3905, "step": 1199 }, { "epoch": 0.16, "grad_norm": 3.9775343829137655, "learning_rate": 9.930444797119348e-06, "loss": 0.4041, "step": 1200 }, { "epoch": 0.16, "grad_norm": 2.9519693589585567, "learning_rate": 9.930261581194739e-06, "loss": 0.3121, "step": 1201 }, { "epoch": 0.16, "grad_norm": 3.4163429702596617, "learning_rate": 9.930078125976767e-06, "loss": 0.3715, "step": 1202 }, { "epoch": 0.16, "grad_norm": 3.67194646447075, "learning_rate": 9.929894431474336e-06, "loss": 0.3651, "step": 1203 }, { "epoch": 0.16, "grad_norm": 3.351365254851614, "learning_rate": 9.929710497696361e-06, "loss": 0.3624, "step": 1204 }, { "epoch": 0.16, "grad_norm": 3.2113466407309725, "learning_rate": 9.929526324651769e-06, "loss": 0.3625, "step": 1205 }, { "epoch": 0.16, "grad_norm": 3.746255344994749, "learning_rate": 9.9293419123495e-06, "loss": 0.4757, "step": 1206 }, { "epoch": 0.16, "grad_norm": 3.1162311829893032, "learning_rate": 9.929157260798504e-06, "loss": 0.3297, "step": 1207 }, { "epoch": 0.16, "grad_norm": 4.327725336921106, "learning_rate": 9.928972370007743e-06, "loss": 0.4012, "step": 1208 }, { "epoch": 0.16, "grad_norm": 3.4572433550523525, "learning_rate": 9.92878723998619e-06, "loss": 0.3845, "step": 1209 }, { "epoch": 0.16, "grad_norm": 3.4130022574885834, "learning_rate": 9.928601870742834e-06, "loss": 0.3852, "step": 1210 }, { "epoch": 0.16, "grad_norm": 3.7867121592612096, "learning_rate": 9.928416262286668e-06, "loss": 0.3463, "step": 1211 }, { "epoch": 0.16, "grad_norm": 3.640171281653871, "learning_rate": 9.9282304146267e-06, "loss": 0.4415, "step": 1212 }, { "epoch": 0.17, "grad_norm": 3.314410307741294, "learning_rate": 9.928044327771955e-06, "loss": 0.4297, "step": 1213 }, { "epoch": 0.17, "grad_norm": 3.2532032968589673, "learning_rate": 9.927858001731461e-06, "loss": 0.3177, "step": 1214 }, { "epoch": 0.17, "grad_norm": 4.346852392769334, "learning_rate": 9.927671436514264e-06, "loss": 0.3681, "step": 1215 }, { "epoch": 0.17, "grad_norm": 3.904039453516431, "learning_rate": 9.927484632129417e-06, "loss": 0.3814, "step": 1216 }, { "epoch": 0.17, "grad_norm": 3.7775365352286947, "learning_rate": 9.927297588585984e-06, "loss": 0.4294, "step": 1217 }, { "epoch": 0.17, "grad_norm": 3.5273555337306512, "learning_rate": 9.927110305893051e-06, "loss": 0.388, "step": 1218 }, { "epoch": 0.17, "grad_norm": 4.057292925888632, "learning_rate": 9.9269227840597e-06, "loss": 0.4509, "step": 1219 }, { "epoch": 0.17, "grad_norm": 3.4122813233573828, "learning_rate": 9.92673502309504e-06, "loss": 0.3909, "step": 1220 }, { "epoch": 0.17, "grad_norm": 4.53290841074288, "learning_rate": 9.926547023008177e-06, "loss": 0.4331, "step": 1221 }, { "epoch": 0.17, "grad_norm": 3.86942081925156, "learning_rate": 9.926358783808238e-06, "loss": 0.3761, "step": 1222 }, { "epoch": 0.17, "grad_norm": 5.0659771481572164, "learning_rate": 9.926170305504362e-06, "loss": 0.3751, "step": 1223 }, { "epoch": 0.17, "grad_norm": 3.9114330518187668, "learning_rate": 9.925981588105695e-06, "loss": 0.419, "step": 1224 }, { "epoch": 0.17, "grad_norm": 3.1547975677131146, "learning_rate": 9.925792631621395e-06, "loss": 0.3897, "step": 1225 }, { "epoch": 0.17, "grad_norm": 3.9178713276402846, "learning_rate": 9.925603436060635e-06, "loss": 0.4506, "step": 1226 }, { "epoch": 0.17, "grad_norm": 3.5038686292964334, "learning_rate": 9.925414001432599e-06, "loss": 0.5023, "step": 1227 }, { "epoch": 0.17, "grad_norm": 3.519563737177596, "learning_rate": 9.925224327746478e-06, "loss": 0.4498, "step": 1228 }, { "epoch": 0.17, "grad_norm": 3.5307118257858106, "learning_rate": 9.925034415011479e-06, "loss": 0.3579, "step": 1229 }, { "epoch": 0.17, "grad_norm": 3.4589172759093674, "learning_rate": 9.92484426323682e-06, "loss": 0.3844, "step": 1230 }, { "epoch": 0.17, "grad_norm": 3.649006258102013, "learning_rate": 9.924653872431733e-06, "loss": 0.3804, "step": 1231 }, { "epoch": 0.17, "grad_norm": 3.3766103351555885, "learning_rate": 9.924463242605454e-06, "loss": 0.3338, "step": 1232 }, { "epoch": 0.17, "grad_norm": 3.587241487655377, "learning_rate": 9.924272373767238e-06, "loss": 0.4144, "step": 1233 }, { "epoch": 0.17, "grad_norm": 3.293076402380575, "learning_rate": 9.924081265926348e-06, "loss": 0.3766, "step": 1234 }, { "epoch": 0.17, "grad_norm": 3.297179753274003, "learning_rate": 9.92388991909206e-06, "loss": 0.3715, "step": 1235 }, { "epoch": 0.17, "grad_norm": 3.907996574139311, "learning_rate": 9.923698333273662e-06, "loss": 0.4088, "step": 1236 }, { "epoch": 0.17, "grad_norm": 3.5705108692676117, "learning_rate": 9.92350650848045e-06, "loss": 0.393, "step": 1237 }, { "epoch": 0.17, "grad_norm": 4.204724924147821, "learning_rate": 9.923314444721735e-06, "loss": 0.3924, "step": 1238 }, { "epoch": 0.17, "grad_norm": 3.629625770815337, "learning_rate": 9.923122142006842e-06, "loss": 0.4204, "step": 1239 }, { "epoch": 0.17, "grad_norm": 3.601069117689726, "learning_rate": 9.922929600345102e-06, "loss": 0.3265, "step": 1240 }, { "epoch": 0.17, "grad_norm": 3.3608794221072142, "learning_rate": 9.922736819745861e-06, "loss": 0.3438, "step": 1241 }, { "epoch": 0.17, "grad_norm": 3.949005939438384, "learning_rate": 9.922543800218474e-06, "loss": 0.4169, "step": 1242 }, { "epoch": 0.17, "grad_norm": 3.6403242021232036, "learning_rate": 9.922350541772312e-06, "loss": 0.3951, "step": 1243 }, { "epoch": 0.17, "grad_norm": 3.059924285805153, "learning_rate": 9.922157044416753e-06, "loss": 0.3898, "step": 1244 }, { "epoch": 0.17, "grad_norm": 3.7795935270505256, "learning_rate": 9.921963308161189e-06, "loss": 0.4091, "step": 1245 }, { "epoch": 0.17, "grad_norm": 3.8235566493798694, "learning_rate": 9.921769333015024e-06, "loss": 0.4255, "step": 1246 }, { "epoch": 0.17, "grad_norm": 3.255196877013241, "learning_rate": 9.921575118987672e-06, "loss": 0.3638, "step": 1247 }, { "epoch": 0.17, "grad_norm": 3.464006692299754, "learning_rate": 9.921380666088558e-06, "loss": 0.3373, "step": 1248 }, { "epoch": 0.17, "grad_norm": 3.8622115565883006, "learning_rate": 9.921185974327122e-06, "loss": 0.3957, "step": 1249 }, { "epoch": 0.17, "grad_norm": 3.492052553837373, "learning_rate": 9.920991043712813e-06, "loss": 0.4244, "step": 1250 }, { "epoch": 0.17, "grad_norm": 3.6191388047913184, "learning_rate": 9.920795874255092e-06, "loss": 0.4413, "step": 1251 }, { "epoch": 0.17, "grad_norm": 3.034237453113676, "learning_rate": 9.92060046596343e-06, "loss": 0.3696, "step": 1252 }, { "epoch": 0.17, "grad_norm": 3.586516330666206, "learning_rate": 9.920404818847313e-06, "loss": 0.4223, "step": 1253 }, { "epoch": 0.17, "grad_norm": 3.709373557483955, "learning_rate": 9.920208932916237e-06, "loss": 0.3567, "step": 1254 }, { "epoch": 0.17, "grad_norm": 3.087666957799985, "learning_rate": 9.920012808179708e-06, "loss": 0.3581, "step": 1255 }, { "epoch": 0.17, "grad_norm": 3.7976762209328303, "learning_rate": 9.919816444647247e-06, "loss": 0.3705, "step": 1256 }, { "epoch": 0.17, "grad_norm": 3.3842672816914052, "learning_rate": 9.919619842328383e-06, "loss": 0.3616, "step": 1257 }, { "epoch": 0.17, "grad_norm": 3.2406210506397914, "learning_rate": 9.91942300123266e-06, "loss": 0.3247, "step": 1258 }, { "epoch": 0.17, "grad_norm": 3.712752981349202, "learning_rate": 9.91922592136963e-06, "loss": 0.4551, "step": 1259 }, { "epoch": 0.17, "grad_norm": 3.73210006088064, "learning_rate": 9.919028602748856e-06, "loss": 0.405, "step": 1260 }, { "epoch": 0.17, "grad_norm": 3.6431145252466783, "learning_rate": 9.918831045379923e-06, "loss": 0.3395, "step": 1261 }, { "epoch": 0.17, "grad_norm": 3.0908569384758615, "learning_rate": 9.918633249272412e-06, "loss": 0.342, "step": 1262 }, { "epoch": 0.17, "grad_norm": 3.603289018994003, "learning_rate": 9.918435214435927e-06, "loss": 0.3357, "step": 1263 }, { "epoch": 0.17, "grad_norm": 3.4345114884322374, "learning_rate": 9.918236940880077e-06, "loss": 0.3865, "step": 1264 }, { "epoch": 0.17, "grad_norm": 3.4694336881659966, "learning_rate": 9.918038428614487e-06, "loss": 0.3518, "step": 1265 }, { "epoch": 0.17, "grad_norm": 3.577997085567747, "learning_rate": 9.917839677648792e-06, "loss": 0.3623, "step": 1266 }, { "epoch": 0.17, "grad_norm": 3.877484757422225, "learning_rate": 9.917640687992638e-06, "loss": 0.4238, "step": 1267 }, { "epoch": 0.17, "grad_norm": 4.079210822408587, "learning_rate": 9.917441459655685e-06, "loss": 0.4265, "step": 1268 }, { "epoch": 0.17, "grad_norm": 3.131564404124772, "learning_rate": 9.917241992647599e-06, "loss": 0.3228, "step": 1269 }, { "epoch": 0.17, "grad_norm": 3.536649618719828, "learning_rate": 9.917042286978064e-06, "loss": 0.3511, "step": 1270 }, { "epoch": 0.17, "grad_norm": 3.7726064646191007, "learning_rate": 9.916842342656772e-06, "loss": 0.3475, "step": 1271 }, { "epoch": 0.17, "grad_norm": 3.507831111072244, "learning_rate": 9.916642159693428e-06, "loss": 0.4143, "step": 1272 }, { "epoch": 0.17, "grad_norm": 4.0810966058908456, "learning_rate": 9.916441738097747e-06, "loss": 0.3918, "step": 1273 }, { "epoch": 0.17, "grad_norm": 3.823724399076645, "learning_rate": 9.916241077879459e-06, "loss": 0.4005, "step": 1274 }, { "epoch": 0.17, "grad_norm": 3.4435593622677736, "learning_rate": 9.916040179048298e-06, "loss": 0.3682, "step": 1275 }, { "epoch": 0.17, "grad_norm": 3.791750573435541, "learning_rate": 9.91583904161402e-06, "loss": 0.3481, "step": 1276 }, { "epoch": 0.17, "grad_norm": 3.338846297103635, "learning_rate": 9.915637665586386e-06, "loss": 0.3549, "step": 1277 }, { "epoch": 0.17, "grad_norm": 4.103333209132796, "learning_rate": 9.915436050975166e-06, "loss": 0.361, "step": 1278 }, { "epoch": 0.17, "grad_norm": 3.95090212011209, "learning_rate": 9.915234197790153e-06, "loss": 0.4204, "step": 1279 }, { "epoch": 0.17, "grad_norm": 3.2666955563528224, "learning_rate": 9.915032106041139e-06, "loss": 0.3566, "step": 1280 }, { "epoch": 0.17, "grad_norm": 3.5498665315773614, "learning_rate": 9.914829775737931e-06, "loss": 0.3733, "step": 1281 }, { "epoch": 0.17, "grad_norm": 3.509364795216015, "learning_rate": 9.914627206890352e-06, "loss": 0.4063, "step": 1282 }, { "epoch": 0.17, "grad_norm": 5.682502836351008, "learning_rate": 9.914424399508236e-06, "loss": 0.4029, "step": 1283 }, { "epoch": 0.17, "grad_norm": 3.3879666138333535, "learning_rate": 9.914221353601424e-06, "loss": 0.3487, "step": 1284 }, { "epoch": 0.17, "grad_norm": 3.7063070306060086, "learning_rate": 9.914018069179769e-06, "loss": 0.377, "step": 1285 }, { "epoch": 0.17, "grad_norm": 4.786265868947307, "learning_rate": 9.91381454625314e-06, "loss": 0.4296, "step": 1286 }, { "epoch": 0.18, "grad_norm": 3.675159332278176, "learning_rate": 9.913610784831415e-06, "loss": 0.3969, "step": 1287 }, { "epoch": 0.18, "grad_norm": 3.6663077253532093, "learning_rate": 9.913406784924482e-06, "loss": 0.3852, "step": 1288 }, { "epoch": 0.18, "grad_norm": 3.4475698292476413, "learning_rate": 9.913202546542245e-06, "loss": 0.3797, "step": 1289 }, { "epoch": 0.18, "grad_norm": 3.445633324954158, "learning_rate": 9.912998069694616e-06, "loss": 0.3552, "step": 1290 }, { "epoch": 0.18, "grad_norm": 3.4784972164534897, "learning_rate": 9.912793354391516e-06, "loss": 0.3863, "step": 1291 }, { "epoch": 0.18, "grad_norm": 25.62588958413826, "learning_rate": 9.912588400642884e-06, "loss": 0.3242, "step": 1292 }, { "epoch": 0.18, "grad_norm": 3.648822378365674, "learning_rate": 9.91238320845867e-06, "loss": 0.3943, "step": 1293 }, { "epoch": 0.18, "grad_norm": 3.623241164417772, "learning_rate": 9.912177777848828e-06, "loss": 0.4038, "step": 1294 }, { "epoch": 0.18, "grad_norm": 3.873672363112047, "learning_rate": 9.911972108823331e-06, "loss": 0.3889, "step": 1295 }, { "epoch": 0.18, "grad_norm": 3.5281552617186156, "learning_rate": 9.911766201392164e-06, "loss": 0.3176, "step": 1296 }, { "epoch": 0.18, "grad_norm": 3.554102150357328, "learning_rate": 9.911560055565316e-06, "loss": 0.3754, "step": 1297 }, { "epoch": 0.18, "grad_norm": 3.6008056860487687, "learning_rate": 9.911353671352796e-06, "loss": 0.3844, "step": 1298 }, { "epoch": 0.18, "grad_norm": 3.8906258490018226, "learning_rate": 9.91114704876462e-06, "loss": 0.4338, "step": 1299 }, { "epoch": 0.18, "grad_norm": 3.9076572138369308, "learning_rate": 9.910940187810814e-06, "loss": 0.3558, "step": 1300 }, { "epoch": 0.18, "grad_norm": 3.210425284966593, "learning_rate": 9.910733088501423e-06, "loss": 0.3739, "step": 1301 }, { "epoch": 0.18, "grad_norm": 3.708991482918067, "learning_rate": 9.910525750846494e-06, "loss": 0.3584, "step": 1302 }, { "epoch": 0.18, "grad_norm": 3.4087148519608004, "learning_rate": 9.910318174856093e-06, "loss": 0.3576, "step": 1303 }, { "epoch": 0.18, "grad_norm": 3.559519593714451, "learning_rate": 9.910110360540293e-06, "loss": 0.4032, "step": 1304 }, { "epoch": 0.18, "grad_norm": 3.5510855528371423, "learning_rate": 9.909902307909184e-06, "loss": 0.4182, "step": 1305 }, { "epoch": 0.18, "grad_norm": 3.6817999506777292, "learning_rate": 9.90969401697286e-06, "loss": 0.3514, "step": 1306 }, { "epoch": 0.18, "grad_norm": 3.6497856215648485, "learning_rate": 9.909485487741432e-06, "loss": 0.3705, "step": 1307 }, { "epoch": 0.18, "grad_norm": 3.488712380809628, "learning_rate": 9.909276720225021e-06, "loss": 0.3656, "step": 1308 }, { "epoch": 0.18, "grad_norm": 3.4464732697567597, "learning_rate": 9.90906771443376e-06, "loss": 0.3588, "step": 1309 }, { "epoch": 0.18, "grad_norm": 3.844743951577511, "learning_rate": 9.908858470377793e-06, "loss": 0.4814, "step": 1310 }, { "epoch": 0.18, "grad_norm": 4.000812509920965, "learning_rate": 9.908648988067277e-06, "loss": 0.374, "step": 1311 }, { "epoch": 0.18, "grad_norm": 3.452673071107478, "learning_rate": 9.908439267512378e-06, "loss": 0.3661, "step": 1312 }, { "epoch": 0.18, "grad_norm": 3.581346615187773, "learning_rate": 9.908229308723274e-06, "loss": 0.4142, "step": 1313 }, { "epoch": 0.18, "grad_norm": 3.5571137111766022, "learning_rate": 9.908019111710157e-06, "loss": 0.4402, "step": 1314 }, { "epoch": 0.18, "grad_norm": 4.108516811779006, "learning_rate": 9.907808676483228e-06, "loss": 0.4182, "step": 1315 }, { "epoch": 0.18, "grad_norm": 3.3024682087691315, "learning_rate": 9.907598003052701e-06, "loss": 0.3371, "step": 1316 }, { "epoch": 0.18, "grad_norm": 3.730353064732616, "learning_rate": 9.907387091428803e-06, "loss": 0.444, "step": 1317 }, { "epoch": 0.18, "grad_norm": 3.8489462609396665, "learning_rate": 9.907175941621768e-06, "loss": 0.4088, "step": 1318 }, { "epoch": 0.18, "grad_norm": 3.8332144853694214, "learning_rate": 9.906964553641846e-06, "loss": 0.3884, "step": 1319 }, { "epoch": 0.18, "grad_norm": 3.673856561666589, "learning_rate": 9.906752927499295e-06, "loss": 0.3711, "step": 1320 }, { "epoch": 0.18, "grad_norm": 3.7180039030929546, "learning_rate": 9.906541063204389e-06, "loss": 0.3573, "step": 1321 }, { "epoch": 0.18, "grad_norm": 3.4131660142972824, "learning_rate": 9.906328960767409e-06, "loss": 0.3123, "step": 1322 }, { "epoch": 0.18, "grad_norm": 3.63164573162617, "learning_rate": 9.906116620198651e-06, "loss": 0.3671, "step": 1323 }, { "epoch": 0.18, "grad_norm": 3.963971834067164, "learning_rate": 9.905904041508421e-06, "loss": 0.3982, "step": 1324 }, { "epoch": 0.18, "grad_norm": 3.666871223324798, "learning_rate": 9.905691224707034e-06, "loss": 0.4536, "step": 1325 }, { "epoch": 0.18, "grad_norm": 3.731946354936295, "learning_rate": 9.90547816980482e-06, "loss": 0.3806, "step": 1326 }, { "epoch": 0.18, "grad_norm": 3.2867775843048324, "learning_rate": 9.905264876812123e-06, "loss": 0.3923, "step": 1327 }, { "epoch": 0.18, "grad_norm": 3.9067603558754533, "learning_rate": 9.905051345739293e-06, "loss": 0.4125, "step": 1328 }, { "epoch": 0.18, "grad_norm": 3.597654150528391, "learning_rate": 9.904837576596694e-06, "loss": 0.3902, "step": 1329 }, { "epoch": 0.18, "grad_norm": 4.3697395523035825, "learning_rate": 9.904623569394702e-06, "loss": 0.3911, "step": 1330 }, { "epoch": 0.18, "grad_norm": 3.2279965193084386, "learning_rate": 9.904409324143702e-06, "loss": 0.3564, "step": 1331 }, { "epoch": 0.18, "grad_norm": 3.187100997239756, "learning_rate": 9.904194840854094e-06, "loss": 0.3136, "step": 1332 }, { "epoch": 0.18, "grad_norm": 3.6603826036737748, "learning_rate": 9.903980119536289e-06, "loss": 0.4626, "step": 1333 }, { "epoch": 0.18, "grad_norm": 3.96121020035925, "learning_rate": 9.903765160200708e-06, "loss": 0.4457, "step": 1334 }, { "epoch": 0.18, "grad_norm": 3.6395042238826814, "learning_rate": 9.903549962857783e-06, "loss": 0.4002, "step": 1335 }, { "epoch": 0.18, "grad_norm": 4.171606459677439, "learning_rate": 9.903334527517959e-06, "loss": 0.4268, "step": 1336 }, { "epoch": 0.18, "grad_norm": 3.596852135070894, "learning_rate": 9.903118854191693e-06, "loss": 0.3561, "step": 1337 }, { "epoch": 0.18, "grad_norm": 3.906114206290507, "learning_rate": 9.902902942889454e-06, "loss": 0.3914, "step": 1338 }, { "epoch": 0.18, "grad_norm": 3.2570940308912832, "learning_rate": 9.90268679362172e-06, "loss": 0.3615, "step": 1339 }, { "epoch": 0.18, "grad_norm": 3.915846640558872, "learning_rate": 9.902470406398981e-06, "loss": 0.4001, "step": 1340 }, { "epoch": 0.18, "grad_norm": 3.3957512050889163, "learning_rate": 9.902253781231741e-06, "loss": 0.3092, "step": 1341 }, { "epoch": 0.18, "grad_norm": 3.3037107675699424, "learning_rate": 9.902036918130514e-06, "loss": 0.3489, "step": 1342 }, { "epoch": 0.18, "grad_norm": 3.5059553908162577, "learning_rate": 9.901819817105824e-06, "loss": 0.3763, "step": 1343 }, { "epoch": 0.18, "grad_norm": 3.9406059089900576, "learning_rate": 9.90160247816821e-06, "loss": 0.3519, "step": 1344 }, { "epoch": 0.18, "grad_norm": 4.062590148630614, "learning_rate": 9.901384901328221e-06, "loss": 0.4401, "step": 1345 }, { "epoch": 0.18, "grad_norm": 3.840307555405805, "learning_rate": 9.901167086596414e-06, "loss": 0.38, "step": 1346 }, { "epoch": 0.18, "grad_norm": 3.5764239102546953, "learning_rate": 9.900949033983366e-06, "loss": 0.3745, "step": 1347 }, { "epoch": 0.18, "grad_norm": 3.7983458830036283, "learning_rate": 9.900730743499656e-06, "loss": 0.3762, "step": 1348 }, { "epoch": 0.18, "grad_norm": 3.568110930792592, "learning_rate": 9.90051221515588e-06, "loss": 0.3743, "step": 1349 }, { "epoch": 0.18, "grad_norm": 3.553678352268365, "learning_rate": 9.900293448962643e-06, "loss": 0.4227, "step": 1350 }, { "epoch": 0.18, "grad_norm": 3.3202835135801454, "learning_rate": 9.900074444930567e-06, "loss": 0.3542, "step": 1351 }, { "epoch": 0.18, "grad_norm": 3.802646356187481, "learning_rate": 9.899855203070278e-06, "loss": 0.421, "step": 1352 }, { "epoch": 0.18, "grad_norm": 3.289080612880311, "learning_rate": 9.899635723392419e-06, "loss": 0.3617, "step": 1353 }, { "epoch": 0.18, "grad_norm": 3.506025928318234, "learning_rate": 9.899416005907639e-06, "loss": 0.3662, "step": 1354 }, { "epoch": 0.18, "grad_norm": 3.9558470829801364, "learning_rate": 9.899196050626608e-06, "loss": 0.4391, "step": 1355 }, { "epoch": 0.18, "grad_norm": 2.8114666464715876, "learning_rate": 9.898975857559996e-06, "loss": 0.3375, "step": 1356 }, { "epoch": 0.18, "grad_norm": 4.360274140817559, "learning_rate": 9.898755426718493e-06, "loss": 0.4084, "step": 1357 }, { "epoch": 0.18, "grad_norm": 3.162598738224981, "learning_rate": 9.898534758112799e-06, "loss": 0.3323, "step": 1358 }, { "epoch": 0.18, "grad_norm": 3.618783711442499, "learning_rate": 9.898313851753623e-06, "loss": 0.4448, "step": 1359 }, { "epoch": 0.18, "grad_norm": 4.497654358084196, "learning_rate": 9.898092707651686e-06, "loss": 0.3883, "step": 1360 }, { "epoch": 0.19, "grad_norm": 3.4873202743226197, "learning_rate": 9.897871325817721e-06, "loss": 0.3726, "step": 1361 }, { "epoch": 0.19, "grad_norm": 3.4273306464071487, "learning_rate": 9.897649706262474e-06, "loss": 0.3386, "step": 1362 }, { "epoch": 0.19, "grad_norm": 3.7740358859455205, "learning_rate": 9.897427848996702e-06, "loss": 0.3678, "step": 1363 }, { "epoch": 0.19, "grad_norm": 3.1355321160131617, "learning_rate": 9.89720575403117e-06, "loss": 0.3244, "step": 1364 }, { "epoch": 0.19, "grad_norm": 3.8160587767859693, "learning_rate": 9.896983421376662e-06, "loss": 0.3898, "step": 1365 }, { "epoch": 0.19, "grad_norm": 3.4222458669643343, "learning_rate": 9.896760851043967e-06, "loss": 0.351, "step": 1366 }, { "epoch": 0.19, "grad_norm": 4.300848141087667, "learning_rate": 9.896538043043887e-06, "loss": 0.4092, "step": 1367 }, { "epoch": 0.19, "grad_norm": 3.3094467894406927, "learning_rate": 9.896314997387235e-06, "loss": 0.3494, "step": 1368 }, { "epoch": 0.19, "grad_norm": 3.7586412538700524, "learning_rate": 9.896091714084842e-06, "loss": 0.357, "step": 1369 }, { "epoch": 0.19, "grad_norm": 3.5693196002833956, "learning_rate": 9.895868193147538e-06, "loss": 0.3842, "step": 1370 }, { "epoch": 0.19, "grad_norm": 3.3787545460371238, "learning_rate": 9.895644434586175e-06, "loss": 0.33, "step": 1371 }, { "epoch": 0.19, "grad_norm": 3.614814240279693, "learning_rate": 9.895420438411616e-06, "loss": 0.3775, "step": 1372 }, { "epoch": 0.19, "grad_norm": 3.465783475349485, "learning_rate": 9.895196204634727e-06, "loss": 0.3802, "step": 1373 }, { "epoch": 0.19, "grad_norm": 4.003748072700043, "learning_rate": 9.894971733266397e-06, "loss": 0.4563, "step": 1374 }, { "epoch": 0.19, "grad_norm": 3.7060080334205114, "learning_rate": 9.894747024317519e-06, "loss": 0.4246, "step": 1375 }, { "epoch": 0.19, "grad_norm": 3.409267902764431, "learning_rate": 9.894522077798995e-06, "loss": 0.3854, "step": 1376 }, { "epoch": 0.19, "grad_norm": 3.6134683392424694, "learning_rate": 9.89429689372175e-06, "loss": 0.3344, "step": 1377 }, { "epoch": 0.19, "grad_norm": 3.252600134452666, "learning_rate": 9.89407147209671e-06, "loss": 0.3233, "step": 1378 }, { "epoch": 0.19, "grad_norm": 3.425274799534015, "learning_rate": 9.893845812934814e-06, "loss": 0.3211, "step": 1379 }, { "epoch": 0.19, "grad_norm": 3.2997288227013377, "learning_rate": 9.89361991624702e-06, "loss": 0.3133, "step": 1380 }, { "epoch": 0.19, "grad_norm": 3.6812868505620235, "learning_rate": 9.893393782044287e-06, "loss": 0.3895, "step": 1381 }, { "epoch": 0.19, "grad_norm": 3.7158947457120255, "learning_rate": 9.893167410337591e-06, "loss": 0.3814, "step": 1382 }, { "epoch": 0.19, "grad_norm": 3.410610257374185, "learning_rate": 9.892940801137923e-06, "loss": 0.3322, "step": 1383 }, { "epoch": 0.19, "grad_norm": 3.1728773519729145, "learning_rate": 9.892713954456276e-06, "loss": 0.3482, "step": 1384 }, { "epoch": 0.19, "grad_norm": 3.9805900346870984, "learning_rate": 9.892486870303665e-06, "loss": 0.3585, "step": 1385 }, { "epoch": 0.19, "grad_norm": 3.6866663928437884, "learning_rate": 9.892259548691109e-06, "loss": 0.3279, "step": 1386 }, { "epoch": 0.19, "grad_norm": 3.732184735838893, "learning_rate": 9.892031989629642e-06, "loss": 0.396, "step": 1387 }, { "epoch": 0.19, "grad_norm": 3.3670261137335453, "learning_rate": 9.891804193130309e-06, "loss": 0.4082, "step": 1388 }, { "epoch": 0.19, "grad_norm": 4.119510336697321, "learning_rate": 9.891576159204165e-06, "loss": 0.4029, "step": 1389 }, { "epoch": 0.19, "grad_norm": 3.7984977866884964, "learning_rate": 9.89134788786228e-06, "loss": 0.3687, "step": 1390 }, { "epoch": 0.19, "grad_norm": 4.137124921809491, "learning_rate": 9.891119379115733e-06, "loss": 0.344, "step": 1391 }, { "epoch": 0.19, "grad_norm": 3.8145720568356114, "learning_rate": 9.890890632975612e-06, "loss": 0.3787, "step": 1392 }, { "epoch": 0.19, "grad_norm": 3.193253380246336, "learning_rate": 9.890661649453022e-06, "loss": 0.2863, "step": 1393 }, { "epoch": 0.19, "grad_norm": 4.219207932850471, "learning_rate": 9.890432428559076e-06, "loss": 0.4036, "step": 1394 }, { "epoch": 0.19, "grad_norm": 4.03696486930997, "learning_rate": 9.890202970304899e-06, "loss": 0.3968, "step": 1395 }, { "epoch": 0.19, "grad_norm": 3.977088054984189, "learning_rate": 9.889973274701627e-06, "loss": 0.4442, "step": 1396 }, { "epoch": 0.19, "grad_norm": 3.277748168534306, "learning_rate": 9.889743341760412e-06, "loss": 0.3391, "step": 1397 }, { "epoch": 0.19, "grad_norm": 3.3390955946549363, "learning_rate": 9.889513171492411e-06, "loss": 0.3894, "step": 1398 }, { "epoch": 0.19, "grad_norm": 3.4306480290256456, "learning_rate": 9.889282763908796e-06, "loss": 0.3444, "step": 1399 }, { "epoch": 0.19, "grad_norm": 3.5684020683709208, "learning_rate": 9.88905211902075e-06, "loss": 0.3792, "step": 1400 }, { "epoch": 0.19, "grad_norm": 3.685409774273693, "learning_rate": 9.888821236839467e-06, "loss": 0.3958, "step": 1401 }, { "epoch": 0.19, "grad_norm": 3.897769548469043, "learning_rate": 9.888590117376154e-06, "loss": 0.4201, "step": 1402 }, { "epoch": 0.19, "grad_norm": 3.5952307779190207, "learning_rate": 9.88835876064203e-06, "loss": 0.3904, "step": 1403 }, { "epoch": 0.19, "grad_norm": 3.388284356780456, "learning_rate": 9.88812716664832e-06, "loss": 0.3542, "step": 1404 }, { "epoch": 0.19, "grad_norm": 3.3383563512591614, "learning_rate": 9.887895335406269e-06, "loss": 0.3512, "step": 1405 }, { "epoch": 0.19, "grad_norm": 3.1137018263560807, "learning_rate": 9.887663266927123e-06, "loss": 0.3805, "step": 1406 }, { "epoch": 0.19, "grad_norm": 3.436446705737145, "learning_rate": 9.887430961222153e-06, "loss": 0.3719, "step": 1407 }, { "epoch": 0.19, "grad_norm": 3.4653983344180848, "learning_rate": 9.887198418302629e-06, "loss": 0.3703, "step": 1408 }, { "epoch": 0.19, "grad_norm": 3.5169501330034434, "learning_rate": 9.88696563817984e-06, "loss": 0.3767, "step": 1409 }, { "epoch": 0.19, "grad_norm": 3.9713753342417246, "learning_rate": 9.886732620865081e-06, "loss": 0.403, "step": 1410 }, { "epoch": 0.19, "grad_norm": 4.837093075048415, "learning_rate": 9.886499366369666e-06, "loss": 0.364, "step": 1411 }, { "epoch": 0.19, "grad_norm": 3.609903927569787, "learning_rate": 9.886265874704914e-06, "loss": 0.3432, "step": 1412 }, { "epoch": 0.19, "grad_norm": 3.4237913257589794, "learning_rate": 9.886032145882158e-06, "loss": 0.3324, "step": 1413 }, { "epoch": 0.19, "grad_norm": 3.886414405987224, "learning_rate": 9.88579817991274e-06, "loss": 0.3807, "step": 1414 }, { "epoch": 0.19, "grad_norm": 3.4359213251790592, "learning_rate": 9.88556397680802e-06, "loss": 0.376, "step": 1415 }, { "epoch": 0.19, "grad_norm": 3.0900943276956383, "learning_rate": 9.88532953657936e-06, "loss": 0.4404, "step": 1416 }, { "epoch": 0.19, "grad_norm": 3.7279801823040035, "learning_rate": 9.885094859238145e-06, "loss": 0.3399, "step": 1417 }, { "epoch": 0.19, "grad_norm": 3.2166714861507217, "learning_rate": 9.884859944795759e-06, "loss": 0.3755, "step": 1418 }, { "epoch": 0.19, "grad_norm": 3.365062655065992, "learning_rate": 9.884624793263607e-06, "loss": 0.3837, "step": 1419 }, { "epoch": 0.19, "grad_norm": 3.888367378344854, "learning_rate": 9.884389404653104e-06, "loss": 0.4068, "step": 1420 }, { "epoch": 0.19, "grad_norm": 3.0085893255757945, "learning_rate": 9.884153778975668e-06, "loss": 0.3131, "step": 1421 }, { "epoch": 0.19, "grad_norm": 3.502223544700426, "learning_rate": 9.883917916242744e-06, "loss": 0.3564, "step": 1422 }, { "epoch": 0.19, "grad_norm": 3.8856368254606224, "learning_rate": 9.883681816465774e-06, "loss": 0.3608, "step": 1423 }, { "epoch": 0.19, "grad_norm": 3.9614017702507907, "learning_rate": 9.883445479656217e-06, "loss": 0.3991, "step": 1424 }, { "epoch": 0.19, "grad_norm": 4.729187051030439, "learning_rate": 9.883208905825547e-06, "loss": 0.3446, "step": 1425 }, { "epoch": 0.19, "grad_norm": 3.6428866253309073, "learning_rate": 9.882972094985243e-06, "loss": 0.3853, "step": 1426 }, { "epoch": 0.19, "grad_norm": 4.3753723925095676, "learning_rate": 9.882735047146803e-06, "loss": 0.3664, "step": 1427 }, { "epoch": 0.19, "grad_norm": 3.79390839410003, "learning_rate": 9.882497762321728e-06, "loss": 0.3788, "step": 1428 }, { "epoch": 0.19, "grad_norm": 3.64533357765429, "learning_rate": 9.882260240521537e-06, "loss": 0.3393, "step": 1429 }, { "epoch": 0.19, "grad_norm": 3.8648339287195372, "learning_rate": 9.882022481757756e-06, "loss": 0.3999, "step": 1430 }, { "epoch": 0.19, "grad_norm": 3.4598198264235034, "learning_rate": 9.881784486041929e-06, "loss": 0.3937, "step": 1431 }, { "epoch": 0.19, "grad_norm": 4.054181841970976, "learning_rate": 9.881546253385603e-06, "loss": 0.3681, "step": 1432 }, { "epoch": 0.19, "grad_norm": 3.4606128336309405, "learning_rate": 9.881307783800344e-06, "loss": 0.3502, "step": 1433 }, { "epoch": 0.2, "grad_norm": 4.482492100345467, "learning_rate": 9.881069077297724e-06, "loss": 0.3719, "step": 1434 }, { "epoch": 0.2, "grad_norm": 3.3301418773001674, "learning_rate": 9.88083013388933e-06, "loss": 0.3762, "step": 1435 }, { "epoch": 0.2, "grad_norm": 4.458395981050994, "learning_rate": 9.880590953586757e-06, "loss": 0.3863, "step": 1436 }, { "epoch": 0.2, "grad_norm": 3.240802646363132, "learning_rate": 9.880351536401617e-06, "loss": 0.3304, "step": 1437 }, { "epoch": 0.2, "grad_norm": 3.788610914515813, "learning_rate": 9.88011188234553e-06, "loss": 0.3668, "step": 1438 }, { "epoch": 0.2, "grad_norm": 3.567162502034271, "learning_rate": 9.879871991430126e-06, "loss": 0.3693, "step": 1439 }, { "epoch": 0.2, "grad_norm": 3.3707692211382843, "learning_rate": 9.879631863667049e-06, "loss": 0.364, "step": 1440 }, { "epoch": 0.2, "grad_norm": 4.011322827590537, "learning_rate": 9.879391499067953e-06, "loss": 0.4052, "step": 1441 }, { "epoch": 0.2, "grad_norm": 3.344956574843724, "learning_rate": 9.879150897644504e-06, "loss": 0.3715, "step": 1442 }, { "epoch": 0.2, "grad_norm": 3.321977937138864, "learning_rate": 9.878910059408382e-06, "loss": 0.383, "step": 1443 }, { "epoch": 0.2, "grad_norm": 3.6136323331403464, "learning_rate": 9.878668984371276e-06, "loss": 0.3779, "step": 1444 }, { "epoch": 0.2, "grad_norm": 3.695022200083563, "learning_rate": 9.878427672544884e-06, "loss": 0.3842, "step": 1445 }, { "epoch": 0.2, "grad_norm": 3.674892006747135, "learning_rate": 9.87818612394092e-06, "loss": 0.3961, "step": 1446 }, { "epoch": 0.2, "grad_norm": 3.5255493212248696, "learning_rate": 9.877944338571108e-06, "loss": 0.3507, "step": 1447 }, { "epoch": 0.2, "grad_norm": 3.491989887290542, "learning_rate": 9.877702316447182e-06, "loss": 0.3832, "step": 1448 }, { "epoch": 0.2, "grad_norm": 3.888103398664497, "learning_rate": 9.87746005758089e-06, "loss": 0.3817, "step": 1449 }, { "epoch": 0.2, "grad_norm": 3.529132321571979, "learning_rate": 9.877217561983991e-06, "loss": 0.4026, "step": 1450 }, { "epoch": 0.2, "grad_norm": 3.684230525980528, "learning_rate": 9.876974829668251e-06, "loss": 0.3653, "step": 1451 }, { "epoch": 0.2, "grad_norm": 3.6704808947937213, "learning_rate": 9.876731860645454e-06, "loss": 0.4216, "step": 1452 }, { "epoch": 0.2, "grad_norm": 3.294521169537819, "learning_rate": 9.876488654927393e-06, "loss": 0.3515, "step": 1453 }, { "epoch": 0.2, "grad_norm": 3.3290893976569818, "learning_rate": 9.87624521252587e-06, "loss": 0.353, "step": 1454 }, { "epoch": 0.2, "grad_norm": 3.6466494892422325, "learning_rate": 9.876001533452703e-06, "loss": 0.391, "step": 1455 }, { "epoch": 0.2, "grad_norm": 3.445633905948606, "learning_rate": 9.875757617719719e-06, "loss": 0.3789, "step": 1456 }, { "epoch": 0.2, "grad_norm": 3.520330087455274, "learning_rate": 9.875513465338754e-06, "loss": 0.3514, "step": 1457 }, { "epoch": 0.2, "grad_norm": 3.3489317867703132, "learning_rate": 9.87526907632166e-06, "loss": 0.3994, "step": 1458 }, { "epoch": 0.2, "grad_norm": 3.431635260855727, "learning_rate": 9.8750244506803e-06, "loss": 0.3444, "step": 1459 }, { "epoch": 0.2, "grad_norm": 3.7332462813770535, "learning_rate": 9.874779588426543e-06, "loss": 0.3855, "step": 1460 }, { "epoch": 0.2, "grad_norm": 3.3850668632861005, "learning_rate": 9.874534489572277e-06, "loss": 0.3886, "step": 1461 }, { "epoch": 0.2, "grad_norm": 3.4441335228080927, "learning_rate": 9.874289154129396e-06, "loss": 0.38, "step": 1462 }, { "epoch": 0.2, "grad_norm": 3.7050944739337046, "learning_rate": 9.874043582109808e-06, "loss": 0.3358, "step": 1463 }, { "epoch": 0.2, "grad_norm": 3.3325524345446276, "learning_rate": 9.873797773525435e-06, "loss": 0.331, "step": 1464 }, { "epoch": 0.2, "grad_norm": 3.183517062952533, "learning_rate": 9.873551728388203e-06, "loss": 0.3019, "step": 1465 }, { "epoch": 0.2, "grad_norm": 3.4749382047702784, "learning_rate": 9.873305446710058e-06, "loss": 0.3967, "step": 1466 }, { "epoch": 0.2, "grad_norm": 3.3694827659466453, "learning_rate": 9.873058928502948e-06, "loss": 0.3218, "step": 1467 }, { "epoch": 0.2, "grad_norm": 3.950604027382564, "learning_rate": 9.872812173778842e-06, "loss": 0.3454, "step": 1468 }, { "epoch": 0.2, "grad_norm": 3.8629786499330416, "learning_rate": 9.872565182549719e-06, "loss": 0.3873, "step": 1469 }, { "epoch": 0.2, "grad_norm": 3.4808045224621353, "learning_rate": 9.87231795482756e-06, "loss": 0.4143, "step": 1470 }, { "epoch": 0.2, "grad_norm": 3.557815536280302, "learning_rate": 9.87207049062437e-06, "loss": 0.4297, "step": 1471 }, { "epoch": 0.2, "grad_norm": 3.6096766125624367, "learning_rate": 9.871822789952155e-06, "loss": 0.3346, "step": 1472 }, { "epoch": 0.2, "grad_norm": 3.8741239785730524, "learning_rate": 9.871574852822942e-06, "loss": 0.4199, "step": 1473 }, { "epoch": 0.2, "grad_norm": 3.4338062095555597, "learning_rate": 9.871326679248762e-06, "loss": 0.4191, "step": 1474 }, { "epoch": 0.2, "grad_norm": 3.8201178381447236, "learning_rate": 9.871078269241661e-06, "loss": 0.4217, "step": 1475 }, { "epoch": 0.2, "grad_norm": 3.4521318980629987, "learning_rate": 9.870829622813697e-06, "loss": 0.3227, "step": 1476 }, { "epoch": 0.2, "grad_norm": 3.5473678517556495, "learning_rate": 9.870580739976936e-06, "loss": 0.3492, "step": 1477 }, { "epoch": 0.2, "grad_norm": 3.420266463101561, "learning_rate": 9.870331620743458e-06, "loss": 0.3247, "step": 1478 }, { "epoch": 0.2, "grad_norm": 4.185101084305181, "learning_rate": 9.870082265125356e-06, "loss": 0.3761, "step": 1479 }, { "epoch": 0.2, "grad_norm": 3.4966974835684748, "learning_rate": 9.869832673134731e-06, "loss": 0.3551, "step": 1480 }, { "epoch": 0.2, "grad_norm": 3.622493634134797, "learning_rate": 9.869582844783699e-06, "loss": 0.3943, "step": 1481 }, { "epoch": 0.2, "grad_norm": 3.5744042794534847, "learning_rate": 9.869332780084383e-06, "loss": 0.3291, "step": 1482 }, { "epoch": 0.2, "grad_norm": 3.4998607596312197, "learning_rate": 9.86908247904892e-06, "loss": 0.3469, "step": 1483 }, { "epoch": 0.2, "grad_norm": 3.2257110320911013, "learning_rate": 9.868831941689461e-06, "loss": 0.3344, "step": 1484 }, { "epoch": 0.2, "grad_norm": 3.885656704515802, "learning_rate": 9.868581168018166e-06, "loss": 0.4399, "step": 1485 }, { "epoch": 0.2, "grad_norm": 3.367939651285891, "learning_rate": 9.868330158047202e-06, "loss": 0.422, "step": 1486 }, { "epoch": 0.2, "grad_norm": 3.77877924545666, "learning_rate": 9.868078911788756e-06, "loss": 0.3928, "step": 1487 }, { "epoch": 0.2, "grad_norm": 3.5187367284782844, "learning_rate": 9.867827429255023e-06, "loss": 0.3883, "step": 1488 }, { "epoch": 0.2, "grad_norm": 3.5460291698025737, "learning_rate": 9.867575710458205e-06, "loss": 0.3197, "step": 1489 }, { "epoch": 0.2, "grad_norm": 3.3227742499007697, "learning_rate": 9.867323755410526e-06, "loss": 0.3503, "step": 1490 }, { "epoch": 0.2, "grad_norm": 3.2097901692265864, "learning_rate": 9.867071564124207e-06, "loss": 0.2899, "step": 1491 }, { "epoch": 0.2, "grad_norm": 3.2045173476774504, "learning_rate": 9.866819136611492e-06, "loss": 0.3634, "step": 1492 }, { "epoch": 0.2, "grad_norm": 3.8610316099515285, "learning_rate": 9.866566472884634e-06, "loss": 0.3584, "step": 1493 }, { "epoch": 0.2, "grad_norm": 3.262303767864237, "learning_rate": 9.866313572955892e-06, "loss": 0.3392, "step": 1494 }, { "epoch": 0.2, "grad_norm": 3.6017634581218414, "learning_rate": 9.866060436837545e-06, "loss": 0.3774, "step": 1495 }, { "epoch": 0.2, "grad_norm": 3.421160202567436, "learning_rate": 9.865807064541878e-06, "loss": 0.362, "step": 1496 }, { "epoch": 0.2, "grad_norm": 3.555687307658706, "learning_rate": 9.865553456081188e-06, "loss": 0.3892, "step": 1497 }, { "epoch": 0.2, "grad_norm": 3.2548778165031242, "learning_rate": 9.865299611467783e-06, "loss": 0.335, "step": 1498 }, { "epoch": 0.2, "grad_norm": 3.8250694862029975, "learning_rate": 9.865045530713985e-06, "loss": 0.37, "step": 1499 }, { "epoch": 0.2, "grad_norm": 3.193302477638719, "learning_rate": 9.864791213832125e-06, "loss": 0.2933, "step": 1500 }, { "epoch": 0.2, "grad_norm": 4.134694527335544, "learning_rate": 9.864536660834548e-06, "loss": 0.3877, "step": 1501 }, { "epoch": 0.2, "grad_norm": 3.517715198881612, "learning_rate": 9.864281871733608e-06, "loss": 0.3816, "step": 1502 }, { "epoch": 0.2, "grad_norm": 15.463412211323611, "learning_rate": 9.86402684654167e-06, "loss": 0.3407, "step": 1503 }, { "epoch": 0.2, "grad_norm": 3.5097122508060745, "learning_rate": 9.863771585271114e-06, "loss": 0.3354, "step": 1504 }, { "epoch": 0.2, "grad_norm": 3.2906821086352473, "learning_rate": 9.863516087934329e-06, "loss": 0.3301, "step": 1505 }, { "epoch": 0.2, "grad_norm": 3.9152552730512267, "learning_rate": 9.863260354543714e-06, "loss": 0.4234, "step": 1506 }, { "epoch": 0.2, "grad_norm": 3.9584286701076277, "learning_rate": 9.863004385111683e-06, "loss": 0.3986, "step": 1507 }, { "epoch": 0.21, "grad_norm": 3.463215769326449, "learning_rate": 9.862748179650656e-06, "loss": 0.3277, "step": 1508 }, { "epoch": 0.21, "grad_norm": 3.5809277097253567, "learning_rate": 9.862491738173074e-06, "loss": 0.3768, "step": 1509 }, { "epoch": 0.21, "grad_norm": 4.041201430859041, "learning_rate": 9.86223506069138e-06, "loss": 0.3841, "step": 1510 }, { "epoch": 0.21, "grad_norm": 3.372584192410698, "learning_rate": 9.861978147218032e-06, "loss": 0.3384, "step": 1511 }, { "epoch": 0.21, "grad_norm": 3.8792598100583926, "learning_rate": 9.8617209977655e-06, "loss": 0.3391, "step": 1512 }, { "epoch": 0.21, "grad_norm": 3.6498717406132655, "learning_rate": 9.861463612346265e-06, "loss": 0.3428, "step": 1513 }, { "epoch": 0.21, "grad_norm": 3.5948829963312043, "learning_rate": 9.861205990972819e-06, "loss": 0.3617, "step": 1514 }, { "epoch": 0.21, "grad_norm": 3.917832554992749, "learning_rate": 9.860948133657666e-06, "loss": 0.346, "step": 1515 }, { "epoch": 0.21, "grad_norm": 4.373529711341608, "learning_rate": 9.86069004041332e-06, "loss": 0.3625, "step": 1516 }, { "epoch": 0.21, "grad_norm": 3.76909670752053, "learning_rate": 9.860431711252312e-06, "loss": 0.4113, "step": 1517 }, { "epoch": 0.21, "grad_norm": 3.5864475730218706, "learning_rate": 9.860173146187175e-06, "loss": 0.3514, "step": 1518 }, { "epoch": 0.21, "grad_norm": 3.6683306675043177, "learning_rate": 9.859914345230462e-06, "loss": 0.3916, "step": 1519 }, { "epoch": 0.21, "grad_norm": 3.4909367960790707, "learning_rate": 9.859655308394733e-06, "loss": 0.397, "step": 1520 }, { "epoch": 0.21, "grad_norm": 3.309075665703501, "learning_rate": 9.859396035692559e-06, "loss": 0.3386, "step": 1521 }, { "epoch": 0.21, "grad_norm": 3.7408012689454657, "learning_rate": 9.859136527136525e-06, "loss": 0.3771, "step": 1522 }, { "epoch": 0.21, "grad_norm": 3.137386610913202, "learning_rate": 9.858876782739229e-06, "loss": 0.36, "step": 1523 }, { "epoch": 0.21, "grad_norm": 3.184377701914103, "learning_rate": 9.858616802513275e-06, "loss": 0.3576, "step": 1524 }, { "epoch": 0.21, "grad_norm": 3.1334119794080966, "learning_rate": 9.858356586471281e-06, "loss": 0.3555, "step": 1525 }, { "epoch": 0.21, "grad_norm": 3.168301153264034, "learning_rate": 9.858096134625879e-06, "loss": 0.3249, "step": 1526 }, { "epoch": 0.21, "grad_norm": 3.004764595453137, "learning_rate": 9.857835446989708e-06, "loss": 0.3053, "step": 1527 }, { "epoch": 0.21, "grad_norm": 3.080663450709156, "learning_rate": 9.857574523575421e-06, "loss": 0.3054, "step": 1528 }, { "epoch": 0.21, "grad_norm": 4.1836587812116655, "learning_rate": 9.857313364395685e-06, "loss": 0.4236, "step": 1529 }, { "epoch": 0.21, "grad_norm": 3.3247383623901445, "learning_rate": 9.857051969463171e-06, "loss": 0.3365, "step": 1530 }, { "epoch": 0.21, "grad_norm": 3.767945071980257, "learning_rate": 9.856790338790568e-06, "loss": 0.3853, "step": 1531 }, { "epoch": 0.21, "grad_norm": 3.7244594678210374, "learning_rate": 9.856528472390576e-06, "loss": 0.376, "step": 1532 }, { "epoch": 0.21, "grad_norm": 4.076962440556855, "learning_rate": 9.856266370275902e-06, "loss": 0.4028, "step": 1533 }, { "epoch": 0.21, "grad_norm": 3.5895231890113277, "learning_rate": 9.85600403245927e-06, "loss": 0.3542, "step": 1534 }, { "epoch": 0.21, "grad_norm": 4.433690886207096, "learning_rate": 9.85574145895341e-06, "loss": 0.4411, "step": 1535 }, { "epoch": 0.21, "grad_norm": 4.349304975799615, "learning_rate": 9.855478649771067e-06, "loss": 0.4206, "step": 1536 }, { "epoch": 0.21, "grad_norm": 3.5312236689440595, "learning_rate": 9.855215604925e-06, "loss": 0.3832, "step": 1537 }, { "epoch": 0.21, "grad_norm": 3.351352344735524, "learning_rate": 9.85495232442797e-06, "loss": 0.3546, "step": 1538 }, { "epoch": 0.21, "grad_norm": 3.8066029782671063, "learning_rate": 9.85468880829276e-06, "loss": 0.3723, "step": 1539 }, { "epoch": 0.21, "grad_norm": 3.2977168932457404, "learning_rate": 9.85442505653216e-06, "loss": 0.3643, "step": 1540 }, { "epoch": 0.21, "grad_norm": 3.6395986930445727, "learning_rate": 9.854161069158967e-06, "loss": 0.397, "step": 1541 }, { "epoch": 0.21, "grad_norm": 3.2272344777219244, "learning_rate": 9.853896846186e-06, "loss": 0.3514, "step": 1542 }, { "epoch": 0.21, "grad_norm": 3.372653677601534, "learning_rate": 9.853632387626077e-06, "loss": 0.3986, "step": 1543 }, { "epoch": 0.21, "grad_norm": 3.187181699506957, "learning_rate": 9.85336769349204e-06, "loss": 0.4272, "step": 1544 }, { "epoch": 0.21, "grad_norm": 3.1053123262323177, "learning_rate": 9.85310276379673e-06, "loss": 0.3113, "step": 1545 }, { "epoch": 0.21, "grad_norm": 3.3780550875772453, "learning_rate": 9.85283759855301e-06, "loss": 0.3571, "step": 1546 }, { "epoch": 0.21, "grad_norm": 3.5015141341702076, "learning_rate": 9.852572197773746e-06, "loss": 0.3363, "step": 1547 }, { "epoch": 0.21, "grad_norm": 3.5616833772806293, "learning_rate": 9.852306561471824e-06, "loss": 0.4104, "step": 1548 }, { "epoch": 0.21, "grad_norm": 3.351301979375071, "learning_rate": 9.852040689660134e-06, "loss": 0.3664, "step": 1549 }, { "epoch": 0.21, "grad_norm": 4.162099051565045, "learning_rate": 9.85177458235158e-06, "loss": 0.3768, "step": 1550 }, { "epoch": 0.21, "grad_norm": 3.44369625610931, "learning_rate": 9.85150823955908e-06, "loss": 0.3711, "step": 1551 }, { "epoch": 0.21, "grad_norm": 3.6191207968385926, "learning_rate": 9.851241661295558e-06, "loss": 0.3299, "step": 1552 }, { "epoch": 0.21, "grad_norm": 3.3639157048163972, "learning_rate": 9.850974847573954e-06, "loss": 0.3468, "step": 1553 }, { "epoch": 0.21, "grad_norm": 3.895727460949913, "learning_rate": 9.850707798407218e-06, "loss": 0.357, "step": 1554 }, { "epoch": 0.21, "grad_norm": 3.4926895151817035, "learning_rate": 9.850440513808313e-06, "loss": 0.4294, "step": 1555 }, { "epoch": 0.21, "grad_norm": 3.322042967576716, "learning_rate": 9.850172993790209e-06, "loss": 0.3875, "step": 1556 }, { "epoch": 0.21, "grad_norm": 3.3834769413253727, "learning_rate": 9.84990523836589e-06, "loss": 0.377, "step": 1557 }, { "epoch": 0.21, "grad_norm": 3.109082985780763, "learning_rate": 9.849637247548356e-06, "loss": 0.3628, "step": 1558 }, { "epoch": 0.21, "grad_norm": 3.4227638274098666, "learning_rate": 9.849369021350612e-06, "loss": 0.3938, "step": 1559 }, { "epoch": 0.21, "grad_norm": 3.1303051901714896, "learning_rate": 9.849100559785674e-06, "loss": 0.3293, "step": 1560 }, { "epoch": 0.21, "grad_norm": 3.4023016393889063, "learning_rate": 9.848831862866574e-06, "loss": 0.3658, "step": 1561 }, { "epoch": 0.21, "grad_norm": 3.829923572350469, "learning_rate": 9.848562930606353e-06, "loss": 0.3898, "step": 1562 }, { "epoch": 0.21, "grad_norm": 3.550721256805146, "learning_rate": 9.848293763018064e-06, "loss": 0.3252, "step": 1563 }, { "epoch": 0.21, "grad_norm": 4.0152343347836075, "learning_rate": 9.848024360114772e-06, "loss": 0.4086, "step": 1564 }, { "epoch": 0.21, "grad_norm": 3.566002786376352, "learning_rate": 9.84775472190955e-06, "loss": 0.4523, "step": 1565 }, { "epoch": 0.21, "grad_norm": 3.8561467325201253, "learning_rate": 9.84748484841549e-06, "loss": 0.417, "step": 1566 }, { "epoch": 0.21, "grad_norm": 3.407607581178228, "learning_rate": 9.847214739645684e-06, "loss": 0.3127, "step": 1567 }, { "epoch": 0.21, "grad_norm": 3.0799365557600313, "learning_rate": 9.846944395613248e-06, "loss": 0.3026, "step": 1568 }, { "epoch": 0.21, "grad_norm": 3.9953607717257045, "learning_rate": 9.846673816331297e-06, "loss": 0.4019, "step": 1569 }, { "epoch": 0.21, "grad_norm": 3.4783273830587293, "learning_rate": 9.846403001812972e-06, "loss": 0.3168, "step": 1570 }, { "epoch": 0.21, "grad_norm": 3.5622954731335774, "learning_rate": 9.84613195207141e-06, "loss": 0.3854, "step": 1571 }, { "epoch": 0.21, "grad_norm": 3.807256995006966, "learning_rate": 9.845860667119769e-06, "loss": 0.4242, "step": 1572 }, { "epoch": 0.21, "grad_norm": 3.8902630173945045, "learning_rate": 9.845589146971217e-06, "loss": 0.3881, "step": 1573 }, { "epoch": 0.21, "grad_norm": 3.4478832070921306, "learning_rate": 9.84531739163893e-06, "loss": 0.3761, "step": 1574 }, { "epoch": 0.21, "grad_norm": 3.738102494456993, "learning_rate": 9.8450454011361e-06, "loss": 0.419, "step": 1575 }, { "epoch": 0.21, "grad_norm": 3.6962806561296437, "learning_rate": 9.844773175475928e-06, "loss": 0.3449, "step": 1576 }, { "epoch": 0.21, "grad_norm": 3.248701704052506, "learning_rate": 9.844500714671625e-06, "loss": 0.3663, "step": 1577 }, { "epoch": 0.21, "grad_norm": 3.2710264458375984, "learning_rate": 9.844228018736418e-06, "loss": 0.3286, "step": 1578 }, { "epoch": 0.21, "grad_norm": 4.458500352606233, "learning_rate": 9.843955087683539e-06, "loss": 0.3523, "step": 1579 }, { "epoch": 0.21, "grad_norm": 3.8634263826611144, "learning_rate": 9.843681921526236e-06, "loss": 0.4374, "step": 1580 }, { "epoch": 0.22, "grad_norm": 3.7444130432300136, "learning_rate": 9.843408520277768e-06, "loss": 0.3425, "step": 1581 }, { "epoch": 0.22, "grad_norm": 2.9340955180902504, "learning_rate": 9.843134883951405e-06, "loss": 0.3066, "step": 1582 }, { "epoch": 0.22, "grad_norm": 3.692083400702191, "learning_rate": 9.84286101256043e-06, "loss": 0.4027, "step": 1583 }, { "epoch": 0.22, "grad_norm": 3.702549582474084, "learning_rate": 9.842586906118132e-06, "loss": 0.3904, "step": 1584 }, { "epoch": 0.22, "grad_norm": 3.4590673508084926, "learning_rate": 9.842312564637814e-06, "loss": 0.3596, "step": 1585 }, { "epoch": 0.22, "grad_norm": 3.572327855590053, "learning_rate": 9.842037988132796e-06, "loss": 0.3432, "step": 1586 }, { "epoch": 0.22, "grad_norm": 3.26872035586442, "learning_rate": 9.8417631766164e-06, "loss": 0.3973, "step": 1587 }, { "epoch": 0.22, "grad_norm": 3.2638465172721567, "learning_rate": 9.841488130101968e-06, "loss": 0.3146, "step": 1588 }, { "epoch": 0.22, "grad_norm": 3.4354830087758383, "learning_rate": 9.841212848602848e-06, "loss": 0.3768, "step": 1589 }, { "epoch": 0.22, "grad_norm": 3.8934146506509735, "learning_rate": 9.840937332132398e-06, "loss": 0.311, "step": 1590 }, { "epoch": 0.22, "grad_norm": 4.218304549457495, "learning_rate": 9.840661580703997e-06, "loss": 0.3315, "step": 1591 }, { "epoch": 0.22, "grad_norm": 4.077997075148733, "learning_rate": 9.840385594331022e-06, "loss": 0.3733, "step": 1592 }, { "epoch": 0.22, "grad_norm": 3.670522709221145, "learning_rate": 9.840109373026873e-06, "loss": 0.4053, "step": 1593 }, { "epoch": 0.22, "grad_norm": 3.148327505033031, "learning_rate": 9.839832916804955e-06, "loss": 0.3077, "step": 1594 }, { "epoch": 0.22, "grad_norm": 3.575206724932285, "learning_rate": 9.839556225678684e-06, "loss": 0.3497, "step": 1595 }, { "epoch": 0.22, "grad_norm": 3.666702478719289, "learning_rate": 9.839279299661492e-06, "loss": 0.378, "step": 1596 }, { "epoch": 0.22, "grad_norm": 3.044218412744507, "learning_rate": 9.839002138766818e-06, "loss": 0.2544, "step": 1597 }, { "epoch": 0.22, "grad_norm": 3.7899494612000177, "learning_rate": 9.838724743008116e-06, "loss": 0.4263, "step": 1598 }, { "epoch": 0.22, "grad_norm": 3.5799464820786584, "learning_rate": 9.838447112398849e-06, "loss": 0.3878, "step": 1599 }, { "epoch": 0.22, "grad_norm": 3.7941983477795915, "learning_rate": 9.838169246952492e-06, "loss": 0.4053, "step": 1600 }, { "epoch": 0.22, "grad_norm": 3.308642245345022, "learning_rate": 9.83789114668253e-06, "loss": 0.341, "step": 1601 }, { "epoch": 0.22, "grad_norm": 3.265867286260555, "learning_rate": 9.837612811602462e-06, "loss": 0.3314, "step": 1602 }, { "epoch": 0.22, "grad_norm": 3.433497248505661, "learning_rate": 9.837334241725796e-06, "loss": 0.3831, "step": 1603 }, { "epoch": 0.22, "grad_norm": 3.1845805131849847, "learning_rate": 9.837055437066054e-06, "loss": 0.3089, "step": 1604 }, { "epoch": 0.22, "grad_norm": 3.6148316179695965, "learning_rate": 9.836776397636769e-06, "loss": 0.3932, "step": 1605 }, { "epoch": 0.22, "grad_norm": 3.499724309377385, "learning_rate": 9.836497123451481e-06, "loss": 0.3661, "step": 1606 }, { "epoch": 0.22, "grad_norm": 4.175212610626864, "learning_rate": 9.836217614523747e-06, "loss": 0.3258, "step": 1607 }, { "epoch": 0.22, "grad_norm": 3.690143615742273, "learning_rate": 9.835937870867133e-06, "loss": 0.4758, "step": 1608 }, { "epoch": 0.22, "grad_norm": 3.21819634800155, "learning_rate": 9.835657892495217e-06, "loss": 0.3238, "step": 1609 }, { "epoch": 0.22, "grad_norm": 3.3282791317686637, "learning_rate": 9.835377679421584e-06, "loss": 0.3579, "step": 1610 }, { "epoch": 0.22, "grad_norm": 3.806565424321965, "learning_rate": 9.83509723165984e-06, "loss": 0.3416, "step": 1611 }, { "epoch": 0.22, "grad_norm": 3.205042567089969, "learning_rate": 9.834816549223595e-06, "loss": 0.3865, "step": 1612 }, { "epoch": 0.22, "grad_norm": 3.0846977749852393, "learning_rate": 9.83453563212647e-06, "loss": 0.3519, "step": 1613 }, { "epoch": 0.22, "grad_norm": 3.8612009730987764, "learning_rate": 9.834254480382101e-06, "loss": 0.3804, "step": 1614 }, { "epoch": 0.22, "grad_norm": 3.5423552688459248, "learning_rate": 9.833973094004134e-06, "loss": 0.3722, "step": 1615 }, { "epoch": 0.22, "grad_norm": 3.131109768311779, "learning_rate": 9.833691473006224e-06, "loss": 0.3419, "step": 1616 }, { "epoch": 0.22, "grad_norm": 3.741359407116507, "learning_rate": 9.833409617402044e-06, "loss": 0.4255, "step": 1617 }, { "epoch": 0.22, "grad_norm": 3.7058567704902754, "learning_rate": 9.833127527205271e-06, "loss": 0.4168, "step": 1618 }, { "epoch": 0.22, "grad_norm": 3.0553161127064197, "learning_rate": 9.832845202429597e-06, "loss": 0.2993, "step": 1619 }, { "epoch": 0.22, "grad_norm": 3.6620994053441676, "learning_rate": 9.832562643088724e-06, "loss": 0.331, "step": 1620 }, { "epoch": 0.22, "grad_norm": 3.6478595752626815, "learning_rate": 9.83227984919637e-06, "loss": 0.3883, "step": 1621 }, { "epoch": 0.22, "grad_norm": 3.2010523016789376, "learning_rate": 9.831996820766255e-06, "loss": 0.3294, "step": 1622 }, { "epoch": 0.22, "grad_norm": 3.5506151502469705, "learning_rate": 9.831713557812118e-06, "loss": 0.3111, "step": 1623 }, { "epoch": 0.22, "grad_norm": 3.7854319455898335, "learning_rate": 9.831430060347709e-06, "loss": 0.3805, "step": 1624 }, { "epoch": 0.22, "grad_norm": 3.4985377161717297, "learning_rate": 9.831146328386785e-06, "loss": 0.2979, "step": 1625 }, { "epoch": 0.22, "grad_norm": 3.9320825546989653, "learning_rate": 9.830862361943122e-06, "loss": 0.3154, "step": 1626 }, { "epoch": 0.22, "grad_norm": 4.377571537457866, "learning_rate": 9.830578161030498e-06, "loss": 0.3585, "step": 1627 }, { "epoch": 0.22, "grad_norm": 3.485571360508608, "learning_rate": 9.830293725662706e-06, "loss": 0.3592, "step": 1628 }, { "epoch": 0.22, "grad_norm": 4.027441573845485, "learning_rate": 9.830009055853556e-06, "loss": 0.3318, "step": 1629 }, { "epoch": 0.22, "grad_norm": 3.2172239562140375, "learning_rate": 9.82972415161686e-06, "loss": 0.3549, "step": 1630 }, { "epoch": 0.22, "grad_norm": 3.595532081583517, "learning_rate": 9.82943901296645e-06, "loss": 0.3446, "step": 1631 }, { "epoch": 0.22, "grad_norm": 3.2097686811203303, "learning_rate": 9.829153639916162e-06, "loss": 0.3461, "step": 1632 }, { "epoch": 0.22, "grad_norm": 5.211814838368004, "learning_rate": 9.828868032479849e-06, "loss": 0.4103, "step": 1633 }, { "epoch": 0.22, "grad_norm": 3.271535146660839, "learning_rate": 9.828582190671372e-06, "loss": 0.3207, "step": 1634 }, { "epoch": 0.22, "grad_norm": 3.054400861513127, "learning_rate": 9.828296114504605e-06, "loss": 0.322, "step": 1635 }, { "epoch": 0.22, "grad_norm": 3.371836628821976, "learning_rate": 9.828009803993432e-06, "loss": 0.2983, "step": 1636 }, { "epoch": 0.22, "grad_norm": 2.9211664125407655, "learning_rate": 9.827723259151752e-06, "loss": 0.3161, "step": 1637 }, { "epoch": 0.22, "grad_norm": 3.762528453569893, "learning_rate": 9.827436479993468e-06, "loss": 0.4155, "step": 1638 }, { "epoch": 0.22, "grad_norm": 3.735299032180317, "learning_rate": 9.827149466532503e-06, "loss": 0.3355, "step": 1639 }, { "epoch": 0.22, "grad_norm": 4.1121169087591865, "learning_rate": 9.826862218782786e-06, "loss": 0.343, "step": 1640 }, { "epoch": 0.22, "grad_norm": 3.2156212777297997, "learning_rate": 9.826574736758259e-06, "loss": 0.3145, "step": 1641 }, { "epoch": 0.22, "grad_norm": 3.4371940882061964, "learning_rate": 9.826287020472873e-06, "loss": 0.388, "step": 1642 }, { "epoch": 0.22, "grad_norm": 3.752966541883302, "learning_rate": 9.825999069940597e-06, "loss": 0.4087, "step": 1643 }, { "epoch": 0.22, "grad_norm": 3.6643718058862054, "learning_rate": 9.825710885175402e-06, "loss": 0.4141, "step": 1644 }, { "epoch": 0.22, "grad_norm": 3.6110475945975513, "learning_rate": 9.825422466191279e-06, "loss": 0.3892, "step": 1645 }, { "epoch": 0.22, "grad_norm": 3.473448725599827, "learning_rate": 9.825133813002225e-06, "loss": 0.3684, "step": 1646 }, { "epoch": 0.22, "grad_norm": 3.3036408822924197, "learning_rate": 9.82484492562225e-06, "loss": 0.3502, "step": 1647 }, { "epoch": 0.22, "grad_norm": 3.175536654257202, "learning_rate": 9.824555804065374e-06, "loss": 0.3417, "step": 1648 }, { "epoch": 0.22, "grad_norm": 4.688628917633621, "learning_rate": 9.824266448345632e-06, "loss": 0.4396, "step": 1649 }, { "epoch": 0.22, "grad_norm": 3.8946347040056284, "learning_rate": 9.823976858477066e-06, "loss": 0.3221, "step": 1650 }, { "epoch": 0.22, "grad_norm": 3.486426020668412, "learning_rate": 9.823687034473734e-06, "loss": 0.4105, "step": 1651 }, { "epoch": 0.22, "grad_norm": 4.542278841996208, "learning_rate": 9.823396976349702e-06, "loss": 0.352, "step": 1652 }, { "epoch": 0.22, "grad_norm": 3.960474592153223, "learning_rate": 9.823106684119046e-06, "loss": 0.4101, "step": 1653 }, { "epoch": 0.22, "grad_norm": 3.385051411285094, "learning_rate": 9.822816157795857e-06, "loss": 0.4156, "step": 1654 }, { "epoch": 0.23, "grad_norm": 3.08633333330911, "learning_rate": 9.822525397394236e-06, "loss": 0.4073, "step": 1655 }, { "epoch": 0.23, "grad_norm": 3.3158590936854853, "learning_rate": 9.822234402928295e-06, "loss": 0.364, "step": 1656 }, { "epoch": 0.23, "grad_norm": 3.0817511859155977, "learning_rate": 9.821943174412159e-06, "loss": 0.3831, "step": 1657 }, { "epoch": 0.23, "grad_norm": 3.483241388917323, "learning_rate": 9.82165171185996e-06, "loss": 0.4162, "step": 1658 }, { "epoch": 0.23, "grad_norm": 3.5480496061787976, "learning_rate": 9.821360015285847e-06, "loss": 0.3258, "step": 1659 }, { "epoch": 0.23, "grad_norm": 3.1036266335248746, "learning_rate": 9.821068084703976e-06, "loss": 0.3483, "step": 1660 }, { "epoch": 0.23, "grad_norm": 3.754161093685211, "learning_rate": 9.820775920128519e-06, "loss": 0.383, "step": 1661 }, { "epoch": 0.23, "grad_norm": 3.1472561368260688, "learning_rate": 9.82048352157365e-06, "loss": 0.3236, "step": 1662 }, { "epoch": 0.23, "grad_norm": 4.058715642372341, "learning_rate": 9.820190889053566e-06, "loss": 0.3546, "step": 1663 }, { "epoch": 0.23, "grad_norm": 3.484495306262926, "learning_rate": 9.819898022582471e-06, "loss": 0.3477, "step": 1664 }, { "epoch": 0.23, "grad_norm": 4.194856653536063, "learning_rate": 9.819604922174578e-06, "loss": 0.3777, "step": 1665 }, { "epoch": 0.23, "grad_norm": 4.122410202502509, "learning_rate": 9.819311587844109e-06, "loss": 0.3537, "step": 1666 }, { "epoch": 0.23, "grad_norm": 3.343160029245745, "learning_rate": 9.819018019605306e-06, "loss": 0.3752, "step": 1667 }, { "epoch": 0.23, "grad_norm": 3.5972249515472563, "learning_rate": 9.818724217472418e-06, "loss": 0.3631, "step": 1668 }, { "epoch": 0.23, "grad_norm": 3.424233881706612, "learning_rate": 9.8184301814597e-06, "loss": 0.3486, "step": 1669 }, { "epoch": 0.23, "grad_norm": 3.6988480564419195, "learning_rate": 9.818135911581426e-06, "loss": 0.3247, "step": 1670 }, { "epoch": 0.23, "grad_norm": 3.3489732651752866, "learning_rate": 9.81784140785188e-06, "loss": 0.4261, "step": 1671 }, { "epoch": 0.23, "grad_norm": 3.168647493579711, "learning_rate": 9.817546670285353e-06, "loss": 0.3525, "step": 1672 }, { "epoch": 0.23, "grad_norm": 3.5265401456750394, "learning_rate": 9.817251698896152e-06, "loss": 0.4165, "step": 1673 }, { "epoch": 0.23, "grad_norm": 3.474367997101576, "learning_rate": 9.816956493698595e-06, "loss": 0.369, "step": 1674 }, { "epoch": 0.23, "grad_norm": 3.3297669886946357, "learning_rate": 9.816661054707007e-06, "loss": 0.3949, "step": 1675 }, { "epoch": 0.23, "grad_norm": 3.279354567182001, "learning_rate": 9.81636538193573e-06, "loss": 0.3863, "step": 1676 }, { "epoch": 0.23, "grad_norm": 3.751801136570614, "learning_rate": 9.816069475399113e-06, "loss": 0.3909, "step": 1677 }, { "epoch": 0.23, "grad_norm": 3.0680405301429543, "learning_rate": 9.815773335111517e-06, "loss": 0.347, "step": 1678 }, { "epoch": 0.23, "grad_norm": 3.2736002224038097, "learning_rate": 9.815476961087318e-06, "loss": 0.3, "step": 1679 }, { "epoch": 0.23, "grad_norm": 3.5048416779298432, "learning_rate": 9.8151803533409e-06, "loss": 0.3535, "step": 1680 }, { "epoch": 0.23, "grad_norm": 3.40789633173699, "learning_rate": 9.814883511886655e-06, "loss": 0.3593, "step": 1681 }, { "epoch": 0.23, "grad_norm": 3.677052236078591, "learning_rate": 9.814586436738998e-06, "loss": 0.3801, "step": 1682 }, { "epoch": 0.23, "grad_norm": 3.4883753490979186, "learning_rate": 9.814289127912343e-06, "loss": 0.3776, "step": 1683 }, { "epoch": 0.23, "grad_norm": 3.468800477533794, "learning_rate": 9.813991585421118e-06, "loss": 0.352, "step": 1684 }, { "epoch": 0.23, "grad_norm": 3.095190294335378, "learning_rate": 9.813693809279769e-06, "loss": 0.3908, "step": 1685 }, { "epoch": 0.23, "grad_norm": 3.4110153236972294, "learning_rate": 9.813395799502747e-06, "loss": 0.3669, "step": 1686 }, { "epoch": 0.23, "grad_norm": 3.371467609424982, "learning_rate": 9.813097556104514e-06, "loss": 0.3606, "step": 1687 }, { "epoch": 0.23, "grad_norm": 3.431935349670912, "learning_rate": 9.81279907909955e-06, "loss": 0.3654, "step": 1688 }, { "epoch": 0.23, "grad_norm": 3.32507442692018, "learning_rate": 9.812500368502337e-06, "loss": 0.339, "step": 1689 }, { "epoch": 0.23, "grad_norm": 3.086008386984285, "learning_rate": 9.812201424327377e-06, "loss": 0.3048, "step": 1690 }, { "epoch": 0.23, "grad_norm": 3.5733845441049583, "learning_rate": 9.811902246589176e-06, "loss": 0.3353, "step": 1691 }, { "epoch": 0.23, "grad_norm": 3.2418770851873346, "learning_rate": 9.811602835302257e-06, "loss": 0.2935, "step": 1692 }, { "epoch": 0.23, "grad_norm": 3.783963941014355, "learning_rate": 9.811303190481152e-06, "loss": 0.3643, "step": 1693 }, { "epoch": 0.23, "grad_norm": 3.965014640175283, "learning_rate": 9.811003312140404e-06, "loss": 0.3236, "step": 1694 }, { "epoch": 0.23, "grad_norm": 3.5840018868372177, "learning_rate": 9.810703200294568e-06, "loss": 0.322, "step": 1695 }, { "epoch": 0.23, "grad_norm": 3.55219864451324, "learning_rate": 9.81040285495821e-06, "loss": 0.3756, "step": 1696 }, { "epoch": 0.23, "grad_norm": 4.156983188691367, "learning_rate": 9.810102276145907e-06, "loss": 0.4122, "step": 1697 }, { "epoch": 0.23, "grad_norm": 3.4639148751314712, "learning_rate": 9.809801463872249e-06, "loss": 0.3381, "step": 1698 }, { "epoch": 0.23, "grad_norm": 3.496767469919019, "learning_rate": 9.809500418151834e-06, "loss": 0.3484, "step": 1699 }, { "epoch": 0.23, "grad_norm": 3.8577713792444928, "learning_rate": 9.809199138999275e-06, "loss": 0.3192, "step": 1700 }, { "epoch": 0.23, "grad_norm": 3.4912488752487563, "learning_rate": 9.808897626429195e-06, "loss": 0.3522, "step": 1701 }, { "epoch": 0.23, "grad_norm": 4.012651696206886, "learning_rate": 9.808595880456226e-06, "loss": 0.4644, "step": 1702 }, { "epoch": 0.23, "grad_norm": 4.144945614632736, "learning_rate": 9.808293901095016e-06, "loss": 0.3076, "step": 1703 }, { "epoch": 0.23, "grad_norm": 3.787013142915524, "learning_rate": 9.80799168836022e-06, "loss": 0.3796, "step": 1704 }, { "epoch": 0.23, "grad_norm": 3.5222047106975936, "learning_rate": 9.80768924226651e-06, "loss": 0.3397, "step": 1705 }, { "epoch": 0.23, "grad_norm": 3.181663969663938, "learning_rate": 9.80738656282856e-06, "loss": 0.3207, "step": 1706 }, { "epoch": 0.23, "grad_norm": 3.6849031357438187, "learning_rate": 9.807083650061063e-06, "loss": 0.3324, "step": 1707 }, { "epoch": 0.23, "grad_norm": 4.036344692836735, "learning_rate": 9.806780503978721e-06, "loss": 0.4062, "step": 1708 }, { "epoch": 0.23, "grad_norm": 3.668201092759312, "learning_rate": 9.806477124596248e-06, "loss": 0.3475, "step": 1709 }, { "epoch": 0.23, "grad_norm": 4.174605260021497, "learning_rate": 9.806173511928369e-06, "loss": 0.3486, "step": 1710 }, { "epoch": 0.23, "grad_norm": 3.412516811594166, "learning_rate": 9.805869665989818e-06, "loss": 0.3347, "step": 1711 }, { "epoch": 0.23, "grad_norm": 2.9912651732090887, "learning_rate": 9.805565586795343e-06, "loss": 0.3085, "step": 1712 }, { "epoch": 0.23, "grad_norm": 3.708447846463784, "learning_rate": 9.805261274359705e-06, "loss": 0.389, "step": 1713 }, { "epoch": 0.23, "grad_norm": 3.8616463080346493, "learning_rate": 9.80495672869767e-06, "loss": 0.3832, "step": 1714 }, { "epoch": 0.23, "grad_norm": 3.510550941910732, "learning_rate": 9.804651949824025e-06, "loss": 0.3432, "step": 1715 }, { "epoch": 0.23, "grad_norm": 3.3250014867737026, "learning_rate": 9.804346937753557e-06, "loss": 0.369, "step": 1716 }, { "epoch": 0.23, "grad_norm": 3.2967900495689118, "learning_rate": 9.804041692501071e-06, "loss": 0.3738, "step": 1717 }, { "epoch": 0.23, "grad_norm": 3.0625655577100575, "learning_rate": 9.803736214081385e-06, "loss": 0.3136, "step": 1718 }, { "epoch": 0.23, "grad_norm": 3.872625979602679, "learning_rate": 9.803430502509323e-06, "loss": 0.3985, "step": 1719 }, { "epoch": 0.23, "grad_norm": 4.003140268126959, "learning_rate": 9.803124557799726e-06, "loss": 0.4209, "step": 1720 }, { "epoch": 0.23, "grad_norm": 3.547116637553752, "learning_rate": 9.802818379967438e-06, "loss": 0.3735, "step": 1721 }, { "epoch": 0.23, "grad_norm": 3.2671765602752942, "learning_rate": 9.802511969027325e-06, "loss": 0.3296, "step": 1722 }, { "epoch": 0.23, "grad_norm": 4.031621642970347, "learning_rate": 9.802205324994256e-06, "loss": 0.4025, "step": 1723 }, { "epoch": 0.23, "grad_norm": 3.0808984637572654, "learning_rate": 9.801898447883114e-06, "loss": 0.3782, "step": 1724 }, { "epoch": 0.23, "grad_norm": 3.4722904425884, "learning_rate": 9.801591337708795e-06, "loss": 0.4037, "step": 1725 }, { "epoch": 0.23, "grad_norm": 3.468354348099712, "learning_rate": 9.801283994486204e-06, "loss": 0.3868, "step": 1726 }, { "epoch": 0.23, "grad_norm": 3.1938796527604505, "learning_rate": 9.800976418230257e-06, "loss": 0.3473, "step": 1727 }, { "epoch": 0.24, "grad_norm": 3.6248285535490443, "learning_rate": 9.800668608955884e-06, "loss": 0.3607, "step": 1728 }, { "epoch": 0.24, "grad_norm": 3.470503340004184, "learning_rate": 9.800360566678024e-06, "loss": 0.3822, "step": 1729 }, { "epoch": 0.24, "grad_norm": 3.4904753540797038, "learning_rate": 9.80005229141163e-06, "loss": 0.3135, "step": 1730 }, { "epoch": 0.24, "grad_norm": 3.8412241059953987, "learning_rate": 9.79974378317166e-06, "loss": 0.3625, "step": 1731 }, { "epoch": 0.24, "grad_norm": 3.2030305365476837, "learning_rate": 9.799435041973092e-06, "loss": 0.3199, "step": 1732 }, { "epoch": 0.24, "grad_norm": 3.718591944878938, "learning_rate": 9.799126067830909e-06, "loss": 0.3172, "step": 1733 }, { "epoch": 0.24, "grad_norm": 3.734377206666006, "learning_rate": 9.798816860760106e-06, "loss": 0.3497, "step": 1734 }, { "epoch": 0.24, "grad_norm": 3.9144469858161277, "learning_rate": 9.798507420775692e-06, "loss": 0.3287, "step": 1735 }, { "epoch": 0.24, "grad_norm": 4.1528566373974005, "learning_rate": 9.798197747892686e-06, "loss": 0.3785, "step": 1736 }, { "epoch": 0.24, "grad_norm": 7.435578667944544, "learning_rate": 9.797887842126119e-06, "loss": 0.3674, "step": 1737 }, { "epoch": 0.24, "grad_norm": 3.718125523339427, "learning_rate": 9.797577703491032e-06, "loss": 0.3784, "step": 1738 }, { "epoch": 0.24, "grad_norm": 3.2882835590233617, "learning_rate": 9.797267332002477e-06, "loss": 0.3574, "step": 1739 }, { "epoch": 0.24, "grad_norm": 3.7939306278210494, "learning_rate": 9.796956727675515e-06, "loss": 0.3686, "step": 1740 }, { "epoch": 0.24, "grad_norm": 3.9101880455359117, "learning_rate": 9.796645890525226e-06, "loss": 0.338, "step": 1741 }, { "epoch": 0.24, "grad_norm": 4.300414308798231, "learning_rate": 9.796334820566697e-06, "loss": 0.4181, "step": 1742 }, { "epoch": 0.24, "grad_norm": 3.783921820387366, "learning_rate": 9.796023517815026e-06, "loss": 0.4088, "step": 1743 }, { "epoch": 0.24, "grad_norm": 3.3535216438757933, "learning_rate": 9.795711982285317e-06, "loss": 0.3463, "step": 1744 }, { "epoch": 0.24, "grad_norm": 3.7111575722382106, "learning_rate": 9.795400213992696e-06, "loss": 0.3812, "step": 1745 }, { "epoch": 0.24, "grad_norm": 3.4294406645590634, "learning_rate": 9.795088212952292e-06, "loss": 0.3452, "step": 1746 }, { "epoch": 0.24, "grad_norm": 3.8120041909222686, "learning_rate": 9.79477597917925e-06, "loss": 0.3217, "step": 1747 }, { "epoch": 0.24, "grad_norm": 3.637307819530362, "learning_rate": 9.794463512688724e-06, "loss": 0.3193, "step": 1748 }, { "epoch": 0.24, "grad_norm": 3.573465653321167, "learning_rate": 9.794150813495877e-06, "loss": 0.3765, "step": 1749 }, { "epoch": 0.24, "grad_norm": 3.5089440165984285, "learning_rate": 9.79383788161589e-06, "loss": 0.3509, "step": 1750 }, { "epoch": 0.24, "grad_norm": 3.7440198324348803, "learning_rate": 9.793524717063952e-06, "loss": 0.401, "step": 1751 }, { "epoch": 0.24, "grad_norm": 3.627062238359795, "learning_rate": 9.793211319855258e-06, "loss": 0.3722, "step": 1752 }, { "epoch": 0.24, "grad_norm": 4.040739690919761, "learning_rate": 9.792897690005022e-06, "loss": 0.3546, "step": 1753 }, { "epoch": 0.24, "grad_norm": 3.1287277938577183, "learning_rate": 9.792583827528465e-06, "loss": 0.3696, "step": 1754 }, { "epoch": 0.24, "grad_norm": 3.4719053472579233, "learning_rate": 9.792269732440822e-06, "loss": 0.3681, "step": 1755 }, { "epoch": 0.24, "grad_norm": 3.172500428475981, "learning_rate": 9.791955404757337e-06, "loss": 0.3168, "step": 1756 }, { "epoch": 0.24, "grad_norm": 3.655284880518049, "learning_rate": 9.791640844493267e-06, "loss": 0.3558, "step": 1757 }, { "epoch": 0.24, "grad_norm": 3.4334101589603594, "learning_rate": 9.791326051663877e-06, "loss": 0.375, "step": 1758 }, { "epoch": 0.24, "grad_norm": 3.6173073877346034, "learning_rate": 9.791011026284447e-06, "loss": 0.3913, "step": 1759 }, { "epoch": 0.24, "grad_norm": 3.598965305341432, "learning_rate": 9.790695768370267e-06, "loss": 0.3563, "step": 1760 }, { "epoch": 0.24, "grad_norm": 3.400755375877299, "learning_rate": 9.79038027793664e-06, "loss": 0.3701, "step": 1761 }, { "epoch": 0.24, "grad_norm": 3.884313335969902, "learning_rate": 9.790064554998875e-06, "loss": 0.3427, "step": 1762 }, { "epoch": 0.24, "grad_norm": 3.333747836578991, "learning_rate": 9.789748599572297e-06, "loss": 0.3411, "step": 1763 }, { "epoch": 0.24, "grad_norm": 3.714025076199745, "learning_rate": 9.789432411672243e-06, "loss": 0.3329, "step": 1764 }, { "epoch": 0.24, "grad_norm": 3.7427671523612203, "learning_rate": 9.789115991314058e-06, "loss": 0.3436, "step": 1765 }, { "epoch": 0.24, "grad_norm": 3.513282382220842, "learning_rate": 9.7887993385131e-06, "loss": 0.3719, "step": 1766 }, { "epoch": 0.24, "grad_norm": 3.3615099495744376, "learning_rate": 9.788482453284737e-06, "loss": 0.3212, "step": 1767 }, { "epoch": 0.24, "grad_norm": 3.869156528418046, "learning_rate": 9.78816533564435e-06, "loss": 0.3613, "step": 1768 }, { "epoch": 0.24, "grad_norm": 3.9665026560088528, "learning_rate": 9.787847985607332e-06, "loss": 0.4061, "step": 1769 }, { "epoch": 0.24, "grad_norm": 3.527105434186619, "learning_rate": 9.787530403189083e-06, "loss": 0.3942, "step": 1770 }, { "epoch": 0.24, "grad_norm": 3.4769730161292207, "learning_rate": 9.787212588405018e-06, "loss": 0.3584, "step": 1771 }, { "epoch": 0.24, "grad_norm": 3.482366633501503, "learning_rate": 9.786894541270563e-06, "loss": 0.3854, "step": 1772 }, { "epoch": 0.24, "grad_norm": 3.0934014122588835, "learning_rate": 9.786576261801155e-06, "loss": 0.3285, "step": 1773 }, { "epoch": 0.24, "grad_norm": 3.5598922657243297, "learning_rate": 9.78625775001224e-06, "loss": 0.3583, "step": 1774 }, { "epoch": 0.24, "grad_norm": 3.5361120186187827, "learning_rate": 9.785939005919279e-06, "loss": 0.2938, "step": 1775 }, { "epoch": 0.24, "grad_norm": 3.6554282423785884, "learning_rate": 9.785620029537741e-06, "loss": 0.3717, "step": 1776 }, { "epoch": 0.24, "grad_norm": 3.2848559303929177, "learning_rate": 9.785300820883108e-06, "loss": 0.3506, "step": 1777 }, { "epoch": 0.24, "grad_norm": 3.3112856173481986, "learning_rate": 9.784981379970874e-06, "loss": 0.4105, "step": 1778 }, { "epoch": 0.24, "grad_norm": 3.130304984973132, "learning_rate": 9.784661706816544e-06, "loss": 0.315, "step": 1779 }, { "epoch": 0.24, "grad_norm": 3.5250585056287598, "learning_rate": 9.784341801435632e-06, "loss": 0.383, "step": 1780 }, { "epoch": 0.24, "grad_norm": 3.7614226112113034, "learning_rate": 9.784021663843664e-06, "loss": 0.3967, "step": 1781 }, { "epoch": 0.24, "grad_norm": 3.4402635642285517, "learning_rate": 9.78370129405618e-06, "loss": 0.3463, "step": 1782 }, { "epoch": 0.24, "grad_norm": 3.8312212410193336, "learning_rate": 9.78338069208873e-06, "loss": 0.3663, "step": 1783 }, { "epoch": 0.24, "grad_norm": 3.947615195967893, "learning_rate": 9.78305985795687e-06, "loss": 0.38, "step": 1784 }, { "epoch": 0.24, "grad_norm": 3.4739962665956643, "learning_rate": 9.782738791676176e-06, "loss": 0.3206, "step": 1785 }, { "epoch": 0.24, "grad_norm": 3.642000002255208, "learning_rate": 9.78241749326223e-06, "loss": 0.4035, "step": 1786 }, { "epoch": 0.24, "grad_norm": 3.6144644074433856, "learning_rate": 9.782095962730628e-06, "loss": 0.3288, "step": 1787 }, { "epoch": 0.24, "grad_norm": 3.278177767238997, "learning_rate": 9.781774200096974e-06, "loss": 0.3022, "step": 1788 }, { "epoch": 0.24, "grad_norm": 4.225142451609505, "learning_rate": 9.781452205376885e-06, "loss": 0.4234, "step": 1789 }, { "epoch": 0.24, "grad_norm": 3.4940405318223915, "learning_rate": 9.781129978585988e-06, "loss": 0.3162, "step": 1790 }, { "epoch": 0.24, "grad_norm": 3.4408147565551914, "learning_rate": 9.780807519739925e-06, "loss": 0.3749, "step": 1791 }, { "epoch": 0.24, "grad_norm": 3.8460903556696007, "learning_rate": 9.780484828854346e-06, "loss": 0.3241, "step": 1792 }, { "epoch": 0.24, "grad_norm": 3.9425449497422984, "learning_rate": 9.780161905944912e-06, "loss": 0.3164, "step": 1793 }, { "epoch": 0.24, "grad_norm": 2.839344661738024, "learning_rate": 9.779838751027297e-06, "loss": 0.2893, "step": 1794 }, { "epoch": 0.24, "grad_norm": 4.041677208909388, "learning_rate": 9.779515364117187e-06, "loss": 0.4401, "step": 1795 }, { "epoch": 0.24, "grad_norm": 3.4746495750436943, "learning_rate": 9.779191745230275e-06, "loss": 0.4263, "step": 1796 }, { "epoch": 0.24, "grad_norm": 3.1092600474607357, "learning_rate": 9.77886789438227e-06, "loss": 0.3358, "step": 1797 }, { "epoch": 0.24, "grad_norm": 3.5607466349031154, "learning_rate": 9.77854381158889e-06, "loss": 0.3604, "step": 1798 }, { "epoch": 0.24, "grad_norm": 2.965286020393845, "learning_rate": 9.778219496865864e-06, "loss": 0.3187, "step": 1799 }, { "epoch": 0.24, "grad_norm": 3.6799111416940318, "learning_rate": 9.777894950228932e-06, "loss": 0.3484, "step": 1800 }, { "epoch": 0.24, "grad_norm": 3.143751713899356, "learning_rate": 9.777570171693847e-06, "loss": 0.3274, "step": 1801 }, { "epoch": 0.25, "grad_norm": 3.1058430945043614, "learning_rate": 9.777245161276372e-06, "loss": 0.3725, "step": 1802 }, { "epoch": 0.25, "grad_norm": 4.115246458071809, "learning_rate": 9.776919918992284e-06, "loss": 0.3861, "step": 1803 }, { "epoch": 0.25, "grad_norm": 3.398553803679445, "learning_rate": 9.776594444857366e-06, "loss": 0.3139, "step": 1804 }, { "epoch": 0.25, "grad_norm": 3.856377004502878, "learning_rate": 9.776268738887415e-06, "loss": 0.3447, "step": 1805 }, { "epoch": 0.25, "grad_norm": 3.4541037058536097, "learning_rate": 9.775942801098241e-06, "loss": 0.3483, "step": 1806 }, { "epoch": 0.25, "grad_norm": 3.327713516132156, "learning_rate": 9.775616631505663e-06, "loss": 0.3344, "step": 1807 }, { "epoch": 0.25, "grad_norm": 3.2313288880231155, "learning_rate": 9.775290230125511e-06, "loss": 0.3384, "step": 1808 }, { "epoch": 0.25, "grad_norm": 3.092657840683117, "learning_rate": 9.77496359697363e-06, "loss": 0.3215, "step": 1809 }, { "epoch": 0.25, "grad_norm": 3.4948248822876393, "learning_rate": 9.77463673206587e-06, "loss": 0.3456, "step": 1810 }, { "epoch": 0.25, "grad_norm": 4.426921558806708, "learning_rate": 9.774309635418095e-06, "loss": 0.3668, "step": 1811 }, { "epoch": 0.25, "grad_norm": 3.2055441057830714, "learning_rate": 9.773982307046185e-06, "loss": 0.2924, "step": 1812 }, { "epoch": 0.25, "grad_norm": 4.108235816232314, "learning_rate": 9.773654746966023e-06, "loss": 0.4442, "step": 1813 }, { "epoch": 0.25, "grad_norm": 3.440760810005909, "learning_rate": 9.77332695519351e-06, "loss": 0.3308, "step": 1814 }, { "epoch": 0.25, "grad_norm": 3.581543527486437, "learning_rate": 9.772998931744555e-06, "loss": 0.3357, "step": 1815 }, { "epoch": 0.25, "grad_norm": 4.338498892826777, "learning_rate": 9.772670676635077e-06, "loss": 0.3849, "step": 1816 }, { "epoch": 0.25, "grad_norm": 4.0354123086694, "learning_rate": 9.772342189881012e-06, "loss": 0.4115, "step": 1817 }, { "epoch": 0.25, "grad_norm": 4.091495017063622, "learning_rate": 9.772013471498298e-06, "loss": 0.3822, "step": 1818 }, { "epoch": 0.25, "grad_norm": 3.4787845239135518, "learning_rate": 9.771684521502893e-06, "loss": 0.3448, "step": 1819 }, { "epoch": 0.25, "grad_norm": 3.3769001037261335, "learning_rate": 9.771355339910764e-06, "loss": 0.3692, "step": 1820 }, { "epoch": 0.25, "grad_norm": 3.9351401685354044, "learning_rate": 9.771025926737885e-06, "loss": 0.3089, "step": 1821 }, { "epoch": 0.25, "grad_norm": 3.910621734948175, "learning_rate": 9.770696282000245e-06, "loss": 0.3779, "step": 1822 }, { "epoch": 0.25, "grad_norm": 3.0809893100174772, "learning_rate": 9.770366405713844e-06, "loss": 0.2965, "step": 1823 }, { "epoch": 0.25, "grad_norm": 3.241970260537631, "learning_rate": 9.770036297894694e-06, "loss": 0.3062, "step": 1824 }, { "epoch": 0.25, "grad_norm": 3.829018386716392, "learning_rate": 9.769705958558815e-06, "loss": 0.3629, "step": 1825 }, { "epoch": 0.25, "grad_norm": 3.934575215729435, "learning_rate": 9.76937538772224e-06, "loss": 0.353, "step": 1826 }, { "epoch": 0.25, "grad_norm": 3.51651366468177, "learning_rate": 9.769044585401017e-06, "loss": 0.3782, "step": 1827 }, { "epoch": 0.25, "grad_norm": 3.26609125099566, "learning_rate": 9.768713551611197e-06, "loss": 0.3537, "step": 1828 }, { "epoch": 0.25, "grad_norm": 3.391220087147713, "learning_rate": 9.76838228636885e-06, "loss": 0.3084, "step": 1829 }, { "epoch": 0.25, "grad_norm": 3.8066833143140233, "learning_rate": 9.768050789690054e-06, "loss": 0.4203, "step": 1830 }, { "epoch": 0.25, "grad_norm": 4.117079039852699, "learning_rate": 9.767719061590895e-06, "loss": 0.3773, "step": 1831 }, { "epoch": 0.25, "grad_norm": 4.25044078978198, "learning_rate": 9.767387102087477e-06, "loss": 0.376, "step": 1832 }, { "epoch": 0.25, "grad_norm": 3.584561850885416, "learning_rate": 9.767054911195912e-06, "loss": 0.3802, "step": 1833 }, { "epoch": 0.25, "grad_norm": 2.951782421303463, "learning_rate": 9.766722488932323e-06, "loss": 0.2814, "step": 1834 }, { "epoch": 0.25, "grad_norm": 3.5992270982399575, "learning_rate": 9.766389835312844e-06, "loss": 0.3721, "step": 1835 }, { "epoch": 0.25, "grad_norm": 3.560979884258195, "learning_rate": 9.766056950353617e-06, "loss": 0.3369, "step": 1836 }, { "epoch": 0.25, "grad_norm": 2.959770950251094, "learning_rate": 9.765723834070805e-06, "loss": 0.3261, "step": 1837 }, { "epoch": 0.25, "grad_norm": 3.2337302476054357, "learning_rate": 9.765390486480572e-06, "loss": 0.2897, "step": 1838 }, { "epoch": 0.25, "grad_norm": 3.0993716509881493, "learning_rate": 9.765056907599098e-06, "loss": 0.3319, "step": 1839 }, { "epoch": 0.25, "grad_norm": 3.6757984030298267, "learning_rate": 9.764723097442573e-06, "loss": 0.3555, "step": 1840 }, { "epoch": 0.25, "grad_norm": 3.738513233343676, "learning_rate": 9.764389056027198e-06, "loss": 0.3633, "step": 1841 }, { "epoch": 0.25, "grad_norm": 3.0538155614035865, "learning_rate": 9.764054783369191e-06, "loss": 0.2907, "step": 1842 }, { "epoch": 0.25, "grad_norm": 3.4464103620967155, "learning_rate": 9.76372027948477e-06, "loss": 0.3753, "step": 1843 }, { "epoch": 0.25, "grad_norm": 3.808768549145614, "learning_rate": 9.763385544390174e-06, "loss": 0.4164, "step": 1844 }, { "epoch": 0.25, "grad_norm": 3.1986161627792407, "learning_rate": 9.763050578101646e-06, "loss": 0.3511, "step": 1845 }, { "epoch": 0.25, "grad_norm": 4.05499361255307, "learning_rate": 9.762715380635448e-06, "loss": 0.445, "step": 1846 }, { "epoch": 0.25, "grad_norm": 3.456550584944299, "learning_rate": 9.762379952007847e-06, "loss": 0.3359, "step": 1847 }, { "epoch": 0.25, "grad_norm": 4.204181617525404, "learning_rate": 9.762044292235123e-06, "loss": 0.3344, "step": 1848 }, { "epoch": 0.25, "grad_norm": 3.804275507428074, "learning_rate": 9.761708401333567e-06, "loss": 0.3527, "step": 1849 }, { "epoch": 0.25, "grad_norm": 3.1778380477181822, "learning_rate": 9.761372279319483e-06, "loss": 0.3213, "step": 1850 }, { "epoch": 0.25, "grad_norm": 3.376947378382084, "learning_rate": 9.761035926209185e-06, "loss": 0.3103, "step": 1851 }, { "epoch": 0.25, "grad_norm": 3.8471864836681045, "learning_rate": 9.760699342018997e-06, "loss": 0.3666, "step": 1852 }, { "epoch": 0.25, "grad_norm": 3.536973503776947, "learning_rate": 9.760362526765257e-06, "loss": 0.2825, "step": 1853 }, { "epoch": 0.25, "grad_norm": 3.4181449204214034, "learning_rate": 9.76002548046431e-06, "loss": 0.2764, "step": 1854 }, { "epoch": 0.25, "grad_norm": 3.8578394423804827, "learning_rate": 9.759688203132515e-06, "loss": 0.3353, "step": 1855 }, { "epoch": 0.25, "grad_norm": 3.946247778541619, "learning_rate": 9.759350694786244e-06, "loss": 0.4045, "step": 1856 }, { "epoch": 0.25, "grad_norm": 3.088109853412147, "learning_rate": 9.759012955441877e-06, "loss": 0.367, "step": 1857 }, { "epoch": 0.25, "grad_norm": 3.4590102747694678, "learning_rate": 9.758674985115809e-06, "loss": 0.3216, "step": 1858 }, { "epoch": 0.25, "grad_norm": 3.745582059766145, "learning_rate": 9.75833678382444e-06, "loss": 0.3548, "step": 1859 }, { "epoch": 0.25, "grad_norm": 3.6406418882564555, "learning_rate": 9.757998351584184e-06, "loss": 0.4257, "step": 1860 }, { "epoch": 0.25, "grad_norm": 3.1032617837214365, "learning_rate": 9.757659688411472e-06, "loss": 0.3444, "step": 1861 }, { "epoch": 0.25, "grad_norm": 3.302094558230847, "learning_rate": 9.757320794322736e-06, "loss": 0.3185, "step": 1862 }, { "epoch": 0.25, "grad_norm": 3.592800621101176, "learning_rate": 9.756981669334428e-06, "loss": 0.3992, "step": 1863 }, { "epoch": 0.25, "grad_norm": 3.2108308222912814, "learning_rate": 9.756642313463008e-06, "loss": 0.3576, "step": 1864 }, { "epoch": 0.25, "grad_norm": 3.196392958552571, "learning_rate": 9.756302726724942e-06, "loss": 0.3159, "step": 1865 }, { "epoch": 0.25, "grad_norm": 3.408299265595955, "learning_rate": 9.755962909136717e-06, "loss": 0.3769, "step": 1866 }, { "epoch": 0.25, "grad_norm": 3.6868715620309627, "learning_rate": 9.755622860714824e-06, "loss": 0.3633, "step": 1867 }, { "epoch": 0.25, "grad_norm": 4.69540675527483, "learning_rate": 9.755282581475769e-06, "loss": 0.3807, "step": 1868 }, { "epoch": 0.25, "grad_norm": 3.743250364809263, "learning_rate": 9.754942071436067e-06, "loss": 0.3795, "step": 1869 }, { "epoch": 0.25, "grad_norm": 3.5048393442060575, "learning_rate": 9.754601330612242e-06, "loss": 0.3156, "step": 1870 }, { "epoch": 0.25, "grad_norm": 2.8404823366209246, "learning_rate": 9.754260359020837e-06, "loss": 0.3395, "step": 1871 }, { "epoch": 0.25, "grad_norm": 3.322554310174297, "learning_rate": 9.753919156678397e-06, "loss": 0.3667, "step": 1872 }, { "epoch": 0.25, "grad_norm": 3.8625200506397546, "learning_rate": 9.753577723601486e-06, "loss": 0.4189, "step": 1873 }, { "epoch": 0.25, "grad_norm": 3.5167703403885957, "learning_rate": 9.753236059806674e-06, "loss": 0.3791, "step": 1874 }, { "epoch": 0.26, "grad_norm": 3.124825775524122, "learning_rate": 9.752894165310544e-06, "loss": 0.2812, "step": 1875 }, { "epoch": 0.26, "grad_norm": 3.1862709038788717, "learning_rate": 9.752552040129691e-06, "loss": 0.3205, "step": 1876 }, { "epoch": 0.26, "grad_norm": 3.4944085087543564, "learning_rate": 9.752209684280717e-06, "loss": 0.3859, "step": 1877 }, { "epoch": 0.26, "grad_norm": 3.7518264731482756, "learning_rate": 9.75186709778024e-06, "loss": 0.3835, "step": 1878 }, { "epoch": 0.26, "grad_norm": 3.531327540439291, "learning_rate": 9.75152428064489e-06, "loss": 0.3891, "step": 1879 }, { "epoch": 0.26, "grad_norm": 3.422882040268158, "learning_rate": 9.751181232891306e-06, "loss": 0.3655, "step": 1880 }, { "epoch": 0.26, "grad_norm": 3.7020167504097747, "learning_rate": 9.750837954536134e-06, "loss": 0.3326, "step": 1881 }, { "epoch": 0.26, "grad_norm": 3.2597492630174254, "learning_rate": 9.750494445596039e-06, "loss": 0.3017, "step": 1882 }, { "epoch": 0.26, "grad_norm": 3.4641950146767724, "learning_rate": 9.75015070608769e-06, "loss": 0.3549, "step": 1883 }, { "epoch": 0.26, "grad_norm": 3.218212902313431, "learning_rate": 9.749806736027774e-06, "loss": 0.3206, "step": 1884 }, { "epoch": 0.26, "grad_norm": 3.212238277829678, "learning_rate": 9.749462535432983e-06, "loss": 0.3387, "step": 1885 }, { "epoch": 0.26, "grad_norm": 3.548796848077671, "learning_rate": 9.749118104320026e-06, "loss": 0.3482, "step": 1886 }, { "epoch": 0.26, "grad_norm": 4.0885290093605615, "learning_rate": 9.748773442705617e-06, "loss": 0.3171, "step": 1887 }, { "epoch": 0.26, "grad_norm": 3.186906428846712, "learning_rate": 9.748428550606486e-06, "loss": 0.344, "step": 1888 }, { "epoch": 0.26, "grad_norm": 3.2697355413069915, "learning_rate": 9.748083428039373e-06, "loss": 0.341, "step": 1889 }, { "epoch": 0.26, "grad_norm": 3.0966126014584834, "learning_rate": 9.747738075021028e-06, "loss": 0.3409, "step": 1890 }, { "epoch": 0.26, "grad_norm": 3.6618912973520836, "learning_rate": 9.747392491568213e-06, "loss": 0.3289, "step": 1891 }, { "epoch": 0.26, "grad_norm": 3.2417968737378455, "learning_rate": 9.747046677697703e-06, "loss": 0.2934, "step": 1892 }, { "epoch": 0.26, "grad_norm": 3.0219529105669336, "learning_rate": 9.746700633426279e-06, "loss": 0.3852, "step": 1893 }, { "epoch": 0.26, "grad_norm": 3.2072130153313125, "learning_rate": 9.746354358770737e-06, "loss": 0.3689, "step": 1894 }, { "epoch": 0.26, "grad_norm": 3.1822801134902465, "learning_rate": 9.746007853747886e-06, "loss": 0.2872, "step": 1895 }, { "epoch": 0.26, "grad_norm": 3.510390892922414, "learning_rate": 9.74566111837454e-06, "loss": 0.3518, "step": 1896 }, { "epoch": 0.26, "grad_norm": 3.770305411976719, "learning_rate": 9.745314152667532e-06, "loss": 0.4082, "step": 1897 }, { "epoch": 0.26, "grad_norm": 3.62525876512084, "learning_rate": 9.744966956643703e-06, "loss": 0.4212, "step": 1898 }, { "epoch": 0.26, "grad_norm": 3.5004379081657566, "learning_rate": 9.7446195303199e-06, "loss": 0.409, "step": 1899 }, { "epoch": 0.26, "grad_norm": 4.444737800116971, "learning_rate": 9.744271873712987e-06, "loss": 0.3955, "step": 1900 }, { "epoch": 0.26, "grad_norm": 4.53966353526624, "learning_rate": 9.743923986839841e-06, "loss": 0.2889, "step": 1901 }, { "epoch": 0.26, "grad_norm": 3.5676357340775557, "learning_rate": 9.743575869717343e-06, "loss": 0.3643, "step": 1902 }, { "epoch": 0.26, "grad_norm": 3.2244020203839656, "learning_rate": 9.743227522362388e-06, "loss": 0.3301, "step": 1903 }, { "epoch": 0.26, "grad_norm": 3.9011165555781644, "learning_rate": 9.742878944791889e-06, "loss": 0.3599, "step": 1904 }, { "epoch": 0.26, "grad_norm": 3.571563259114754, "learning_rate": 9.742530137022759e-06, "loss": 0.3062, "step": 1905 }, { "epoch": 0.26, "grad_norm": 3.472711200289608, "learning_rate": 9.74218109907193e-06, "loss": 0.3997, "step": 1906 }, { "epoch": 0.26, "grad_norm": 3.325855232669742, "learning_rate": 9.741831830956344e-06, "loss": 0.357, "step": 1907 }, { "epoch": 0.26, "grad_norm": 3.213731242648025, "learning_rate": 9.741482332692948e-06, "loss": 0.3487, "step": 1908 }, { "epoch": 0.26, "grad_norm": 3.3297808195420937, "learning_rate": 9.741132604298712e-06, "loss": 0.3461, "step": 1909 }, { "epoch": 0.26, "grad_norm": 3.0588560880546187, "learning_rate": 9.740782645790605e-06, "loss": 0.3299, "step": 1910 }, { "epoch": 0.26, "grad_norm": 2.7245173243729566, "learning_rate": 9.740432457185617e-06, "loss": 0.2712, "step": 1911 }, { "epoch": 0.26, "grad_norm": 3.613192011549721, "learning_rate": 9.740082038500738e-06, "loss": 0.3359, "step": 1912 }, { "epoch": 0.26, "grad_norm": 3.157070947386184, "learning_rate": 9.739731389752982e-06, "loss": 0.3281, "step": 1913 }, { "epoch": 0.26, "grad_norm": 2.960994985609924, "learning_rate": 9.739380510959365e-06, "loss": 0.3432, "step": 1914 }, { "epoch": 0.26, "grad_norm": 3.4047017131132256, "learning_rate": 9.739029402136917e-06, "loss": 0.3426, "step": 1915 }, { "epoch": 0.26, "grad_norm": 3.834881668276771, "learning_rate": 9.73867806330268e-06, "loss": 0.4018, "step": 1916 }, { "epoch": 0.26, "grad_norm": 3.454857588372101, "learning_rate": 9.738326494473708e-06, "loss": 0.356, "step": 1917 }, { "epoch": 0.26, "grad_norm": 4.228820531347088, "learning_rate": 9.737974695667063e-06, "loss": 0.453, "step": 1918 }, { "epoch": 0.26, "grad_norm": 3.37295615510254, "learning_rate": 9.737622666899819e-06, "loss": 0.347, "step": 1919 }, { "epoch": 0.26, "grad_norm": 3.3463568506592183, "learning_rate": 9.737270408189064e-06, "loss": 0.3803, "step": 1920 }, { "epoch": 0.26, "grad_norm": 3.1968883284521095, "learning_rate": 9.736917919551891e-06, "loss": 0.2761, "step": 1921 }, { "epoch": 0.26, "grad_norm": 3.014430821497962, "learning_rate": 9.736565201005415e-06, "loss": 0.3562, "step": 1922 }, { "epoch": 0.26, "grad_norm": 3.201297115325336, "learning_rate": 9.736212252566749e-06, "loss": 0.3659, "step": 1923 }, { "epoch": 0.26, "grad_norm": 3.3026617885290332, "learning_rate": 9.735859074253026e-06, "loss": 0.3321, "step": 1924 }, { "epoch": 0.26, "grad_norm": 3.522082845505956, "learning_rate": 9.735505666081389e-06, "loss": 0.3046, "step": 1925 }, { "epoch": 0.26, "grad_norm": 3.4398015585256267, "learning_rate": 9.73515202806899e-06, "loss": 0.2961, "step": 1926 }, { "epoch": 0.26, "grad_norm": 3.9250693414062003, "learning_rate": 9.734798160232994e-06, "loss": 0.3869, "step": 1927 }, { "epoch": 0.26, "grad_norm": 3.2881862598380938, "learning_rate": 9.734444062590573e-06, "loss": 0.3707, "step": 1928 }, { "epoch": 0.26, "grad_norm": 3.2757802578834387, "learning_rate": 9.734089735158916e-06, "loss": 0.3713, "step": 1929 }, { "epoch": 0.26, "grad_norm": 2.7871250632766764, "learning_rate": 9.733735177955219e-06, "loss": 0.3067, "step": 1930 }, { "epoch": 0.26, "grad_norm": 3.4263376608045233, "learning_rate": 9.733380390996694e-06, "loss": 0.3405, "step": 1931 }, { "epoch": 0.26, "grad_norm": 3.19894836370517, "learning_rate": 9.733025374300556e-06, "loss": 0.3248, "step": 1932 }, { "epoch": 0.26, "grad_norm": 3.4766176207423767, "learning_rate": 9.732670127884039e-06, "loss": 0.289, "step": 1933 }, { "epoch": 0.26, "grad_norm": 3.7812160045320553, "learning_rate": 9.732314651764385e-06, "loss": 0.3627, "step": 1934 }, { "epoch": 0.26, "grad_norm": 2.817760254091848, "learning_rate": 9.731958945958846e-06, "loss": 0.2688, "step": 1935 }, { "epoch": 0.26, "grad_norm": 3.6571192696999626, "learning_rate": 9.731603010484687e-06, "loss": 0.384, "step": 1936 }, { "epoch": 0.26, "grad_norm": 3.068237649231404, "learning_rate": 9.731246845359187e-06, "loss": 0.2796, "step": 1937 }, { "epoch": 0.26, "grad_norm": 3.7344558442143136, "learning_rate": 9.730890450599623e-06, "loss": 0.3054, "step": 1938 }, { "epoch": 0.26, "grad_norm": 3.8650704926810997, "learning_rate": 9.730533826223304e-06, "loss": 0.3827, "step": 1939 }, { "epoch": 0.26, "grad_norm": 3.4529033971127134, "learning_rate": 9.730176972247534e-06, "loss": 0.2721, "step": 1940 }, { "epoch": 0.26, "grad_norm": 3.487903842989575, "learning_rate": 9.729819888689631e-06, "loss": 0.3575, "step": 1941 }, { "epoch": 0.26, "grad_norm": 3.6057395617162973, "learning_rate": 9.729462575566931e-06, "loss": 0.3554, "step": 1942 }, { "epoch": 0.26, "grad_norm": 3.9055865174046445, "learning_rate": 9.729105032896773e-06, "loss": 0.3371, "step": 1943 }, { "epoch": 0.26, "grad_norm": 4.0423278480650735, "learning_rate": 9.72874726069651e-06, "loss": 0.3969, "step": 1944 }, { "epoch": 0.26, "grad_norm": 3.3803139207145745, "learning_rate": 9.72838925898351e-06, "loss": 0.3743, "step": 1945 }, { "epoch": 0.26, "grad_norm": 3.3356228134904296, "learning_rate": 9.728031027775146e-06, "loss": 0.3051, "step": 1946 }, { "epoch": 0.26, "grad_norm": 3.5246103533326076, "learning_rate": 9.727672567088809e-06, "loss": 0.3383, "step": 1947 }, { "epoch": 0.26, "grad_norm": 3.892988051883675, "learning_rate": 9.72731387694189e-06, "loss": 0.3871, "step": 1948 }, { "epoch": 0.27, "grad_norm": 4.042188013209971, "learning_rate": 9.726954957351807e-06, "loss": 0.4107, "step": 1949 }, { "epoch": 0.27, "grad_norm": 3.7201274481465414, "learning_rate": 9.726595808335972e-06, "loss": 0.4256, "step": 1950 }, { "epoch": 0.27, "grad_norm": 3.8815439342070497, "learning_rate": 9.726236429911823e-06, "loss": 0.4436, "step": 1951 }, { "epoch": 0.27, "grad_norm": 3.3390489035197506, "learning_rate": 9.725876822096798e-06, "loss": 0.3271, "step": 1952 }, { "epoch": 0.27, "grad_norm": 3.3306082954098755, "learning_rate": 9.725516984908355e-06, "loss": 0.3344, "step": 1953 }, { "epoch": 0.27, "grad_norm": 3.534563278915659, "learning_rate": 9.725156918363954e-06, "loss": 0.3568, "step": 1954 }, { "epoch": 0.27, "grad_norm": 3.2802631203934287, "learning_rate": 9.724796622481077e-06, "loss": 0.3763, "step": 1955 }, { "epoch": 0.27, "grad_norm": 3.762503271764268, "learning_rate": 9.724436097277205e-06, "loss": 0.4298, "step": 1956 }, { "epoch": 0.27, "grad_norm": 3.7392543061618047, "learning_rate": 9.724075342769841e-06, "loss": 0.3908, "step": 1957 }, { "epoch": 0.27, "grad_norm": 3.7866048142993005, "learning_rate": 9.723714358976493e-06, "loss": 0.3845, "step": 1958 }, { "epoch": 0.27, "grad_norm": 2.9660338789035525, "learning_rate": 9.72335314591468e-06, "loss": 0.3104, "step": 1959 }, { "epoch": 0.27, "grad_norm": 3.744827956831934, "learning_rate": 9.722991703601936e-06, "loss": 0.3885, "step": 1960 }, { "epoch": 0.27, "grad_norm": 3.4151054560988165, "learning_rate": 9.722630032055804e-06, "loss": 0.3717, "step": 1961 }, { "epoch": 0.27, "grad_norm": 3.2186685868059755, "learning_rate": 9.722268131293835e-06, "loss": 0.3207, "step": 1962 }, { "epoch": 0.27, "grad_norm": 3.2818289593871492, "learning_rate": 9.721906001333596e-06, "loss": 0.3317, "step": 1963 }, { "epoch": 0.27, "grad_norm": 3.177393383224234, "learning_rate": 9.721543642192665e-06, "loss": 0.3559, "step": 1964 }, { "epoch": 0.27, "grad_norm": 3.3814331719658712, "learning_rate": 9.721181053888625e-06, "loss": 0.3579, "step": 1965 }, { "epoch": 0.27, "grad_norm": 3.4549197180106748, "learning_rate": 9.720818236439077e-06, "loss": 0.3823, "step": 1966 }, { "epoch": 0.27, "grad_norm": 3.3630391371091286, "learning_rate": 9.720455189861634e-06, "loss": 0.3457, "step": 1967 }, { "epoch": 0.27, "grad_norm": 4.2999807126140395, "learning_rate": 9.72009191417391e-06, "loss": 0.2988, "step": 1968 }, { "epoch": 0.27, "grad_norm": 3.5628475269732998, "learning_rate": 9.71972840939354e-06, "loss": 0.3771, "step": 1969 }, { "epoch": 0.27, "grad_norm": 3.275550605985853, "learning_rate": 9.71936467553817e-06, "loss": 0.3239, "step": 1970 }, { "epoch": 0.27, "grad_norm": 3.1979324815477765, "learning_rate": 9.719000712625447e-06, "loss": 0.3804, "step": 1971 }, { "epoch": 0.27, "grad_norm": 3.504312487380276, "learning_rate": 9.718636520673042e-06, "loss": 0.3346, "step": 1972 }, { "epoch": 0.27, "grad_norm": 2.887863140513751, "learning_rate": 9.718272099698629e-06, "loss": 0.2923, "step": 1973 }, { "epoch": 0.27, "grad_norm": 3.366224786866524, "learning_rate": 9.717907449719897e-06, "loss": 0.2873, "step": 1974 }, { "epoch": 0.27, "grad_norm": 3.572412480986972, "learning_rate": 9.717542570754542e-06, "loss": 0.3409, "step": 1975 }, { "epoch": 0.27, "grad_norm": 3.3749010598463016, "learning_rate": 9.717177462820276e-06, "loss": 0.2894, "step": 1976 }, { "epoch": 0.27, "grad_norm": 3.2016207638176586, "learning_rate": 9.716812125934818e-06, "loss": 0.3239, "step": 1977 }, { "epoch": 0.27, "grad_norm": 3.639975760167683, "learning_rate": 9.716446560115901e-06, "loss": 0.3106, "step": 1978 }, { "epoch": 0.27, "grad_norm": 3.597055964299524, "learning_rate": 9.716080765381267e-06, "loss": 0.4108, "step": 1979 }, { "epoch": 0.27, "grad_norm": 3.5418331259294664, "learning_rate": 9.715714741748672e-06, "loss": 0.3711, "step": 1980 }, { "epoch": 0.27, "grad_norm": 3.4683595306732635, "learning_rate": 9.71534848923588e-06, "loss": 0.4131, "step": 1981 }, { "epoch": 0.27, "grad_norm": 2.9345235748113003, "learning_rate": 9.714982007860666e-06, "loss": 0.2635, "step": 1982 }, { "epoch": 0.27, "grad_norm": 3.392023283753395, "learning_rate": 9.714615297640818e-06, "loss": 0.3853, "step": 1983 }, { "epoch": 0.27, "grad_norm": 5.110905429476173, "learning_rate": 9.714248358594133e-06, "loss": 0.3083, "step": 1984 }, { "epoch": 0.27, "grad_norm": 3.3307830927427933, "learning_rate": 9.713881190738426e-06, "loss": 0.2951, "step": 1985 }, { "epoch": 0.27, "grad_norm": 3.302019778563027, "learning_rate": 9.713513794091514e-06, "loss": 0.3231, "step": 1986 }, { "epoch": 0.27, "grad_norm": 3.4679523240488836, "learning_rate": 9.713146168671229e-06, "loss": 0.298, "step": 1987 }, { "epoch": 0.27, "grad_norm": 3.4703657128860836, "learning_rate": 9.712778314495415e-06, "loss": 0.3965, "step": 1988 }, { "epoch": 0.27, "grad_norm": 3.526032986421355, "learning_rate": 9.712410231581924e-06, "loss": 0.3751, "step": 1989 }, { "epoch": 0.27, "grad_norm": 3.1396764766479057, "learning_rate": 9.712041919948623e-06, "loss": 0.3049, "step": 1990 }, { "epoch": 0.27, "grad_norm": 3.565771552903686, "learning_rate": 9.711673379613387e-06, "loss": 0.4028, "step": 1991 }, { "epoch": 0.27, "grad_norm": 2.9551705408566415, "learning_rate": 9.711304610594104e-06, "loss": 0.2614, "step": 1992 }, { "epoch": 0.27, "grad_norm": 3.089825253191854, "learning_rate": 9.710935612908671e-06, "loss": 0.3089, "step": 1993 }, { "epoch": 0.27, "grad_norm": 3.880165962828579, "learning_rate": 9.710566386575e-06, "loss": 0.3637, "step": 1994 }, { "epoch": 0.27, "grad_norm": 3.4564962760525026, "learning_rate": 9.710196931611011e-06, "loss": 0.3101, "step": 1995 }, { "epoch": 0.27, "grad_norm": 3.3013719941351223, "learning_rate": 9.709827248034636e-06, "loss": 0.3645, "step": 1996 }, { "epoch": 0.27, "grad_norm": 3.6489162407866926, "learning_rate": 9.709457335863815e-06, "loss": 0.2827, "step": 1997 }, { "epoch": 0.27, "grad_norm": 3.5897519328653966, "learning_rate": 9.709087195116505e-06, "loss": 0.3166, "step": 1998 }, { "epoch": 0.27, "grad_norm": 2.88192833324684, "learning_rate": 9.708716825810672e-06, "loss": 0.3106, "step": 1999 }, { "epoch": 0.27, "grad_norm": 3.2801726313870168, "learning_rate": 9.708346227964287e-06, "loss": 0.3322, "step": 2000 }, { "epoch": 0.27, "grad_norm": 2.7307886583355376, "learning_rate": 9.707975401595341e-06, "loss": 0.2675, "step": 2001 }, { "epoch": 0.27, "grad_norm": 3.2771390625468775, "learning_rate": 9.707604346721833e-06, "loss": 0.3477, "step": 2002 }, { "epoch": 0.27, "grad_norm": 2.9226967444910072, "learning_rate": 9.707233063361769e-06, "loss": 0.3022, "step": 2003 }, { "epoch": 0.27, "grad_norm": 3.4252023729768064, "learning_rate": 9.706861551533173e-06, "loss": 0.3351, "step": 2004 }, { "epoch": 0.27, "grad_norm": 3.265350322484085, "learning_rate": 9.706489811254073e-06, "loss": 0.3356, "step": 2005 }, { "epoch": 0.27, "grad_norm": 3.0241360259046357, "learning_rate": 9.706117842542517e-06, "loss": 0.3322, "step": 2006 }, { "epoch": 0.27, "grad_norm": 3.3083224273448395, "learning_rate": 9.705745645416553e-06, "loss": 0.348, "step": 2007 }, { "epoch": 0.27, "grad_norm": 3.969569568952289, "learning_rate": 9.705373219894249e-06, "loss": 0.4288, "step": 2008 }, { "epoch": 0.27, "grad_norm": 3.4379620503338417, "learning_rate": 9.705000565993681e-06, "loss": 0.4086, "step": 2009 }, { "epoch": 0.27, "grad_norm": 3.4630212932087616, "learning_rate": 9.704627683732934e-06, "loss": 0.3272, "step": 2010 }, { "epoch": 0.27, "grad_norm": 3.6826598523392846, "learning_rate": 9.704254573130107e-06, "loss": 0.3502, "step": 2011 }, { "epoch": 0.27, "grad_norm": 4.474196032349486, "learning_rate": 9.703881234203309e-06, "loss": 0.3954, "step": 2012 }, { "epoch": 0.27, "grad_norm": 3.3389771114631794, "learning_rate": 9.703507666970662e-06, "loss": 0.2718, "step": 2013 }, { "epoch": 0.27, "grad_norm": 3.156497860884775, "learning_rate": 9.703133871450294e-06, "loss": 0.3472, "step": 2014 }, { "epoch": 0.27, "grad_norm": 3.8782483960800183, "learning_rate": 9.70275984766035e-06, "loss": 0.3178, "step": 2015 }, { "epoch": 0.27, "grad_norm": 3.3157347342795793, "learning_rate": 9.702385595618985e-06, "loss": 0.3709, "step": 2016 }, { "epoch": 0.27, "grad_norm": 3.287656959556907, "learning_rate": 9.702011115344359e-06, "loss": 0.3511, "step": 2017 }, { "epoch": 0.27, "grad_norm": 3.542840314499278, "learning_rate": 9.701636406854652e-06, "loss": 0.4072, "step": 2018 }, { "epoch": 0.27, "grad_norm": 3.3077493127512794, "learning_rate": 9.701261470168046e-06, "loss": 0.3248, "step": 2019 }, { "epoch": 0.27, "grad_norm": 3.454263714340931, "learning_rate": 9.700886305302743e-06, "loss": 0.3064, "step": 2020 }, { "epoch": 0.27, "grad_norm": 3.303356913906008, "learning_rate": 9.70051091227695e-06, "loss": 0.3274, "step": 2021 }, { "epoch": 0.28, "grad_norm": 3.3430899117976374, "learning_rate": 9.70013529110889e-06, "loss": 0.3069, "step": 2022 }, { "epoch": 0.28, "grad_norm": 3.2203560925718246, "learning_rate": 9.699759441816788e-06, "loss": 0.3387, "step": 2023 }, { "epoch": 0.28, "grad_norm": 3.405176804270259, "learning_rate": 9.699383364418891e-06, "loss": 0.3597, "step": 2024 }, { "epoch": 0.28, "grad_norm": 3.8421826210891123, "learning_rate": 9.69900705893345e-06, "loss": 0.4063, "step": 2025 }, { "epoch": 0.28, "grad_norm": 3.240059333975737, "learning_rate": 9.698630525378729e-06, "loss": 0.3698, "step": 2026 }, { "epoch": 0.28, "grad_norm": 3.3644263083368013, "learning_rate": 9.698253763773005e-06, "loss": 0.3249, "step": 2027 }, { "epoch": 0.28, "grad_norm": 3.893170742349812, "learning_rate": 9.697876774134563e-06, "loss": 0.3503, "step": 2028 }, { "epoch": 0.28, "grad_norm": 3.8575597827653705, "learning_rate": 9.6974995564817e-06, "loss": 0.3749, "step": 2029 }, { "epoch": 0.28, "grad_norm": 4.726257316378083, "learning_rate": 9.697122110832727e-06, "loss": 0.3864, "step": 2030 }, { "epoch": 0.28, "grad_norm": 3.2098065827661637, "learning_rate": 9.696744437205961e-06, "loss": 0.3427, "step": 2031 }, { "epoch": 0.28, "grad_norm": 4.2119417581958585, "learning_rate": 9.696366535619735e-06, "loss": 0.4096, "step": 2032 }, { "epoch": 0.28, "grad_norm": 3.4807034244954096, "learning_rate": 9.695988406092388e-06, "loss": 0.3178, "step": 2033 }, { "epoch": 0.28, "grad_norm": 3.7616538029471345, "learning_rate": 9.695610048642275e-06, "loss": 0.3559, "step": 2034 }, { "epoch": 0.28, "grad_norm": 3.460818596206866, "learning_rate": 9.695231463287757e-06, "loss": 0.3305, "step": 2035 }, { "epoch": 0.28, "grad_norm": 3.564576006909528, "learning_rate": 9.694852650047212e-06, "loss": 0.3629, "step": 2036 }, { "epoch": 0.28, "grad_norm": 17.18633253261886, "learning_rate": 9.694473608939024e-06, "loss": 0.3564, "step": 2037 }, { "epoch": 0.28, "grad_norm": 3.1226251286312543, "learning_rate": 9.694094339981593e-06, "loss": 0.3098, "step": 2038 }, { "epoch": 0.28, "grad_norm": 2.971920525405135, "learning_rate": 9.693714843193321e-06, "loss": 0.2665, "step": 2039 }, { "epoch": 0.28, "grad_norm": 2.8925623005009764, "learning_rate": 9.693335118592632e-06, "loss": 0.2686, "step": 2040 }, { "epoch": 0.28, "grad_norm": 3.4279130215187426, "learning_rate": 9.692955166197956e-06, "loss": 0.2996, "step": 2041 }, { "epoch": 0.28, "grad_norm": 3.369964549325946, "learning_rate": 9.692574986027733e-06, "loss": 0.3833, "step": 2042 }, { "epoch": 0.28, "grad_norm": 3.901493184810338, "learning_rate": 9.692194578100415e-06, "loss": 0.3998, "step": 2043 }, { "epoch": 0.28, "grad_norm": 3.53064895506673, "learning_rate": 9.691813942434468e-06, "loss": 0.3174, "step": 2044 }, { "epoch": 0.28, "grad_norm": 3.7381208086599504, "learning_rate": 9.69143307904836e-06, "loss": 0.398, "step": 2045 }, { "epoch": 0.28, "grad_norm": 3.2069657285446174, "learning_rate": 9.691051987960585e-06, "loss": 0.3467, "step": 2046 }, { "epoch": 0.28, "grad_norm": 3.592168530939964, "learning_rate": 9.690670669189632e-06, "loss": 0.3682, "step": 2047 }, { "epoch": 0.28, "grad_norm": 3.169834627182799, "learning_rate": 9.690289122754015e-06, "loss": 0.3548, "step": 2048 }, { "epoch": 0.28, "grad_norm": 3.1238524031904, "learning_rate": 9.689907348672245e-06, "loss": 0.3279, "step": 2049 }, { "epoch": 0.28, "grad_norm": 3.689930665472042, "learning_rate": 9.68952534696286e-06, "loss": 0.3679, "step": 2050 }, { "epoch": 0.28, "grad_norm": 3.631884277753636, "learning_rate": 9.689143117644395e-06, "loss": 0.3592, "step": 2051 }, { "epoch": 0.28, "grad_norm": 3.3945806310763573, "learning_rate": 9.688760660735403e-06, "loss": 0.2931, "step": 2052 }, { "epoch": 0.28, "grad_norm": 3.263981693408903, "learning_rate": 9.688377976254448e-06, "loss": 0.3359, "step": 2053 }, { "epoch": 0.28, "grad_norm": 3.605848580925574, "learning_rate": 9.687995064220102e-06, "loss": 0.3286, "step": 2054 }, { "epoch": 0.28, "grad_norm": 3.1266872436902675, "learning_rate": 9.687611924650952e-06, "loss": 0.3872, "step": 2055 }, { "epoch": 0.28, "grad_norm": 3.356030767642699, "learning_rate": 9.687228557565593e-06, "loss": 0.3491, "step": 2056 }, { "epoch": 0.28, "grad_norm": 3.2283331745616066, "learning_rate": 9.68684496298263e-06, "loss": 0.2973, "step": 2057 }, { "epoch": 0.28, "grad_norm": 2.963589823183119, "learning_rate": 9.686461140920685e-06, "loss": 0.3447, "step": 2058 }, { "epoch": 0.28, "grad_norm": 3.566191607385433, "learning_rate": 9.686077091398384e-06, "loss": 0.3456, "step": 2059 }, { "epoch": 0.28, "grad_norm": 4.064050974832813, "learning_rate": 9.685692814434367e-06, "loss": 0.3994, "step": 2060 }, { "epoch": 0.28, "grad_norm": 3.1315881076001686, "learning_rate": 9.685308310047288e-06, "loss": 0.3218, "step": 2061 }, { "epoch": 0.28, "grad_norm": 4.0427265507527945, "learning_rate": 9.684923578255806e-06, "loss": 0.4404, "step": 2062 }, { "epoch": 0.28, "grad_norm": 3.3343895996776545, "learning_rate": 9.684538619078594e-06, "loss": 0.2728, "step": 2063 }, { "epoch": 0.28, "grad_norm": 3.6023802845652915, "learning_rate": 9.68415343253434e-06, "loss": 0.3838, "step": 2064 }, { "epoch": 0.28, "grad_norm": 3.585041074581868, "learning_rate": 9.683768018641736e-06, "loss": 0.3561, "step": 2065 }, { "epoch": 0.28, "grad_norm": 3.9926955849252668, "learning_rate": 9.68338237741949e-06, "loss": 0.4177, "step": 2066 }, { "epoch": 0.28, "grad_norm": 3.75386316926973, "learning_rate": 9.682996508886318e-06, "loss": 0.3132, "step": 2067 }, { "epoch": 0.28, "grad_norm": 3.4285937194303333, "learning_rate": 9.682610413060948e-06, "loss": 0.2973, "step": 2068 }, { "epoch": 0.28, "grad_norm": 3.4637535185416115, "learning_rate": 9.682224089962122e-06, "loss": 0.3054, "step": 2069 }, { "epoch": 0.28, "grad_norm": 3.90302616773956, "learning_rate": 9.681837539608587e-06, "loss": 0.3592, "step": 2070 }, { "epoch": 0.28, "grad_norm": 3.237191181778077, "learning_rate": 9.681450762019109e-06, "loss": 0.3004, "step": 2071 }, { "epoch": 0.28, "grad_norm": 3.0025089333192274, "learning_rate": 9.681063757212455e-06, "loss": 0.2858, "step": 2072 }, { "epoch": 0.28, "grad_norm": 3.8783267687109646, "learning_rate": 9.680676525207414e-06, "loss": 0.3782, "step": 2073 }, { "epoch": 0.28, "grad_norm": 3.2560077846750852, "learning_rate": 9.680289066022774e-06, "loss": 0.3327, "step": 2074 }, { "epoch": 0.28, "grad_norm": 3.4762071290798446, "learning_rate": 9.679901379677348e-06, "loss": 0.3459, "step": 2075 }, { "epoch": 0.28, "grad_norm": 3.8890216922364402, "learning_rate": 9.679513466189948e-06, "loss": 0.3306, "step": 2076 }, { "epoch": 0.28, "grad_norm": 4.000578696272132, "learning_rate": 9.679125325579402e-06, "loss": 0.4115, "step": 2077 }, { "epoch": 0.28, "grad_norm": 3.0205751780838646, "learning_rate": 9.67873695786455e-06, "loss": 0.2824, "step": 2078 }, { "epoch": 0.28, "grad_norm": 3.665671236743963, "learning_rate": 9.67834836306424e-06, "loss": 0.3135, "step": 2079 }, { "epoch": 0.28, "grad_norm": 3.535692827352044, "learning_rate": 9.677959541197333e-06, "loss": 0.3018, "step": 2080 }, { "epoch": 0.28, "grad_norm": 3.475520520676468, "learning_rate": 9.677570492282703e-06, "loss": 0.3488, "step": 2081 }, { "epoch": 0.28, "grad_norm": 3.653922644967812, "learning_rate": 9.67718121633923e-06, "loss": 0.3432, "step": 2082 }, { "epoch": 0.28, "grad_norm": 3.684139911274434, "learning_rate": 9.676791713385811e-06, "loss": 0.3492, "step": 2083 }, { "epoch": 0.28, "grad_norm": 3.559580215205325, "learning_rate": 9.676401983441346e-06, "loss": 0.3332, "step": 2084 }, { "epoch": 0.28, "grad_norm": 3.73996167846277, "learning_rate": 9.676012026524755e-06, "loss": 0.3398, "step": 2085 }, { "epoch": 0.28, "grad_norm": 4.404907714411046, "learning_rate": 9.675621842654963e-06, "loss": 0.3407, "step": 2086 }, { "epoch": 0.28, "grad_norm": 4.070273254023153, "learning_rate": 9.675231431850907e-06, "loss": 0.3312, "step": 2087 }, { "epoch": 0.28, "grad_norm": 3.560663473096966, "learning_rate": 9.674840794131539e-06, "loss": 0.3529, "step": 2088 }, { "epoch": 0.28, "grad_norm": 3.133294890227466, "learning_rate": 9.674449929515815e-06, "loss": 0.3, "step": 2089 }, { "epoch": 0.28, "grad_norm": 3.2491515130718227, "learning_rate": 9.674058838022707e-06, "loss": 0.3708, "step": 2090 }, { "epoch": 0.28, "grad_norm": 3.6461301931092205, "learning_rate": 9.6736675196712e-06, "loss": 0.2999, "step": 2091 }, { "epoch": 0.28, "grad_norm": 3.4242895455733726, "learning_rate": 9.673275974480282e-06, "loss": 0.3364, "step": 2092 }, { "epoch": 0.28, "grad_norm": 3.8360845162815194, "learning_rate": 9.67288420246896e-06, "loss": 0.3795, "step": 2093 }, { "epoch": 0.28, "grad_norm": 3.653559409075135, "learning_rate": 9.672492203656247e-06, "loss": 0.3169, "step": 2094 }, { "epoch": 0.28, "grad_norm": 3.469609189195369, "learning_rate": 9.672099978061173e-06, "loss": 0.333, "step": 2095 }, { "epoch": 0.29, "grad_norm": 4.3057091469571125, "learning_rate": 9.671707525702769e-06, "loss": 0.405, "step": 2096 }, { "epoch": 0.29, "grad_norm": 2.8051221700028375, "learning_rate": 9.671314846600088e-06, "loss": 0.2973, "step": 2097 }, { "epoch": 0.29, "grad_norm": 3.5218627215180844, "learning_rate": 9.670921940772186e-06, "loss": 0.2929, "step": 2098 }, { "epoch": 0.29, "grad_norm": 3.797092435639217, "learning_rate": 9.670528808238133e-06, "loss": 0.336, "step": 2099 }, { "epoch": 0.29, "grad_norm": 3.2125698402338596, "learning_rate": 9.670135449017012e-06, "loss": 0.3561, "step": 2100 }, { "epoch": 0.29, "grad_norm": 3.1731326520504792, "learning_rate": 9.669741863127912e-06, "loss": 0.2866, "step": 2101 }, { "epoch": 0.29, "grad_norm": 3.230361446714494, "learning_rate": 9.66934805058994e-06, "loss": 0.3132, "step": 2102 }, { "epoch": 0.29, "grad_norm": 3.1394955801715034, "learning_rate": 9.668954011422206e-06, "loss": 0.3134, "step": 2103 }, { "epoch": 0.29, "grad_norm": 3.022105115267478, "learning_rate": 9.668559745643836e-06, "loss": 0.327, "step": 2104 }, { "epoch": 0.29, "grad_norm": 3.5151452687001448, "learning_rate": 9.668165253273967e-06, "loss": 0.3256, "step": 2105 }, { "epoch": 0.29, "grad_norm": 3.4530460944824592, "learning_rate": 9.667770534331746e-06, "loss": 0.3644, "step": 2106 }, { "epoch": 0.29, "grad_norm": 2.9405788740930436, "learning_rate": 9.667375588836329e-06, "loss": 0.2986, "step": 2107 }, { "epoch": 0.29, "grad_norm": 3.6811933325039057, "learning_rate": 9.666980416806888e-06, "loss": 0.3068, "step": 2108 }, { "epoch": 0.29, "grad_norm": 3.8206563079501215, "learning_rate": 9.666585018262598e-06, "loss": 0.2658, "step": 2109 }, { "epoch": 0.29, "grad_norm": 3.515774046407313, "learning_rate": 9.666189393222656e-06, "loss": 0.3211, "step": 2110 }, { "epoch": 0.29, "grad_norm": 3.509243532497938, "learning_rate": 9.66579354170626e-06, "loss": 0.377, "step": 2111 }, { "epoch": 0.29, "grad_norm": 3.0704087286673336, "learning_rate": 9.665397463732623e-06, "loss": 0.3038, "step": 2112 }, { "epoch": 0.29, "grad_norm": 3.432930657295209, "learning_rate": 9.66500115932097e-06, "loss": 0.3025, "step": 2113 }, { "epoch": 0.29, "grad_norm": 3.633457281670093, "learning_rate": 9.664604628490537e-06, "loss": 0.3283, "step": 2114 }, { "epoch": 0.29, "grad_norm": 3.806480518882434, "learning_rate": 9.664207871260567e-06, "loss": 0.364, "step": 2115 }, { "epoch": 0.29, "grad_norm": 4.090816730563407, "learning_rate": 9.66381088765032e-06, "loss": 0.4066, "step": 2116 }, { "epoch": 0.29, "grad_norm": 3.6899928644129365, "learning_rate": 9.66341367767906e-06, "loss": 0.4082, "step": 2117 }, { "epoch": 0.29, "grad_norm": 3.6830101035285834, "learning_rate": 9.66301624136607e-06, "loss": 0.3686, "step": 2118 }, { "epoch": 0.29, "grad_norm": 3.5115179563081074, "learning_rate": 9.662618578730635e-06, "loss": 0.3961, "step": 2119 }, { "epoch": 0.29, "grad_norm": 3.3246605821510276, "learning_rate": 9.662220689792062e-06, "loss": 0.2906, "step": 2120 }, { "epoch": 0.29, "grad_norm": 3.661119943085185, "learning_rate": 9.66182257456966e-06, "loss": 0.3263, "step": 2121 }, { "epoch": 0.29, "grad_norm": 3.336443591589339, "learning_rate": 9.661424233082748e-06, "loss": 0.2727, "step": 2122 }, { "epoch": 0.29, "grad_norm": 3.4884184710259456, "learning_rate": 9.661025665350665e-06, "loss": 0.3289, "step": 2123 }, { "epoch": 0.29, "grad_norm": 3.307996550404924, "learning_rate": 9.660626871392752e-06, "loss": 0.3283, "step": 2124 }, { "epoch": 0.29, "grad_norm": 2.885181569550149, "learning_rate": 9.660227851228367e-06, "loss": 0.301, "step": 2125 }, { "epoch": 0.29, "grad_norm": 3.195394377017159, "learning_rate": 9.659828604876877e-06, "loss": 0.3239, "step": 2126 }, { "epoch": 0.29, "grad_norm": 3.130645995707832, "learning_rate": 9.65942913235766e-06, "loss": 0.2938, "step": 2127 }, { "epoch": 0.29, "grad_norm": 3.389658950955741, "learning_rate": 9.659029433690101e-06, "loss": 0.3432, "step": 2128 }, { "epoch": 0.29, "grad_norm": 3.3127701720960574, "learning_rate": 9.658629508893604e-06, "loss": 0.3417, "step": 2129 }, { "epoch": 0.29, "grad_norm": 3.362949743618081, "learning_rate": 9.658229357987577e-06, "loss": 0.3556, "step": 2130 }, { "epoch": 0.29, "grad_norm": 4.093749013272989, "learning_rate": 9.65782898099144e-06, "loss": 0.3753, "step": 2131 }, { "epoch": 0.29, "grad_norm": 3.914389268560356, "learning_rate": 9.657428377924632e-06, "loss": 0.3536, "step": 2132 }, { "epoch": 0.29, "grad_norm": 3.2096163656014953, "learning_rate": 9.65702754880659e-06, "loss": 0.3656, "step": 2133 }, { "epoch": 0.29, "grad_norm": 3.2825953585084275, "learning_rate": 9.656626493656773e-06, "loss": 0.2737, "step": 2134 }, { "epoch": 0.29, "grad_norm": 4.286913269896477, "learning_rate": 9.656225212494641e-06, "loss": 0.2756, "step": 2135 }, { "epoch": 0.29, "grad_norm": 3.662666646186503, "learning_rate": 9.655823705339676e-06, "loss": 0.3387, "step": 2136 }, { "epoch": 0.29, "grad_norm": 3.763280150133477, "learning_rate": 9.655421972211362e-06, "loss": 0.3183, "step": 2137 }, { "epoch": 0.29, "grad_norm": 3.4922197608137946, "learning_rate": 9.6550200131292e-06, "loss": 0.3139, "step": 2138 }, { "epoch": 0.29, "grad_norm": 3.7971466216282206, "learning_rate": 9.654617828112697e-06, "loss": 0.3299, "step": 2139 }, { "epoch": 0.29, "grad_norm": 3.812691565312145, "learning_rate": 9.654215417181374e-06, "loss": 0.353, "step": 2140 }, { "epoch": 0.29, "grad_norm": 3.679240620207408, "learning_rate": 9.653812780354763e-06, "loss": 0.323, "step": 2141 }, { "epoch": 0.29, "grad_norm": 4.041371199240893, "learning_rate": 9.653409917652406e-06, "loss": 0.4004, "step": 2142 }, { "epoch": 0.29, "grad_norm": 3.3577203653533982, "learning_rate": 9.653006829093854e-06, "loss": 0.2937, "step": 2143 }, { "epoch": 0.29, "grad_norm": 3.4802631153517494, "learning_rate": 9.652603514698674e-06, "loss": 0.3442, "step": 2144 }, { "epoch": 0.29, "grad_norm": 3.404675598020327, "learning_rate": 9.65219997448644e-06, "loss": 0.3248, "step": 2145 }, { "epoch": 0.29, "grad_norm": 6.7169046495372955, "learning_rate": 9.651796208476738e-06, "loss": 0.3767, "step": 2146 }, { "epoch": 0.29, "grad_norm": 4.63003396000059, "learning_rate": 9.651392216689167e-06, "loss": 0.4142, "step": 2147 }, { "epoch": 0.29, "grad_norm": 5.212509078117971, "learning_rate": 9.650987999143331e-06, "loss": 0.3979, "step": 2148 }, { "epoch": 0.29, "grad_norm": 3.5972555278938905, "learning_rate": 9.650583555858852e-06, "loss": 0.3689, "step": 2149 }, { "epoch": 0.29, "grad_norm": 3.2917455042409065, "learning_rate": 9.65017888685536e-06, "loss": 0.3513, "step": 2150 }, { "epoch": 0.29, "grad_norm": 3.3922352114294054, "learning_rate": 9.649773992152495e-06, "loss": 0.3054, "step": 2151 }, { "epoch": 0.29, "grad_norm": 3.078977456478015, "learning_rate": 9.649368871769908e-06, "loss": 0.3031, "step": 2152 }, { "epoch": 0.29, "grad_norm": 3.306050495755109, "learning_rate": 9.648963525727262e-06, "loss": 0.3538, "step": 2153 }, { "epoch": 0.29, "grad_norm": 3.3934990069387685, "learning_rate": 9.648557954044234e-06, "loss": 0.3446, "step": 2154 }, { "epoch": 0.29, "grad_norm": 3.441812419416839, "learning_rate": 9.648152156740502e-06, "loss": 0.3674, "step": 2155 }, { "epoch": 0.29, "grad_norm": 3.398073219339027, "learning_rate": 9.64774613383577e-06, "loss": 0.3784, "step": 2156 }, { "epoch": 0.29, "grad_norm": 3.096487576787053, "learning_rate": 9.647339885349736e-06, "loss": 0.2777, "step": 2157 }, { "epoch": 0.29, "grad_norm": 3.2274077650489064, "learning_rate": 9.646933411302124e-06, "loss": 0.3336, "step": 2158 }, { "epoch": 0.29, "grad_norm": 3.173408253400741, "learning_rate": 9.64652671171266e-06, "loss": 0.2707, "step": 2159 }, { "epoch": 0.29, "grad_norm": 3.0674948008594276, "learning_rate": 9.646119786601083e-06, "loss": 0.2961, "step": 2160 }, { "epoch": 0.29, "grad_norm": 3.5778048057155383, "learning_rate": 9.645712635987146e-06, "loss": 0.3607, "step": 2161 }, { "epoch": 0.29, "grad_norm": 3.2680874078096127, "learning_rate": 9.645305259890606e-06, "loss": 0.3289, "step": 2162 }, { "epoch": 0.29, "grad_norm": 3.168440091323415, "learning_rate": 9.644897658331239e-06, "loss": 0.3357, "step": 2163 }, { "epoch": 0.29, "grad_norm": 3.7350998005541913, "learning_rate": 9.644489831328825e-06, "loss": 0.3047, "step": 2164 }, { "epoch": 0.29, "grad_norm": 3.5434545452986472, "learning_rate": 9.644081778903161e-06, "loss": 0.3441, "step": 2165 }, { "epoch": 0.29, "grad_norm": 4.160094138100419, "learning_rate": 9.643673501074051e-06, "loss": 0.3557, "step": 2166 }, { "epoch": 0.29, "grad_norm": 3.3361674002887196, "learning_rate": 9.643264997861312e-06, "loss": 0.2982, "step": 2167 }, { "epoch": 0.29, "grad_norm": 3.6804321874309163, "learning_rate": 9.642856269284768e-06, "loss": 0.337, "step": 2168 }, { "epoch": 0.3, "grad_norm": 4.272423564715503, "learning_rate": 9.64244731536426e-06, "loss": 0.3067, "step": 2169 }, { "epoch": 0.3, "grad_norm": 3.4398784499743207, "learning_rate": 9.642038136119635e-06, "loss": 0.3286, "step": 2170 }, { "epoch": 0.3, "grad_norm": 3.4030887067466966, "learning_rate": 9.641628731570757e-06, "loss": 0.2964, "step": 2171 }, { "epoch": 0.3, "grad_norm": 3.0616658979357405, "learning_rate": 9.641219101737489e-06, "loss": 0.2712, "step": 2172 }, { "epoch": 0.3, "grad_norm": 4.106843472065243, "learning_rate": 9.640809246639718e-06, "loss": 0.3257, "step": 2173 }, { "epoch": 0.3, "grad_norm": 4.114748297347267, "learning_rate": 9.640399166297335e-06, "loss": 0.3131, "step": 2174 }, { "epoch": 0.3, "grad_norm": 3.7296880922436233, "learning_rate": 9.639988860730246e-06, "loss": 0.3114, "step": 2175 }, { "epoch": 0.3, "grad_norm": 3.374212612741855, "learning_rate": 9.63957832995836e-06, "loss": 0.2935, "step": 2176 }, { "epoch": 0.3, "grad_norm": 3.536617316914755, "learning_rate": 9.639167574001608e-06, "loss": 0.3422, "step": 2177 }, { "epoch": 0.3, "grad_norm": 3.6682089965715385, "learning_rate": 9.638756592879923e-06, "loss": 0.2919, "step": 2178 }, { "epoch": 0.3, "grad_norm": 2.9312811480972223, "learning_rate": 9.638345386613254e-06, "loss": 0.2779, "step": 2179 }, { "epoch": 0.3, "grad_norm": 3.679955228218515, "learning_rate": 9.637933955221559e-06, "loss": 0.4121, "step": 2180 }, { "epoch": 0.3, "grad_norm": 4.06493003489283, "learning_rate": 9.637522298724806e-06, "loss": 0.3472, "step": 2181 }, { "epoch": 0.3, "grad_norm": 3.350457744169883, "learning_rate": 9.637110417142975e-06, "loss": 0.3587, "step": 2182 }, { "epoch": 0.3, "grad_norm": 3.5406312925879138, "learning_rate": 9.636698310496056e-06, "loss": 0.3354, "step": 2183 }, { "epoch": 0.3, "grad_norm": 3.107263930042939, "learning_rate": 9.636285978804054e-06, "loss": 0.3209, "step": 2184 }, { "epoch": 0.3, "grad_norm": 3.570398040515925, "learning_rate": 9.635873422086982e-06, "loss": 0.3841, "step": 2185 }, { "epoch": 0.3, "grad_norm": 3.3245226541563744, "learning_rate": 9.635460640364858e-06, "loss": 0.3297, "step": 2186 }, { "epoch": 0.3, "grad_norm": 3.826976945515226, "learning_rate": 9.635047633657723e-06, "loss": 0.3516, "step": 2187 }, { "epoch": 0.3, "grad_norm": 3.6809602502764993, "learning_rate": 9.634634401985618e-06, "loss": 0.32, "step": 2188 }, { "epoch": 0.3, "grad_norm": 2.8917109513745793, "learning_rate": 9.634220945368603e-06, "loss": 0.2908, "step": 2189 }, { "epoch": 0.3, "grad_norm": 2.5515917767238823, "learning_rate": 9.633807263826745e-06, "loss": 0.2536, "step": 2190 }, { "epoch": 0.3, "grad_norm": 3.369997082150038, "learning_rate": 9.633393357380118e-06, "loss": 0.3435, "step": 2191 }, { "epoch": 0.3, "grad_norm": 3.7577604805637552, "learning_rate": 9.632979226048816e-06, "loss": 0.4014, "step": 2192 }, { "epoch": 0.3, "grad_norm": 3.757590182870404, "learning_rate": 9.632564869852936e-06, "loss": 0.3677, "step": 2193 }, { "epoch": 0.3, "grad_norm": 3.2252232305247905, "learning_rate": 9.632150288812593e-06, "loss": 0.3025, "step": 2194 }, { "epoch": 0.3, "grad_norm": 3.3124846295249726, "learning_rate": 9.631735482947904e-06, "loss": 0.3533, "step": 2195 }, { "epoch": 0.3, "grad_norm": 3.2511111708800375, "learning_rate": 9.631320452279006e-06, "loss": 0.3898, "step": 2196 }, { "epoch": 0.3, "grad_norm": 3.395371618421636, "learning_rate": 9.630905196826039e-06, "loss": 0.3294, "step": 2197 }, { "epoch": 0.3, "grad_norm": 3.810165607797084, "learning_rate": 9.630489716609162e-06, "loss": 0.4134, "step": 2198 }, { "epoch": 0.3, "grad_norm": 11.290217115795649, "learning_rate": 9.630074011648536e-06, "loss": 0.3194, "step": 2199 }, { "epoch": 0.3, "grad_norm": 3.5062717716535845, "learning_rate": 9.629658081964341e-06, "loss": 0.22, "step": 2200 }, { "epoch": 0.3, "grad_norm": 2.8710861108641144, "learning_rate": 9.629241927576764e-06, "loss": 0.3066, "step": 2201 }, { "epoch": 0.3, "grad_norm": 3.0446320617259257, "learning_rate": 9.628825548506002e-06, "loss": 0.3305, "step": 2202 }, { "epoch": 0.3, "grad_norm": 3.4206035561387256, "learning_rate": 9.628408944772265e-06, "loss": 0.2938, "step": 2203 }, { "epoch": 0.3, "grad_norm": 3.5845055425579995, "learning_rate": 9.627992116395773e-06, "loss": 0.3506, "step": 2204 }, { "epoch": 0.3, "grad_norm": 2.973898529079843, "learning_rate": 9.627575063396756e-06, "loss": 0.3197, "step": 2205 }, { "epoch": 0.3, "grad_norm": 3.627740748178962, "learning_rate": 9.627157785795458e-06, "loss": 0.3588, "step": 2206 }, { "epoch": 0.3, "grad_norm": 3.2038806870548404, "learning_rate": 9.62674028361213e-06, "loss": 0.3354, "step": 2207 }, { "epoch": 0.3, "grad_norm": 3.5751578253527807, "learning_rate": 9.626322556867035e-06, "loss": 0.3408, "step": 2208 }, { "epoch": 0.3, "grad_norm": 3.5229188231399156, "learning_rate": 9.625904605580452e-06, "loss": 0.2889, "step": 2209 }, { "epoch": 0.3, "grad_norm": 3.3036632694617944, "learning_rate": 9.625486429772662e-06, "loss": 0.3686, "step": 2210 }, { "epoch": 0.3, "grad_norm": 2.912962976200562, "learning_rate": 9.625068029463963e-06, "loss": 0.3129, "step": 2211 }, { "epoch": 0.3, "grad_norm": 3.3622211844784466, "learning_rate": 9.624649404674661e-06, "loss": 0.3041, "step": 2212 }, { "epoch": 0.3, "grad_norm": 3.411389621874764, "learning_rate": 9.624230555425078e-06, "loss": 0.3665, "step": 2213 }, { "epoch": 0.3, "grad_norm": 4.224134005606847, "learning_rate": 9.62381148173554e-06, "loss": 0.3195, "step": 2214 }, { "epoch": 0.3, "grad_norm": 3.869235500282523, "learning_rate": 9.623392183626387e-06, "loss": 0.351, "step": 2215 }, { "epoch": 0.3, "grad_norm": 3.0129259271574935, "learning_rate": 9.622972661117971e-06, "loss": 0.3188, "step": 2216 }, { "epoch": 0.3, "grad_norm": 3.1726376518917334, "learning_rate": 9.622552914230655e-06, "loss": 0.3077, "step": 2217 }, { "epoch": 0.3, "grad_norm": 4.129613196359941, "learning_rate": 9.622132942984807e-06, "loss": 0.3332, "step": 2218 }, { "epoch": 0.3, "grad_norm": 4.128671653907828, "learning_rate": 9.621712747400815e-06, "loss": 0.375, "step": 2219 }, { "epoch": 0.3, "grad_norm": 3.549582806354396, "learning_rate": 9.621292327499072e-06, "loss": 0.3473, "step": 2220 }, { "epoch": 0.3, "grad_norm": 3.3903112777506097, "learning_rate": 9.620871683299983e-06, "loss": 0.3124, "step": 2221 }, { "epoch": 0.3, "grad_norm": 3.1650459903370924, "learning_rate": 9.620450814823966e-06, "loss": 0.333, "step": 2222 }, { "epoch": 0.3, "grad_norm": 3.1357106701330024, "learning_rate": 9.620029722091446e-06, "loss": 0.3133, "step": 2223 }, { "epoch": 0.3, "grad_norm": 4.250781489500259, "learning_rate": 9.61960840512286e-06, "loss": 0.3932, "step": 2224 }, { "epoch": 0.3, "grad_norm": 3.7095607213240855, "learning_rate": 9.61918686393866e-06, "loss": 0.3354, "step": 2225 }, { "epoch": 0.3, "grad_norm": 3.6144951438577166, "learning_rate": 9.618765098559306e-06, "loss": 0.401, "step": 2226 }, { "epoch": 0.3, "grad_norm": 3.512791548737835, "learning_rate": 9.618343109005266e-06, "loss": 0.3423, "step": 2227 }, { "epoch": 0.3, "grad_norm": 3.7297049301521925, "learning_rate": 9.617920895297023e-06, "loss": 0.3183, "step": 2228 }, { "epoch": 0.3, "grad_norm": 3.562875109979445, "learning_rate": 9.617498457455068e-06, "loss": 0.3053, "step": 2229 }, { "epoch": 0.3, "grad_norm": 3.2045054607850956, "learning_rate": 9.617075795499906e-06, "loss": 0.3628, "step": 2230 }, { "epoch": 0.3, "grad_norm": 3.5474897597628017, "learning_rate": 9.61665290945205e-06, "loss": 0.3888, "step": 2231 }, { "epoch": 0.3, "grad_norm": 3.7900494206991806, "learning_rate": 9.616229799332026e-06, "loss": 0.3852, "step": 2232 }, { "epoch": 0.3, "grad_norm": 4.088880117878681, "learning_rate": 9.615806465160369e-06, "loss": 0.3604, "step": 2233 }, { "epoch": 0.3, "grad_norm": 3.6284615505163202, "learning_rate": 9.615382906957627e-06, "loss": 0.2983, "step": 2234 }, { "epoch": 0.3, "grad_norm": 3.345587141073643, "learning_rate": 9.614959124744356e-06, "loss": 0.3015, "step": 2235 }, { "epoch": 0.3, "grad_norm": 3.5844306266850365, "learning_rate": 9.614535118541126e-06, "loss": 0.3526, "step": 2236 }, { "epoch": 0.3, "grad_norm": 3.6174786773288954, "learning_rate": 9.614110888368515e-06, "loss": 0.3682, "step": 2237 }, { "epoch": 0.3, "grad_norm": 2.818282834675107, "learning_rate": 9.613686434247115e-06, "loss": 0.3274, "step": 2238 }, { "epoch": 0.3, "grad_norm": 3.0885398918928852, "learning_rate": 9.613261756197525e-06, "loss": 0.2949, "step": 2239 }, { "epoch": 0.3, "grad_norm": 4.243260379766727, "learning_rate": 9.61283685424036e-06, "loss": 0.3992, "step": 2240 }, { "epoch": 0.3, "grad_norm": 3.4553609473474283, "learning_rate": 9.61241172839624e-06, "loss": 0.3907, "step": 2241 }, { "epoch": 0.3, "grad_norm": 3.5964084799733764, "learning_rate": 9.6119863786858e-06, "loss": 0.3093, "step": 2242 }, { "epoch": 0.31, "grad_norm": 3.617176801878641, "learning_rate": 9.611560805129685e-06, "loss": 0.3312, "step": 2243 }, { "epoch": 0.31, "grad_norm": 3.3417441571381445, "learning_rate": 9.61113500774855e-06, "loss": 0.3502, "step": 2244 }, { "epoch": 0.31, "grad_norm": 3.445079969616812, "learning_rate": 9.610708986563061e-06, "loss": 0.351, "step": 2245 }, { "epoch": 0.31, "grad_norm": 3.277573868171076, "learning_rate": 9.610282741593896e-06, "loss": 0.379, "step": 2246 }, { "epoch": 0.31, "grad_norm": 3.4539854325587966, "learning_rate": 9.609856272861742e-06, "loss": 0.3549, "step": 2247 }, { "epoch": 0.31, "grad_norm": 3.3612379597000044, "learning_rate": 9.6094295803873e-06, "loss": 0.3355, "step": 2248 }, { "epoch": 0.31, "grad_norm": 3.4223087109890082, "learning_rate": 9.609002664191276e-06, "loss": 0.3144, "step": 2249 }, { "epoch": 0.31, "grad_norm": 3.8927903819414484, "learning_rate": 9.608575524294395e-06, "loss": 0.3903, "step": 2250 }, { "epoch": 0.31, "grad_norm": 4.167045133857706, "learning_rate": 9.608148160717385e-06, "loss": 0.3286, "step": 2251 }, { "epoch": 0.31, "grad_norm": 3.0931531994417196, "learning_rate": 9.607720573480991e-06, "loss": 0.3185, "step": 2252 }, { "epoch": 0.31, "grad_norm": 3.4719573800157404, "learning_rate": 9.607292762605965e-06, "loss": 0.3385, "step": 2253 }, { "epoch": 0.31, "grad_norm": 3.784292435676924, "learning_rate": 9.606864728113072e-06, "loss": 0.3804, "step": 2254 }, { "epoch": 0.31, "grad_norm": 3.37669124183241, "learning_rate": 9.606436470023084e-06, "loss": 0.3579, "step": 2255 }, { "epoch": 0.31, "grad_norm": 3.247801077946735, "learning_rate": 9.60600798835679e-06, "loss": 0.2746, "step": 2256 }, { "epoch": 0.31, "grad_norm": 3.950863623089795, "learning_rate": 9.605579283134985e-06, "loss": 0.3673, "step": 2257 }, { "epoch": 0.31, "grad_norm": 3.434446891524887, "learning_rate": 9.605150354378477e-06, "loss": 0.3016, "step": 2258 }, { "epoch": 0.31, "grad_norm": 3.003501346080088, "learning_rate": 9.604721202108084e-06, "loss": 0.2662, "step": 2259 }, { "epoch": 0.31, "grad_norm": 3.577600383418439, "learning_rate": 9.604291826344638e-06, "loss": 0.3292, "step": 2260 }, { "epoch": 0.31, "grad_norm": 3.03924927662915, "learning_rate": 9.603862227108973e-06, "loss": 0.3064, "step": 2261 }, { "epoch": 0.31, "grad_norm": 3.899359819191869, "learning_rate": 9.603432404421947e-06, "loss": 0.3544, "step": 2262 }, { "epoch": 0.31, "grad_norm": 3.4493922310454086, "learning_rate": 9.603002358304415e-06, "loss": 0.2893, "step": 2263 }, { "epoch": 0.31, "grad_norm": 3.3017456170053765, "learning_rate": 9.602572088777254e-06, "loss": 0.3166, "step": 2264 }, { "epoch": 0.31, "grad_norm": 3.5102360288328107, "learning_rate": 9.602141595861345e-06, "loss": 0.3056, "step": 2265 }, { "epoch": 0.31, "grad_norm": 3.9486428326517964, "learning_rate": 9.601710879577583e-06, "loss": 0.3367, "step": 2266 }, { "epoch": 0.31, "grad_norm": 2.9634746355343395, "learning_rate": 9.601279939946874e-06, "loss": 0.282, "step": 2267 }, { "epoch": 0.31, "grad_norm": 3.8620972323036273, "learning_rate": 9.600848776990135e-06, "loss": 0.3892, "step": 2268 }, { "epoch": 0.31, "grad_norm": 4.032605783482466, "learning_rate": 9.60041739072829e-06, "loss": 0.4416, "step": 2269 }, { "epoch": 0.31, "grad_norm": 3.56097741862032, "learning_rate": 9.599985781182277e-06, "loss": 0.2905, "step": 2270 }, { "epoch": 0.31, "grad_norm": 3.991423443923368, "learning_rate": 9.599553948373047e-06, "loss": 0.3532, "step": 2271 }, { "epoch": 0.31, "grad_norm": 4.05221093146065, "learning_rate": 9.599121892321554e-06, "loss": 0.3746, "step": 2272 }, { "epoch": 0.31, "grad_norm": 3.5449439869678274, "learning_rate": 9.598689613048774e-06, "loss": 0.3025, "step": 2273 }, { "epoch": 0.31, "grad_norm": 3.6519655290271973, "learning_rate": 9.598257110575686e-06, "loss": 0.3699, "step": 2274 }, { "epoch": 0.31, "grad_norm": 3.8835128306226068, "learning_rate": 9.59782438492328e-06, "loss": 0.3632, "step": 2275 }, { "epoch": 0.31, "grad_norm": 3.9513153352971253, "learning_rate": 9.59739143611256e-06, "loss": 0.301, "step": 2276 }, { "epoch": 0.31, "grad_norm": 3.690760802441199, "learning_rate": 9.59695826416454e-06, "loss": 0.3126, "step": 2277 }, { "epoch": 0.31, "grad_norm": 3.9197376339722196, "learning_rate": 9.596524869100242e-06, "loss": 0.4266, "step": 2278 }, { "epoch": 0.31, "grad_norm": 3.368788005125385, "learning_rate": 9.596091250940705e-06, "loss": 0.3542, "step": 2279 }, { "epoch": 0.31, "grad_norm": 3.2446827627539943, "learning_rate": 9.59565740970697e-06, "loss": 0.3229, "step": 2280 }, { "epoch": 0.31, "grad_norm": 3.3274915060240975, "learning_rate": 9.595223345420097e-06, "loss": 0.3558, "step": 2281 }, { "epoch": 0.31, "grad_norm": 3.548928911048484, "learning_rate": 9.594789058101154e-06, "loss": 0.3137, "step": 2282 }, { "epoch": 0.31, "grad_norm": 3.6255090302783852, "learning_rate": 9.594354547771218e-06, "loss": 0.3551, "step": 2283 }, { "epoch": 0.31, "grad_norm": 3.6967326254395063, "learning_rate": 9.593919814451378e-06, "loss": 0.2753, "step": 2284 }, { "epoch": 0.31, "grad_norm": 3.969418687344694, "learning_rate": 9.593484858162737e-06, "loss": 0.319, "step": 2285 }, { "epoch": 0.31, "grad_norm": 3.2359544808758836, "learning_rate": 9.593049678926401e-06, "loss": 0.288, "step": 2286 }, { "epoch": 0.31, "grad_norm": 3.1663724928184025, "learning_rate": 9.592614276763494e-06, "loss": 0.301, "step": 2287 }, { "epoch": 0.31, "grad_norm": 3.4456262979239476, "learning_rate": 9.59217865169515e-06, "loss": 0.3444, "step": 2288 }, { "epoch": 0.31, "grad_norm": 3.4418556814750536, "learning_rate": 9.59174280374251e-06, "loss": 0.303, "step": 2289 }, { "epoch": 0.31, "grad_norm": 3.9730946758949486, "learning_rate": 9.59130673292673e-06, "loss": 0.3738, "step": 2290 }, { "epoch": 0.31, "grad_norm": 3.2661310957030554, "learning_rate": 9.590870439268975e-06, "loss": 0.3219, "step": 2291 }, { "epoch": 0.31, "grad_norm": 3.36353572736976, "learning_rate": 9.590433922790418e-06, "loss": 0.3162, "step": 2292 }, { "epoch": 0.31, "grad_norm": 3.2737716962904857, "learning_rate": 9.58999718351225e-06, "loss": 0.3398, "step": 2293 }, { "epoch": 0.31, "grad_norm": 3.5731877882771887, "learning_rate": 9.589560221455666e-06, "loss": 0.357, "step": 2294 }, { "epoch": 0.31, "grad_norm": 3.586854742829287, "learning_rate": 9.589123036641872e-06, "loss": 0.3851, "step": 2295 }, { "epoch": 0.31, "grad_norm": 2.9999012721639264, "learning_rate": 9.588685629092089e-06, "loss": 0.2942, "step": 2296 }, { "epoch": 0.31, "grad_norm": 3.971218705256527, "learning_rate": 9.58824799882755e-06, "loss": 0.3389, "step": 2297 }, { "epoch": 0.31, "grad_norm": 3.759512797066232, "learning_rate": 9.587810145869491e-06, "loss": 0.3586, "step": 2298 }, { "epoch": 0.31, "grad_norm": 4.00277585117829, "learning_rate": 9.587372070239164e-06, "loss": 0.3552, "step": 2299 }, { "epoch": 0.31, "grad_norm": 3.820707123850192, "learning_rate": 9.586933771957835e-06, "loss": 0.309, "step": 2300 }, { "epoch": 0.31, "grad_norm": 3.3309604548151377, "learning_rate": 9.586495251046774e-06, "loss": 0.2796, "step": 2301 }, { "epoch": 0.31, "grad_norm": 4.001696332104719, "learning_rate": 9.586056507527266e-06, "loss": 0.3797, "step": 2302 }, { "epoch": 0.31, "grad_norm": 3.322571897412361, "learning_rate": 9.585617541420604e-06, "loss": 0.3059, "step": 2303 }, { "epoch": 0.31, "grad_norm": 3.3759663587865236, "learning_rate": 9.585178352748097e-06, "loss": 0.3232, "step": 2304 }, { "epoch": 0.31, "grad_norm": 4.1846468034957045, "learning_rate": 9.584738941531055e-06, "loss": 0.3565, "step": 2305 }, { "epoch": 0.31, "grad_norm": 3.4383874965796553, "learning_rate": 9.584299307790812e-06, "loss": 0.307, "step": 2306 }, { "epoch": 0.31, "grad_norm": 3.356099201230768, "learning_rate": 9.583859451548703e-06, "loss": 0.3294, "step": 2307 }, { "epoch": 0.31, "grad_norm": 5.425636877588579, "learning_rate": 9.583419372826077e-06, "loss": 0.4041, "step": 2308 }, { "epoch": 0.31, "grad_norm": 4.055394384670599, "learning_rate": 9.582979071644292e-06, "loss": 0.4053, "step": 2309 }, { "epoch": 0.31, "grad_norm": 3.830263114712799, "learning_rate": 9.58253854802472e-06, "loss": 0.3517, "step": 2310 }, { "epoch": 0.31, "grad_norm": 3.4277476094720414, "learning_rate": 9.58209780198874e-06, "loss": 0.3724, "step": 2311 }, { "epoch": 0.31, "grad_norm": 3.608965134571338, "learning_rate": 9.581656833557749e-06, "loss": 0.3583, "step": 2312 }, { "epoch": 0.31, "grad_norm": 3.489289055519575, "learning_rate": 9.581215642753144e-06, "loss": 0.3284, "step": 2313 }, { "epoch": 0.31, "grad_norm": 3.4693239537735763, "learning_rate": 9.580774229596341e-06, "loss": 0.3147, "step": 2314 }, { "epoch": 0.31, "grad_norm": 3.1159791505155154, "learning_rate": 9.580332594108763e-06, "loss": 0.3213, "step": 2315 }, { "epoch": 0.32, "grad_norm": 3.420954141149516, "learning_rate": 9.579890736311847e-06, "loss": 0.3537, "step": 2316 }, { "epoch": 0.32, "grad_norm": 3.638858273011892, "learning_rate": 9.57944865622704e-06, "loss": 0.337, "step": 2317 }, { "epoch": 0.32, "grad_norm": 3.3213792156060027, "learning_rate": 9.579006353875792e-06, "loss": 0.3679, "step": 2318 }, { "epoch": 0.32, "grad_norm": 3.569919271502192, "learning_rate": 9.578563829279578e-06, "loss": 0.3549, "step": 2319 }, { "epoch": 0.32, "grad_norm": 3.252599086439841, "learning_rate": 9.578121082459872e-06, "loss": 0.3082, "step": 2320 }, { "epoch": 0.32, "grad_norm": 4.069723991695278, "learning_rate": 9.577678113438164e-06, "loss": 0.4149, "step": 2321 }, { "epoch": 0.32, "grad_norm": 3.7691288005806762, "learning_rate": 9.577234922235954e-06, "loss": 0.3987, "step": 2322 }, { "epoch": 0.32, "grad_norm": 3.504123710409951, "learning_rate": 9.576791508874752e-06, "loss": 0.4094, "step": 2323 }, { "epoch": 0.32, "grad_norm": 3.401778476923812, "learning_rate": 9.57634787337608e-06, "loss": 0.2837, "step": 2324 }, { "epoch": 0.32, "grad_norm": 3.7138826114780152, "learning_rate": 9.57590401576147e-06, "loss": 0.3504, "step": 2325 }, { "epoch": 0.32, "grad_norm": 3.6398111206903168, "learning_rate": 9.575459936052467e-06, "loss": 0.362, "step": 2326 }, { "epoch": 0.32, "grad_norm": 3.251396636515415, "learning_rate": 9.575015634270619e-06, "loss": 0.2464, "step": 2327 }, { "epoch": 0.32, "grad_norm": 3.0927379367557655, "learning_rate": 9.574571110437496e-06, "loss": 0.3295, "step": 2328 }, { "epoch": 0.32, "grad_norm": 3.058209014909298, "learning_rate": 9.574126364574671e-06, "loss": 0.3549, "step": 2329 }, { "epoch": 0.32, "grad_norm": 3.683982505705345, "learning_rate": 9.573681396703731e-06, "loss": 0.3615, "step": 2330 }, { "epoch": 0.32, "grad_norm": 3.462474083301709, "learning_rate": 9.573236206846271e-06, "loss": 0.4061, "step": 2331 }, { "epoch": 0.32, "grad_norm": 3.583063667037495, "learning_rate": 9.5727907950239e-06, "loss": 0.3834, "step": 2332 }, { "epoch": 0.32, "grad_norm": 3.861529747346894, "learning_rate": 9.572345161258235e-06, "loss": 0.3771, "step": 2333 }, { "epoch": 0.32, "grad_norm": 4.759124173718573, "learning_rate": 9.571899305570906e-06, "loss": 0.3592, "step": 2334 }, { "epoch": 0.32, "grad_norm": 3.198412091523014, "learning_rate": 9.571453227983556e-06, "loss": 0.3215, "step": 2335 }, { "epoch": 0.32, "grad_norm": 3.2400680936953097, "learning_rate": 9.57100692851783e-06, "loss": 0.2824, "step": 2336 }, { "epoch": 0.32, "grad_norm": 3.362644049659171, "learning_rate": 9.570560407195392e-06, "loss": 0.3354, "step": 2337 }, { "epoch": 0.32, "grad_norm": 4.404398810130689, "learning_rate": 9.570113664037914e-06, "loss": 0.3111, "step": 2338 }, { "epoch": 0.32, "grad_norm": 3.8071458116850487, "learning_rate": 9.569666699067082e-06, "loss": 0.3048, "step": 2339 }, { "epoch": 0.32, "grad_norm": 3.5748916231810917, "learning_rate": 9.569219512304583e-06, "loss": 0.3505, "step": 2340 }, { "epoch": 0.32, "grad_norm": 3.604214286373558, "learning_rate": 9.568772103772128e-06, "loss": 0.3253, "step": 2341 }, { "epoch": 0.32, "grad_norm": 3.902961656804498, "learning_rate": 9.568324473491431e-06, "loss": 0.4033, "step": 2342 }, { "epoch": 0.32, "grad_norm": 3.8569953372828265, "learning_rate": 9.567876621484214e-06, "loss": 0.4029, "step": 2343 }, { "epoch": 0.32, "grad_norm": 3.5122483296301925, "learning_rate": 9.567428547772217e-06, "loss": 0.3327, "step": 2344 }, { "epoch": 0.32, "grad_norm": 3.282345478238041, "learning_rate": 9.566980252377187e-06, "loss": 0.3014, "step": 2345 }, { "epoch": 0.32, "grad_norm": 3.6172296109246833, "learning_rate": 9.566531735320883e-06, "loss": 0.3171, "step": 2346 }, { "epoch": 0.32, "grad_norm": 3.8802306335279813, "learning_rate": 9.566082996625072e-06, "loss": 0.331, "step": 2347 }, { "epoch": 0.32, "grad_norm": 3.2124474837486257, "learning_rate": 9.565634036311537e-06, "loss": 0.2742, "step": 2348 }, { "epoch": 0.32, "grad_norm": 3.513271705901283, "learning_rate": 9.565184854402067e-06, "loss": 0.3124, "step": 2349 }, { "epoch": 0.32, "grad_norm": 4.042859878146324, "learning_rate": 9.564735450918461e-06, "loss": 0.3298, "step": 2350 }, { "epoch": 0.32, "grad_norm": 3.510045845584844, "learning_rate": 9.564285825882536e-06, "loss": 0.3642, "step": 2351 }, { "epoch": 0.32, "grad_norm": 4.488687755622603, "learning_rate": 9.56383597931611e-06, "loss": 0.4835, "step": 2352 }, { "epoch": 0.32, "grad_norm": 4.225784247219549, "learning_rate": 9.563385911241019e-06, "loss": 0.3682, "step": 2353 }, { "epoch": 0.32, "grad_norm": 3.549993444966526, "learning_rate": 9.562935621679105e-06, "loss": 0.2773, "step": 2354 }, { "epoch": 0.32, "grad_norm": 3.7153221921731467, "learning_rate": 9.562485110652228e-06, "loss": 0.3263, "step": 2355 }, { "epoch": 0.32, "grad_norm": 4.594342894056878, "learning_rate": 9.56203437818225e-06, "loss": 0.3596, "step": 2356 }, { "epoch": 0.32, "grad_norm": 3.8308416522905304, "learning_rate": 9.561583424291048e-06, "loss": 0.3294, "step": 2357 }, { "epoch": 0.32, "grad_norm": 3.541919867978602, "learning_rate": 9.56113224900051e-06, "loss": 0.3148, "step": 2358 }, { "epoch": 0.32, "grad_norm": 3.190494463211217, "learning_rate": 9.560680852332534e-06, "loss": 0.3077, "step": 2359 }, { "epoch": 0.32, "grad_norm": 3.2026166931162816, "learning_rate": 9.56022923430903e-06, "loss": 0.2944, "step": 2360 }, { "epoch": 0.32, "grad_norm": 3.6027918712187836, "learning_rate": 9.559777394951915e-06, "loss": 0.3575, "step": 2361 }, { "epoch": 0.32, "grad_norm": 2.8923309116357654, "learning_rate": 9.55932533428312e-06, "loss": 0.2763, "step": 2362 }, { "epoch": 0.32, "grad_norm": 3.5759221461347463, "learning_rate": 9.558873052324587e-06, "loss": 0.3094, "step": 2363 }, { "epoch": 0.32, "grad_norm": 2.9773332120522573, "learning_rate": 9.558420549098269e-06, "loss": 0.2833, "step": 2364 }, { "epoch": 0.32, "grad_norm": 3.9003483619055848, "learning_rate": 9.557967824626124e-06, "loss": 0.3029, "step": 2365 }, { "epoch": 0.32, "grad_norm": 4.2072342134968705, "learning_rate": 9.557514878930129e-06, "loss": 0.3543, "step": 2366 }, { "epoch": 0.32, "grad_norm": 3.385580133942015, "learning_rate": 9.557061712032269e-06, "loss": 0.3347, "step": 2367 }, { "epoch": 0.32, "grad_norm": 3.702221947673088, "learning_rate": 9.556608323954537e-06, "loss": 0.339, "step": 2368 }, { "epoch": 0.32, "grad_norm": 3.6226490432820193, "learning_rate": 9.556154714718938e-06, "loss": 0.3118, "step": 2369 }, { "epoch": 0.32, "grad_norm": 3.3632890381412595, "learning_rate": 9.555700884347487e-06, "loss": 0.323, "step": 2370 }, { "epoch": 0.32, "grad_norm": 3.362613250947677, "learning_rate": 9.555246832862214e-06, "loss": 0.2962, "step": 2371 }, { "epoch": 0.32, "grad_norm": 3.59895252376486, "learning_rate": 9.554792560285152e-06, "loss": 0.3831, "step": 2372 }, { "epoch": 0.32, "grad_norm": 3.452988144320104, "learning_rate": 9.554338066638357e-06, "loss": 0.3101, "step": 2373 }, { "epoch": 0.32, "grad_norm": 3.8003926360021567, "learning_rate": 9.553883351943882e-06, "loss": 0.3621, "step": 2374 }, { "epoch": 0.32, "grad_norm": 3.5067071497045346, "learning_rate": 9.5534284162238e-06, "loss": 0.3228, "step": 2375 }, { "epoch": 0.32, "grad_norm": 3.4673788191110386, "learning_rate": 9.55297325950019e-06, "loss": 0.3613, "step": 2376 }, { "epoch": 0.32, "grad_norm": 3.0741508590726765, "learning_rate": 9.552517881795142e-06, "loss": 0.2861, "step": 2377 }, { "epoch": 0.32, "grad_norm": 4.352903227568402, "learning_rate": 9.55206228313076e-06, "loss": 0.458, "step": 2378 }, { "epoch": 0.32, "grad_norm": 3.4737835648683744, "learning_rate": 9.551606463529158e-06, "loss": 0.3796, "step": 2379 }, { "epoch": 0.32, "grad_norm": 3.383717212210397, "learning_rate": 9.551150423012456e-06, "loss": 0.2831, "step": 2380 }, { "epoch": 0.32, "grad_norm": 3.108437544997103, "learning_rate": 9.550694161602792e-06, "loss": 0.3177, "step": 2381 }, { "epoch": 0.32, "grad_norm": 3.5689486795270686, "learning_rate": 9.550237679322308e-06, "loss": 0.3468, "step": 2382 }, { "epoch": 0.32, "grad_norm": 3.1412184001077357, "learning_rate": 9.54978097619316e-06, "loss": 0.3499, "step": 2383 }, { "epoch": 0.32, "grad_norm": 3.2501487288533957, "learning_rate": 9.549324052237517e-06, "loss": 0.2891, "step": 2384 }, { "epoch": 0.32, "grad_norm": 3.4981146998807797, "learning_rate": 9.548866907477552e-06, "loss": 0.3977, "step": 2385 }, { "epoch": 0.32, "grad_norm": 4.280737083351014, "learning_rate": 9.548409541935457e-06, "loss": 0.3342, "step": 2386 }, { "epoch": 0.32, "grad_norm": 3.465918589005146, "learning_rate": 9.547951955633428e-06, "loss": 0.3052, "step": 2387 }, { "epoch": 0.32, "grad_norm": 3.4728169057678193, "learning_rate": 9.547494148593674e-06, "loss": 0.3401, "step": 2388 }, { "epoch": 0.32, "grad_norm": 3.715888019392495, "learning_rate": 9.547036120838415e-06, "loss": 0.4255, "step": 2389 }, { "epoch": 0.33, "grad_norm": 3.3921110032474173, "learning_rate": 9.546577872389885e-06, "loss": 0.3412, "step": 2390 }, { "epoch": 0.33, "grad_norm": 3.2114696603097506, "learning_rate": 9.54611940327032e-06, "loss": 0.3025, "step": 2391 }, { "epoch": 0.33, "grad_norm": 3.8937732223696737, "learning_rate": 9.545660713501975e-06, "loss": 0.2843, "step": 2392 }, { "epoch": 0.33, "grad_norm": 3.141969115055242, "learning_rate": 9.545201803107114e-06, "loss": 0.3512, "step": 2393 }, { "epoch": 0.33, "grad_norm": 3.651866319270897, "learning_rate": 9.54474267210801e-06, "loss": 0.3777, "step": 2394 }, { "epoch": 0.33, "grad_norm": 3.9960521901657198, "learning_rate": 9.544283320526943e-06, "loss": 0.3219, "step": 2395 }, { "epoch": 0.33, "grad_norm": 3.5282617677447585, "learning_rate": 9.543823748386214e-06, "loss": 0.3066, "step": 2396 }, { "epoch": 0.33, "grad_norm": 3.574373990958657, "learning_rate": 9.543363955708124e-06, "loss": 0.3316, "step": 2397 }, { "epoch": 0.33, "grad_norm": 3.596704182554139, "learning_rate": 9.542903942514993e-06, "loss": 0.347, "step": 2398 }, { "epoch": 0.33, "grad_norm": 3.5741522017452305, "learning_rate": 9.542443708829147e-06, "loss": 0.2855, "step": 2399 }, { "epoch": 0.33, "grad_norm": 3.495989080230568, "learning_rate": 9.541983254672922e-06, "loss": 0.3448, "step": 2400 }, { "epoch": 0.33, "grad_norm": 3.4949272242438285, "learning_rate": 9.541522580068666e-06, "loss": 0.3173, "step": 2401 }, { "epoch": 0.33, "grad_norm": 3.5625045092726735, "learning_rate": 9.541061685038742e-06, "loss": 0.2542, "step": 2402 }, { "epoch": 0.33, "grad_norm": 3.3999923858061374, "learning_rate": 9.540600569605516e-06, "loss": 0.366, "step": 2403 }, { "epoch": 0.33, "grad_norm": 3.3458252527624563, "learning_rate": 9.540139233791368e-06, "loss": 0.3245, "step": 2404 }, { "epoch": 0.33, "grad_norm": 3.5911235724564494, "learning_rate": 9.539677677618695e-06, "loss": 0.3296, "step": 2405 }, { "epoch": 0.33, "grad_norm": 3.5378838851593635, "learning_rate": 9.539215901109893e-06, "loss": 0.3112, "step": 2406 }, { "epoch": 0.33, "grad_norm": 3.4049788775304455, "learning_rate": 9.538753904287376e-06, "loss": 0.2914, "step": 2407 }, { "epoch": 0.33, "grad_norm": 4.394179506214795, "learning_rate": 9.538291687173568e-06, "loss": 0.3643, "step": 2408 }, { "epoch": 0.33, "grad_norm": 3.246285845495235, "learning_rate": 9.537829249790905e-06, "loss": 0.3058, "step": 2409 }, { "epoch": 0.33, "grad_norm": 3.6321938888777128, "learning_rate": 9.53736659216183e-06, "loss": 0.3541, "step": 2410 }, { "epoch": 0.33, "grad_norm": 3.3430351186656133, "learning_rate": 9.536903714308795e-06, "loss": 0.328, "step": 2411 }, { "epoch": 0.33, "grad_norm": 3.364604313424274, "learning_rate": 9.53644061625427e-06, "loss": 0.3609, "step": 2412 }, { "epoch": 0.33, "grad_norm": 3.5725077031375254, "learning_rate": 9.535977298020733e-06, "loss": 0.3307, "step": 2413 }, { "epoch": 0.33, "grad_norm": 3.6420608464025865, "learning_rate": 9.535513759630668e-06, "loss": 0.3591, "step": 2414 }, { "epoch": 0.33, "grad_norm": 4.089875277930854, "learning_rate": 9.535050001106573e-06, "loss": 0.3787, "step": 2415 }, { "epoch": 0.33, "grad_norm": 3.6796340650970536, "learning_rate": 9.534586022470961e-06, "loss": 0.3301, "step": 2416 }, { "epoch": 0.33, "grad_norm": 3.247228180612793, "learning_rate": 9.534121823746348e-06, "loss": 0.3155, "step": 2417 }, { "epoch": 0.33, "grad_norm": 3.0562364243480653, "learning_rate": 9.533657404955263e-06, "loss": 0.2488, "step": 2418 }, { "epoch": 0.33, "grad_norm": 3.7473779307608694, "learning_rate": 9.53319276612025e-06, "loss": 0.3375, "step": 2419 }, { "epoch": 0.33, "grad_norm": 3.9434024767257676, "learning_rate": 9.532727907263861e-06, "loss": 0.3409, "step": 2420 }, { "epoch": 0.33, "grad_norm": 3.415674979342953, "learning_rate": 9.532262828408654e-06, "loss": 0.3729, "step": 2421 }, { "epoch": 0.33, "grad_norm": 3.6398750210418953, "learning_rate": 9.531797529577205e-06, "loss": 0.3347, "step": 2422 }, { "epoch": 0.33, "grad_norm": 3.576496874592366, "learning_rate": 9.531332010792098e-06, "loss": 0.4249, "step": 2423 }, { "epoch": 0.33, "grad_norm": 2.9033008646239122, "learning_rate": 9.530866272075926e-06, "loss": 0.3176, "step": 2424 }, { "epoch": 0.33, "grad_norm": 2.8060137773032094, "learning_rate": 9.530400313451294e-06, "loss": 0.2678, "step": 2425 }, { "epoch": 0.33, "grad_norm": 3.260739570390811, "learning_rate": 9.529934134940819e-06, "loss": 0.3727, "step": 2426 }, { "epoch": 0.33, "grad_norm": 3.3361668081778473, "learning_rate": 9.529467736567124e-06, "loss": 0.3479, "step": 2427 }, { "epoch": 0.33, "grad_norm": 3.527159722397584, "learning_rate": 9.52900111835285e-06, "loss": 0.3129, "step": 2428 }, { "epoch": 0.33, "grad_norm": 2.9459885329911026, "learning_rate": 9.528534280320641e-06, "loss": 0.3085, "step": 2429 }, { "epoch": 0.33, "grad_norm": 3.750127159622899, "learning_rate": 9.528067222493159e-06, "loss": 0.3363, "step": 2430 }, { "epoch": 0.33, "grad_norm": 3.6559050867712135, "learning_rate": 9.527599944893068e-06, "loss": 0.353, "step": 2431 }, { "epoch": 0.33, "grad_norm": 3.365058391220238, "learning_rate": 9.527132447543051e-06, "loss": 0.2936, "step": 2432 }, { "epoch": 0.33, "grad_norm": 3.2119888329753246, "learning_rate": 9.526664730465799e-06, "loss": 0.3194, "step": 2433 }, { "epoch": 0.33, "grad_norm": 3.9878370825656284, "learning_rate": 9.52619679368401e-06, "loss": 0.3921, "step": 2434 }, { "epoch": 0.33, "grad_norm": 3.0257797955003767, "learning_rate": 9.5257286372204e-06, "loss": 0.3177, "step": 2435 }, { "epoch": 0.33, "grad_norm": 3.9960782975638094, "learning_rate": 9.525260261097688e-06, "loss": 0.296, "step": 2436 }, { "epoch": 0.33, "grad_norm": 3.2400785194827986, "learning_rate": 9.524791665338606e-06, "loss": 0.3446, "step": 2437 }, { "epoch": 0.33, "grad_norm": 3.225011682785851, "learning_rate": 9.524322849965899e-06, "loss": 0.3275, "step": 2438 }, { "epoch": 0.33, "grad_norm": 3.0641379408723846, "learning_rate": 9.523853815002322e-06, "loss": 0.3149, "step": 2439 }, { "epoch": 0.33, "grad_norm": 3.2389118326773705, "learning_rate": 9.52338456047064e-06, "loss": 0.3457, "step": 2440 }, { "epoch": 0.33, "grad_norm": 3.9128946723174325, "learning_rate": 9.522915086393628e-06, "loss": 0.4109, "step": 2441 }, { "epoch": 0.33, "grad_norm": 3.4206892989803204, "learning_rate": 9.522445392794069e-06, "loss": 0.3431, "step": 2442 }, { "epoch": 0.33, "grad_norm": 3.527337085738484, "learning_rate": 9.521975479694766e-06, "loss": 0.3696, "step": 2443 }, { "epoch": 0.33, "grad_norm": 3.2025934774537625, "learning_rate": 9.521505347118522e-06, "loss": 0.32, "step": 2444 }, { "epoch": 0.33, "grad_norm": 3.420885182675037, "learning_rate": 9.521034995088157e-06, "loss": 0.3075, "step": 2445 }, { "epoch": 0.33, "grad_norm": 2.7996014236405165, "learning_rate": 9.5205644236265e-06, "loss": 0.3265, "step": 2446 }, { "epoch": 0.33, "grad_norm": 3.638603483081791, "learning_rate": 9.520093632756388e-06, "loss": 0.3304, "step": 2447 }, { "epoch": 0.33, "grad_norm": 3.2541623686079517, "learning_rate": 9.519622622500676e-06, "loss": 0.2602, "step": 2448 }, { "epoch": 0.33, "grad_norm": 3.1639605757000266, "learning_rate": 9.51915139288222e-06, "loss": 0.3225, "step": 2449 }, { "epoch": 0.33, "grad_norm": 3.6587684197552255, "learning_rate": 9.518679943923894e-06, "loss": 0.312, "step": 2450 }, { "epoch": 0.33, "grad_norm": 3.3976898389250962, "learning_rate": 9.51820827564858e-06, "loss": 0.2857, "step": 2451 }, { "epoch": 0.33, "grad_norm": 3.914144220808181, "learning_rate": 9.517736388079169e-06, "loss": 0.3453, "step": 2452 }, { "epoch": 0.33, "grad_norm": 3.197237433887797, "learning_rate": 9.517264281238563e-06, "loss": 0.2729, "step": 2453 }, { "epoch": 0.33, "grad_norm": 3.7936014205166515, "learning_rate": 9.516791955149682e-06, "loss": 0.2658, "step": 2454 }, { "epoch": 0.33, "grad_norm": 3.4157664314448652, "learning_rate": 9.516319409835446e-06, "loss": 0.3374, "step": 2455 }, { "epoch": 0.33, "grad_norm": 4.6667895180620995, "learning_rate": 9.515846645318792e-06, "loss": 0.3755, "step": 2456 }, { "epoch": 0.33, "grad_norm": 3.755442508468413, "learning_rate": 9.515373661622665e-06, "loss": 0.3485, "step": 2457 }, { "epoch": 0.33, "grad_norm": 3.3789065096447617, "learning_rate": 9.514900458770021e-06, "loss": 0.3259, "step": 2458 }, { "epoch": 0.33, "grad_norm": 3.168350969482604, "learning_rate": 9.51442703678383e-06, "loss": 0.3049, "step": 2459 }, { "epoch": 0.33, "grad_norm": 3.721659761739798, "learning_rate": 9.513953395687066e-06, "loss": 0.3148, "step": 2460 }, { "epoch": 0.33, "grad_norm": 3.3603799643162633, "learning_rate": 9.513479535502719e-06, "loss": 0.2481, "step": 2461 }, { "epoch": 0.33, "grad_norm": 7.218341156021362, "learning_rate": 9.51300545625379e-06, "loss": 0.2427, "step": 2462 }, { "epoch": 0.34, "grad_norm": 5.405826174348986, "learning_rate": 9.512531157963285e-06, "loss": 0.2892, "step": 2463 }, { "epoch": 0.34, "grad_norm": 3.4848964125112243, "learning_rate": 9.512056640654228e-06, "loss": 0.3309, "step": 2464 }, { "epoch": 0.34, "grad_norm": 3.519797377933815, "learning_rate": 9.51158190434965e-06, "loss": 0.3545, "step": 2465 }, { "epoch": 0.34, "grad_norm": 3.8875191046136797, "learning_rate": 9.511106949072588e-06, "loss": 0.3468, "step": 2466 }, { "epoch": 0.34, "grad_norm": 3.4955878614705065, "learning_rate": 9.510631774846099e-06, "loss": 0.3786, "step": 2467 }, { "epoch": 0.34, "grad_norm": 2.819100127230006, "learning_rate": 9.510156381693245e-06, "loss": 0.2928, "step": 2468 }, { "epoch": 0.34, "grad_norm": 3.77193668173726, "learning_rate": 9.509680769637099e-06, "loss": 0.2683, "step": 2469 }, { "epoch": 0.34, "grad_norm": 3.6337566586205514, "learning_rate": 9.509204938700745e-06, "loss": 0.3135, "step": 2470 }, { "epoch": 0.34, "grad_norm": 3.64085486315168, "learning_rate": 9.508728888907277e-06, "loss": 0.292, "step": 2471 }, { "epoch": 0.34, "grad_norm": 3.660384841299868, "learning_rate": 9.5082526202798e-06, "loss": 0.322, "step": 2472 }, { "epoch": 0.34, "grad_norm": 3.36963751568736, "learning_rate": 9.507776132841432e-06, "loss": 0.2775, "step": 2473 }, { "epoch": 0.34, "grad_norm": 4.211105621446265, "learning_rate": 9.5072994266153e-06, "loss": 0.2709, "step": 2474 }, { "epoch": 0.34, "grad_norm": 3.552194376888417, "learning_rate": 9.506822501624538e-06, "loss": 0.3614, "step": 2475 }, { "epoch": 0.34, "grad_norm": 3.30638055985548, "learning_rate": 9.506345357892297e-06, "loss": 0.3315, "step": 2476 }, { "epoch": 0.34, "grad_norm": 3.3614167449597145, "learning_rate": 9.505867995441734e-06, "loss": 0.3321, "step": 2477 }, { "epoch": 0.34, "grad_norm": 3.7712431338877366, "learning_rate": 9.505390414296018e-06, "loss": 0.3158, "step": 2478 }, { "epoch": 0.34, "grad_norm": 3.7291264639510753, "learning_rate": 9.50491261447833e-06, "loss": 0.3393, "step": 2479 }, { "epoch": 0.34, "grad_norm": 3.4553782739761414, "learning_rate": 9.504434596011858e-06, "loss": 0.3552, "step": 2480 }, { "epoch": 0.34, "grad_norm": 3.7541720632511355, "learning_rate": 9.503956358919805e-06, "loss": 0.3726, "step": 2481 }, { "epoch": 0.34, "grad_norm": 3.0630360090983912, "learning_rate": 9.503477903225382e-06, "loss": 0.2289, "step": 2482 }, { "epoch": 0.34, "grad_norm": 3.633434834360419, "learning_rate": 9.502999228951812e-06, "loss": 0.2907, "step": 2483 }, { "epoch": 0.34, "grad_norm": 3.0260619270652294, "learning_rate": 9.502520336122327e-06, "loss": 0.2718, "step": 2484 }, { "epoch": 0.34, "grad_norm": 2.9086530108122695, "learning_rate": 9.502041224760167e-06, "loss": 0.2818, "step": 2485 }, { "epoch": 0.34, "grad_norm": 3.2083161594591663, "learning_rate": 9.501561894888592e-06, "loss": 0.3174, "step": 2486 }, { "epoch": 0.34, "grad_norm": 3.6719376879452588, "learning_rate": 9.501082346530864e-06, "loss": 0.3384, "step": 2487 }, { "epoch": 0.34, "grad_norm": 3.362199524345544, "learning_rate": 9.500602579710256e-06, "loss": 0.3147, "step": 2488 }, { "epoch": 0.34, "grad_norm": 3.1143304784415293, "learning_rate": 9.500122594450056e-06, "loss": 0.3219, "step": 2489 }, { "epoch": 0.34, "grad_norm": 3.947692823239979, "learning_rate": 9.49964239077356e-06, "loss": 0.2824, "step": 2490 }, { "epoch": 0.34, "grad_norm": 3.225390367137568, "learning_rate": 9.499161968704076e-06, "loss": 0.2951, "step": 2491 }, { "epoch": 0.34, "grad_norm": 3.4445709290112014, "learning_rate": 9.498681328264919e-06, "loss": 0.3071, "step": 2492 }, { "epoch": 0.34, "grad_norm": 4.997741761577481, "learning_rate": 9.498200469479419e-06, "loss": 0.3311, "step": 2493 }, { "epoch": 0.34, "grad_norm": 3.34510140731967, "learning_rate": 9.497719392370917e-06, "loss": 0.3088, "step": 2494 }, { "epoch": 0.34, "grad_norm": 3.672692950307285, "learning_rate": 9.497238096962757e-06, "loss": 0.3537, "step": 2495 }, { "epoch": 0.34, "grad_norm": 3.6290331733852534, "learning_rate": 9.496756583278304e-06, "loss": 0.328, "step": 2496 }, { "epoch": 0.34, "grad_norm": 3.385032409084255, "learning_rate": 9.496274851340926e-06, "loss": 0.3463, "step": 2497 }, { "epoch": 0.34, "grad_norm": 3.3554272473518405, "learning_rate": 9.495792901174005e-06, "loss": 0.2974, "step": 2498 }, { "epoch": 0.34, "grad_norm": 3.9904397418330757, "learning_rate": 9.495310732800931e-06, "loss": 0.3476, "step": 2499 }, { "epoch": 0.34, "grad_norm": 3.745748676115922, "learning_rate": 9.494828346245108e-06, "loss": 0.3151, "step": 2500 }, { "epoch": 0.34, "grad_norm": 3.4128264102507107, "learning_rate": 9.494345741529952e-06, "loss": 0.365, "step": 2501 }, { "epoch": 0.34, "grad_norm": 3.7152463050148143, "learning_rate": 9.49386291867888e-06, "loss": 0.25, "step": 2502 }, { "epoch": 0.34, "grad_norm": 3.4224123429454094, "learning_rate": 9.493379877715329e-06, "loss": 0.2998, "step": 2503 }, { "epoch": 0.34, "grad_norm": 3.144382535598247, "learning_rate": 9.492896618662745e-06, "loss": 0.2562, "step": 2504 }, { "epoch": 0.34, "grad_norm": 3.57001409354432, "learning_rate": 9.492413141544584e-06, "loss": 0.2736, "step": 2505 }, { "epoch": 0.34, "grad_norm": 3.7976758893980644, "learning_rate": 9.491929446384307e-06, "loss": 0.3897, "step": 2506 }, { "epoch": 0.34, "grad_norm": 2.9005213721360827, "learning_rate": 9.491445533205397e-06, "loss": 0.2759, "step": 2507 }, { "epoch": 0.34, "grad_norm": 3.7177480908904323, "learning_rate": 9.490961402031335e-06, "loss": 0.2912, "step": 2508 }, { "epoch": 0.34, "grad_norm": 3.6936461427274967, "learning_rate": 9.490477052885622e-06, "loss": 0.3073, "step": 2509 }, { "epoch": 0.34, "grad_norm": 4.250627054421376, "learning_rate": 9.489992485791766e-06, "loss": 0.2703, "step": 2510 }, { "epoch": 0.34, "grad_norm": 3.695585822714089, "learning_rate": 9.489507700773287e-06, "loss": 0.348, "step": 2511 }, { "epoch": 0.34, "grad_norm": 3.842445617851256, "learning_rate": 9.48902269785371e-06, "loss": 0.3525, "step": 2512 }, { "epoch": 0.34, "grad_norm": 4.161974051402449, "learning_rate": 9.488537477056578e-06, "loss": 0.3304, "step": 2513 }, { "epoch": 0.34, "grad_norm": 3.2850833637625394, "learning_rate": 9.488052038405441e-06, "loss": 0.2921, "step": 2514 }, { "epoch": 0.34, "grad_norm": 4.79333101045154, "learning_rate": 9.48756638192386e-06, "loss": 0.2939, "step": 2515 }, { "epoch": 0.34, "grad_norm": 3.8806896942826703, "learning_rate": 9.487080507635408e-06, "loss": 0.3341, "step": 2516 }, { "epoch": 0.34, "grad_norm": 3.430085841407778, "learning_rate": 9.486594415563665e-06, "loss": 0.3213, "step": 2517 }, { "epoch": 0.34, "grad_norm": 3.4953710292503164, "learning_rate": 9.486108105732225e-06, "loss": 0.3239, "step": 2518 }, { "epoch": 0.34, "grad_norm": 3.1844109446398883, "learning_rate": 9.48562157816469e-06, "loss": 0.2472, "step": 2519 }, { "epoch": 0.34, "grad_norm": 3.6050014333068248, "learning_rate": 9.485134832884678e-06, "loss": 0.3751, "step": 2520 }, { "epoch": 0.34, "grad_norm": 3.7963296709623036, "learning_rate": 9.484647869915807e-06, "loss": 0.3296, "step": 2521 }, { "epoch": 0.34, "grad_norm": 3.777349789104605, "learning_rate": 9.484160689281718e-06, "loss": 0.3579, "step": 2522 }, { "epoch": 0.34, "grad_norm": 4.0258916916402905, "learning_rate": 9.483673291006053e-06, "loss": 0.3117, "step": 2523 }, { "epoch": 0.34, "grad_norm": 3.169908357285852, "learning_rate": 9.483185675112469e-06, "loss": 0.3365, "step": 2524 }, { "epoch": 0.34, "grad_norm": 3.246536779709645, "learning_rate": 9.482697841624635e-06, "loss": 0.3379, "step": 2525 }, { "epoch": 0.34, "grad_norm": 3.5714942826462446, "learning_rate": 9.482209790566224e-06, "loss": 0.3064, "step": 2526 }, { "epoch": 0.34, "grad_norm": 3.575557453277341, "learning_rate": 9.48172152196093e-06, "loss": 0.3011, "step": 2527 }, { "epoch": 0.34, "grad_norm": 4.528785929354005, "learning_rate": 9.481233035832445e-06, "loss": 0.3515, "step": 2528 }, { "epoch": 0.34, "grad_norm": 3.8788391728817277, "learning_rate": 9.48074433220448e-06, "loss": 0.3187, "step": 2529 }, { "epoch": 0.34, "grad_norm": 3.133907479584155, "learning_rate": 9.480255411100756e-06, "loss": 0.2837, "step": 2530 }, { "epoch": 0.34, "grad_norm": 3.6587629483944246, "learning_rate": 9.479766272545003e-06, "loss": 0.3563, "step": 2531 }, { "epoch": 0.34, "grad_norm": 3.726281708946879, "learning_rate": 9.47927691656096e-06, "loss": 0.2716, "step": 2532 }, { "epoch": 0.34, "grad_norm": 3.6216721898986513, "learning_rate": 9.47878734317238e-06, "loss": 0.237, "step": 2533 }, { "epoch": 0.34, "grad_norm": 2.8984842565979574, "learning_rate": 9.478297552403023e-06, "loss": 0.2842, "step": 2534 }, { "epoch": 0.34, "grad_norm": 3.1056537308511234, "learning_rate": 9.477807544276665e-06, "loss": 0.2647, "step": 2535 }, { "epoch": 0.34, "grad_norm": 3.4869392228988385, "learning_rate": 9.477317318817084e-06, "loss": 0.305, "step": 2536 }, { "epoch": 0.35, "grad_norm": 2.989023785177858, "learning_rate": 9.476826876048076e-06, "loss": 0.3106, "step": 2537 }, { "epoch": 0.35, "grad_norm": 3.3801149173360456, "learning_rate": 9.476336215993445e-06, "loss": 0.3618, "step": 2538 }, { "epoch": 0.35, "grad_norm": 3.4166348149640906, "learning_rate": 9.475845338677003e-06, "loss": 0.2958, "step": 2539 }, { "epoch": 0.35, "grad_norm": 3.8244823214791546, "learning_rate": 9.475354244122578e-06, "loss": 0.3459, "step": 2540 }, { "epoch": 0.35, "grad_norm": 4.229185159022408, "learning_rate": 9.474862932354006e-06, "loss": 0.3414, "step": 2541 }, { "epoch": 0.35, "grad_norm": 3.9311599004701163, "learning_rate": 9.474371403395129e-06, "loss": 0.3556, "step": 2542 }, { "epoch": 0.35, "grad_norm": 3.680695471231166, "learning_rate": 9.473879657269808e-06, "loss": 0.3141, "step": 2543 }, { "epoch": 0.35, "grad_norm": 3.878538141894439, "learning_rate": 9.47338769400191e-06, "loss": 0.2975, "step": 2544 }, { "epoch": 0.35, "grad_norm": 3.9072524315217607, "learning_rate": 9.472895513615309e-06, "loss": 0.3165, "step": 2545 }, { "epoch": 0.35, "grad_norm": 3.5318178937623497, "learning_rate": 9.472403116133896e-06, "loss": 0.3004, "step": 2546 }, { "epoch": 0.35, "grad_norm": 4.027355553511843, "learning_rate": 9.47191050158157e-06, "loss": 0.3414, "step": 2547 }, { "epoch": 0.35, "grad_norm": 3.273354445157702, "learning_rate": 9.47141766998224e-06, "loss": 0.3129, "step": 2548 }, { "epoch": 0.35, "grad_norm": 3.883527375228978, "learning_rate": 9.470924621359824e-06, "loss": 0.3439, "step": 2549 }, { "epoch": 0.35, "grad_norm": 3.774794323963319, "learning_rate": 9.470431355738257e-06, "loss": 0.2417, "step": 2550 }, { "epoch": 0.35, "grad_norm": 3.0481827532560777, "learning_rate": 9.469937873141474e-06, "loss": 0.3115, "step": 2551 }, { "epoch": 0.35, "grad_norm": 3.34400192150364, "learning_rate": 9.469444173593433e-06, "loss": 0.3104, "step": 2552 }, { "epoch": 0.35, "grad_norm": 3.22835007519981, "learning_rate": 9.46895025711809e-06, "loss": 0.2986, "step": 2553 }, { "epoch": 0.35, "grad_norm": 3.895862640660134, "learning_rate": 9.468456123739422e-06, "loss": 0.3926, "step": 2554 }, { "epoch": 0.35, "grad_norm": 3.0543738559071523, "learning_rate": 9.467961773481409e-06, "loss": 0.2953, "step": 2555 }, { "epoch": 0.35, "grad_norm": 3.0370752891806583, "learning_rate": 9.467467206368046e-06, "loss": 0.3274, "step": 2556 }, { "epoch": 0.35, "grad_norm": 3.336665520237808, "learning_rate": 9.466972422423338e-06, "loss": 0.3289, "step": 2557 }, { "epoch": 0.35, "grad_norm": 3.184041388662976, "learning_rate": 9.466477421671296e-06, "loss": 0.3075, "step": 2558 }, { "epoch": 0.35, "grad_norm": 3.5760863212644134, "learning_rate": 9.46598220413595e-06, "loss": 0.3125, "step": 2559 }, { "epoch": 0.35, "grad_norm": 3.3776949937565264, "learning_rate": 9.465486769841332e-06, "loss": 0.3246, "step": 2560 }, { "epoch": 0.35, "grad_norm": 3.0441462178678376, "learning_rate": 9.46499111881149e-06, "loss": 0.2589, "step": 2561 }, { "epoch": 0.35, "grad_norm": 3.204328583682982, "learning_rate": 9.464495251070483e-06, "loss": 0.2726, "step": 2562 }, { "epoch": 0.35, "grad_norm": 6.324894708715439, "learning_rate": 9.463999166642372e-06, "loss": 0.3182, "step": 2563 }, { "epoch": 0.35, "grad_norm": 3.842626440811825, "learning_rate": 9.463502865551241e-06, "loss": 0.3194, "step": 2564 }, { "epoch": 0.35, "grad_norm": 3.401645444147183, "learning_rate": 9.463006347821175e-06, "loss": 0.2946, "step": 2565 }, { "epoch": 0.35, "grad_norm": 6.790747625663712, "learning_rate": 9.462509613476273e-06, "loss": 0.3248, "step": 2566 }, { "epoch": 0.35, "grad_norm": 3.2587902501867694, "learning_rate": 9.462012662540645e-06, "loss": 0.2716, "step": 2567 }, { "epoch": 0.35, "grad_norm": 3.361066683574132, "learning_rate": 9.461515495038411e-06, "loss": 0.3083, "step": 2568 }, { "epoch": 0.35, "grad_norm": 3.7141154368484472, "learning_rate": 9.4610181109937e-06, "loss": 0.2623, "step": 2569 }, { "epoch": 0.35, "grad_norm": 4.20043548592031, "learning_rate": 9.460520510430655e-06, "loss": 0.3879, "step": 2570 }, { "epoch": 0.35, "grad_norm": 3.594289097702332, "learning_rate": 9.460022693373428e-06, "loss": 0.3126, "step": 2571 }, { "epoch": 0.35, "grad_norm": 4.011768296769084, "learning_rate": 9.459524659846176e-06, "loss": 0.3166, "step": 2572 }, { "epoch": 0.35, "grad_norm": 4.153389413726448, "learning_rate": 9.459026409873075e-06, "loss": 0.3732, "step": 2573 }, { "epoch": 0.35, "grad_norm": 3.1524051656309275, "learning_rate": 9.458527943478309e-06, "loss": 0.2783, "step": 2574 }, { "epoch": 0.35, "grad_norm": 3.1273206638160183, "learning_rate": 9.458029260686068e-06, "loss": 0.2693, "step": 2575 }, { "epoch": 0.35, "grad_norm": 3.1174689670041094, "learning_rate": 9.457530361520558e-06, "loss": 0.2856, "step": 2576 }, { "epoch": 0.35, "grad_norm": 3.3323639920030117, "learning_rate": 9.457031246005994e-06, "loss": 0.3413, "step": 2577 }, { "epoch": 0.35, "grad_norm": 3.727215925007012, "learning_rate": 9.456531914166598e-06, "loss": 0.3866, "step": 2578 }, { "epoch": 0.35, "grad_norm": 4.811954742715354, "learning_rate": 9.45603236602661e-06, "loss": 0.3723, "step": 2579 }, { "epoch": 0.35, "grad_norm": 4.119900457380265, "learning_rate": 9.45553260161027e-06, "loss": 0.287, "step": 2580 }, { "epoch": 0.35, "grad_norm": 3.5375924227414552, "learning_rate": 9.45503262094184e-06, "loss": 0.2651, "step": 2581 }, { "epoch": 0.35, "grad_norm": 3.8123533398355525, "learning_rate": 9.454532424045585e-06, "loss": 0.3373, "step": 2582 }, { "epoch": 0.35, "grad_norm": 3.504831014805764, "learning_rate": 9.454032010945779e-06, "loss": 0.2731, "step": 2583 }, { "epoch": 0.35, "grad_norm": 3.516832644547503, "learning_rate": 9.453531381666714e-06, "loss": 0.3322, "step": 2584 }, { "epoch": 0.35, "grad_norm": 3.603042388682409, "learning_rate": 9.453030536232687e-06, "loss": 0.2882, "step": 2585 }, { "epoch": 0.35, "grad_norm": 3.0485414318135637, "learning_rate": 9.452529474668008e-06, "loss": 0.3013, "step": 2586 }, { "epoch": 0.35, "grad_norm": 3.7344184925464736, "learning_rate": 9.452028196996994e-06, "loss": 0.3362, "step": 2587 }, { "epoch": 0.35, "grad_norm": 4.375795512167435, "learning_rate": 9.451526703243977e-06, "loss": 0.334, "step": 2588 }, { "epoch": 0.35, "grad_norm": 3.472273504714847, "learning_rate": 9.451024993433294e-06, "loss": 0.284, "step": 2589 }, { "epoch": 0.35, "grad_norm": 4.07112968047582, "learning_rate": 9.450523067589299e-06, "loss": 0.3762, "step": 2590 }, { "epoch": 0.35, "grad_norm": 3.943038743274586, "learning_rate": 9.450020925736354e-06, "loss": 0.3569, "step": 2591 }, { "epoch": 0.35, "grad_norm": 3.5936211233938713, "learning_rate": 9.449518567898827e-06, "loss": 0.3297, "step": 2592 }, { "epoch": 0.35, "grad_norm": 2.8203548116880888, "learning_rate": 9.449015994101104e-06, "loss": 0.2581, "step": 2593 }, { "epoch": 0.35, "grad_norm": 2.7881644803403347, "learning_rate": 9.448513204367578e-06, "loss": 0.213, "step": 2594 }, { "epoch": 0.35, "grad_norm": 3.443507794452085, "learning_rate": 9.448010198722648e-06, "loss": 0.3662, "step": 2595 }, { "epoch": 0.35, "grad_norm": 3.025803375967303, "learning_rate": 9.44750697719073e-06, "loss": 0.3097, "step": 2596 }, { "epoch": 0.35, "grad_norm": 3.4013689689333613, "learning_rate": 9.44700353979625e-06, "loss": 0.3014, "step": 2597 }, { "epoch": 0.35, "grad_norm": 2.789567942456384, "learning_rate": 9.446499886563638e-06, "loss": 0.2912, "step": 2598 }, { "epoch": 0.35, "grad_norm": 4.071462264127469, "learning_rate": 9.445996017517344e-06, "loss": 0.3248, "step": 2599 }, { "epoch": 0.35, "grad_norm": 3.678251211516993, "learning_rate": 9.445491932681822e-06, "loss": 0.2931, "step": 2600 }, { "epoch": 0.35, "grad_norm": 3.106501687148944, "learning_rate": 9.444987632081536e-06, "loss": 0.3, "step": 2601 }, { "epoch": 0.35, "grad_norm": 3.336415564906388, "learning_rate": 9.444483115740968e-06, "loss": 0.3256, "step": 2602 }, { "epoch": 0.35, "grad_norm": 3.5051894106762393, "learning_rate": 9.443978383684598e-06, "loss": 0.3564, "step": 2603 }, { "epoch": 0.35, "grad_norm": 3.836617185653977, "learning_rate": 9.44347343593693e-06, "loss": 0.298, "step": 2604 }, { "epoch": 0.35, "grad_norm": 2.980221447491443, "learning_rate": 9.442968272522464e-06, "loss": 0.2973, "step": 2605 }, { "epoch": 0.35, "grad_norm": 3.2747895996220735, "learning_rate": 9.442462893465727e-06, "loss": 0.3029, "step": 2606 }, { "epoch": 0.35, "grad_norm": 3.9641101729547623, "learning_rate": 9.441957298791243e-06, "loss": 0.3761, "step": 2607 }, { "epoch": 0.35, "grad_norm": 3.1315754804043685, "learning_rate": 9.441451488523552e-06, "loss": 0.3101, "step": 2608 }, { "epoch": 0.35, "grad_norm": 2.555446949106363, "learning_rate": 9.440945462687205e-06, "loss": 0.2123, "step": 2609 }, { "epoch": 0.36, "grad_norm": 3.4535377608561397, "learning_rate": 9.440439221306762e-06, "loss": 0.3634, "step": 2610 }, { "epoch": 0.36, "grad_norm": 3.4565108778209805, "learning_rate": 9.439932764406793e-06, "loss": 0.3833, "step": 2611 }, { "epoch": 0.36, "grad_norm": 5.282375226751719, "learning_rate": 9.439426092011877e-06, "loss": 0.3578, "step": 2612 }, { "epoch": 0.36, "grad_norm": 2.90228245722118, "learning_rate": 9.43891920414661e-06, "loss": 0.2783, "step": 2613 }, { "epoch": 0.36, "grad_norm": 3.3503904094537944, "learning_rate": 9.438412100835592e-06, "loss": 0.3432, "step": 2614 }, { "epoch": 0.36, "grad_norm": 3.9671680764385004, "learning_rate": 9.437904782103438e-06, "loss": 0.3148, "step": 2615 }, { "epoch": 0.36, "grad_norm": 3.832074975801602, "learning_rate": 9.437397247974766e-06, "loss": 0.3322, "step": 2616 }, { "epoch": 0.36, "grad_norm": 3.5560224793995165, "learning_rate": 9.436889498474213e-06, "loss": 0.298, "step": 2617 }, { "epoch": 0.36, "grad_norm": 3.659803472016776, "learning_rate": 9.436381533626425e-06, "loss": 0.4277, "step": 2618 }, { "epoch": 0.36, "grad_norm": 3.4375779331243805, "learning_rate": 9.435873353456051e-06, "loss": 0.319, "step": 2619 }, { "epoch": 0.36, "grad_norm": 3.6620142202134907, "learning_rate": 9.435364957987757e-06, "loss": 0.337, "step": 2620 }, { "epoch": 0.36, "grad_norm": 3.6458322075818552, "learning_rate": 9.434856347246223e-06, "loss": 0.3032, "step": 2621 }, { "epoch": 0.36, "grad_norm": 4.039920438739919, "learning_rate": 9.434347521256131e-06, "loss": 0.3878, "step": 2622 }, { "epoch": 0.36, "grad_norm": 2.95063735530626, "learning_rate": 9.433838480042179e-06, "loss": 0.2814, "step": 2623 }, { "epoch": 0.36, "grad_norm": 3.2027055152912847, "learning_rate": 9.43332922362907e-06, "loss": 0.3199, "step": 2624 }, { "epoch": 0.36, "grad_norm": 3.368102333063383, "learning_rate": 9.432819752041525e-06, "loss": 0.3517, "step": 2625 }, { "epoch": 0.36, "grad_norm": 4.280123535896667, "learning_rate": 9.432310065304269e-06, "loss": 0.3409, "step": 2626 }, { "epoch": 0.36, "grad_norm": 3.787342479474491, "learning_rate": 9.431800163442043e-06, "loss": 0.2849, "step": 2627 }, { "epoch": 0.36, "grad_norm": 3.436196825020012, "learning_rate": 9.431290046479591e-06, "loss": 0.3595, "step": 2628 }, { "epoch": 0.36, "grad_norm": 3.617811501673642, "learning_rate": 9.430779714441676e-06, "loss": 0.3251, "step": 2629 }, { "epoch": 0.36, "grad_norm": 3.509715394554404, "learning_rate": 9.430269167353065e-06, "loss": 0.3869, "step": 2630 }, { "epoch": 0.36, "grad_norm": 3.896867869454114, "learning_rate": 9.429758405238539e-06, "loss": 0.3198, "step": 2631 }, { "epoch": 0.36, "grad_norm": 3.420952784873707, "learning_rate": 9.429247428122886e-06, "loss": 0.2674, "step": 2632 }, { "epoch": 0.36, "grad_norm": 3.431856931037137, "learning_rate": 9.428736236030908e-06, "loss": 0.3344, "step": 2633 }, { "epoch": 0.36, "grad_norm": 3.5770914829102116, "learning_rate": 9.428224828987418e-06, "loss": 0.3227, "step": 2634 }, { "epoch": 0.36, "grad_norm": 3.9536215955153073, "learning_rate": 9.427713207017233e-06, "loss": 0.3102, "step": 2635 }, { "epoch": 0.36, "grad_norm": 4.034500468632103, "learning_rate": 9.42720137014519e-06, "loss": 0.3932, "step": 2636 }, { "epoch": 0.36, "grad_norm": 3.6752070380467616, "learning_rate": 9.426689318396128e-06, "loss": 0.3468, "step": 2637 }, { "epoch": 0.36, "grad_norm": 3.7620588461811093, "learning_rate": 9.426177051794901e-06, "loss": 0.303, "step": 2638 }, { "epoch": 0.36, "grad_norm": 3.3020643704827086, "learning_rate": 9.425664570366369e-06, "loss": 0.2987, "step": 2639 }, { "epoch": 0.36, "grad_norm": 4.043791952382103, "learning_rate": 9.425151874135411e-06, "loss": 0.3408, "step": 2640 }, { "epoch": 0.36, "grad_norm": 3.812338651857478, "learning_rate": 9.424638963126907e-06, "loss": 0.3525, "step": 2641 }, { "epoch": 0.36, "grad_norm": 3.1253389934527354, "learning_rate": 9.424125837365754e-06, "loss": 0.3249, "step": 2642 }, { "epoch": 0.36, "grad_norm": 2.962306410059906, "learning_rate": 9.423612496876856e-06, "loss": 0.3111, "step": 2643 }, { "epoch": 0.36, "grad_norm": 3.532935726402942, "learning_rate": 9.423098941685126e-06, "loss": 0.3393, "step": 2644 }, { "epoch": 0.36, "grad_norm": 19.106702576000885, "learning_rate": 9.422585171815493e-06, "loss": 0.3425, "step": 2645 }, { "epoch": 0.36, "grad_norm": 3.5946730083482774, "learning_rate": 9.422071187292891e-06, "loss": 0.3034, "step": 2646 }, { "epoch": 0.36, "grad_norm": 3.7507434305210547, "learning_rate": 9.42155698814227e-06, "loss": 0.3556, "step": 2647 }, { "epoch": 0.36, "grad_norm": 3.8746976856870217, "learning_rate": 9.421042574388581e-06, "loss": 0.3021, "step": 2648 }, { "epoch": 0.36, "grad_norm": 3.6327119945953084, "learning_rate": 9.420527946056797e-06, "loss": 0.3416, "step": 2649 }, { "epoch": 0.36, "grad_norm": 3.6203144964386498, "learning_rate": 9.420013103171893e-06, "loss": 0.3563, "step": 2650 }, { "epoch": 0.36, "grad_norm": 3.1771874821438395, "learning_rate": 9.419498045758857e-06, "loss": 0.3271, "step": 2651 }, { "epoch": 0.36, "grad_norm": 3.207016325368593, "learning_rate": 9.41898277384269e-06, "loss": 0.3724, "step": 2652 }, { "epoch": 0.36, "grad_norm": 2.943623891296714, "learning_rate": 9.418467287448399e-06, "loss": 0.2431, "step": 2653 }, { "epoch": 0.36, "grad_norm": 3.5428133154654535, "learning_rate": 9.417951586601004e-06, "loss": 0.2981, "step": 2654 }, { "epoch": 0.36, "grad_norm": 3.2313117919308305, "learning_rate": 9.417435671325535e-06, "loss": 0.2859, "step": 2655 }, { "epoch": 0.36, "grad_norm": 3.868821267931807, "learning_rate": 9.416919541647032e-06, "loss": 0.3021, "step": 2656 }, { "epoch": 0.36, "grad_norm": 3.7132892516745555, "learning_rate": 9.416403197590547e-06, "loss": 0.3147, "step": 2657 }, { "epoch": 0.36, "grad_norm": 3.553516509940827, "learning_rate": 9.415886639181137e-06, "loss": 0.3303, "step": 2658 }, { "epoch": 0.36, "grad_norm": 5.947906305431066, "learning_rate": 9.41536986644388e-06, "loss": 0.3255, "step": 2659 }, { "epoch": 0.36, "grad_norm": 3.7471488497696046, "learning_rate": 9.414852879403852e-06, "loss": 0.4056, "step": 2660 }, { "epoch": 0.36, "grad_norm": 3.44277668037896, "learning_rate": 9.414335678086149e-06, "loss": 0.2735, "step": 2661 }, { "epoch": 0.36, "grad_norm": 4.274241695035886, "learning_rate": 9.41381826251587e-06, "loss": 0.3422, "step": 2662 }, { "epoch": 0.36, "grad_norm": 4.133233201578972, "learning_rate": 9.413300632718131e-06, "loss": 0.3356, "step": 2663 }, { "epoch": 0.36, "grad_norm": 3.553140183457307, "learning_rate": 9.412782788718056e-06, "loss": 0.3583, "step": 2664 }, { "epoch": 0.36, "grad_norm": 3.827589643074106, "learning_rate": 9.412264730540778e-06, "loss": 0.3715, "step": 2665 }, { "epoch": 0.36, "grad_norm": 3.3090678769996185, "learning_rate": 9.411746458211439e-06, "loss": 0.3272, "step": 2666 }, { "epoch": 0.36, "grad_norm": 3.33198422406773, "learning_rate": 9.411227971755197e-06, "loss": 0.3587, "step": 2667 }, { "epoch": 0.36, "grad_norm": 3.513887734860876, "learning_rate": 9.410709271197216e-06, "loss": 0.3152, "step": 2668 }, { "epoch": 0.36, "grad_norm": 3.581337027047426, "learning_rate": 9.410190356562672e-06, "loss": 0.3485, "step": 2669 }, { "epoch": 0.36, "grad_norm": 3.0559594312476563, "learning_rate": 9.409671227876748e-06, "loss": 0.2959, "step": 2670 }, { "epoch": 0.36, "grad_norm": 3.2020958796366723, "learning_rate": 9.409151885164645e-06, "loss": 0.369, "step": 2671 }, { "epoch": 0.36, "grad_norm": 3.7562239375582505, "learning_rate": 9.408632328451565e-06, "loss": 0.3462, "step": 2672 }, { "epoch": 0.36, "grad_norm": 3.540753544406325, "learning_rate": 9.408112557762727e-06, "loss": 0.2757, "step": 2673 }, { "epoch": 0.36, "grad_norm": 3.108529858337664, "learning_rate": 9.407592573123359e-06, "loss": 0.2806, "step": 2674 }, { "epoch": 0.36, "grad_norm": 4.384661938872271, "learning_rate": 9.407072374558698e-06, "loss": 0.3058, "step": 2675 }, { "epoch": 0.36, "grad_norm": 5.379984428998722, "learning_rate": 9.406551962093993e-06, "loss": 0.2985, "step": 2676 }, { "epoch": 0.36, "grad_norm": 3.406460710109471, "learning_rate": 9.4060313357545e-06, "loss": 0.2992, "step": 2677 }, { "epoch": 0.36, "grad_norm": 3.476940582645142, "learning_rate": 9.405510495565492e-06, "loss": 0.3241, "step": 2678 }, { "epoch": 0.36, "grad_norm": 3.4699970572924324, "learning_rate": 9.404989441552243e-06, "loss": 0.2926, "step": 2679 }, { "epoch": 0.36, "grad_norm": 4.041961215552666, "learning_rate": 9.40446817374005e-06, "loss": 0.2781, "step": 2680 }, { "epoch": 0.36, "grad_norm": 3.4471650582429585, "learning_rate": 9.403946692154205e-06, "loss": 0.3169, "step": 2681 }, { "epoch": 0.36, "grad_norm": 3.253749787699659, "learning_rate": 9.403424996820024e-06, "loss": 0.3269, "step": 2682 }, { "epoch": 0.36, "grad_norm": 3.350952997632391, "learning_rate": 9.402903087762826e-06, "loss": 0.2524, "step": 2683 }, { "epoch": 0.37, "grad_norm": 3.703128421454789, "learning_rate": 9.402380965007942e-06, "loss": 0.3218, "step": 2684 }, { "epoch": 0.37, "grad_norm": 3.1605825369885796, "learning_rate": 9.401858628580713e-06, "loss": 0.3204, "step": 2685 }, { "epoch": 0.37, "grad_norm": 4.1269089336394424, "learning_rate": 9.401336078506493e-06, "loss": 0.3553, "step": 2686 }, { "epoch": 0.37, "grad_norm": 3.3007531560198062, "learning_rate": 9.400813314810644e-06, "loss": 0.29, "step": 2687 }, { "epoch": 0.37, "grad_norm": 3.680379452289437, "learning_rate": 9.400290337518536e-06, "loss": 0.2716, "step": 2688 }, { "epoch": 0.37, "grad_norm": 3.15621660512861, "learning_rate": 9.399767146655555e-06, "loss": 0.3652, "step": 2689 }, { "epoch": 0.37, "grad_norm": 3.936229091626894, "learning_rate": 9.399243742247092e-06, "loss": 0.3268, "step": 2690 }, { "epoch": 0.37, "grad_norm": 3.9904743756308108, "learning_rate": 9.398720124318551e-06, "loss": 0.2941, "step": 2691 }, { "epoch": 0.37, "grad_norm": 4.064335435218328, "learning_rate": 9.39819629289535e-06, "loss": 0.3215, "step": 2692 }, { "epoch": 0.37, "grad_norm": 3.6624243031848702, "learning_rate": 9.397672248002909e-06, "loss": 0.3163, "step": 2693 }, { "epoch": 0.37, "grad_norm": 3.4406584933890367, "learning_rate": 9.397147989666664e-06, "loss": 0.337, "step": 2694 }, { "epoch": 0.37, "grad_norm": 3.281890472026991, "learning_rate": 9.39662351791206e-06, "loss": 0.2768, "step": 2695 }, { "epoch": 0.37, "grad_norm": 3.2363972333771933, "learning_rate": 9.396098832764555e-06, "loss": 0.276, "step": 2696 }, { "epoch": 0.37, "grad_norm": 4.069586132949836, "learning_rate": 9.395573934249614e-06, "loss": 0.3494, "step": 2697 }, { "epoch": 0.37, "grad_norm": 4.158993097175403, "learning_rate": 9.39504882239271e-06, "loss": 0.3673, "step": 2698 }, { "epoch": 0.37, "grad_norm": 3.8812409563106653, "learning_rate": 9.394523497219332e-06, "loss": 0.3896, "step": 2699 }, { "epoch": 0.37, "grad_norm": 4.087398810750772, "learning_rate": 9.393997958754977e-06, "loss": 0.3622, "step": 2700 }, { "epoch": 0.37, "grad_norm": 2.9906853849555675, "learning_rate": 9.393472207025153e-06, "loss": 0.2597, "step": 2701 }, { "epoch": 0.37, "grad_norm": 3.164155998086177, "learning_rate": 9.392946242055379e-06, "loss": 0.2475, "step": 2702 }, { "epoch": 0.37, "grad_norm": 3.6518726250062032, "learning_rate": 9.392420063871178e-06, "loss": 0.3284, "step": 2703 }, { "epoch": 0.37, "grad_norm": 3.812487609482167, "learning_rate": 9.391893672498093e-06, "loss": 0.4338, "step": 2704 }, { "epoch": 0.37, "grad_norm": 3.6605447755953278, "learning_rate": 9.39136706796167e-06, "loss": 0.3173, "step": 2705 }, { "epoch": 0.37, "grad_norm": 3.9846424130004268, "learning_rate": 9.390840250287471e-06, "loss": 0.3391, "step": 2706 }, { "epoch": 0.37, "grad_norm": 3.1903701447043185, "learning_rate": 9.390313219501061e-06, "loss": 0.2856, "step": 2707 }, { "epoch": 0.37, "grad_norm": 3.131325736723797, "learning_rate": 9.389785975628026e-06, "loss": 0.2696, "step": 2708 }, { "epoch": 0.37, "grad_norm": 3.1177634989456844, "learning_rate": 9.389258518693951e-06, "loss": 0.2929, "step": 2709 }, { "epoch": 0.37, "grad_norm": 3.1775406737430427, "learning_rate": 9.388730848724437e-06, "loss": 0.3047, "step": 2710 }, { "epoch": 0.37, "grad_norm": 2.9296420418168285, "learning_rate": 9.388202965745097e-06, "loss": 0.2763, "step": 2711 }, { "epoch": 0.37, "grad_norm": 3.5861994768895276, "learning_rate": 9.38767486978155e-06, "loss": 0.3018, "step": 2712 }, { "epoch": 0.37, "grad_norm": 3.301436776617449, "learning_rate": 9.38714656085943e-06, "loss": 0.3116, "step": 2713 }, { "epoch": 0.37, "grad_norm": 2.959530588066331, "learning_rate": 9.386618039004378e-06, "loss": 0.2993, "step": 2714 }, { "epoch": 0.37, "grad_norm": 3.579986947639408, "learning_rate": 9.386089304242044e-06, "loss": 0.3533, "step": 2715 }, { "epoch": 0.37, "grad_norm": 3.4557575257436777, "learning_rate": 9.385560356598092e-06, "loss": 0.293, "step": 2716 }, { "epoch": 0.37, "grad_norm": 3.278901581855382, "learning_rate": 9.385031196098194e-06, "loss": 0.2946, "step": 2717 }, { "epoch": 0.37, "grad_norm": 3.1443355941605002, "learning_rate": 9.384501822768036e-06, "loss": 0.3065, "step": 2718 }, { "epoch": 0.37, "grad_norm": 3.9068231197682253, "learning_rate": 9.383972236633308e-06, "loss": 0.3578, "step": 2719 }, { "epoch": 0.37, "grad_norm": 3.6018742719350363, "learning_rate": 9.383442437719714e-06, "loss": 0.3788, "step": 2720 }, { "epoch": 0.37, "grad_norm": 3.3358840452150416, "learning_rate": 9.38291242605297e-06, "loss": 0.2766, "step": 2721 }, { "epoch": 0.37, "grad_norm": 3.5433023437888873, "learning_rate": 9.3823822016588e-06, "loss": 0.3048, "step": 2722 }, { "epoch": 0.37, "grad_norm": 3.5111240293207584, "learning_rate": 9.381851764562938e-06, "loss": 0.2755, "step": 2723 }, { "epoch": 0.37, "grad_norm": 3.7951229228442953, "learning_rate": 9.38132111479113e-06, "loss": 0.3282, "step": 2724 }, { "epoch": 0.37, "grad_norm": 3.0419687035030267, "learning_rate": 9.380790252369131e-06, "loss": 0.3058, "step": 2725 }, { "epoch": 0.37, "grad_norm": 3.6589632171314013, "learning_rate": 9.380259177322705e-06, "loss": 0.2941, "step": 2726 }, { "epoch": 0.37, "grad_norm": 3.397785438042012, "learning_rate": 9.379727889677632e-06, "loss": 0.329, "step": 2727 }, { "epoch": 0.37, "grad_norm": 4.141909047814298, "learning_rate": 9.379196389459697e-06, "loss": 0.29, "step": 2728 }, { "epoch": 0.37, "grad_norm": 3.5332700073666468, "learning_rate": 9.378664676694693e-06, "loss": 0.3191, "step": 2729 }, { "epoch": 0.37, "grad_norm": 4.223129435494906, "learning_rate": 9.378132751408433e-06, "loss": 0.3258, "step": 2730 }, { "epoch": 0.37, "grad_norm": 3.248076995166896, "learning_rate": 9.37760061362673e-06, "loss": 0.2841, "step": 2731 }, { "epoch": 0.37, "grad_norm": 4.087482390031128, "learning_rate": 9.377068263375411e-06, "loss": 0.3205, "step": 2732 }, { "epoch": 0.37, "grad_norm": 3.573443405050588, "learning_rate": 9.376535700680318e-06, "loss": 0.2898, "step": 2733 }, { "epoch": 0.37, "grad_norm": 4.095450346538232, "learning_rate": 9.376002925567296e-06, "loss": 0.2408, "step": 2734 }, { "epoch": 0.37, "grad_norm": 3.2023259432512425, "learning_rate": 9.375469938062205e-06, "loss": 0.3222, "step": 2735 }, { "epoch": 0.37, "grad_norm": 3.152126780290029, "learning_rate": 9.374936738190913e-06, "loss": 0.2881, "step": 2736 }, { "epoch": 0.37, "grad_norm": 11.043565947134844, "learning_rate": 9.374403325979301e-06, "loss": 0.3686, "step": 2737 }, { "epoch": 0.37, "grad_norm": 3.893818295774342, "learning_rate": 9.373869701453256e-06, "loss": 0.3354, "step": 2738 }, { "epoch": 0.37, "grad_norm": 2.9182239127368312, "learning_rate": 9.373335864638679e-06, "loss": 0.3273, "step": 2739 }, { "epoch": 0.37, "grad_norm": 3.7957454013090652, "learning_rate": 9.372801815561481e-06, "loss": 0.237, "step": 2740 }, { "epoch": 0.37, "grad_norm": 3.1416030002283337, "learning_rate": 9.372267554247583e-06, "loss": 0.2729, "step": 2741 }, { "epoch": 0.37, "grad_norm": 3.439178269962087, "learning_rate": 9.371733080722911e-06, "loss": 0.3446, "step": 2742 }, { "epoch": 0.37, "grad_norm": 3.864944335999969, "learning_rate": 9.37119839501341e-06, "loss": 0.3454, "step": 2743 }, { "epoch": 0.37, "grad_norm": 3.554163686188431, "learning_rate": 9.370663497145033e-06, "loss": 0.3132, "step": 2744 }, { "epoch": 0.37, "grad_norm": 3.0263065230242767, "learning_rate": 9.370128387143737e-06, "loss": 0.2581, "step": 2745 }, { "epoch": 0.37, "grad_norm": 3.6828466409309883, "learning_rate": 9.369593065035496e-06, "loss": 0.3196, "step": 2746 }, { "epoch": 0.37, "grad_norm": 3.7663280369717596, "learning_rate": 9.369057530846294e-06, "loss": 0.3695, "step": 2747 }, { "epoch": 0.37, "grad_norm": 3.7977354576905396, "learning_rate": 9.36852178460212e-06, "loss": 0.3357, "step": 2748 }, { "epoch": 0.37, "grad_norm": 3.3440604612256366, "learning_rate": 9.36798582632898e-06, "loss": 0.3555, "step": 2749 }, { "epoch": 0.37, "grad_norm": 4.231181407515953, "learning_rate": 9.367449656052885e-06, "loss": 0.356, "step": 2750 }, { "epoch": 0.37, "grad_norm": 3.708136478813516, "learning_rate": 9.366913273799859e-06, "loss": 0.3362, "step": 2751 }, { "epoch": 0.37, "grad_norm": 3.5927474946694633, "learning_rate": 9.366376679595936e-06, "loss": 0.3212, "step": 2752 }, { "epoch": 0.37, "grad_norm": 3.5378173967736974, "learning_rate": 9.36583987346716e-06, "loss": 0.3531, "step": 2753 }, { "epoch": 0.37, "grad_norm": 3.944013378807077, "learning_rate": 9.365302855439584e-06, "loss": 0.368, "step": 2754 }, { "epoch": 0.37, "grad_norm": 4.058086386109027, "learning_rate": 9.364765625539273e-06, "loss": 0.3211, "step": 2755 }, { "epoch": 0.37, "grad_norm": 3.1701900370140605, "learning_rate": 9.364228183792303e-06, "loss": 0.33, "step": 2756 }, { "epoch": 0.38, "grad_norm": 3.549217419461763, "learning_rate": 9.363690530224757e-06, "loss": 0.3969, "step": 2757 }, { "epoch": 0.38, "grad_norm": 3.1368870971010443, "learning_rate": 9.363152664862734e-06, "loss": 0.3328, "step": 2758 }, { "epoch": 0.38, "grad_norm": 3.386454254634835, "learning_rate": 9.362614587732334e-06, "loss": 0.2923, "step": 2759 }, { "epoch": 0.38, "grad_norm": 3.1530263691124047, "learning_rate": 9.362076298859677e-06, "loss": 0.2976, "step": 2760 }, { "epoch": 0.38, "grad_norm": 3.5504139631249325, "learning_rate": 9.36153779827089e-06, "loss": 0.2934, "step": 2761 }, { "epoch": 0.38, "grad_norm": 3.0079305449607623, "learning_rate": 9.360999085992106e-06, "loss": 0.29, "step": 2762 }, { "epoch": 0.38, "grad_norm": 3.0716309980044385, "learning_rate": 9.360460162049473e-06, "loss": 0.3065, "step": 2763 }, { "epoch": 0.38, "grad_norm": 3.781451741383236, "learning_rate": 9.359921026469152e-06, "loss": 0.2794, "step": 2764 }, { "epoch": 0.38, "grad_norm": 3.904421598243603, "learning_rate": 9.359381679277303e-06, "loss": 0.3642, "step": 2765 }, { "epoch": 0.38, "grad_norm": 3.272000612647564, "learning_rate": 9.358842120500109e-06, "loss": 0.2978, "step": 2766 }, { "epoch": 0.38, "grad_norm": 4.132260894550544, "learning_rate": 9.358302350163758e-06, "loss": 0.354, "step": 2767 }, { "epoch": 0.38, "grad_norm": 3.7513666692080436, "learning_rate": 9.357762368294443e-06, "loss": 0.3667, "step": 2768 }, { "epoch": 0.38, "grad_norm": 3.8395775963933607, "learning_rate": 9.357222174918376e-06, "loss": 0.3306, "step": 2769 }, { "epoch": 0.38, "grad_norm": 3.089401824822482, "learning_rate": 9.356681770061777e-06, "loss": 0.2849, "step": 2770 }, { "epoch": 0.38, "grad_norm": 4.158928963006987, "learning_rate": 9.356141153750873e-06, "loss": 0.3327, "step": 2771 }, { "epoch": 0.38, "grad_norm": 3.809880092678141, "learning_rate": 9.355600326011903e-06, "loss": 0.2837, "step": 2772 }, { "epoch": 0.38, "grad_norm": 3.718616532308441, "learning_rate": 9.355059286871115e-06, "loss": 0.3288, "step": 2773 }, { "epoch": 0.38, "grad_norm": 3.782001253758118, "learning_rate": 9.354518036354774e-06, "loss": 0.3066, "step": 2774 }, { "epoch": 0.38, "grad_norm": 3.353596125528348, "learning_rate": 9.353976574489143e-06, "loss": 0.3198, "step": 2775 }, { "epoch": 0.38, "grad_norm": 3.157375349555216, "learning_rate": 9.353434901300507e-06, "loss": 0.2845, "step": 2776 }, { "epoch": 0.38, "grad_norm": 4.242750935948519, "learning_rate": 9.352893016815155e-06, "loss": 0.2923, "step": 2777 }, { "epoch": 0.38, "grad_norm": 4.169899237999498, "learning_rate": 9.352350921059387e-06, "loss": 0.3618, "step": 2778 }, { "epoch": 0.38, "grad_norm": 3.3838924826669983, "learning_rate": 9.351808614059516e-06, "loss": 0.3326, "step": 2779 }, { "epoch": 0.38, "grad_norm": 3.3602677134618855, "learning_rate": 9.351266095841861e-06, "loss": 0.3007, "step": 2780 }, { "epoch": 0.38, "grad_norm": 3.359277732135042, "learning_rate": 9.350723366432755e-06, "loss": 0.2895, "step": 2781 }, { "epoch": 0.38, "grad_norm": 3.5635188098600072, "learning_rate": 9.350180425858538e-06, "loss": 0.3433, "step": 2782 }, { "epoch": 0.38, "grad_norm": 3.757666238043574, "learning_rate": 9.349637274145563e-06, "loss": 0.2791, "step": 2783 }, { "epoch": 0.38, "grad_norm": 3.7887367073753686, "learning_rate": 9.349093911320192e-06, "loss": 0.2719, "step": 2784 }, { "epoch": 0.38, "grad_norm": 3.2475617697822643, "learning_rate": 9.348550337408799e-06, "loss": 0.3168, "step": 2785 }, { "epoch": 0.38, "grad_norm": 4.335421393830996, "learning_rate": 9.348006552437765e-06, "loss": 0.3035, "step": 2786 }, { "epoch": 0.38, "grad_norm": 3.7292233047836736, "learning_rate": 9.347462556433483e-06, "loss": 0.3395, "step": 2787 }, { "epoch": 0.38, "grad_norm": 3.1912019278405923, "learning_rate": 9.346918349422356e-06, "loss": 0.3075, "step": 2788 }, { "epoch": 0.38, "grad_norm": 3.5554721015685913, "learning_rate": 9.346373931430799e-06, "loss": 0.2983, "step": 2789 }, { "epoch": 0.38, "grad_norm": 2.9270015609473523, "learning_rate": 9.345829302485234e-06, "loss": 0.2359, "step": 2790 }, { "epoch": 0.38, "grad_norm": 3.8442183283179023, "learning_rate": 9.345284462612097e-06, "loss": 0.2876, "step": 2791 }, { "epoch": 0.38, "grad_norm": 3.093329897842043, "learning_rate": 9.34473941183783e-06, "loss": 0.2662, "step": 2792 }, { "epoch": 0.38, "grad_norm": 3.3617431556321655, "learning_rate": 9.344194150188885e-06, "loss": 0.3469, "step": 2793 }, { "epoch": 0.38, "grad_norm": 3.6759946871142684, "learning_rate": 9.343648677691733e-06, "loss": 0.2961, "step": 2794 }, { "epoch": 0.38, "grad_norm": 3.8472230076216336, "learning_rate": 9.343102994372845e-06, "loss": 0.3775, "step": 2795 }, { "epoch": 0.38, "grad_norm": 3.616111192103147, "learning_rate": 9.342557100258708e-06, "loss": 0.3211, "step": 2796 }, { "epoch": 0.38, "grad_norm": 3.2156289644392784, "learning_rate": 9.342010995375811e-06, "loss": 0.3192, "step": 2797 }, { "epoch": 0.38, "grad_norm": 3.769664031639845, "learning_rate": 9.341464679750669e-06, "loss": 0.3225, "step": 2798 }, { "epoch": 0.38, "grad_norm": 3.5659359405898394, "learning_rate": 9.340918153409793e-06, "loss": 0.3143, "step": 2799 }, { "epoch": 0.38, "grad_norm": 3.5651457891463516, "learning_rate": 9.340371416379708e-06, "loss": 0.2708, "step": 2800 }, { "epoch": 0.38, "grad_norm": 3.4892936705990008, "learning_rate": 9.33982446868695e-06, "loss": 0.2325, "step": 2801 }, { "epoch": 0.38, "grad_norm": 4.7043057602942095, "learning_rate": 9.33927731035807e-06, "loss": 0.3132, "step": 2802 }, { "epoch": 0.38, "grad_norm": 3.7384529358364498, "learning_rate": 9.338729941419621e-06, "loss": 0.3349, "step": 2803 }, { "epoch": 0.38, "grad_norm": 3.8746795999468135, "learning_rate": 9.33818236189817e-06, "loss": 0.2991, "step": 2804 }, { "epoch": 0.38, "grad_norm": 2.997773877500825, "learning_rate": 9.337634571820294e-06, "loss": 0.3007, "step": 2805 }, { "epoch": 0.38, "grad_norm": 3.76550317306074, "learning_rate": 9.337086571212581e-06, "loss": 0.367, "step": 2806 }, { "epoch": 0.38, "grad_norm": 4.29158941230933, "learning_rate": 9.336538360101631e-06, "loss": 0.2816, "step": 2807 }, { "epoch": 0.38, "grad_norm": 3.0784419238374143, "learning_rate": 9.335989938514045e-06, "loss": 0.288, "step": 2808 }, { "epoch": 0.38, "grad_norm": 3.4356981345352122, "learning_rate": 9.335441306476449e-06, "loss": 0.3007, "step": 2809 }, { "epoch": 0.38, "grad_norm": 3.795659109168898, "learning_rate": 9.334892464015466e-06, "loss": 0.3345, "step": 2810 }, { "epoch": 0.38, "grad_norm": 3.427083135346511, "learning_rate": 9.334343411157735e-06, "loss": 0.3112, "step": 2811 }, { "epoch": 0.38, "grad_norm": 3.7284227306137483, "learning_rate": 9.333794147929907e-06, "loss": 0.2591, "step": 2812 }, { "epoch": 0.38, "grad_norm": 3.907883064735523, "learning_rate": 9.33324467435864e-06, "loss": 0.3488, "step": 2813 }, { "epoch": 0.38, "grad_norm": 3.4735437635716853, "learning_rate": 9.3326949904706e-06, "loss": 0.3739, "step": 2814 }, { "epoch": 0.38, "grad_norm": 3.554647009719693, "learning_rate": 9.33214509629247e-06, "loss": 0.3281, "step": 2815 }, { "epoch": 0.38, "grad_norm": 3.431618279267299, "learning_rate": 9.33159499185094e-06, "loss": 0.2802, "step": 2816 }, { "epoch": 0.38, "grad_norm": 3.0473597434218793, "learning_rate": 9.331044677172705e-06, "loss": 0.3237, "step": 2817 }, { "epoch": 0.38, "grad_norm": 3.2042546262385647, "learning_rate": 9.33049415228448e-06, "loss": 0.3133, "step": 2818 }, { "epoch": 0.38, "grad_norm": 3.6265307221792757, "learning_rate": 9.329943417212981e-06, "loss": 0.3125, "step": 2819 }, { "epoch": 0.38, "grad_norm": 3.151436122084683, "learning_rate": 9.329392471984941e-06, "loss": 0.26, "step": 2820 }, { "epoch": 0.38, "grad_norm": 3.6882877142224904, "learning_rate": 9.328841316627099e-06, "loss": 0.2817, "step": 2821 }, { "epoch": 0.38, "grad_norm": 2.820947076366334, "learning_rate": 9.328289951166205e-06, "loss": 0.2474, "step": 2822 }, { "epoch": 0.38, "grad_norm": 3.607941764840144, "learning_rate": 9.327738375629023e-06, "loss": 0.2861, "step": 2823 }, { "epoch": 0.38, "grad_norm": 3.5349576511137917, "learning_rate": 9.32718659004232e-06, "loss": 0.3089, "step": 2824 }, { "epoch": 0.38, "grad_norm": 2.9208949312962686, "learning_rate": 9.326634594432882e-06, "loss": 0.2875, "step": 2825 }, { "epoch": 0.38, "grad_norm": 3.178283421068841, "learning_rate": 9.326082388827496e-06, "loss": 0.2911, "step": 2826 }, { "epoch": 0.38, "grad_norm": 3.3022805367981123, "learning_rate": 9.325529973252967e-06, "loss": 0.3084, "step": 2827 }, { "epoch": 0.38, "grad_norm": 3.6843644560670383, "learning_rate": 9.324977347736104e-06, "loss": 0.336, "step": 2828 }, { "epoch": 0.38, "grad_norm": 4.067390235694695, "learning_rate": 9.32442451230373e-06, "loss": 0.3123, "step": 2829 }, { "epoch": 0.38, "grad_norm": 3.210551892699556, "learning_rate": 9.323871466982679e-06, "loss": 0.2633, "step": 2830 }, { "epoch": 0.39, "grad_norm": 4.304041762234377, "learning_rate": 9.32331821179979e-06, "loss": 0.3051, "step": 2831 }, { "epoch": 0.39, "grad_norm": 3.6488328809240356, "learning_rate": 9.32276474678192e-06, "loss": 0.2927, "step": 2832 }, { "epoch": 0.39, "grad_norm": 3.734740981466433, "learning_rate": 9.322211071955927e-06, "loss": 0.2516, "step": 2833 }, { "epoch": 0.39, "grad_norm": 3.44596091855582, "learning_rate": 9.321657187348689e-06, "loss": 0.2783, "step": 2834 }, { "epoch": 0.39, "grad_norm": 3.166492159320205, "learning_rate": 9.321103092987085e-06, "loss": 0.2435, "step": 2835 }, { "epoch": 0.39, "grad_norm": 3.471476340091668, "learning_rate": 9.32054878889801e-06, "loss": 0.3053, "step": 2836 }, { "epoch": 0.39, "grad_norm": 3.167921388327324, "learning_rate": 9.319994275108365e-06, "loss": 0.285, "step": 2837 }, { "epoch": 0.39, "grad_norm": 3.885456227701532, "learning_rate": 9.319439551645068e-06, "loss": 0.3109, "step": 2838 }, { "epoch": 0.39, "grad_norm": 5.2987824635824925, "learning_rate": 9.318884618535042e-06, "loss": 0.2989, "step": 2839 }, { "epoch": 0.39, "grad_norm": 3.7404351306319588, "learning_rate": 9.31832947580522e-06, "loss": 0.3091, "step": 2840 }, { "epoch": 0.39, "grad_norm": 3.062713731922783, "learning_rate": 9.317774123482543e-06, "loss": 0.3137, "step": 2841 }, { "epoch": 0.39, "grad_norm": 3.539780444086311, "learning_rate": 9.31721856159397e-06, "loss": 0.324, "step": 2842 }, { "epoch": 0.39, "grad_norm": 3.7520807267974345, "learning_rate": 9.316662790166463e-06, "loss": 0.3361, "step": 2843 }, { "epoch": 0.39, "grad_norm": 3.4756110937995617, "learning_rate": 9.316106809226999e-06, "loss": 0.2703, "step": 2844 }, { "epoch": 0.39, "grad_norm": 4.586259149712066, "learning_rate": 9.31555061880256e-06, "loss": 0.2999, "step": 2845 }, { "epoch": 0.39, "grad_norm": 3.38079020288487, "learning_rate": 9.314994218920144e-06, "loss": 0.3216, "step": 2846 }, { "epoch": 0.39, "grad_norm": 3.2746172021365263, "learning_rate": 9.314437609606754e-06, "loss": 0.3085, "step": 2847 }, { "epoch": 0.39, "grad_norm": 3.8877668483064096, "learning_rate": 9.313880790889407e-06, "loss": 0.3223, "step": 2848 }, { "epoch": 0.39, "grad_norm": 2.969878985932587, "learning_rate": 9.313323762795128e-06, "loss": 0.3237, "step": 2849 }, { "epoch": 0.39, "grad_norm": 3.110689049401815, "learning_rate": 9.31276652535095e-06, "loss": 0.3047, "step": 2850 }, { "epoch": 0.39, "grad_norm": 3.676518943421437, "learning_rate": 9.312209078583926e-06, "loss": 0.3378, "step": 2851 }, { "epoch": 0.39, "grad_norm": 2.7567928952423526, "learning_rate": 9.311651422521103e-06, "loss": 0.2778, "step": 2852 }, { "epoch": 0.39, "grad_norm": 3.797765676109861, "learning_rate": 9.311093557189554e-06, "loss": 0.3199, "step": 2853 }, { "epoch": 0.39, "grad_norm": 3.7667720938687994, "learning_rate": 9.310535482616352e-06, "loss": 0.3048, "step": 2854 }, { "epoch": 0.39, "grad_norm": 4.203013397829128, "learning_rate": 9.309977198828583e-06, "loss": 0.3271, "step": 2855 }, { "epoch": 0.39, "grad_norm": 3.8656785090925756, "learning_rate": 9.309418705853348e-06, "loss": 0.3664, "step": 2856 }, { "epoch": 0.39, "grad_norm": 3.568453827111757, "learning_rate": 9.308860003717748e-06, "loss": 0.2492, "step": 2857 }, { "epoch": 0.39, "grad_norm": 3.9345848098887797, "learning_rate": 9.308301092448906e-06, "loss": 0.2958, "step": 2858 }, { "epoch": 0.39, "grad_norm": 3.335573160939399, "learning_rate": 9.307741972073944e-06, "loss": 0.2868, "step": 2859 }, { "epoch": 0.39, "grad_norm": 3.746960019823314, "learning_rate": 9.307182642620001e-06, "loss": 0.2949, "step": 2860 }, { "epoch": 0.39, "grad_norm": 2.9190011300206318, "learning_rate": 9.306623104114225e-06, "loss": 0.261, "step": 2861 }, { "epoch": 0.39, "grad_norm": 3.2978838579533627, "learning_rate": 9.306063356583772e-06, "loss": 0.272, "step": 2862 }, { "epoch": 0.39, "grad_norm": 3.2676848045120703, "learning_rate": 9.305503400055813e-06, "loss": 0.3347, "step": 2863 }, { "epoch": 0.39, "grad_norm": 3.1008122126157898, "learning_rate": 9.304943234557523e-06, "loss": 0.2845, "step": 2864 }, { "epoch": 0.39, "grad_norm": 3.5182190966034983, "learning_rate": 9.30438286011609e-06, "loss": 0.2593, "step": 2865 }, { "epoch": 0.39, "grad_norm": 4.086440384237418, "learning_rate": 9.303822276758714e-06, "loss": 0.3162, "step": 2866 }, { "epoch": 0.39, "grad_norm": 2.9512728422782777, "learning_rate": 9.3032614845126e-06, "loss": 0.276, "step": 2867 }, { "epoch": 0.39, "grad_norm": 3.4383370035679306, "learning_rate": 9.30270048340497e-06, "loss": 0.3101, "step": 2868 }, { "epoch": 0.39, "grad_norm": 3.8619233195777727, "learning_rate": 9.30213927346305e-06, "loss": 0.3551, "step": 2869 }, { "epoch": 0.39, "grad_norm": 3.5300496623588327, "learning_rate": 9.301577854714082e-06, "loss": 0.2646, "step": 2870 }, { "epoch": 0.39, "grad_norm": 3.600408641822414, "learning_rate": 9.30101622718531e-06, "loss": 0.3455, "step": 2871 }, { "epoch": 0.39, "grad_norm": 3.731732233343748, "learning_rate": 9.300454390903999e-06, "loss": 0.3443, "step": 2872 }, { "epoch": 0.39, "grad_norm": 3.150828932897379, "learning_rate": 9.299892345897412e-06, "loss": 0.2856, "step": 2873 }, { "epoch": 0.39, "grad_norm": 3.713078003104897, "learning_rate": 9.29933009219283e-06, "loss": 0.2293, "step": 2874 }, { "epoch": 0.39, "grad_norm": 7.029076869159636, "learning_rate": 9.298767629817544e-06, "loss": 0.3663, "step": 2875 }, { "epoch": 0.39, "grad_norm": 3.66357373201581, "learning_rate": 9.298204958798853e-06, "loss": 0.3051, "step": 2876 }, { "epoch": 0.39, "grad_norm": 3.4525845047048604, "learning_rate": 9.297642079164067e-06, "loss": 0.2827, "step": 2877 }, { "epoch": 0.39, "grad_norm": 4.358356583781535, "learning_rate": 9.297078990940504e-06, "loss": 0.3795, "step": 2878 }, { "epoch": 0.39, "grad_norm": 4.551471817815703, "learning_rate": 9.296515694155495e-06, "loss": 0.3631, "step": 2879 }, { "epoch": 0.39, "grad_norm": 3.4533084151454982, "learning_rate": 9.29595218883638e-06, "loss": 0.3296, "step": 2880 }, { "epoch": 0.39, "grad_norm": 3.557108619665075, "learning_rate": 9.295388475010506e-06, "loss": 0.2777, "step": 2881 }, { "epoch": 0.39, "grad_norm": 3.8637478757055703, "learning_rate": 9.294824552705238e-06, "loss": 0.3204, "step": 2882 }, { "epoch": 0.39, "grad_norm": 4.123153461271816, "learning_rate": 9.294260421947945e-06, "loss": 0.3727, "step": 2883 }, { "epoch": 0.39, "grad_norm": 5.6782624077147075, "learning_rate": 9.293696082766006e-06, "loss": 0.3418, "step": 2884 }, { "epoch": 0.39, "grad_norm": 3.2293632328545163, "learning_rate": 9.293131535186813e-06, "loss": 0.3049, "step": 2885 }, { "epoch": 0.39, "grad_norm": 3.0710724865124894, "learning_rate": 9.292566779237765e-06, "loss": 0.2793, "step": 2886 }, { "epoch": 0.39, "grad_norm": 4.1074347166454865, "learning_rate": 9.292001814946275e-06, "loss": 0.3623, "step": 2887 }, { "epoch": 0.39, "grad_norm": 3.3863444013011894, "learning_rate": 9.291436642339762e-06, "loss": 0.2851, "step": 2888 }, { "epoch": 0.39, "grad_norm": 3.456011155695491, "learning_rate": 9.290871261445657e-06, "loss": 0.3185, "step": 2889 }, { "epoch": 0.39, "grad_norm": 3.9939119923534974, "learning_rate": 9.290305672291402e-06, "loss": 0.3449, "step": 2890 }, { "epoch": 0.39, "grad_norm": 3.6527695616293974, "learning_rate": 9.289739874904448e-06, "loss": 0.308, "step": 2891 }, { "epoch": 0.39, "grad_norm": 3.2807430174837635, "learning_rate": 9.289173869312259e-06, "loss": 0.33, "step": 2892 }, { "epoch": 0.39, "grad_norm": 3.5802362208950975, "learning_rate": 9.288607655542302e-06, "loss": 0.3393, "step": 2893 }, { "epoch": 0.39, "grad_norm": 3.1524418997218304, "learning_rate": 9.288041233622058e-06, "loss": 0.3032, "step": 2894 }, { "epoch": 0.39, "grad_norm": 3.15823754338483, "learning_rate": 9.287474603579023e-06, "loss": 0.2826, "step": 2895 }, { "epoch": 0.39, "grad_norm": 3.2723906709543917, "learning_rate": 9.286907765440698e-06, "loss": 0.312, "step": 2896 }, { "epoch": 0.39, "grad_norm": 3.247618543231338, "learning_rate": 9.286340719234592e-06, "loss": 0.2882, "step": 2897 }, { "epoch": 0.39, "grad_norm": 3.6740315251162494, "learning_rate": 9.28577346498823e-06, "loss": 0.3513, "step": 2898 }, { "epoch": 0.39, "grad_norm": 3.2852440863883543, "learning_rate": 9.285206002729142e-06, "loss": 0.3221, "step": 2899 }, { "epoch": 0.39, "grad_norm": 3.466783065689903, "learning_rate": 9.28463833248487e-06, "loss": 0.3262, "step": 2900 }, { "epoch": 0.39, "grad_norm": 2.6464899736430914, "learning_rate": 9.284070454282968e-06, "loss": 0.2839, "step": 2901 }, { "epoch": 0.39, "grad_norm": 3.600195199403884, "learning_rate": 9.283502368150996e-06, "loss": 0.3303, "step": 2902 }, { "epoch": 0.39, "grad_norm": 3.1694419985098548, "learning_rate": 9.282934074116528e-06, "loss": 0.2852, "step": 2903 }, { "epoch": 0.4, "grad_norm": 3.2586985121870957, "learning_rate": 9.282365572207146e-06, "loss": 0.3207, "step": 2904 }, { "epoch": 0.4, "grad_norm": 3.4316771288013403, "learning_rate": 9.281796862450443e-06, "loss": 0.3242, "step": 2905 }, { "epoch": 0.4, "grad_norm": 3.7863992439414185, "learning_rate": 9.281227944874024e-06, "loss": 0.3502, "step": 2906 }, { "epoch": 0.4, "grad_norm": 4.160069376357478, "learning_rate": 9.280658819505495e-06, "loss": 0.2964, "step": 2907 }, { "epoch": 0.4, "grad_norm": 3.3425132103117052, "learning_rate": 9.280089486372486e-06, "loss": 0.2986, "step": 2908 }, { "epoch": 0.4, "grad_norm": 3.698615361509277, "learning_rate": 9.279519945502626e-06, "loss": 0.3469, "step": 2909 }, { "epoch": 0.4, "grad_norm": 3.4027025489251885, "learning_rate": 9.27895019692356e-06, "loss": 0.3664, "step": 2910 }, { "epoch": 0.4, "grad_norm": 4.025620946337271, "learning_rate": 9.278380240662939e-06, "loss": 0.2714, "step": 2911 }, { "epoch": 0.4, "grad_norm": 3.691451860171158, "learning_rate": 9.277810076748427e-06, "loss": 0.3419, "step": 2912 }, { "epoch": 0.4, "grad_norm": 3.7883431082055403, "learning_rate": 9.277239705207702e-06, "loss": 0.3587, "step": 2913 }, { "epoch": 0.4, "grad_norm": 3.4037961023873313, "learning_rate": 9.276669126068439e-06, "loss": 0.3391, "step": 2914 }, { "epoch": 0.4, "grad_norm": 3.129509685016227, "learning_rate": 9.276098339358335e-06, "loss": 0.2608, "step": 2915 }, { "epoch": 0.4, "grad_norm": 3.552310169570235, "learning_rate": 9.275527345105096e-06, "loss": 0.3779, "step": 2916 }, { "epoch": 0.4, "grad_norm": 3.4732046869893693, "learning_rate": 9.274956143336433e-06, "loss": 0.3258, "step": 2917 }, { "epoch": 0.4, "grad_norm": 5.039545956385628, "learning_rate": 9.27438473408007e-06, "loss": 0.3945, "step": 2918 }, { "epoch": 0.4, "grad_norm": 3.6940122986309496, "learning_rate": 9.27381311736374e-06, "loss": 0.3321, "step": 2919 }, { "epoch": 0.4, "grad_norm": 2.9847056878648486, "learning_rate": 9.27324129321519e-06, "loss": 0.3197, "step": 2920 }, { "epoch": 0.4, "grad_norm": 3.525827442681831, "learning_rate": 9.272669261662168e-06, "loss": 0.3644, "step": 2921 }, { "epoch": 0.4, "grad_norm": 3.232695237913099, "learning_rate": 9.272097022732444e-06, "loss": 0.3582, "step": 2922 }, { "epoch": 0.4, "grad_norm": 3.4486154756010627, "learning_rate": 9.271524576453788e-06, "loss": 0.2998, "step": 2923 }, { "epoch": 0.4, "grad_norm": 3.5686928467931676, "learning_rate": 9.270951922853986e-06, "loss": 0.3697, "step": 2924 }, { "epoch": 0.4, "grad_norm": 4.5179174510511375, "learning_rate": 9.270379061960832e-06, "loss": 0.3112, "step": 2925 }, { "epoch": 0.4, "grad_norm": 3.0911635885640973, "learning_rate": 9.26980599380213e-06, "loss": 0.2974, "step": 2926 }, { "epoch": 0.4, "grad_norm": 3.5581451940730777, "learning_rate": 9.269232718405692e-06, "loss": 0.3278, "step": 2927 }, { "epoch": 0.4, "grad_norm": 3.9767052725464143, "learning_rate": 9.268659235799344e-06, "loss": 0.3283, "step": 2928 }, { "epoch": 0.4, "grad_norm": 3.322614195639491, "learning_rate": 9.268085546010923e-06, "loss": 0.2867, "step": 2929 }, { "epoch": 0.4, "grad_norm": 2.9180478941695274, "learning_rate": 9.26751164906827e-06, "loss": 0.2153, "step": 2930 }, { "epoch": 0.4, "grad_norm": 3.2512363082067464, "learning_rate": 9.266937544999239e-06, "loss": 0.3542, "step": 2931 }, { "epoch": 0.4, "grad_norm": 3.882546673604499, "learning_rate": 9.266363233831697e-06, "loss": 0.3309, "step": 2932 }, { "epoch": 0.4, "grad_norm": 3.7168328866882745, "learning_rate": 9.265788715593518e-06, "loss": 0.3169, "step": 2933 }, { "epoch": 0.4, "grad_norm": 3.427302178821499, "learning_rate": 9.265213990312585e-06, "loss": 0.3335, "step": 2934 }, { "epoch": 0.4, "grad_norm": 3.5901950807622787, "learning_rate": 9.264639058016794e-06, "loss": 0.3397, "step": 2935 }, { "epoch": 0.4, "grad_norm": 3.154288879816114, "learning_rate": 9.26406391873405e-06, "loss": 0.2755, "step": 2936 }, { "epoch": 0.4, "grad_norm": 3.345008294569031, "learning_rate": 9.263488572492267e-06, "loss": 0.3645, "step": 2937 }, { "epoch": 0.4, "grad_norm": 3.3969859639042945, "learning_rate": 9.26291301931937e-06, "loss": 0.2976, "step": 2938 }, { "epoch": 0.4, "grad_norm": 3.694954057099213, "learning_rate": 9.262337259243292e-06, "loss": 0.2557, "step": 2939 }, { "epoch": 0.4, "grad_norm": 3.1944645388838593, "learning_rate": 9.261761292291981e-06, "loss": 0.3291, "step": 2940 }, { "epoch": 0.4, "grad_norm": 2.813527371539373, "learning_rate": 9.261185118493392e-06, "loss": 0.2665, "step": 2941 }, { "epoch": 0.4, "grad_norm": 3.3064722484153277, "learning_rate": 9.260608737875487e-06, "loss": 0.3345, "step": 2942 }, { "epoch": 0.4, "grad_norm": 2.9552892654796787, "learning_rate": 9.260032150466242e-06, "loss": 0.2619, "step": 2943 }, { "epoch": 0.4, "grad_norm": 2.719987905473901, "learning_rate": 9.259455356293644e-06, "loss": 0.2176, "step": 2944 }, { "epoch": 0.4, "grad_norm": 3.7906314973255326, "learning_rate": 9.258878355385685e-06, "loss": 0.2658, "step": 2945 }, { "epoch": 0.4, "grad_norm": 3.2134120537129025, "learning_rate": 9.258301147770375e-06, "loss": 0.2969, "step": 2946 }, { "epoch": 0.4, "grad_norm": 4.290899610633465, "learning_rate": 9.257723733475723e-06, "loss": 0.2868, "step": 2947 }, { "epoch": 0.4, "grad_norm": 5.090521721614074, "learning_rate": 9.257146112529758e-06, "loss": 0.3269, "step": 2948 }, { "epoch": 0.4, "grad_norm": 3.1515968878960634, "learning_rate": 9.256568284960513e-06, "loss": 0.2507, "step": 2949 }, { "epoch": 0.4, "grad_norm": 4.080117369092223, "learning_rate": 9.255990250796036e-06, "loss": 0.3397, "step": 2950 }, { "epoch": 0.4, "grad_norm": 3.7684502366218435, "learning_rate": 9.25541201006438e-06, "loss": 0.3258, "step": 2951 }, { "epoch": 0.4, "grad_norm": 3.75051356751554, "learning_rate": 9.25483356279361e-06, "loss": 0.309, "step": 2952 }, { "epoch": 0.4, "grad_norm": 3.8365969753564526, "learning_rate": 9.254254909011805e-06, "loss": 0.3835, "step": 2953 }, { "epoch": 0.4, "grad_norm": 3.6468592120046206, "learning_rate": 9.253676048747044e-06, "loss": 0.2883, "step": 2954 }, { "epoch": 0.4, "grad_norm": 3.3717271734027627, "learning_rate": 9.25309698202743e-06, "loss": 0.2789, "step": 2955 }, { "epoch": 0.4, "grad_norm": 3.671943219521924, "learning_rate": 9.252517708881062e-06, "loss": 0.2788, "step": 2956 }, { "epoch": 0.4, "grad_norm": 3.1406796711380327, "learning_rate": 9.251938229336057e-06, "loss": 0.2461, "step": 2957 }, { "epoch": 0.4, "grad_norm": 3.640718501853126, "learning_rate": 9.251358543420543e-06, "loss": 0.3291, "step": 2958 }, { "epoch": 0.4, "grad_norm": 4.300231631677044, "learning_rate": 9.250778651162653e-06, "loss": 0.3227, "step": 2959 }, { "epoch": 0.4, "grad_norm": 3.295346678313004, "learning_rate": 9.250198552590532e-06, "loss": 0.3377, "step": 2960 }, { "epoch": 0.4, "grad_norm": 3.2923099239880442, "learning_rate": 9.249618247732337e-06, "loss": 0.337, "step": 2961 }, { "epoch": 0.4, "grad_norm": 3.5604194940033014, "learning_rate": 9.249037736616235e-06, "loss": 0.3769, "step": 2962 }, { "epoch": 0.4, "grad_norm": 3.304040393844614, "learning_rate": 9.248457019270396e-06, "loss": 0.3329, "step": 2963 }, { "epoch": 0.4, "grad_norm": 3.6620414875578766, "learning_rate": 9.247876095723011e-06, "loss": 0.3467, "step": 2964 }, { "epoch": 0.4, "grad_norm": 4.510564493353915, "learning_rate": 9.247294966002274e-06, "loss": 0.3297, "step": 2965 }, { "epoch": 0.4, "grad_norm": 3.001247919385904, "learning_rate": 9.246713630136388e-06, "loss": 0.2806, "step": 2966 }, { "epoch": 0.4, "grad_norm": 3.252556302188951, "learning_rate": 9.24613208815357e-06, "loss": 0.3181, "step": 2967 }, { "epoch": 0.4, "grad_norm": 3.612827793487503, "learning_rate": 9.245550340082048e-06, "loss": 0.2787, "step": 2968 }, { "epoch": 0.4, "grad_norm": 4.432012256901483, "learning_rate": 9.244968385950055e-06, "loss": 0.2868, "step": 2969 }, { "epoch": 0.4, "grad_norm": 3.064294842186907, "learning_rate": 9.244386225785837e-06, "loss": 0.3122, "step": 2970 }, { "epoch": 0.4, "grad_norm": 3.1896140735644183, "learning_rate": 9.243803859617649e-06, "loss": 0.3193, "step": 2971 }, { "epoch": 0.4, "grad_norm": 3.2879663518748146, "learning_rate": 9.243221287473755e-06, "loss": 0.2893, "step": 2972 }, { "epoch": 0.4, "grad_norm": 3.372765462492879, "learning_rate": 9.242638509382434e-06, "loss": 0.2488, "step": 2973 }, { "epoch": 0.4, "grad_norm": 3.6794537404155077, "learning_rate": 9.242055525371971e-06, "loss": 0.2737, "step": 2974 }, { "epoch": 0.4, "grad_norm": 4.48925401658578, "learning_rate": 9.24147233547066e-06, "loss": 0.2744, "step": 2975 }, { "epoch": 0.4, "grad_norm": 3.012282493823136, "learning_rate": 9.240888939706805e-06, "loss": 0.3211, "step": 2976 }, { "epoch": 0.4, "grad_norm": 3.9698774904479346, "learning_rate": 9.240305338108726e-06, "loss": 0.3447, "step": 2977 }, { "epoch": 0.41, "grad_norm": 4.062577826577079, "learning_rate": 9.239721530704744e-06, "loss": 0.3262, "step": 2978 }, { "epoch": 0.41, "grad_norm": 2.9589807335670755, "learning_rate": 9.239137517523198e-06, "loss": 0.2868, "step": 2979 }, { "epoch": 0.41, "grad_norm": 3.610680742888123, "learning_rate": 9.238553298592429e-06, "loss": 0.301, "step": 2980 }, { "epoch": 0.41, "grad_norm": 3.559466379863997, "learning_rate": 9.2379688739408e-06, "loss": 0.3078, "step": 2981 }, { "epoch": 0.41, "grad_norm": 3.6062204486066323, "learning_rate": 9.237384243596667e-06, "loss": 0.2811, "step": 2982 }, { "epoch": 0.41, "grad_norm": 3.3469628025283282, "learning_rate": 9.236799407588412e-06, "loss": 0.2889, "step": 2983 }, { "epoch": 0.41, "grad_norm": 3.303983492874655, "learning_rate": 9.236214365944418e-06, "loss": 0.3488, "step": 2984 }, { "epoch": 0.41, "grad_norm": 3.3032975531557183, "learning_rate": 9.235629118693081e-06, "loss": 0.264, "step": 2985 }, { "epoch": 0.41, "grad_norm": 3.0753216953706324, "learning_rate": 9.235043665862806e-06, "loss": 0.3303, "step": 2986 }, { "epoch": 0.41, "grad_norm": 3.623629068715079, "learning_rate": 9.23445800748201e-06, "loss": 0.3093, "step": 2987 }, { "epoch": 0.41, "grad_norm": 3.1151334161055546, "learning_rate": 9.233872143579115e-06, "loss": 0.2867, "step": 2988 }, { "epoch": 0.41, "grad_norm": 3.444605695022197, "learning_rate": 9.23328607418256e-06, "loss": 0.2563, "step": 2989 }, { "epoch": 0.41, "grad_norm": 3.809047481283885, "learning_rate": 9.232699799320785e-06, "loss": 0.3126, "step": 2990 }, { "epoch": 0.41, "grad_norm": 2.854275330191533, "learning_rate": 9.23211331902225e-06, "loss": 0.2637, "step": 2991 }, { "epoch": 0.41, "grad_norm": 3.418952704746495, "learning_rate": 9.231526633315419e-06, "loss": 0.2903, "step": 2992 }, { "epoch": 0.41, "grad_norm": 3.0337268519590483, "learning_rate": 9.230939742228768e-06, "loss": 0.2537, "step": 2993 }, { "epoch": 0.41, "grad_norm": 3.294033596948967, "learning_rate": 9.23035264579078e-06, "loss": 0.2825, "step": 2994 }, { "epoch": 0.41, "grad_norm": 4.447983275820442, "learning_rate": 9.22976534402995e-06, "loss": 0.3419, "step": 2995 }, { "epoch": 0.41, "grad_norm": 3.6189499673148884, "learning_rate": 9.229177836974785e-06, "loss": 0.3365, "step": 2996 }, { "epoch": 0.41, "grad_norm": 4.02901878223799, "learning_rate": 9.2285901246538e-06, "loss": 0.3236, "step": 2997 }, { "epoch": 0.41, "grad_norm": 3.5318984172354257, "learning_rate": 9.22800220709552e-06, "loss": 0.3303, "step": 2998 }, { "epoch": 0.41, "grad_norm": 3.2210717901850283, "learning_rate": 9.227414084328477e-06, "loss": 0.3665, "step": 2999 }, { "epoch": 0.41, "grad_norm": 3.369622416635978, "learning_rate": 9.226825756381218e-06, "loss": 0.3193, "step": 3000 }, { "epoch": 0.41, "grad_norm": 3.4868363594716043, "learning_rate": 9.226237223282299e-06, "loss": 0.3433, "step": 3001 }, { "epoch": 0.41, "grad_norm": 3.2265913925943397, "learning_rate": 9.225648485060283e-06, "loss": 0.3077, "step": 3002 }, { "epoch": 0.41, "grad_norm": 3.0694751273356857, "learning_rate": 9.225059541743745e-06, "loss": 0.2582, "step": 3003 }, { "epoch": 0.41, "grad_norm": 5.127703467226999, "learning_rate": 9.224470393361271e-06, "loss": 0.3545, "step": 3004 }, { "epoch": 0.41, "grad_norm": 3.359487442590105, "learning_rate": 9.223881039941455e-06, "loss": 0.2644, "step": 3005 }, { "epoch": 0.41, "grad_norm": 3.561172671372699, "learning_rate": 9.223291481512903e-06, "loss": 0.3066, "step": 3006 }, { "epoch": 0.41, "grad_norm": 3.125023753323209, "learning_rate": 9.222701718104226e-06, "loss": 0.3046, "step": 3007 }, { "epoch": 0.41, "grad_norm": 3.3091391443264544, "learning_rate": 9.222111749744052e-06, "loss": 0.3472, "step": 3008 }, { "epoch": 0.41, "grad_norm": 3.032502733929489, "learning_rate": 9.221521576461014e-06, "loss": 0.3342, "step": 3009 }, { "epoch": 0.41, "grad_norm": 3.7262690506905143, "learning_rate": 9.220931198283757e-06, "loss": 0.2757, "step": 3010 }, { "epoch": 0.41, "grad_norm": 3.768546515774847, "learning_rate": 9.220340615240934e-06, "loss": 0.2821, "step": 3011 }, { "epoch": 0.41, "grad_norm": 3.2763823839432917, "learning_rate": 9.21974982736121e-06, "loss": 0.2832, "step": 3012 }, { "epoch": 0.41, "grad_norm": 3.9091820436579736, "learning_rate": 9.219158834673261e-06, "loss": 0.331, "step": 3013 }, { "epoch": 0.41, "grad_norm": 3.3056642457425034, "learning_rate": 9.21856763720577e-06, "loss": 0.3124, "step": 3014 }, { "epoch": 0.41, "grad_norm": 4.120637497806574, "learning_rate": 9.217976234987429e-06, "loss": 0.3266, "step": 3015 }, { "epoch": 0.41, "grad_norm": 4.356522282060987, "learning_rate": 9.217384628046947e-06, "loss": 0.3751, "step": 3016 }, { "epoch": 0.41, "grad_norm": 3.1411971664975744, "learning_rate": 9.21679281641303e-06, "loss": 0.2435, "step": 3017 }, { "epoch": 0.41, "grad_norm": 3.207139827416924, "learning_rate": 9.216200800114412e-06, "loss": 0.3583, "step": 3018 }, { "epoch": 0.41, "grad_norm": 3.7271453238792933, "learning_rate": 9.21560857917982e-06, "loss": 0.323, "step": 3019 }, { "epoch": 0.41, "grad_norm": 3.8215435777126525, "learning_rate": 9.215016153637998e-06, "loss": 0.3322, "step": 3020 }, { "epoch": 0.41, "grad_norm": 3.7978897747203595, "learning_rate": 9.214423523517704e-06, "loss": 0.354, "step": 3021 }, { "epoch": 0.41, "grad_norm": 2.8773239545822977, "learning_rate": 9.2138306888477e-06, "loss": 0.3187, "step": 3022 }, { "epoch": 0.41, "grad_norm": 3.627581699940949, "learning_rate": 9.213237649656757e-06, "loss": 0.324, "step": 3023 }, { "epoch": 0.41, "grad_norm": 3.264741188084682, "learning_rate": 9.212644405973658e-06, "loss": 0.3178, "step": 3024 }, { "epoch": 0.41, "grad_norm": 3.1476808747094944, "learning_rate": 9.212050957827202e-06, "loss": 0.3003, "step": 3025 }, { "epoch": 0.41, "grad_norm": 3.689854872313485, "learning_rate": 9.211457305246189e-06, "loss": 0.3254, "step": 3026 }, { "epoch": 0.41, "grad_norm": 3.01775617212241, "learning_rate": 9.21086344825943e-06, "loss": 0.2906, "step": 3027 }, { "epoch": 0.41, "grad_norm": 3.5868421245893587, "learning_rate": 9.210269386895752e-06, "loss": 0.3205, "step": 3028 }, { "epoch": 0.41, "grad_norm": 3.518868835985213, "learning_rate": 9.209675121183988e-06, "loss": 0.2945, "step": 3029 }, { "epoch": 0.41, "grad_norm": 3.68392745428086, "learning_rate": 9.209080651152978e-06, "loss": 0.2983, "step": 3030 }, { "epoch": 0.41, "grad_norm": 3.772656837007922, "learning_rate": 9.208485976831579e-06, "loss": 0.3579, "step": 3031 }, { "epoch": 0.41, "grad_norm": 3.366646379717961, "learning_rate": 9.207891098248648e-06, "loss": 0.3143, "step": 3032 }, { "epoch": 0.41, "grad_norm": 3.765211503948153, "learning_rate": 9.207296015433065e-06, "loss": 0.3004, "step": 3033 }, { "epoch": 0.41, "grad_norm": 3.3492160076546047, "learning_rate": 9.206700728413707e-06, "loss": 0.2817, "step": 3034 }, { "epoch": 0.41, "grad_norm": 3.499210837683972, "learning_rate": 9.206105237219471e-06, "loss": 0.3375, "step": 3035 }, { "epoch": 0.41, "grad_norm": 3.99561777141976, "learning_rate": 9.205509541879257e-06, "loss": 0.3097, "step": 3036 }, { "epoch": 0.41, "grad_norm": 4.000063158612441, "learning_rate": 9.204913642421977e-06, "loss": 0.3269, "step": 3037 }, { "epoch": 0.41, "grad_norm": 3.4706861258725485, "learning_rate": 9.204317538876555e-06, "loss": 0.3287, "step": 3038 }, { "epoch": 0.41, "grad_norm": 3.8753901887823887, "learning_rate": 9.20372123127192e-06, "loss": 0.2741, "step": 3039 }, { "epoch": 0.41, "grad_norm": 3.799014498287299, "learning_rate": 9.20312471963702e-06, "loss": 0.3716, "step": 3040 }, { "epoch": 0.41, "grad_norm": 3.443291035780327, "learning_rate": 9.202528004000801e-06, "loss": 0.2891, "step": 3041 }, { "epoch": 0.41, "grad_norm": 3.558863704690877, "learning_rate": 9.20193108439223e-06, "loss": 0.3069, "step": 3042 }, { "epoch": 0.41, "grad_norm": 3.6516113003924295, "learning_rate": 9.201333960840275e-06, "loss": 0.308, "step": 3043 }, { "epoch": 0.41, "grad_norm": 3.970859122031615, "learning_rate": 9.20073663337392e-06, "loss": 0.3004, "step": 3044 }, { "epoch": 0.41, "grad_norm": 3.5377854078904267, "learning_rate": 9.200139102022154e-06, "loss": 0.3341, "step": 3045 }, { "epoch": 0.41, "grad_norm": 3.631399600314653, "learning_rate": 9.199541366813984e-06, "loss": 0.3165, "step": 3046 }, { "epoch": 0.41, "grad_norm": 3.0811697145311734, "learning_rate": 9.198943427778415e-06, "loss": 0.2924, "step": 3047 }, { "epoch": 0.41, "grad_norm": 4.265765280374754, "learning_rate": 9.198345284944473e-06, "loss": 0.3566, "step": 3048 }, { "epoch": 0.41, "grad_norm": 3.6544897168686967, "learning_rate": 9.197746938341187e-06, "loss": 0.3019, "step": 3049 }, { "epoch": 0.41, "grad_norm": 2.973817258821886, "learning_rate": 9.197148387997599e-06, "loss": 0.2731, "step": 3050 }, { "epoch": 0.42, "grad_norm": 3.9025447912895186, "learning_rate": 9.19654963394276e-06, "loss": 0.2639, "step": 3051 }, { "epoch": 0.42, "grad_norm": 3.711481636086182, "learning_rate": 9.19595067620573e-06, "loss": 0.2664, "step": 3052 }, { "epoch": 0.42, "grad_norm": 3.78536525865468, "learning_rate": 9.195351514815581e-06, "loss": 0.3619, "step": 3053 }, { "epoch": 0.42, "grad_norm": 3.807916175743485, "learning_rate": 9.194752149801394e-06, "loss": 0.3654, "step": 3054 }, { "epoch": 0.42, "grad_norm": 3.286878805194883, "learning_rate": 9.194152581192258e-06, "loss": 0.3153, "step": 3055 }, { "epoch": 0.42, "grad_norm": 3.4748116415255743, "learning_rate": 9.193552809017273e-06, "loss": 0.2518, "step": 3056 }, { "epoch": 0.42, "grad_norm": 3.6561544047077876, "learning_rate": 9.19295283330555e-06, "loss": 0.3145, "step": 3057 }, { "epoch": 0.42, "grad_norm": 3.1679491539023235, "learning_rate": 9.192352654086212e-06, "loss": 0.3262, "step": 3058 }, { "epoch": 0.42, "grad_norm": 3.3291218002259804, "learning_rate": 9.191752271388385e-06, "loss": 0.3059, "step": 3059 }, { "epoch": 0.42, "grad_norm": 3.523161845386632, "learning_rate": 9.191151685241211e-06, "loss": 0.2654, "step": 3060 }, { "epoch": 0.42, "grad_norm": 3.897270961719186, "learning_rate": 9.190550895673841e-06, "loss": 0.331, "step": 3061 }, { "epoch": 0.42, "grad_norm": 3.2196010891187146, "learning_rate": 9.189949902715432e-06, "loss": 0.2769, "step": 3062 }, { "epoch": 0.42, "grad_norm": 3.565493785398354, "learning_rate": 9.189348706395154e-06, "loss": 0.2919, "step": 3063 }, { "epoch": 0.42, "grad_norm": 3.4330162647391194, "learning_rate": 9.18874730674219e-06, "loss": 0.3609, "step": 3064 }, { "epoch": 0.42, "grad_norm": 3.906675487411138, "learning_rate": 9.188145703785724e-06, "loss": 0.291, "step": 3065 }, { "epoch": 0.42, "grad_norm": 3.5043854039668676, "learning_rate": 9.187543897554958e-06, "loss": 0.3147, "step": 3066 }, { "epoch": 0.42, "grad_norm": 3.704895566372291, "learning_rate": 9.1869418880791e-06, "loss": 0.3483, "step": 3067 }, { "epoch": 0.42, "grad_norm": 3.8475268333792534, "learning_rate": 9.18633967538737e-06, "loss": 0.3621, "step": 3068 }, { "epoch": 0.42, "grad_norm": 4.071691840663559, "learning_rate": 9.185737259509e-06, "loss": 0.3432, "step": 3069 }, { "epoch": 0.42, "grad_norm": 4.812376315664203, "learning_rate": 9.185134640473222e-06, "loss": 0.265, "step": 3070 }, { "epoch": 0.42, "grad_norm": 3.0211911157098004, "learning_rate": 9.184531818309287e-06, "loss": 0.2745, "step": 3071 }, { "epoch": 0.42, "grad_norm": 3.5605618125837113, "learning_rate": 9.183928793046456e-06, "loss": 0.3449, "step": 3072 }, { "epoch": 0.42, "grad_norm": 2.953598422568759, "learning_rate": 9.183325564713994e-06, "loss": 0.2848, "step": 3073 }, { "epoch": 0.42, "grad_norm": 3.4797887141163986, "learning_rate": 9.182722133341178e-06, "loss": 0.3402, "step": 3074 }, { "epoch": 0.42, "grad_norm": 4.009411860034554, "learning_rate": 9.182118498957301e-06, "loss": 0.3392, "step": 3075 }, { "epoch": 0.42, "grad_norm": 4.827909434950316, "learning_rate": 9.181514661591657e-06, "loss": 0.3712, "step": 3076 }, { "epoch": 0.42, "grad_norm": 3.6087930728175595, "learning_rate": 9.180910621273555e-06, "loss": 0.3259, "step": 3077 }, { "epoch": 0.42, "grad_norm": 3.6986372398109877, "learning_rate": 9.180306378032313e-06, "loss": 0.3145, "step": 3078 }, { "epoch": 0.42, "grad_norm": 4.112147703037047, "learning_rate": 9.179701931897255e-06, "loss": 0.2729, "step": 3079 }, { "epoch": 0.42, "grad_norm": 3.518517306498294, "learning_rate": 9.179097282897722e-06, "loss": 0.256, "step": 3080 }, { "epoch": 0.42, "grad_norm": 3.6765867645689614, "learning_rate": 9.17849243106306e-06, "loss": 0.3535, "step": 3081 }, { "epoch": 0.42, "grad_norm": 3.1233833400776425, "learning_rate": 9.177887376422624e-06, "loss": 0.3042, "step": 3082 }, { "epoch": 0.42, "grad_norm": 3.83854187596778, "learning_rate": 9.177282119005783e-06, "loss": 0.3045, "step": 3083 }, { "epoch": 0.42, "grad_norm": 2.8829102429682885, "learning_rate": 9.176676658841912e-06, "loss": 0.2611, "step": 3084 }, { "epoch": 0.42, "grad_norm": 3.6658545965509517, "learning_rate": 9.176070995960399e-06, "loss": 0.3151, "step": 3085 }, { "epoch": 0.42, "grad_norm": 3.42104373916461, "learning_rate": 9.175465130390639e-06, "loss": 0.3724, "step": 3086 }, { "epoch": 0.42, "grad_norm": 3.9910466678635355, "learning_rate": 9.174859062162037e-06, "loss": 0.3771, "step": 3087 }, { "epoch": 0.42, "grad_norm": 3.815551540297001, "learning_rate": 9.174252791304012e-06, "loss": 0.2794, "step": 3088 }, { "epoch": 0.42, "grad_norm": 3.1917495058298773, "learning_rate": 9.173646317845988e-06, "loss": 0.3019, "step": 3089 }, { "epoch": 0.42, "grad_norm": 3.491303822396133, "learning_rate": 9.1730396418174e-06, "loss": 0.3629, "step": 3090 }, { "epoch": 0.42, "grad_norm": 3.204521822594564, "learning_rate": 9.172432763247694e-06, "loss": 0.2925, "step": 3091 }, { "epoch": 0.42, "grad_norm": 3.4434825753335154, "learning_rate": 9.171825682166325e-06, "loss": 0.3023, "step": 3092 }, { "epoch": 0.42, "grad_norm": 3.8965204540509024, "learning_rate": 9.171218398602758e-06, "loss": 0.3584, "step": 3093 }, { "epoch": 0.42, "grad_norm": 4.296262250904497, "learning_rate": 9.17061091258647e-06, "loss": 0.3465, "step": 3094 }, { "epoch": 0.42, "grad_norm": 3.279350972940552, "learning_rate": 9.170003224146941e-06, "loss": 0.2649, "step": 3095 }, { "epoch": 0.42, "grad_norm": 3.1748363785845215, "learning_rate": 9.16939533331367e-06, "loss": 0.3147, "step": 3096 }, { "epoch": 0.42, "grad_norm": 3.5724539158098074, "learning_rate": 9.168787240116162e-06, "loss": 0.2942, "step": 3097 }, { "epoch": 0.42, "grad_norm": 3.18197666130981, "learning_rate": 9.168178944583927e-06, "loss": 0.2813, "step": 3098 }, { "epoch": 0.42, "grad_norm": 3.5273965896561856, "learning_rate": 9.167570446746489e-06, "loss": 0.2897, "step": 3099 }, { "epoch": 0.42, "grad_norm": 3.4796693807507713, "learning_rate": 9.166961746633387e-06, "loss": 0.2872, "step": 3100 }, { "epoch": 0.42, "grad_norm": 3.761957990806958, "learning_rate": 9.16635284427416e-06, "loss": 0.2889, "step": 3101 }, { "epoch": 0.42, "grad_norm": 3.787839896965025, "learning_rate": 9.165743739698364e-06, "loss": 0.3664, "step": 3102 }, { "epoch": 0.42, "grad_norm": 3.5838753466398243, "learning_rate": 9.16513443293556e-06, "loss": 0.3, "step": 3103 }, { "epoch": 0.42, "grad_norm": 4.184051070337082, "learning_rate": 9.164524924015324e-06, "loss": 0.3488, "step": 3104 }, { "epoch": 0.42, "grad_norm": 4.417930673594007, "learning_rate": 9.163915212967236e-06, "loss": 0.3261, "step": 3105 }, { "epoch": 0.42, "grad_norm": 4.772950159013205, "learning_rate": 9.163305299820887e-06, "loss": 0.2992, "step": 3106 }, { "epoch": 0.42, "grad_norm": 3.1800496634826034, "learning_rate": 9.162695184605887e-06, "loss": 0.2949, "step": 3107 }, { "epoch": 0.42, "grad_norm": 5.839453640657351, "learning_rate": 9.16208486735184e-06, "loss": 0.284, "step": 3108 }, { "epoch": 0.42, "grad_norm": 3.0421187776373095, "learning_rate": 9.161474348088377e-06, "loss": 0.259, "step": 3109 }, { "epoch": 0.42, "grad_norm": 3.3297920920111252, "learning_rate": 9.16086362684512e-06, "loss": 0.3547, "step": 3110 }, { "epoch": 0.42, "grad_norm": 3.7127101324614635, "learning_rate": 9.160252703651718e-06, "loss": 0.3106, "step": 3111 }, { "epoch": 0.42, "grad_norm": 4.067129352789907, "learning_rate": 9.15964157853782e-06, "loss": 0.3341, "step": 3112 }, { "epoch": 0.42, "grad_norm": 3.308489886033917, "learning_rate": 9.159030251533087e-06, "loss": 0.3057, "step": 3113 }, { "epoch": 0.42, "grad_norm": 3.4883793678706256, "learning_rate": 9.15841872266719e-06, "loss": 0.296, "step": 3114 }, { "epoch": 0.42, "grad_norm": 3.7930619736743614, "learning_rate": 9.15780699196981e-06, "loss": 0.2531, "step": 3115 }, { "epoch": 0.42, "grad_norm": 2.996519217420893, "learning_rate": 9.157195059470641e-06, "loss": 0.2993, "step": 3116 }, { "epoch": 0.42, "grad_norm": 3.8560266496575175, "learning_rate": 9.15658292519938e-06, "loss": 0.2822, "step": 3117 }, { "epoch": 0.42, "grad_norm": 3.27213750854163, "learning_rate": 9.155970589185737e-06, "loss": 0.2704, "step": 3118 }, { "epoch": 0.42, "grad_norm": 3.4707614715872994, "learning_rate": 9.155358051459432e-06, "loss": 0.2877, "step": 3119 }, { "epoch": 0.42, "grad_norm": 3.2566851511104744, "learning_rate": 9.154745312050198e-06, "loss": 0.3478, "step": 3120 }, { "epoch": 0.42, "grad_norm": 3.681437499832049, "learning_rate": 9.154132370987773e-06, "loss": 0.2598, "step": 3121 }, { "epoch": 0.42, "grad_norm": 4.118367238695721, "learning_rate": 9.153519228301907e-06, "loss": 0.3065, "step": 3122 }, { "epoch": 0.42, "grad_norm": 3.0919654676837167, "learning_rate": 9.152905884022356e-06, "loss": 0.3502, "step": 3123 }, { "epoch": 0.42, "grad_norm": 3.2937380710744284, "learning_rate": 9.152292338178894e-06, "loss": 0.3413, "step": 3124 }, { "epoch": 0.43, "grad_norm": 3.4773067038151733, "learning_rate": 9.151678590801297e-06, "loss": 0.3303, "step": 3125 }, { "epoch": 0.43, "grad_norm": 3.3569834816764725, "learning_rate": 9.151064641919355e-06, "loss": 0.2786, "step": 3126 }, { "epoch": 0.43, "grad_norm": 3.4587741044226434, "learning_rate": 9.150450491562864e-06, "loss": 0.35, "step": 3127 }, { "epoch": 0.43, "grad_norm": 3.817566842084153, "learning_rate": 9.149836139761633e-06, "loss": 0.3305, "step": 3128 }, { "epoch": 0.43, "grad_norm": 3.225571084914654, "learning_rate": 9.149221586545482e-06, "loss": 0.2613, "step": 3129 }, { "epoch": 0.43, "grad_norm": 3.22442471924784, "learning_rate": 9.148606831944235e-06, "loss": 0.2751, "step": 3130 }, { "epoch": 0.43, "grad_norm": 2.979842991472142, "learning_rate": 9.147991875987735e-06, "loss": 0.2598, "step": 3131 }, { "epoch": 0.43, "grad_norm": 3.330330728801712, "learning_rate": 9.147376718705825e-06, "loss": 0.2962, "step": 3132 }, { "epoch": 0.43, "grad_norm": 3.9629963388211213, "learning_rate": 9.146761360128363e-06, "loss": 0.3325, "step": 3133 }, { "epoch": 0.43, "grad_norm": 3.8096264292177966, "learning_rate": 9.146145800285216e-06, "loss": 0.2677, "step": 3134 }, { "epoch": 0.43, "grad_norm": 4.00980520660621, "learning_rate": 9.145530039206261e-06, "loss": 0.3195, "step": 3135 }, { "epoch": 0.43, "grad_norm": 4.338609358497998, "learning_rate": 9.144914076921386e-06, "loss": 0.3478, "step": 3136 }, { "epoch": 0.43, "grad_norm": 3.1874781442489435, "learning_rate": 9.144297913460481e-06, "loss": 0.3184, "step": 3137 }, { "epoch": 0.43, "grad_norm": 3.216272635260998, "learning_rate": 9.143681548853458e-06, "loss": 0.2518, "step": 3138 }, { "epoch": 0.43, "grad_norm": 3.499934645844375, "learning_rate": 9.14306498313023e-06, "loss": 0.2615, "step": 3139 }, { "epoch": 0.43, "grad_norm": 4.035606840600295, "learning_rate": 9.142448216320724e-06, "loss": 0.3089, "step": 3140 }, { "epoch": 0.43, "grad_norm": 2.8204156302526395, "learning_rate": 9.141831248454873e-06, "loss": 0.2409, "step": 3141 }, { "epoch": 0.43, "grad_norm": 3.0594871711784024, "learning_rate": 9.141214079562624e-06, "loss": 0.2765, "step": 3142 }, { "epoch": 0.43, "grad_norm": 3.6526993647321633, "learning_rate": 9.140596709673932e-06, "loss": 0.2331, "step": 3143 }, { "epoch": 0.43, "grad_norm": 3.108754488548819, "learning_rate": 9.139979138818757e-06, "loss": 0.2557, "step": 3144 }, { "epoch": 0.43, "grad_norm": 3.0388689233545345, "learning_rate": 9.13936136702708e-06, "loss": 0.2529, "step": 3145 }, { "epoch": 0.43, "grad_norm": 3.8816030115348377, "learning_rate": 9.138743394328879e-06, "loss": 0.3724, "step": 3146 }, { "epoch": 0.43, "grad_norm": 3.482751247124158, "learning_rate": 9.13812522075415e-06, "loss": 0.3222, "step": 3147 }, { "epoch": 0.43, "grad_norm": 3.745428464940513, "learning_rate": 9.137506846332898e-06, "loss": 0.3196, "step": 3148 }, { "epoch": 0.43, "grad_norm": 3.2882394152708625, "learning_rate": 9.136888271095132e-06, "loss": 0.3027, "step": 3149 }, { "epoch": 0.43, "grad_norm": 2.8639749345675973, "learning_rate": 9.136269495070879e-06, "loss": 0.2804, "step": 3150 }, { "epoch": 0.43, "grad_norm": 3.2700298148049987, "learning_rate": 9.135650518290171e-06, "loss": 0.2899, "step": 3151 }, { "epoch": 0.43, "grad_norm": 3.4675017528360517, "learning_rate": 9.13503134078305e-06, "loss": 0.2891, "step": 3152 }, { "epoch": 0.43, "grad_norm": 3.6392179769125477, "learning_rate": 9.134411962579568e-06, "loss": 0.3297, "step": 3153 }, { "epoch": 0.43, "grad_norm": 4.412825455479258, "learning_rate": 9.133792383709785e-06, "loss": 0.2682, "step": 3154 }, { "epoch": 0.43, "grad_norm": 3.644684243760564, "learning_rate": 9.133172604203776e-06, "loss": 0.3088, "step": 3155 }, { "epoch": 0.43, "grad_norm": 3.7554001207460175, "learning_rate": 9.13255262409162e-06, "loss": 0.2814, "step": 3156 }, { "epoch": 0.43, "grad_norm": 3.644962296335754, "learning_rate": 9.13193244340341e-06, "loss": 0.3208, "step": 3157 }, { "epoch": 0.43, "grad_norm": 3.0148763471038538, "learning_rate": 9.131312062169247e-06, "loss": 0.3016, "step": 3158 }, { "epoch": 0.43, "grad_norm": 3.1812173049950996, "learning_rate": 9.130691480419238e-06, "loss": 0.2796, "step": 3159 }, { "epoch": 0.43, "grad_norm": 2.991840941642122, "learning_rate": 9.130070698183506e-06, "loss": 0.2824, "step": 3160 }, { "epoch": 0.43, "grad_norm": 4.120294720243241, "learning_rate": 9.129449715492183e-06, "loss": 0.2776, "step": 3161 }, { "epoch": 0.43, "grad_norm": 3.490205596043655, "learning_rate": 9.128828532375404e-06, "loss": 0.296, "step": 3162 }, { "epoch": 0.43, "grad_norm": 3.2715967895439526, "learning_rate": 9.128207148863322e-06, "loss": 0.2827, "step": 3163 }, { "epoch": 0.43, "grad_norm": 3.1858348521047697, "learning_rate": 9.127585564986095e-06, "loss": 0.24, "step": 3164 }, { "epoch": 0.43, "grad_norm": 3.856429748355323, "learning_rate": 9.126963780773895e-06, "loss": 0.2878, "step": 3165 }, { "epoch": 0.43, "grad_norm": 3.351945164516175, "learning_rate": 9.126341796256895e-06, "loss": 0.2875, "step": 3166 }, { "epoch": 0.43, "grad_norm": 3.4952761044716936, "learning_rate": 9.125719611465287e-06, "loss": 0.2639, "step": 3167 }, { "epoch": 0.43, "grad_norm": 3.5649696469480867, "learning_rate": 9.12509722642927e-06, "loss": 0.3338, "step": 3168 }, { "epoch": 0.43, "grad_norm": 3.1447925689096294, "learning_rate": 9.124474641179048e-06, "loss": 0.2891, "step": 3169 }, { "epoch": 0.43, "grad_norm": 3.2499939292575912, "learning_rate": 9.123851855744842e-06, "loss": 0.258, "step": 3170 }, { "epoch": 0.43, "grad_norm": 5.0193606005298586, "learning_rate": 9.12322887015688e-06, "loss": 0.3269, "step": 3171 }, { "epoch": 0.43, "grad_norm": 4.874456259198068, "learning_rate": 9.122605684445397e-06, "loss": 0.3156, "step": 3172 }, { "epoch": 0.43, "grad_norm": 3.521695199627063, "learning_rate": 9.12198229864064e-06, "loss": 0.2891, "step": 3173 }, { "epoch": 0.43, "grad_norm": 3.433571167572371, "learning_rate": 9.121358712772862e-06, "loss": 0.3074, "step": 3174 }, { "epoch": 0.43, "grad_norm": 3.7230233391784378, "learning_rate": 9.120734926872337e-06, "loss": 0.3129, "step": 3175 }, { "epoch": 0.43, "grad_norm": 3.544096652639471, "learning_rate": 9.120110940969333e-06, "loss": 0.278, "step": 3176 }, { "epoch": 0.43, "grad_norm": 3.4882007351228865, "learning_rate": 9.119486755094143e-06, "loss": 0.3142, "step": 3177 }, { "epoch": 0.43, "grad_norm": 3.5119461304952413, "learning_rate": 9.118862369277055e-06, "loss": 0.2776, "step": 3178 }, { "epoch": 0.43, "grad_norm": 2.7586331484629456, "learning_rate": 9.118237783548376e-06, "loss": 0.2294, "step": 3179 }, { "epoch": 0.43, "grad_norm": 3.410046261939833, "learning_rate": 9.117612997938425e-06, "loss": 0.3187, "step": 3180 }, { "epoch": 0.43, "grad_norm": 3.3843642777284955, "learning_rate": 9.116988012477522e-06, "loss": 0.2727, "step": 3181 }, { "epoch": 0.43, "grad_norm": 3.948314976445244, "learning_rate": 9.116362827196002e-06, "loss": 0.2742, "step": 3182 }, { "epoch": 0.43, "grad_norm": 3.3587362103921117, "learning_rate": 9.11573744212421e-06, "loss": 0.3078, "step": 3183 }, { "epoch": 0.43, "grad_norm": 3.9664943121542344, "learning_rate": 9.115111857292499e-06, "loss": 0.2868, "step": 3184 }, { "epoch": 0.43, "grad_norm": 3.324983968691708, "learning_rate": 9.114486072731228e-06, "loss": 0.2533, "step": 3185 }, { "epoch": 0.43, "grad_norm": 3.189797328860412, "learning_rate": 9.113860088470777e-06, "loss": 0.318, "step": 3186 }, { "epoch": 0.43, "grad_norm": 3.629861902495209, "learning_rate": 9.113233904541524e-06, "loss": 0.3148, "step": 3187 }, { "epoch": 0.43, "grad_norm": 2.934511254848689, "learning_rate": 9.112607520973863e-06, "loss": 0.2619, "step": 3188 }, { "epoch": 0.43, "grad_norm": 3.7744140493237874, "learning_rate": 9.111980937798193e-06, "loss": 0.2843, "step": 3189 }, { "epoch": 0.43, "grad_norm": 4.044291644747398, "learning_rate": 9.111354155044929e-06, "loss": 0.2863, "step": 3190 }, { "epoch": 0.43, "grad_norm": 3.3621048946362797, "learning_rate": 9.110727172744492e-06, "loss": 0.3085, "step": 3191 }, { "epoch": 0.43, "grad_norm": 3.477019043609793, "learning_rate": 9.110099990927311e-06, "loss": 0.323, "step": 3192 }, { "epoch": 0.43, "grad_norm": 3.247470687255231, "learning_rate": 9.109472609623829e-06, "loss": 0.304, "step": 3193 }, { "epoch": 0.43, "grad_norm": 3.975735602293111, "learning_rate": 9.108845028864492e-06, "loss": 0.2365, "step": 3194 }, { "epoch": 0.43, "grad_norm": 3.1802098076763037, "learning_rate": 9.108217248679768e-06, "loss": 0.3299, "step": 3195 }, { "epoch": 0.43, "grad_norm": 3.486103127025599, "learning_rate": 9.107589269100118e-06, "loss": 0.298, "step": 3196 }, { "epoch": 0.43, "grad_norm": 4.234924881744724, "learning_rate": 9.106961090156026e-06, "loss": 0.2675, "step": 3197 }, { "epoch": 0.44, "grad_norm": 3.47207873613142, "learning_rate": 9.10633271187798e-06, "loss": 0.3458, "step": 3198 }, { "epoch": 0.44, "grad_norm": 3.213605630524654, "learning_rate": 9.10570413429648e-06, "loss": 0.3052, "step": 3199 }, { "epoch": 0.44, "grad_norm": 3.8700027989173815, "learning_rate": 9.105075357442034e-06, "loss": 0.2718, "step": 3200 }, { "epoch": 0.44, "grad_norm": 3.154252043939407, "learning_rate": 9.10444638134516e-06, "loss": 0.2314, "step": 3201 }, { "epoch": 0.44, "grad_norm": 4.369085418250159, "learning_rate": 9.103817206036383e-06, "loss": 0.2669, "step": 3202 }, { "epoch": 0.44, "grad_norm": 3.4680850302667494, "learning_rate": 9.103187831546243e-06, "loss": 0.251, "step": 3203 }, { "epoch": 0.44, "grad_norm": 3.2111617165268815, "learning_rate": 9.102558257905288e-06, "loss": 0.297, "step": 3204 }, { "epoch": 0.44, "grad_norm": 21.713237111973317, "learning_rate": 9.101928485144072e-06, "loss": 0.324, "step": 3205 }, { "epoch": 0.44, "grad_norm": 3.9559373300533944, "learning_rate": 9.101298513293164e-06, "loss": 0.3526, "step": 3206 }, { "epoch": 0.44, "grad_norm": 3.288539221905602, "learning_rate": 9.100668342383138e-06, "loss": 0.3573, "step": 3207 }, { "epoch": 0.44, "grad_norm": 3.8364710590405844, "learning_rate": 9.100037972444583e-06, "loss": 0.3274, "step": 3208 }, { "epoch": 0.44, "grad_norm": 6.427476530389878, "learning_rate": 9.099407403508092e-06, "loss": 0.3511, "step": 3209 }, { "epoch": 0.44, "grad_norm": 3.652063522173932, "learning_rate": 9.098776635604267e-06, "loss": 0.2809, "step": 3210 }, { "epoch": 0.44, "grad_norm": 2.846231883840211, "learning_rate": 9.09814566876373e-06, "loss": 0.2509, "step": 3211 }, { "epoch": 0.44, "grad_norm": 3.733886444044056, "learning_rate": 9.097514503017098e-06, "loss": 0.2442, "step": 3212 }, { "epoch": 0.44, "grad_norm": 3.6165986749223316, "learning_rate": 9.09688313839501e-06, "loss": 0.3096, "step": 3213 }, { "epoch": 0.44, "grad_norm": 3.5079274698996814, "learning_rate": 9.096251574928108e-06, "loss": 0.2845, "step": 3214 }, { "epoch": 0.44, "grad_norm": 3.3655680770262686, "learning_rate": 9.095619812647045e-06, "loss": 0.2808, "step": 3215 }, { "epoch": 0.44, "grad_norm": 3.9984517831073663, "learning_rate": 9.094987851582484e-06, "loss": 0.299, "step": 3216 }, { "epoch": 0.44, "grad_norm": 3.6949547784437673, "learning_rate": 9.0943556917651e-06, "loss": 0.3259, "step": 3217 }, { "epoch": 0.44, "grad_norm": 3.771276512965171, "learning_rate": 9.093723333225572e-06, "loss": 0.33, "step": 3218 }, { "epoch": 0.44, "grad_norm": 3.001324965085365, "learning_rate": 9.093090775994593e-06, "loss": 0.27, "step": 3219 }, { "epoch": 0.44, "grad_norm": 3.7617331058712367, "learning_rate": 9.092458020102863e-06, "loss": 0.2786, "step": 3220 }, { "epoch": 0.44, "grad_norm": 3.3338042047252685, "learning_rate": 9.091825065581097e-06, "loss": 0.2563, "step": 3221 }, { "epoch": 0.44, "grad_norm": 3.4240746464106464, "learning_rate": 9.091191912460014e-06, "loss": 0.3335, "step": 3222 }, { "epoch": 0.44, "grad_norm": 3.2031288831335587, "learning_rate": 9.090558560770344e-06, "loss": 0.3009, "step": 3223 }, { "epoch": 0.44, "grad_norm": 3.2042669718532633, "learning_rate": 9.089925010542829e-06, "loss": 0.292, "step": 3224 }, { "epoch": 0.44, "grad_norm": 3.779778433500634, "learning_rate": 9.089291261808212e-06, "loss": 0.2652, "step": 3225 }, { "epoch": 0.44, "grad_norm": 3.8771990128849865, "learning_rate": 9.088657314597263e-06, "loss": 0.3002, "step": 3226 }, { "epoch": 0.44, "grad_norm": 3.868108633066953, "learning_rate": 9.088023168940743e-06, "loss": 0.2455, "step": 3227 }, { "epoch": 0.44, "grad_norm": 3.594971299458237, "learning_rate": 9.087388824869435e-06, "loss": 0.315, "step": 3228 }, { "epoch": 0.44, "grad_norm": 4.162877382415757, "learning_rate": 9.086754282414124e-06, "loss": 0.2999, "step": 3229 }, { "epoch": 0.44, "grad_norm": 3.5724717778906663, "learning_rate": 9.086119541605609e-06, "loss": 0.3157, "step": 3230 }, { "epoch": 0.44, "grad_norm": 3.29108609466217, "learning_rate": 9.085484602474699e-06, "loss": 0.2774, "step": 3231 }, { "epoch": 0.44, "grad_norm": 3.3701296151661553, "learning_rate": 9.08484946505221e-06, "loss": 0.2702, "step": 3232 }, { "epoch": 0.44, "grad_norm": 3.266145755727634, "learning_rate": 9.08421412936897e-06, "loss": 0.32, "step": 3233 }, { "epoch": 0.44, "grad_norm": 4.338980296464482, "learning_rate": 9.083578595455812e-06, "loss": 0.3073, "step": 3234 }, { "epoch": 0.44, "grad_norm": 2.7695174556071227, "learning_rate": 9.082942863343587e-06, "loss": 0.2589, "step": 3235 }, { "epoch": 0.44, "grad_norm": 5.141486608613338, "learning_rate": 9.082306933063147e-06, "loss": 0.3201, "step": 3236 }, { "epoch": 0.44, "grad_norm": 3.4124691254459605, "learning_rate": 9.08167080464536e-06, "loss": 0.258, "step": 3237 }, { "epoch": 0.44, "grad_norm": 4.025420568272342, "learning_rate": 9.081034478121098e-06, "loss": 0.3731, "step": 3238 }, { "epoch": 0.44, "grad_norm": 3.610781283021532, "learning_rate": 9.080397953521245e-06, "loss": 0.3179, "step": 3239 }, { "epoch": 0.44, "grad_norm": 3.533240977406321, "learning_rate": 9.0797612308767e-06, "loss": 0.2703, "step": 3240 }, { "epoch": 0.44, "grad_norm": 3.269182415408117, "learning_rate": 9.079124310218361e-06, "loss": 0.3485, "step": 3241 }, { "epoch": 0.44, "grad_norm": 3.3920863018627863, "learning_rate": 9.078487191577146e-06, "loss": 0.3183, "step": 3242 }, { "epoch": 0.44, "grad_norm": 3.6736269213322084, "learning_rate": 9.077849874983974e-06, "loss": 0.3248, "step": 3243 }, { "epoch": 0.44, "grad_norm": 3.4118304562534494, "learning_rate": 9.077212360469781e-06, "loss": 0.3272, "step": 3244 }, { "epoch": 0.44, "grad_norm": 3.42244645964787, "learning_rate": 9.076574648065509e-06, "loss": 0.3059, "step": 3245 }, { "epoch": 0.44, "grad_norm": 3.5422240657457964, "learning_rate": 9.075936737802105e-06, "loss": 0.3001, "step": 3246 }, { "epoch": 0.44, "grad_norm": 3.720732027105139, "learning_rate": 9.075298629710536e-06, "loss": 0.3349, "step": 3247 }, { "epoch": 0.44, "grad_norm": 3.1334384194566294, "learning_rate": 9.074660323821772e-06, "loss": 0.2103, "step": 3248 }, { "epoch": 0.44, "grad_norm": 3.3582435768509646, "learning_rate": 9.074021820166789e-06, "loss": 0.31, "step": 3249 }, { "epoch": 0.44, "grad_norm": 3.6349459142157623, "learning_rate": 9.073383118776581e-06, "loss": 0.3471, "step": 3250 }, { "epoch": 0.44, "grad_norm": 3.1735742119042314, "learning_rate": 9.07274421968215e-06, "loss": 0.24, "step": 3251 }, { "epoch": 0.44, "grad_norm": 4.0236028043897525, "learning_rate": 9.072105122914502e-06, "loss": 0.3587, "step": 3252 }, { "epoch": 0.44, "grad_norm": 3.192672939486223, "learning_rate": 9.071465828504656e-06, "loss": 0.2846, "step": 3253 }, { "epoch": 0.44, "grad_norm": 3.5100811794787754, "learning_rate": 9.070826336483643e-06, "loss": 0.291, "step": 3254 }, { "epoch": 0.44, "grad_norm": 3.9467809561249214, "learning_rate": 9.070186646882497e-06, "loss": 0.309, "step": 3255 }, { "epoch": 0.44, "grad_norm": 3.5269554359500583, "learning_rate": 9.069546759732268e-06, "loss": 0.342, "step": 3256 }, { "epoch": 0.44, "grad_norm": 3.388754486125784, "learning_rate": 9.068906675064016e-06, "loss": 0.2966, "step": 3257 }, { "epoch": 0.44, "grad_norm": 3.6754875030578353, "learning_rate": 9.068266392908804e-06, "loss": 0.2788, "step": 3258 }, { "epoch": 0.44, "grad_norm": 3.315721419424054, "learning_rate": 9.067625913297708e-06, "loss": 0.3058, "step": 3259 }, { "epoch": 0.44, "grad_norm": 3.83587593112205, "learning_rate": 9.066985236261818e-06, "loss": 0.3325, "step": 3260 }, { "epoch": 0.44, "grad_norm": 3.7524536127187207, "learning_rate": 9.06634436183223e-06, "loss": 0.294, "step": 3261 }, { "epoch": 0.44, "grad_norm": 3.422296108647952, "learning_rate": 9.065703290040043e-06, "loss": 0.2842, "step": 3262 }, { "epoch": 0.44, "grad_norm": 3.1772014872182277, "learning_rate": 9.065062020916376e-06, "loss": 0.2275, "step": 3263 }, { "epoch": 0.44, "grad_norm": 3.5908574420132013, "learning_rate": 9.064420554492355e-06, "loss": 0.3706, "step": 3264 }, { "epoch": 0.44, "grad_norm": 4.745580400225671, "learning_rate": 9.06377889079911e-06, "loss": 0.3056, "step": 3265 }, { "epoch": 0.44, "grad_norm": 3.117107439846378, "learning_rate": 9.063137029867789e-06, "loss": 0.2887, "step": 3266 }, { "epoch": 0.44, "grad_norm": 3.003828602097744, "learning_rate": 9.062494971729542e-06, "loss": 0.2412, "step": 3267 }, { "epoch": 0.44, "grad_norm": 4.1305336195115245, "learning_rate": 9.061852716415531e-06, "loss": 0.3754, "step": 3268 }, { "epoch": 0.44, "grad_norm": 3.5900082436981937, "learning_rate": 9.061210263956931e-06, "loss": 0.3201, "step": 3269 }, { "epoch": 0.44, "grad_norm": 3.2916741966655887, "learning_rate": 9.060567614384924e-06, "loss": 0.2581, "step": 3270 }, { "epoch": 0.44, "grad_norm": 3.6400163328852804, "learning_rate": 9.059924767730697e-06, "loss": 0.2984, "step": 3271 }, { "epoch": 0.45, "grad_norm": 3.0744458207552703, "learning_rate": 9.059281724025455e-06, "loss": 0.3004, "step": 3272 }, { "epoch": 0.45, "grad_norm": 3.3741969858731014, "learning_rate": 9.058638483300409e-06, "loss": 0.2627, "step": 3273 }, { "epoch": 0.45, "grad_norm": 4.701123610753166, "learning_rate": 9.057995045586775e-06, "loss": 0.337, "step": 3274 }, { "epoch": 0.45, "grad_norm": 3.2726200989169856, "learning_rate": 9.057351410915786e-06, "loss": 0.2796, "step": 3275 }, { "epoch": 0.45, "grad_norm": 4.077835619292597, "learning_rate": 9.056707579318682e-06, "loss": 0.2757, "step": 3276 }, { "epoch": 0.45, "grad_norm": 4.212053931313147, "learning_rate": 9.056063550826708e-06, "loss": 0.3166, "step": 3277 }, { "epoch": 0.45, "grad_norm": 2.8093418853131453, "learning_rate": 9.055419325471126e-06, "loss": 0.2418, "step": 3278 }, { "epoch": 0.45, "grad_norm": 3.175359431831383, "learning_rate": 9.054774903283203e-06, "loss": 0.2726, "step": 3279 }, { "epoch": 0.45, "grad_norm": 3.672841307785759, "learning_rate": 9.054130284294214e-06, "loss": 0.3202, "step": 3280 }, { "epoch": 0.45, "grad_norm": 3.794955408547322, "learning_rate": 9.05348546853545e-06, "loss": 0.3329, "step": 3281 }, { "epoch": 0.45, "grad_norm": 3.160360959991028, "learning_rate": 9.052840456038204e-06, "loss": 0.3071, "step": 3282 }, { "epoch": 0.45, "grad_norm": 3.5166124995077896, "learning_rate": 9.052195246833784e-06, "loss": 0.3111, "step": 3283 }, { "epoch": 0.45, "grad_norm": 3.4007438922723625, "learning_rate": 9.051549840953504e-06, "loss": 0.2535, "step": 3284 }, { "epoch": 0.45, "grad_norm": 3.9833810516193626, "learning_rate": 9.05090423842869e-06, "loss": 0.3338, "step": 3285 }, { "epoch": 0.45, "grad_norm": 4.077339209330636, "learning_rate": 9.05025843929068e-06, "loss": 0.3753, "step": 3286 }, { "epoch": 0.45, "grad_norm": 4.10469709065947, "learning_rate": 9.049612443570814e-06, "loss": 0.3531, "step": 3287 }, { "epoch": 0.45, "grad_norm": 3.4695138690328924, "learning_rate": 9.048966251300447e-06, "loss": 0.2736, "step": 3288 }, { "epoch": 0.45, "grad_norm": 3.6043154357158436, "learning_rate": 9.048319862510941e-06, "loss": 0.3177, "step": 3289 }, { "epoch": 0.45, "grad_norm": 3.5972277567031594, "learning_rate": 9.047673277233671e-06, "loss": 0.3185, "step": 3290 }, { "epoch": 0.45, "grad_norm": 3.6567080146488657, "learning_rate": 9.04702649550002e-06, "loss": 0.3304, "step": 3291 }, { "epoch": 0.45, "grad_norm": 3.4633352418574916, "learning_rate": 9.046379517341378e-06, "loss": 0.3279, "step": 3292 }, { "epoch": 0.45, "grad_norm": 4.062004698823781, "learning_rate": 9.045732342789146e-06, "loss": 0.3508, "step": 3293 }, { "epoch": 0.45, "grad_norm": 3.591984663204401, "learning_rate": 9.045084971874738e-06, "loss": 0.3416, "step": 3294 }, { "epoch": 0.45, "grad_norm": 3.1193413531529854, "learning_rate": 9.04443740462957e-06, "loss": 0.2416, "step": 3295 }, { "epoch": 0.45, "grad_norm": 3.2356905045978013, "learning_rate": 9.043789641085078e-06, "loss": 0.3132, "step": 3296 }, { "epoch": 0.45, "grad_norm": 3.617736491475342, "learning_rate": 9.0431416812727e-06, "loss": 0.2943, "step": 3297 }, { "epoch": 0.45, "grad_norm": 3.745129431405722, "learning_rate": 9.042493525223881e-06, "loss": 0.3045, "step": 3298 }, { "epoch": 0.45, "grad_norm": 3.575899057380259, "learning_rate": 9.041845172970083e-06, "loss": 0.3043, "step": 3299 }, { "epoch": 0.45, "grad_norm": 3.089380958177393, "learning_rate": 9.041196624542772e-06, "loss": 0.2938, "step": 3300 }, { "epoch": 0.45, "grad_norm": 3.7895070748609414, "learning_rate": 9.04054787997343e-06, "loss": 0.2647, "step": 3301 }, { "epoch": 0.45, "grad_norm": 3.6842078571405708, "learning_rate": 9.039898939293539e-06, "loss": 0.3326, "step": 3302 }, { "epoch": 0.45, "grad_norm": 3.2910222529526916, "learning_rate": 9.039249802534597e-06, "loss": 0.222, "step": 3303 }, { "epoch": 0.45, "grad_norm": 4.120298645692047, "learning_rate": 9.038600469728115e-06, "loss": 0.3082, "step": 3304 }, { "epoch": 0.45, "grad_norm": 4.280556543922331, "learning_rate": 9.037950940905603e-06, "loss": 0.304, "step": 3305 }, { "epoch": 0.45, "grad_norm": 3.6435396383279124, "learning_rate": 9.03730121609859e-06, "loss": 0.3556, "step": 3306 }, { "epoch": 0.45, "grad_norm": 4.212581577415645, "learning_rate": 9.036651295338608e-06, "loss": 0.3524, "step": 3307 }, { "epoch": 0.45, "grad_norm": 3.735672692492366, "learning_rate": 9.036001178657203e-06, "loss": 0.284, "step": 3308 }, { "epoch": 0.45, "grad_norm": 3.08867039543905, "learning_rate": 9.035350866085928e-06, "loss": 0.2542, "step": 3309 }, { "epoch": 0.45, "grad_norm": 3.663557758475049, "learning_rate": 9.034700357656347e-06, "loss": 0.3101, "step": 3310 }, { "epoch": 0.45, "grad_norm": 2.9825357770630685, "learning_rate": 9.034049653400034e-06, "loss": 0.2459, "step": 3311 }, { "epoch": 0.45, "grad_norm": 3.53778164485549, "learning_rate": 9.033398753348569e-06, "loss": 0.2886, "step": 3312 }, { "epoch": 0.45, "grad_norm": 3.7168286268809263, "learning_rate": 9.032747657533545e-06, "loss": 0.2564, "step": 3313 }, { "epoch": 0.45, "grad_norm": 4.025288194293369, "learning_rate": 9.032096365986562e-06, "loss": 0.3724, "step": 3314 }, { "epoch": 0.45, "grad_norm": 3.2084171116955353, "learning_rate": 9.031444878739234e-06, "loss": 0.3103, "step": 3315 }, { "epoch": 0.45, "grad_norm": 3.1476387354528588, "learning_rate": 9.030793195823178e-06, "loss": 0.3048, "step": 3316 }, { "epoch": 0.45, "grad_norm": 3.08600494917626, "learning_rate": 9.030141317270026e-06, "loss": 0.2108, "step": 3317 }, { "epoch": 0.45, "grad_norm": 4.141224645689776, "learning_rate": 9.029489243111416e-06, "loss": 0.3729, "step": 3318 }, { "epoch": 0.45, "grad_norm": 3.0234325412017076, "learning_rate": 9.028836973378998e-06, "loss": 0.211, "step": 3319 }, { "epoch": 0.45, "grad_norm": 3.1523793149463937, "learning_rate": 9.02818450810443e-06, "loss": 0.322, "step": 3320 }, { "epoch": 0.45, "grad_norm": 4.067841409006454, "learning_rate": 9.027531847319378e-06, "loss": 0.3328, "step": 3321 }, { "epoch": 0.45, "grad_norm": 3.58359777969385, "learning_rate": 9.026878991055521e-06, "loss": 0.2364, "step": 3322 }, { "epoch": 0.45, "grad_norm": 3.560204357644649, "learning_rate": 9.026225939344546e-06, "loss": 0.302, "step": 3323 }, { "epoch": 0.45, "grad_norm": 3.376573590273032, "learning_rate": 9.02557269221815e-06, "loss": 0.2897, "step": 3324 }, { "epoch": 0.45, "grad_norm": 3.40304788043316, "learning_rate": 9.024919249708034e-06, "loss": 0.2891, "step": 3325 }, { "epoch": 0.45, "grad_norm": 3.7317794248754903, "learning_rate": 9.024265611845922e-06, "loss": 0.2135, "step": 3326 }, { "epoch": 0.45, "grad_norm": 3.372134353802569, "learning_rate": 9.02361177866353e-06, "loss": 0.2411, "step": 3327 }, { "epoch": 0.45, "grad_norm": 3.3073926432467795, "learning_rate": 9.022957750192598e-06, "loss": 0.2838, "step": 3328 }, { "epoch": 0.45, "grad_norm": 3.4855999196303125, "learning_rate": 9.022303526464866e-06, "loss": 0.2423, "step": 3329 }, { "epoch": 0.45, "grad_norm": 3.1522098058025807, "learning_rate": 9.021649107512089e-06, "loss": 0.2744, "step": 3330 }, { "epoch": 0.45, "grad_norm": 4.755426677175798, "learning_rate": 9.02099449336603e-06, "loss": 0.3125, "step": 3331 }, { "epoch": 0.45, "grad_norm": 2.858535567885278, "learning_rate": 9.020339684058459e-06, "loss": 0.2418, "step": 3332 }, { "epoch": 0.45, "grad_norm": 3.369241676078469, "learning_rate": 9.019684679621162e-06, "loss": 0.2808, "step": 3333 }, { "epoch": 0.45, "grad_norm": 3.7717364081256815, "learning_rate": 9.019029480085924e-06, "loss": 0.2388, "step": 3334 }, { "epoch": 0.45, "grad_norm": 3.9427580725509594, "learning_rate": 9.01837408548455e-06, "loss": 0.2921, "step": 3335 }, { "epoch": 0.45, "grad_norm": 3.4414652040205187, "learning_rate": 9.017718495848848e-06, "loss": 0.3346, "step": 3336 }, { "epoch": 0.45, "grad_norm": 4.2444926918066725, "learning_rate": 9.017062711210638e-06, "loss": 0.2802, "step": 3337 }, { "epoch": 0.45, "grad_norm": 3.624607438708206, "learning_rate": 9.016406731601749e-06, "loss": 0.2857, "step": 3338 }, { "epoch": 0.45, "grad_norm": 3.5431898402315154, "learning_rate": 9.015750557054019e-06, "loss": 0.2737, "step": 3339 }, { "epoch": 0.45, "grad_norm": 3.1717583516995145, "learning_rate": 9.015094187599297e-06, "loss": 0.2682, "step": 3340 }, { "epoch": 0.45, "grad_norm": 3.3755075792778957, "learning_rate": 9.014437623269439e-06, "loss": 0.2081, "step": 3341 }, { "epoch": 0.45, "grad_norm": 4.113325622481237, "learning_rate": 9.013780864096313e-06, "loss": 0.3624, "step": 3342 }, { "epoch": 0.45, "grad_norm": 3.66345767809991, "learning_rate": 9.013123910111794e-06, "loss": 0.2963, "step": 3343 }, { "epoch": 0.45, "grad_norm": 4.124292496852031, "learning_rate": 9.012466761347766e-06, "loss": 0.2615, "step": 3344 }, { "epoch": 0.46, "grad_norm": 3.170736074559607, "learning_rate": 9.01180941783613e-06, "loss": 0.2961, "step": 3345 }, { "epoch": 0.46, "grad_norm": 3.9930542192161758, "learning_rate": 9.011151879608786e-06, "loss": 0.3698, "step": 3346 }, { "epoch": 0.46, "grad_norm": 3.718596606468606, "learning_rate": 9.010494146697648e-06, "loss": 0.3906, "step": 3347 }, { "epoch": 0.46, "grad_norm": 3.99485432437199, "learning_rate": 9.00983621913464e-06, "loss": 0.2831, "step": 3348 }, { "epoch": 0.46, "grad_norm": 3.494646168004021, "learning_rate": 9.009178096951696e-06, "loss": 0.2825, "step": 3349 }, { "epoch": 0.46, "grad_norm": 2.8612909868892853, "learning_rate": 9.008519780180757e-06, "loss": 0.2536, "step": 3350 }, { "epoch": 0.46, "grad_norm": 3.218752918943194, "learning_rate": 9.007861268853777e-06, "loss": 0.2813, "step": 3351 }, { "epoch": 0.46, "grad_norm": 3.133120557723259, "learning_rate": 9.007202563002715e-06, "loss": 0.2625, "step": 3352 }, { "epoch": 0.46, "grad_norm": 3.2484733868124063, "learning_rate": 9.006543662659542e-06, "loss": 0.2818, "step": 3353 }, { "epoch": 0.46, "grad_norm": 2.922656377125525, "learning_rate": 9.005884567856239e-06, "loss": 0.2619, "step": 3354 }, { "epoch": 0.46, "grad_norm": 2.9762283529002147, "learning_rate": 9.005225278624795e-06, "loss": 0.2545, "step": 3355 }, { "epoch": 0.46, "grad_norm": 3.3508567581313033, "learning_rate": 9.004565794997209e-06, "loss": 0.2332, "step": 3356 }, { "epoch": 0.46, "grad_norm": 3.0835135110750906, "learning_rate": 9.003906117005489e-06, "loss": 0.2838, "step": 3357 }, { "epoch": 0.46, "grad_norm": 3.8719169719542266, "learning_rate": 9.003246244681657e-06, "loss": 0.3406, "step": 3358 }, { "epoch": 0.46, "grad_norm": 3.301986157708332, "learning_rate": 9.002586178057734e-06, "loss": 0.2673, "step": 3359 }, { "epoch": 0.46, "grad_norm": 3.435477604013651, "learning_rate": 9.00192591716576e-06, "loss": 0.296, "step": 3360 }, { "epoch": 0.46, "grad_norm": 4.209898627091975, "learning_rate": 9.001265462037781e-06, "loss": 0.3737, "step": 3361 }, { "epoch": 0.46, "grad_norm": 3.671234015162857, "learning_rate": 9.000604812705854e-06, "loss": 0.2766, "step": 3362 }, { "epoch": 0.46, "grad_norm": 3.6311907447130336, "learning_rate": 8.99994396920204e-06, "loss": 0.37, "step": 3363 }, { "epoch": 0.46, "grad_norm": 3.279080663350196, "learning_rate": 8.999282931558417e-06, "loss": 0.2921, "step": 3364 }, { "epoch": 0.46, "grad_norm": 3.4468424825162427, "learning_rate": 8.998621699807067e-06, "loss": 0.2929, "step": 3365 }, { "epoch": 0.46, "grad_norm": 3.2798933702964286, "learning_rate": 8.997960273980086e-06, "loss": 0.2854, "step": 3366 }, { "epoch": 0.46, "grad_norm": 3.211700761947628, "learning_rate": 8.997298654109573e-06, "loss": 0.2926, "step": 3367 }, { "epoch": 0.46, "grad_norm": 3.647468518741931, "learning_rate": 8.996636840227642e-06, "loss": 0.3407, "step": 3368 }, { "epoch": 0.46, "grad_norm": 3.755545800184851, "learning_rate": 8.995974832366415e-06, "loss": 0.3147, "step": 3369 }, { "epoch": 0.46, "grad_norm": 3.6486388487895414, "learning_rate": 8.995312630558022e-06, "loss": 0.3398, "step": 3370 }, { "epoch": 0.46, "grad_norm": 3.954447532976496, "learning_rate": 8.994650234834605e-06, "loss": 0.2982, "step": 3371 }, { "epoch": 0.46, "grad_norm": 3.6895332108047407, "learning_rate": 8.993987645228313e-06, "loss": 0.296, "step": 3372 }, { "epoch": 0.46, "grad_norm": 3.3320607450718644, "learning_rate": 8.993324861771305e-06, "loss": 0.3497, "step": 3373 }, { "epoch": 0.46, "grad_norm": 3.3770223708318765, "learning_rate": 8.992661884495747e-06, "loss": 0.2861, "step": 3374 }, { "epoch": 0.46, "grad_norm": 4.418326136268954, "learning_rate": 8.991998713433821e-06, "loss": 0.3389, "step": 3375 }, { "epoch": 0.46, "grad_norm": 3.4338514489385723, "learning_rate": 8.991335348617713e-06, "loss": 0.2603, "step": 3376 }, { "epoch": 0.46, "grad_norm": 3.08508987032152, "learning_rate": 8.99067179007962e-06, "loss": 0.3058, "step": 3377 }, { "epoch": 0.46, "grad_norm": 3.676800964736662, "learning_rate": 8.990008037851748e-06, "loss": 0.3293, "step": 3378 }, { "epoch": 0.46, "grad_norm": 3.372263529545427, "learning_rate": 8.989344091966312e-06, "loss": 0.2484, "step": 3379 }, { "epoch": 0.46, "grad_norm": 3.704934858715451, "learning_rate": 8.988679952455539e-06, "loss": 0.2911, "step": 3380 }, { "epoch": 0.46, "grad_norm": 3.6307364919731353, "learning_rate": 8.98801561935166e-06, "loss": 0.3007, "step": 3381 }, { "epoch": 0.46, "grad_norm": 3.3880332773096824, "learning_rate": 8.987351092686923e-06, "loss": 0.2851, "step": 3382 }, { "epoch": 0.46, "grad_norm": 3.08110276129312, "learning_rate": 8.986686372493579e-06, "loss": 0.2175, "step": 3383 }, { "epoch": 0.46, "grad_norm": 3.011543131229212, "learning_rate": 8.986021458803889e-06, "loss": 0.2558, "step": 3384 }, { "epoch": 0.46, "grad_norm": 3.477308075406346, "learning_rate": 8.985356351650128e-06, "loss": 0.2971, "step": 3385 }, { "epoch": 0.46, "grad_norm": 4.018791095466086, "learning_rate": 8.984691051064576e-06, "loss": 0.2842, "step": 3386 }, { "epoch": 0.46, "grad_norm": 3.2737497229174757, "learning_rate": 8.984025557079523e-06, "loss": 0.3363, "step": 3387 }, { "epoch": 0.46, "grad_norm": 3.8458665012915536, "learning_rate": 8.983359869727271e-06, "loss": 0.381, "step": 3388 }, { "epoch": 0.46, "grad_norm": 5.080768685579047, "learning_rate": 8.98269398904013e-06, "loss": 0.3296, "step": 3389 }, { "epoch": 0.46, "grad_norm": 3.1551742864990717, "learning_rate": 8.982027915050415e-06, "loss": 0.3261, "step": 3390 }, { "epoch": 0.46, "grad_norm": 3.5691056555447336, "learning_rate": 8.98136164779046e-06, "loss": 0.2755, "step": 3391 }, { "epoch": 0.46, "grad_norm": 3.878618787495031, "learning_rate": 8.980695187292598e-06, "loss": 0.2935, "step": 3392 }, { "epoch": 0.46, "grad_norm": 3.667786428922581, "learning_rate": 8.980028533589176e-06, "loss": 0.3146, "step": 3393 }, { "epoch": 0.46, "grad_norm": 4.032958042402591, "learning_rate": 8.979361686712554e-06, "loss": 0.2886, "step": 3394 }, { "epoch": 0.46, "grad_norm": 3.352317484220682, "learning_rate": 8.978694646695096e-06, "loss": 0.2718, "step": 3395 }, { "epoch": 0.46, "grad_norm": 3.4986014071129117, "learning_rate": 8.978027413569176e-06, "loss": 0.3007, "step": 3396 }, { "epoch": 0.46, "grad_norm": 3.6046607399001998, "learning_rate": 8.977359987367182e-06, "loss": 0.3132, "step": 3397 }, { "epoch": 0.46, "grad_norm": 3.4367308035719555, "learning_rate": 8.976692368121506e-06, "loss": 0.3037, "step": 3398 }, { "epoch": 0.46, "grad_norm": 3.7923614438849014, "learning_rate": 8.976024555864549e-06, "loss": 0.3645, "step": 3399 }, { "epoch": 0.46, "grad_norm": 3.1994185282887986, "learning_rate": 8.975356550628726e-06, "loss": 0.2323, "step": 3400 }, { "epoch": 0.46, "grad_norm": 3.5814567244826567, "learning_rate": 8.97468835244646e-06, "loss": 0.2694, "step": 3401 }, { "epoch": 0.46, "grad_norm": 3.6926189112504844, "learning_rate": 8.97401996135018e-06, "loss": 0.3229, "step": 3402 }, { "epoch": 0.46, "grad_norm": 3.4366719472509946, "learning_rate": 8.97335137737233e-06, "loss": 0.2888, "step": 3403 }, { "epoch": 0.46, "grad_norm": 3.49956119075141, "learning_rate": 8.972682600545357e-06, "loss": 0.3298, "step": 3404 }, { "epoch": 0.46, "grad_norm": 3.486644931797766, "learning_rate": 8.972013630901722e-06, "loss": 0.3417, "step": 3405 }, { "epoch": 0.46, "grad_norm": 4.34017387884803, "learning_rate": 8.971344468473893e-06, "loss": 0.3408, "step": 3406 }, { "epoch": 0.46, "grad_norm": 3.4044226077083777, "learning_rate": 8.970675113294348e-06, "loss": 0.3035, "step": 3407 }, { "epoch": 0.46, "grad_norm": 3.6805487848269314, "learning_rate": 8.970005565395576e-06, "loss": 0.288, "step": 3408 }, { "epoch": 0.46, "grad_norm": 3.1663128636899907, "learning_rate": 8.969335824810075e-06, "loss": 0.3195, "step": 3409 }, { "epoch": 0.46, "grad_norm": 3.797118614427176, "learning_rate": 8.968665891570346e-06, "loss": 0.2995, "step": 3410 }, { "epoch": 0.46, "grad_norm": 3.473986540721862, "learning_rate": 8.967995765708912e-06, "loss": 0.3376, "step": 3411 }, { "epoch": 0.46, "grad_norm": 3.3265065735038135, "learning_rate": 8.967325447258292e-06, "loss": 0.3079, "step": 3412 }, { "epoch": 0.46, "grad_norm": 3.1990687780350817, "learning_rate": 8.966654936251022e-06, "loss": 0.2474, "step": 3413 }, { "epoch": 0.46, "grad_norm": 3.3218151515260272, "learning_rate": 8.965984232719649e-06, "loss": 0.3206, "step": 3414 }, { "epoch": 0.46, "grad_norm": 3.4297791307187495, "learning_rate": 8.965313336696722e-06, "loss": 0.2478, "step": 3415 }, { "epoch": 0.46, "grad_norm": 4.1863942699491865, "learning_rate": 8.964642248214803e-06, "loss": 0.2917, "step": 3416 }, { "epoch": 0.46, "grad_norm": 3.767540737023742, "learning_rate": 8.963970967306466e-06, "loss": 0.2578, "step": 3417 }, { "epoch": 0.46, "grad_norm": 3.2037789779497206, "learning_rate": 8.963299494004292e-06, "loss": 0.2607, "step": 3418 }, { "epoch": 0.47, "grad_norm": 3.359508089531975, "learning_rate": 8.962627828340872e-06, "loss": 0.3215, "step": 3419 }, { "epoch": 0.47, "grad_norm": 4.943117712788573, "learning_rate": 8.961955970348801e-06, "loss": 0.2989, "step": 3420 }, { "epoch": 0.47, "grad_norm": 3.5755577967617613, "learning_rate": 8.961283920060693e-06, "loss": 0.3568, "step": 3421 }, { "epoch": 0.47, "grad_norm": 3.394885176579378, "learning_rate": 8.960611677509166e-06, "loss": 0.3196, "step": 3422 }, { "epoch": 0.47, "grad_norm": 3.7256316627520163, "learning_rate": 8.959939242726844e-06, "loss": 0.2779, "step": 3423 }, { "epoch": 0.47, "grad_norm": 4.751503021720271, "learning_rate": 8.959266615746369e-06, "loss": 0.3833, "step": 3424 }, { "epoch": 0.47, "grad_norm": 3.3901510685225813, "learning_rate": 8.958593796600385e-06, "loss": 0.3456, "step": 3425 }, { "epoch": 0.47, "grad_norm": 3.502565330292588, "learning_rate": 8.957920785321546e-06, "loss": 0.3379, "step": 3426 }, { "epoch": 0.47, "grad_norm": 2.801496500403003, "learning_rate": 8.95724758194252e-06, "loss": 0.2688, "step": 3427 }, { "epoch": 0.47, "grad_norm": 3.161353743079003, "learning_rate": 8.956574186495981e-06, "loss": 0.2866, "step": 3428 }, { "epoch": 0.47, "grad_norm": 3.672250273359282, "learning_rate": 8.955900599014609e-06, "loss": 0.3675, "step": 3429 }, { "epoch": 0.47, "grad_norm": 3.041010305723784, "learning_rate": 8.955226819531102e-06, "loss": 0.2564, "step": 3430 }, { "epoch": 0.47, "grad_norm": 3.5742277366955273, "learning_rate": 8.95455284807816e-06, "loss": 0.3039, "step": 3431 }, { "epoch": 0.47, "grad_norm": 4.022885867449215, "learning_rate": 8.953878684688492e-06, "loss": 0.2879, "step": 3432 }, { "epoch": 0.47, "grad_norm": 3.2521011513375853, "learning_rate": 8.953204329394825e-06, "loss": 0.3025, "step": 3433 }, { "epoch": 0.47, "grad_norm": 3.324591364037485, "learning_rate": 8.952529782229886e-06, "loss": 0.2773, "step": 3434 }, { "epoch": 0.47, "grad_norm": 3.297234310177385, "learning_rate": 8.951855043226412e-06, "loss": 0.2787, "step": 3435 }, { "epoch": 0.47, "grad_norm": 3.858568921435422, "learning_rate": 8.951180112417154e-06, "loss": 0.3265, "step": 3436 }, { "epoch": 0.47, "grad_norm": 2.991803767231958, "learning_rate": 8.950504989834873e-06, "loss": 0.3119, "step": 3437 }, { "epoch": 0.47, "grad_norm": 4.051354090067382, "learning_rate": 8.949829675512334e-06, "loss": 0.3055, "step": 3438 }, { "epoch": 0.47, "grad_norm": 3.3908999225627725, "learning_rate": 8.949154169482312e-06, "loss": 0.2669, "step": 3439 }, { "epoch": 0.47, "grad_norm": 3.3008068941168136, "learning_rate": 8.948478471777598e-06, "loss": 0.2998, "step": 3440 }, { "epoch": 0.47, "grad_norm": 3.1007925567038295, "learning_rate": 8.947802582430981e-06, "loss": 0.3108, "step": 3441 }, { "epoch": 0.47, "grad_norm": 3.6394299534115477, "learning_rate": 8.94712650147527e-06, "loss": 0.2729, "step": 3442 }, { "epoch": 0.47, "grad_norm": 3.2400711384807046, "learning_rate": 8.94645022894328e-06, "loss": 0.2833, "step": 3443 }, { "epoch": 0.47, "grad_norm": 3.737392472694631, "learning_rate": 8.94577376486783e-06, "loss": 0.3758, "step": 3444 }, { "epoch": 0.47, "grad_norm": 3.562187683607563, "learning_rate": 8.945097109281757e-06, "loss": 0.2782, "step": 3445 }, { "epoch": 0.47, "grad_norm": 3.082591248211479, "learning_rate": 8.944420262217899e-06, "loss": 0.2296, "step": 3446 }, { "epoch": 0.47, "grad_norm": 3.3148262993670197, "learning_rate": 8.943743223709109e-06, "loss": 0.2829, "step": 3447 }, { "epoch": 0.47, "grad_norm": 3.748080086127276, "learning_rate": 8.94306599378825e-06, "loss": 0.2735, "step": 3448 }, { "epoch": 0.47, "grad_norm": 4.187739278647283, "learning_rate": 8.942388572488188e-06, "loss": 0.3495, "step": 3449 }, { "epoch": 0.47, "grad_norm": 3.4883621872128856, "learning_rate": 8.941710959841803e-06, "loss": 0.3112, "step": 3450 }, { "epoch": 0.47, "grad_norm": 3.3046465579881423, "learning_rate": 8.941033155881985e-06, "loss": 0.2744, "step": 3451 }, { "epoch": 0.47, "grad_norm": 3.3550044900905696, "learning_rate": 8.94035516064163e-06, "loss": 0.3295, "step": 3452 }, { "epoch": 0.47, "grad_norm": 4.2523439546020185, "learning_rate": 8.939676974153646e-06, "loss": 0.3137, "step": 3453 }, { "epoch": 0.47, "grad_norm": 3.0620838346481243, "learning_rate": 8.938998596450947e-06, "loss": 0.3319, "step": 3454 }, { "epoch": 0.47, "grad_norm": 4.145382193637166, "learning_rate": 8.938320027566462e-06, "loss": 0.3039, "step": 3455 }, { "epoch": 0.47, "grad_norm": 3.879446966871803, "learning_rate": 8.937641267533123e-06, "loss": 0.3249, "step": 3456 }, { "epoch": 0.47, "grad_norm": 3.4493580684189045, "learning_rate": 8.936962316383876e-06, "loss": 0.309, "step": 3457 }, { "epoch": 0.47, "grad_norm": 3.3513126689985304, "learning_rate": 8.936283174151673e-06, "loss": 0.3256, "step": 3458 }, { "epoch": 0.47, "grad_norm": 3.5239550466578677, "learning_rate": 8.935603840869476e-06, "loss": 0.2725, "step": 3459 }, { "epoch": 0.47, "grad_norm": 3.65123916277256, "learning_rate": 8.934924316570257e-06, "loss": 0.2548, "step": 3460 }, { "epoch": 0.47, "grad_norm": 3.6968906880815005, "learning_rate": 8.934244601287e-06, "loss": 0.262, "step": 3461 }, { "epoch": 0.47, "grad_norm": 3.7014683121162704, "learning_rate": 8.933564695052692e-06, "loss": 0.2921, "step": 3462 }, { "epoch": 0.47, "grad_norm": 3.1144162044727937, "learning_rate": 8.932884597900334e-06, "loss": 0.2712, "step": 3463 }, { "epoch": 0.47, "grad_norm": 3.69212123535998, "learning_rate": 8.932204309862934e-06, "loss": 0.3271, "step": 3464 }, { "epoch": 0.47, "grad_norm": 4.02441604411544, "learning_rate": 8.931523830973515e-06, "loss": 0.3264, "step": 3465 }, { "epoch": 0.47, "grad_norm": 3.3722532226634563, "learning_rate": 8.930843161265097e-06, "loss": 0.3142, "step": 3466 }, { "epoch": 0.47, "grad_norm": 3.982969718695919, "learning_rate": 8.930162300770721e-06, "loss": 0.284, "step": 3467 }, { "epoch": 0.47, "grad_norm": 2.8244744975973823, "learning_rate": 8.929481249523432e-06, "loss": 0.2467, "step": 3468 }, { "epoch": 0.47, "grad_norm": 3.936788773379481, "learning_rate": 8.928800007556286e-06, "loss": 0.287, "step": 3469 }, { "epoch": 0.47, "grad_norm": 3.290587329614966, "learning_rate": 8.928118574902347e-06, "loss": 0.3023, "step": 3470 }, { "epoch": 0.47, "grad_norm": 3.2835098464645407, "learning_rate": 8.927436951594691e-06, "loss": 0.2686, "step": 3471 }, { "epoch": 0.47, "grad_norm": 4.163151125900567, "learning_rate": 8.926755137666396e-06, "loss": 0.2741, "step": 3472 }, { "epoch": 0.47, "grad_norm": 3.293552188059256, "learning_rate": 8.926073133150557e-06, "loss": 0.2761, "step": 3473 }, { "epoch": 0.47, "grad_norm": 3.6058268773021105, "learning_rate": 8.925390938080277e-06, "loss": 0.2956, "step": 3474 }, { "epoch": 0.47, "grad_norm": 3.452777109338167, "learning_rate": 8.924708552488666e-06, "loss": 0.2836, "step": 3475 }, { "epoch": 0.47, "grad_norm": 3.5932465794488553, "learning_rate": 8.924025976408842e-06, "loss": 0.2608, "step": 3476 }, { "epoch": 0.47, "grad_norm": 3.562909261978353, "learning_rate": 8.923343209873937e-06, "loss": 0.3138, "step": 3477 }, { "epoch": 0.47, "grad_norm": 3.5944297439177437, "learning_rate": 8.922660252917088e-06, "loss": 0.2614, "step": 3478 }, { "epoch": 0.47, "grad_norm": 3.277709458997763, "learning_rate": 8.921977105571443e-06, "loss": 0.2295, "step": 3479 }, { "epoch": 0.47, "grad_norm": 3.1147301752396936, "learning_rate": 8.921293767870157e-06, "loss": 0.2676, "step": 3480 }, { "epoch": 0.47, "grad_norm": 3.2458802933984106, "learning_rate": 8.9206102398464e-06, "loss": 0.3065, "step": 3481 }, { "epoch": 0.47, "grad_norm": 3.43050112734937, "learning_rate": 8.919926521533346e-06, "loss": 0.2645, "step": 3482 }, { "epoch": 0.47, "grad_norm": 2.9659837615622946, "learning_rate": 8.919242612964179e-06, "loss": 0.2309, "step": 3483 }, { "epoch": 0.47, "grad_norm": 3.2083710143828417, "learning_rate": 8.918558514172092e-06, "loss": 0.2534, "step": 3484 }, { "epoch": 0.47, "grad_norm": 4.167897509473162, "learning_rate": 8.91787422519029e-06, "loss": 0.3164, "step": 3485 }, { "epoch": 0.47, "grad_norm": 3.4150638210686237, "learning_rate": 8.917189746051986e-06, "loss": 0.2795, "step": 3486 }, { "epoch": 0.47, "grad_norm": 3.5287274793513697, "learning_rate": 8.9165050767904e-06, "loss": 0.2735, "step": 3487 }, { "epoch": 0.47, "grad_norm": 3.8816846002094794, "learning_rate": 8.915820217438763e-06, "loss": 0.2968, "step": 3488 }, { "epoch": 0.47, "grad_norm": 2.8390200975850766, "learning_rate": 8.915135168030316e-06, "loss": 0.2663, "step": 3489 }, { "epoch": 0.47, "grad_norm": 3.38237546825336, "learning_rate": 8.914449928598306e-06, "loss": 0.2604, "step": 3490 }, { "epoch": 0.47, "grad_norm": 3.3878125814459046, "learning_rate": 8.913764499175995e-06, "loss": 0.2465, "step": 3491 }, { "epoch": 0.48, "grad_norm": 3.6041648913288875, "learning_rate": 8.913078879796648e-06, "loss": 0.2708, "step": 3492 }, { "epoch": 0.48, "grad_norm": 3.5302507147666047, "learning_rate": 8.912393070493543e-06, "loss": 0.2844, "step": 3493 }, { "epoch": 0.48, "grad_norm": 3.051335063064145, "learning_rate": 8.911707071299968e-06, "loss": 0.2588, "step": 3494 }, { "epoch": 0.48, "grad_norm": 4.116256914662863, "learning_rate": 8.911020882249216e-06, "loss": 0.2953, "step": 3495 }, { "epoch": 0.48, "grad_norm": 3.605805014178956, "learning_rate": 8.91033450337459e-06, "loss": 0.2829, "step": 3496 }, { "epoch": 0.48, "grad_norm": 2.6782137997909583, "learning_rate": 8.90964793470941e-06, "loss": 0.2528, "step": 3497 }, { "epoch": 0.48, "grad_norm": 3.07424592132726, "learning_rate": 8.908961176286991e-06, "loss": 0.2853, "step": 3498 }, { "epoch": 0.48, "grad_norm": 3.699094543468601, "learning_rate": 8.90827422814067e-06, "loss": 0.296, "step": 3499 }, { "epoch": 0.48, "grad_norm": 3.320172214682342, "learning_rate": 8.907587090303789e-06, "loss": 0.2521, "step": 3500 }, { "epoch": 0.48, "grad_norm": 3.3588685302404313, "learning_rate": 8.906899762809698e-06, "loss": 0.3098, "step": 3501 }, { "epoch": 0.48, "grad_norm": 3.895878147655629, "learning_rate": 8.906212245691755e-06, "loss": 0.2613, "step": 3502 }, { "epoch": 0.48, "grad_norm": 3.612717094509527, "learning_rate": 8.90552453898333e-06, "loss": 0.2248, "step": 3503 }, { "epoch": 0.48, "grad_norm": 3.577689974836532, "learning_rate": 8.904836642717802e-06, "loss": 0.3056, "step": 3504 }, { "epoch": 0.48, "grad_norm": 3.3008719311067307, "learning_rate": 8.90414855692856e-06, "loss": 0.2959, "step": 3505 }, { "epoch": 0.48, "grad_norm": 3.0347006462039134, "learning_rate": 8.903460281648996e-06, "loss": 0.234, "step": 3506 }, { "epoch": 0.48, "grad_norm": 3.4878662572158277, "learning_rate": 8.902771816912521e-06, "loss": 0.3444, "step": 3507 }, { "epoch": 0.48, "grad_norm": 4.256306604364036, "learning_rate": 8.902083162752547e-06, "loss": 0.3434, "step": 3508 }, { "epoch": 0.48, "grad_norm": 3.827605893708808, "learning_rate": 8.901394319202499e-06, "loss": 0.2758, "step": 3509 }, { "epoch": 0.48, "grad_norm": 3.873045413180642, "learning_rate": 8.90070528629581e-06, "loss": 0.3532, "step": 3510 }, { "epoch": 0.48, "grad_norm": 3.5513627168940203, "learning_rate": 8.900016064065923e-06, "loss": 0.3021, "step": 3511 }, { "epoch": 0.48, "grad_norm": 3.0696927055948042, "learning_rate": 8.899326652546292e-06, "loss": 0.3005, "step": 3512 }, { "epoch": 0.48, "grad_norm": 3.116074285880984, "learning_rate": 8.898637051770375e-06, "loss": 0.3056, "step": 3513 }, { "epoch": 0.48, "grad_norm": 3.365341352468792, "learning_rate": 8.897947261771643e-06, "loss": 0.2718, "step": 3514 }, { "epoch": 0.48, "grad_norm": 3.334378162649772, "learning_rate": 8.897257282583575e-06, "loss": 0.2228, "step": 3515 }, { "epoch": 0.48, "grad_norm": 3.3160958152072655, "learning_rate": 8.89656711423966e-06, "loss": 0.3062, "step": 3516 }, { "epoch": 0.48, "grad_norm": 3.8696319649037787, "learning_rate": 8.895876756773398e-06, "loss": 0.2899, "step": 3517 }, { "epoch": 0.48, "grad_norm": 3.4778175484959366, "learning_rate": 8.895186210218293e-06, "loss": 0.2941, "step": 3518 }, { "epoch": 0.48, "grad_norm": 3.5174325173632113, "learning_rate": 8.89449547460786e-06, "loss": 0.2601, "step": 3519 }, { "epoch": 0.48, "grad_norm": 2.8790390558239904, "learning_rate": 8.893804549975628e-06, "loss": 0.2554, "step": 3520 }, { "epoch": 0.48, "grad_norm": 3.6616349122472602, "learning_rate": 8.89311343635513e-06, "loss": 0.3016, "step": 3521 }, { "epoch": 0.48, "grad_norm": 5.841429568571127, "learning_rate": 8.89242213377991e-06, "loss": 0.2833, "step": 3522 }, { "epoch": 0.48, "grad_norm": 3.3172275174269217, "learning_rate": 8.891730642283519e-06, "loss": 0.3003, "step": 3523 }, { "epoch": 0.48, "grad_norm": 3.3757016191412452, "learning_rate": 8.891038961899521e-06, "loss": 0.2824, "step": 3524 }, { "epoch": 0.48, "grad_norm": 3.0950098451739296, "learning_rate": 8.890347092661485e-06, "loss": 0.3076, "step": 3525 }, { "epoch": 0.48, "grad_norm": 3.243474862645363, "learning_rate": 8.889655034602993e-06, "loss": 0.2767, "step": 3526 }, { "epoch": 0.48, "grad_norm": 3.015599785804988, "learning_rate": 8.888962787757636e-06, "loss": 0.3, "step": 3527 }, { "epoch": 0.48, "grad_norm": 3.5925623279310606, "learning_rate": 8.888270352159009e-06, "loss": 0.3058, "step": 3528 }, { "epoch": 0.48, "grad_norm": 4.209010097830384, "learning_rate": 8.887577727840722e-06, "loss": 0.3151, "step": 3529 }, { "epoch": 0.48, "grad_norm": 3.4224061259851664, "learning_rate": 8.88688491483639e-06, "loss": 0.3228, "step": 3530 }, { "epoch": 0.48, "grad_norm": 3.2565841014094254, "learning_rate": 8.886191913179642e-06, "loss": 0.2163, "step": 3531 }, { "epoch": 0.48, "grad_norm": 3.0571281705424322, "learning_rate": 8.885498722904114e-06, "loss": 0.3267, "step": 3532 }, { "epoch": 0.48, "grad_norm": 3.4145809475279343, "learning_rate": 8.884805344043445e-06, "loss": 0.3055, "step": 3533 }, { "epoch": 0.48, "grad_norm": 3.3653124163782944, "learning_rate": 8.884111776631293e-06, "loss": 0.3072, "step": 3534 }, { "epoch": 0.48, "grad_norm": 3.5437835802526423, "learning_rate": 8.88341802070132e-06, "loss": 0.2803, "step": 3535 }, { "epoch": 0.48, "grad_norm": 2.688790031459041, "learning_rate": 8.882724076287197e-06, "loss": 0.2372, "step": 3536 }, { "epoch": 0.48, "grad_norm": 3.8720942513925594, "learning_rate": 8.882029943422605e-06, "loss": 0.3497, "step": 3537 }, { "epoch": 0.48, "grad_norm": 3.7519841162803296, "learning_rate": 8.881335622141236e-06, "loss": 0.2853, "step": 3538 }, { "epoch": 0.48, "grad_norm": 3.5153782653878514, "learning_rate": 8.880641112476788e-06, "loss": 0.2638, "step": 3539 }, { "epoch": 0.48, "grad_norm": 4.096713418420055, "learning_rate": 8.879946414462969e-06, "loss": 0.3645, "step": 3540 }, { "epoch": 0.48, "grad_norm": 3.9224510224912237, "learning_rate": 8.879251528133499e-06, "loss": 0.3448, "step": 3541 }, { "epoch": 0.48, "grad_norm": 3.756611847696216, "learning_rate": 8.8785564535221e-06, "loss": 0.2918, "step": 3542 }, { "epoch": 0.48, "grad_norm": 3.6480598167933818, "learning_rate": 8.87786119066251e-06, "loss": 0.2766, "step": 3543 }, { "epoch": 0.48, "grad_norm": 2.894995376219301, "learning_rate": 8.877165739588478e-06, "loss": 0.2724, "step": 3544 }, { "epoch": 0.48, "grad_norm": 3.553374807165972, "learning_rate": 8.876470100333755e-06, "loss": 0.3099, "step": 3545 }, { "epoch": 0.48, "grad_norm": 3.5124364241237362, "learning_rate": 8.875774272932103e-06, "loss": 0.2886, "step": 3546 }, { "epoch": 0.48, "grad_norm": 3.677007143177038, "learning_rate": 8.875078257417294e-06, "loss": 0.2992, "step": 3547 }, { "epoch": 0.48, "grad_norm": 3.913436710216213, "learning_rate": 8.874382053823115e-06, "loss": 0.303, "step": 3548 }, { "epoch": 0.48, "grad_norm": 3.6333766318451493, "learning_rate": 8.87368566218335e-06, "loss": 0.3511, "step": 3549 }, { "epoch": 0.48, "grad_norm": 3.6954469843039144, "learning_rate": 8.872989082531802e-06, "loss": 0.2374, "step": 3550 }, { "epoch": 0.48, "grad_norm": 4.253784568387391, "learning_rate": 8.87229231490228e-06, "loss": 0.3438, "step": 3551 }, { "epoch": 0.48, "grad_norm": 2.821850808543835, "learning_rate": 8.871595359328603e-06, "loss": 0.2796, "step": 3552 }, { "epoch": 0.48, "grad_norm": 3.7038597799964235, "learning_rate": 8.870898215844594e-06, "loss": 0.2812, "step": 3553 }, { "epoch": 0.48, "grad_norm": 3.54413602019436, "learning_rate": 8.870200884484094e-06, "loss": 0.2987, "step": 3554 }, { "epoch": 0.48, "grad_norm": 3.1187807451997065, "learning_rate": 8.869503365280946e-06, "loss": 0.2552, "step": 3555 }, { "epoch": 0.48, "grad_norm": 3.2522081548207367, "learning_rate": 8.868805658269005e-06, "loss": 0.2538, "step": 3556 }, { "epoch": 0.48, "grad_norm": 5.081833288721144, "learning_rate": 8.868107763482137e-06, "loss": 0.3321, "step": 3557 }, { "epoch": 0.48, "grad_norm": 3.8731593456662092, "learning_rate": 8.86740968095421e-06, "loss": 0.3216, "step": 3558 }, { "epoch": 0.48, "grad_norm": 3.435558694493149, "learning_rate": 8.86671141071911e-06, "loss": 0.2985, "step": 3559 }, { "epoch": 0.48, "grad_norm": 3.8427437106676083, "learning_rate": 8.866012952810724e-06, "loss": 0.2556, "step": 3560 }, { "epoch": 0.48, "grad_norm": 3.75936684209667, "learning_rate": 8.865314307262956e-06, "loss": 0.2938, "step": 3561 }, { "epoch": 0.48, "grad_norm": 4.200148404985688, "learning_rate": 8.864615474109715e-06, "loss": 0.3005, "step": 3562 }, { "epoch": 0.48, "grad_norm": 3.2552855124226854, "learning_rate": 8.863916453384916e-06, "loss": 0.2386, "step": 3563 }, { "epoch": 0.48, "grad_norm": 3.8132833879968078, "learning_rate": 8.86321724512249e-06, "loss": 0.2416, "step": 3564 }, { "epoch": 0.48, "grad_norm": 3.8130980679420468, "learning_rate": 8.86251784935637e-06, "loss": 0.2784, "step": 3565 }, { "epoch": 0.49, "grad_norm": 3.468259153477433, "learning_rate": 8.861818266120506e-06, "loss": 0.2998, "step": 3566 }, { "epoch": 0.49, "grad_norm": 2.8593448833533506, "learning_rate": 8.861118495448847e-06, "loss": 0.2771, "step": 3567 }, { "epoch": 0.49, "grad_norm": 3.244449739181952, "learning_rate": 8.860418537375364e-06, "loss": 0.2746, "step": 3568 }, { "epoch": 0.49, "grad_norm": 3.274641487055879, "learning_rate": 8.859718391934024e-06, "loss": 0.301, "step": 3569 }, { "epoch": 0.49, "grad_norm": 3.647549237825786, "learning_rate": 8.85901805915881e-06, "loss": 0.3281, "step": 3570 }, { "epoch": 0.49, "grad_norm": 3.264165561247552, "learning_rate": 8.858317539083715e-06, "loss": 0.3033, "step": 3571 }, { "epoch": 0.49, "grad_norm": 3.400548846277208, "learning_rate": 8.857616831742739e-06, "loss": 0.2862, "step": 3572 }, { "epoch": 0.49, "grad_norm": 3.435064731124362, "learning_rate": 8.85691593716989e-06, "loss": 0.3227, "step": 3573 }, { "epoch": 0.49, "grad_norm": 3.403004756820864, "learning_rate": 8.856214855399188e-06, "loss": 0.2654, "step": 3574 }, { "epoch": 0.49, "grad_norm": 3.1177490387570654, "learning_rate": 8.855513586464658e-06, "loss": 0.2216, "step": 3575 }, { "epoch": 0.49, "grad_norm": 3.0424734414570196, "learning_rate": 8.85481213040034e-06, "loss": 0.2837, "step": 3576 }, { "epoch": 0.49, "grad_norm": 3.601615316282444, "learning_rate": 8.854110487240275e-06, "loss": 0.2741, "step": 3577 }, { "epoch": 0.49, "grad_norm": 3.661574828416806, "learning_rate": 8.853408657018521e-06, "loss": 0.3201, "step": 3578 }, { "epoch": 0.49, "grad_norm": 3.7525087027960304, "learning_rate": 8.852706639769139e-06, "loss": 0.3265, "step": 3579 }, { "epoch": 0.49, "grad_norm": 3.901909134605903, "learning_rate": 8.852004435526207e-06, "loss": 0.331, "step": 3580 }, { "epoch": 0.49, "grad_norm": 2.9756354822151203, "learning_rate": 8.851302044323803e-06, "loss": 0.2547, "step": 3581 }, { "epoch": 0.49, "grad_norm": 3.312906074411149, "learning_rate": 8.850599466196018e-06, "loss": 0.3031, "step": 3582 }, { "epoch": 0.49, "grad_norm": 3.198552430920695, "learning_rate": 8.849896701176952e-06, "loss": 0.3272, "step": 3583 }, { "epoch": 0.49, "grad_norm": 3.931823549595083, "learning_rate": 8.849193749300716e-06, "loss": 0.3049, "step": 3584 }, { "epoch": 0.49, "grad_norm": 2.7558073397316667, "learning_rate": 8.848490610601428e-06, "loss": 0.2206, "step": 3585 }, { "epoch": 0.49, "grad_norm": 3.633813613483365, "learning_rate": 8.847787285113212e-06, "loss": 0.3009, "step": 3586 }, { "epoch": 0.49, "grad_norm": 4.250204093235515, "learning_rate": 8.847083772870209e-06, "loss": 0.2866, "step": 3587 }, { "epoch": 0.49, "grad_norm": 3.5319260646293844, "learning_rate": 8.846380073906561e-06, "loss": 0.3082, "step": 3588 }, { "epoch": 0.49, "grad_norm": 3.4783362595317575, "learning_rate": 8.845676188256425e-06, "loss": 0.2645, "step": 3589 }, { "epoch": 0.49, "grad_norm": 3.438469889116636, "learning_rate": 8.844972115953963e-06, "loss": 0.2773, "step": 3590 }, { "epoch": 0.49, "grad_norm": 3.803615631190731, "learning_rate": 8.844267857033348e-06, "loss": 0.2855, "step": 3591 }, { "epoch": 0.49, "grad_norm": 3.5003157901968582, "learning_rate": 8.84356341152876e-06, "loss": 0.2613, "step": 3592 }, { "epoch": 0.49, "grad_norm": 3.8351789662459113, "learning_rate": 8.842858779474392e-06, "loss": 0.2687, "step": 3593 }, { "epoch": 0.49, "grad_norm": 4.087110529439265, "learning_rate": 8.842153960904445e-06, "loss": 0.3355, "step": 3594 }, { "epoch": 0.49, "grad_norm": 3.274615045926965, "learning_rate": 8.841448955853123e-06, "loss": 0.3041, "step": 3595 }, { "epoch": 0.49, "grad_norm": 3.4313218789607727, "learning_rate": 8.840743764354647e-06, "loss": 0.3265, "step": 3596 }, { "epoch": 0.49, "grad_norm": 3.4803252476695534, "learning_rate": 8.840038386443243e-06, "loss": 0.3271, "step": 3597 }, { "epoch": 0.49, "grad_norm": 3.3453418263122567, "learning_rate": 8.839332822153151e-06, "loss": 0.3051, "step": 3598 }, { "epoch": 0.49, "grad_norm": 3.269181026184391, "learning_rate": 8.838627071518608e-06, "loss": 0.2706, "step": 3599 }, { "epoch": 0.49, "grad_norm": 3.1415731966708984, "learning_rate": 8.837921134573875e-06, "loss": 0.273, "step": 3600 }, { "epoch": 0.49, "grad_norm": 4.378057201307205, "learning_rate": 8.837215011353213e-06, "loss": 0.3369, "step": 3601 }, { "epoch": 0.49, "grad_norm": 3.3686010279838334, "learning_rate": 8.836508701890892e-06, "loss": 0.3233, "step": 3602 }, { "epoch": 0.49, "grad_norm": 2.9467223397092623, "learning_rate": 8.835802206221197e-06, "loss": 0.2774, "step": 3603 }, { "epoch": 0.49, "grad_norm": 3.5030060762492377, "learning_rate": 8.835095524378413e-06, "loss": 0.2714, "step": 3604 }, { "epoch": 0.49, "grad_norm": 3.5355361210892524, "learning_rate": 8.834388656396846e-06, "loss": 0.2931, "step": 3605 }, { "epoch": 0.49, "grad_norm": 3.584746630117638, "learning_rate": 8.8336816023108e-06, "loss": 0.3936, "step": 3606 }, { "epoch": 0.49, "grad_norm": 4.494570953383951, "learning_rate": 8.832974362154592e-06, "loss": 0.3478, "step": 3607 }, { "epoch": 0.49, "grad_norm": 9.209263884014339, "learning_rate": 8.83226693596255e-06, "loss": 0.2671, "step": 3608 }, { "epoch": 0.49, "grad_norm": 3.1389330913631577, "learning_rate": 8.83155932376901e-06, "loss": 0.2646, "step": 3609 }, { "epoch": 0.49, "grad_norm": 3.06804762120088, "learning_rate": 8.830851525608313e-06, "loss": 0.2863, "step": 3610 }, { "epoch": 0.49, "grad_norm": 3.578841375084879, "learning_rate": 8.830143541514815e-06, "loss": 0.2812, "step": 3611 }, { "epoch": 0.49, "grad_norm": 3.405210600374847, "learning_rate": 8.829435371522879e-06, "loss": 0.2485, "step": 3612 }, { "epoch": 0.49, "grad_norm": 3.1144192343170003, "learning_rate": 8.828727015666873e-06, "loss": 0.3246, "step": 3613 }, { "epoch": 0.49, "grad_norm": 4.260035753450584, "learning_rate": 8.828018473981184e-06, "loss": 0.2646, "step": 3614 }, { "epoch": 0.49, "grad_norm": 3.9314637222292097, "learning_rate": 8.827309746500196e-06, "loss": 0.2536, "step": 3615 }, { "epoch": 0.49, "grad_norm": 3.6974463584109634, "learning_rate": 8.826600833258307e-06, "loss": 0.2766, "step": 3616 }, { "epoch": 0.49, "grad_norm": 4.186318359660291, "learning_rate": 8.82589173428993e-06, "loss": 0.2691, "step": 3617 }, { "epoch": 0.49, "grad_norm": 3.462657485711332, "learning_rate": 8.825182449629477e-06, "loss": 0.2611, "step": 3618 }, { "epoch": 0.49, "grad_norm": 3.6097223341983193, "learning_rate": 8.824472979311374e-06, "loss": 0.3302, "step": 3619 }, { "epoch": 0.49, "grad_norm": 3.32700416989697, "learning_rate": 8.823763323370056e-06, "loss": 0.2925, "step": 3620 }, { "epoch": 0.49, "grad_norm": 4.068459183983432, "learning_rate": 8.823053481839967e-06, "loss": 0.2608, "step": 3621 }, { "epoch": 0.49, "grad_norm": 3.633216987352601, "learning_rate": 8.822343454755562e-06, "loss": 0.3598, "step": 3622 }, { "epoch": 0.49, "grad_norm": 3.6310947990890945, "learning_rate": 8.821633242151297e-06, "loss": 0.2843, "step": 3623 }, { "epoch": 0.49, "grad_norm": 3.4708726444809903, "learning_rate": 8.820922844061648e-06, "loss": 0.2759, "step": 3624 }, { "epoch": 0.49, "grad_norm": 2.8310557675894654, "learning_rate": 8.820212260521092e-06, "loss": 0.2891, "step": 3625 }, { "epoch": 0.49, "grad_norm": 3.6837265885422785, "learning_rate": 8.819501491564118e-06, "loss": 0.3097, "step": 3626 }, { "epoch": 0.49, "grad_norm": 3.3358009585667534, "learning_rate": 8.818790537225224e-06, "loss": 0.2937, "step": 3627 }, { "epoch": 0.49, "grad_norm": 3.4226941883616178, "learning_rate": 8.818079397538915e-06, "loss": 0.2878, "step": 3628 }, { "epoch": 0.49, "grad_norm": 3.632489405126844, "learning_rate": 8.817368072539709e-06, "loss": 0.2506, "step": 3629 }, { "epoch": 0.49, "grad_norm": 3.74887255878439, "learning_rate": 8.81665656226213e-06, "loss": 0.277, "step": 3630 }, { "epoch": 0.49, "grad_norm": 2.951920007148453, "learning_rate": 8.81594486674071e-06, "loss": 0.2898, "step": 3631 }, { "epoch": 0.49, "grad_norm": 4.266060074757242, "learning_rate": 8.815232986009994e-06, "loss": 0.2962, "step": 3632 }, { "epoch": 0.49, "grad_norm": 3.528802990235292, "learning_rate": 8.81452092010453e-06, "loss": 0.274, "step": 3633 }, { "epoch": 0.49, "grad_norm": 4.371113854272213, "learning_rate": 8.813808669058883e-06, "loss": 0.3495, "step": 3634 }, { "epoch": 0.49, "grad_norm": 3.726839946802292, "learning_rate": 8.81309623290762e-06, "loss": 0.3178, "step": 3635 }, { "epoch": 0.49, "grad_norm": 3.2944929192588526, "learning_rate": 8.81238361168532e-06, "loss": 0.2268, "step": 3636 }, { "epoch": 0.49, "grad_norm": 3.1999810349430002, "learning_rate": 8.81167080542657e-06, "loss": 0.2383, "step": 3637 }, { "epoch": 0.49, "grad_norm": 3.9596640564796175, "learning_rate": 8.810957814165968e-06, "loss": 0.3074, "step": 3638 }, { "epoch": 0.5, "grad_norm": 3.1155147949319963, "learning_rate": 8.810244637938117e-06, "loss": 0.2563, "step": 3639 }, { "epoch": 0.5, "grad_norm": 3.79508928733521, "learning_rate": 8.809531276777633e-06, "loss": 0.3114, "step": 3640 }, { "epoch": 0.5, "grad_norm": 3.9289657853340354, "learning_rate": 8.808817730719141e-06, "loss": 0.2937, "step": 3641 }, { "epoch": 0.5, "grad_norm": 3.757617090958288, "learning_rate": 8.80810399979727e-06, "loss": 0.3228, "step": 3642 }, { "epoch": 0.5, "grad_norm": 3.1795653568689173, "learning_rate": 8.807390084046665e-06, "loss": 0.2596, "step": 3643 }, { "epoch": 0.5, "grad_norm": 3.3352604369811196, "learning_rate": 8.806675983501973e-06, "loss": 0.2937, "step": 3644 }, { "epoch": 0.5, "grad_norm": 4.120425019466423, "learning_rate": 8.805961698197856e-06, "loss": 0.3763, "step": 3645 }, { "epoch": 0.5, "grad_norm": 3.1082399246161154, "learning_rate": 8.80524722816898e-06, "loss": 0.2619, "step": 3646 }, { "epoch": 0.5, "grad_norm": 3.90583296942362, "learning_rate": 8.804532573450024e-06, "loss": 0.3291, "step": 3647 }, { "epoch": 0.5, "grad_norm": 3.2006147974596058, "learning_rate": 8.803817734075674e-06, "loss": 0.2478, "step": 3648 }, { "epoch": 0.5, "grad_norm": 3.4113032227864166, "learning_rate": 8.803102710080623e-06, "loss": 0.2452, "step": 3649 }, { "epoch": 0.5, "grad_norm": 3.950117857864946, "learning_rate": 8.802387501499578e-06, "loss": 0.3512, "step": 3650 }, { "epoch": 0.5, "grad_norm": 3.444549868266838, "learning_rate": 8.801672108367251e-06, "loss": 0.3555, "step": 3651 }, { "epoch": 0.5, "grad_norm": 3.964994475121369, "learning_rate": 8.800956530718365e-06, "loss": 0.3173, "step": 3652 }, { "epoch": 0.5, "grad_norm": 3.7659449625080654, "learning_rate": 8.800240768587649e-06, "loss": 0.342, "step": 3653 }, { "epoch": 0.5, "grad_norm": 2.876942462146189, "learning_rate": 8.799524822009844e-06, "loss": 0.2701, "step": 3654 }, { "epoch": 0.5, "grad_norm": 3.5175076001482206, "learning_rate": 8.7988086910197e-06, "loss": 0.3157, "step": 3655 }, { "epoch": 0.5, "grad_norm": 3.2507003403689105, "learning_rate": 8.798092375651973e-06, "loss": 0.2636, "step": 3656 }, { "epoch": 0.5, "grad_norm": 3.375634181539502, "learning_rate": 8.797375875941431e-06, "loss": 0.286, "step": 3657 }, { "epoch": 0.5, "grad_norm": 3.317630459538184, "learning_rate": 8.796659191922849e-06, "loss": 0.2845, "step": 3658 }, { "epoch": 0.5, "grad_norm": 3.4843904168268254, "learning_rate": 8.795942323631013e-06, "loss": 0.2885, "step": 3659 }, { "epoch": 0.5, "grad_norm": 3.1484418607413986, "learning_rate": 8.795225271100716e-06, "loss": 0.2964, "step": 3660 }, { "epoch": 0.5, "grad_norm": 3.44423634559877, "learning_rate": 8.794508034366758e-06, "loss": 0.2797, "step": 3661 }, { "epoch": 0.5, "grad_norm": 3.6120306300049827, "learning_rate": 8.793790613463956e-06, "loss": 0.3521, "step": 3662 }, { "epoch": 0.5, "grad_norm": 3.2487550295516905, "learning_rate": 8.793073008427126e-06, "loss": 0.2546, "step": 3663 }, { "epoch": 0.5, "grad_norm": 3.1069035724631116, "learning_rate": 8.792355219291098e-06, "loss": 0.2823, "step": 3664 }, { "epoch": 0.5, "grad_norm": 3.9860991786437725, "learning_rate": 8.791637246090712e-06, "loss": 0.3126, "step": 3665 }, { "epoch": 0.5, "grad_norm": 3.4924420699945, "learning_rate": 8.790919088860815e-06, "loss": 0.2468, "step": 3666 }, { "epoch": 0.5, "grad_norm": 3.551008613647354, "learning_rate": 8.790200747636261e-06, "loss": 0.2939, "step": 3667 }, { "epoch": 0.5, "grad_norm": 4.226186514074771, "learning_rate": 8.78948222245192e-06, "loss": 0.2727, "step": 3668 }, { "epoch": 0.5, "grad_norm": 3.790098664116247, "learning_rate": 8.78876351334266e-06, "loss": 0.3285, "step": 3669 }, { "epoch": 0.5, "grad_norm": 3.5344298463698953, "learning_rate": 8.788044620343368e-06, "loss": 0.2657, "step": 3670 }, { "epoch": 0.5, "grad_norm": 2.9112617943568013, "learning_rate": 8.787325543488934e-06, "loss": 0.2385, "step": 3671 }, { "epoch": 0.5, "grad_norm": 3.2859957208512873, "learning_rate": 8.78660628281426e-06, "loss": 0.2739, "step": 3672 }, { "epoch": 0.5, "grad_norm": 3.40554426227948, "learning_rate": 8.785886838354256e-06, "loss": 0.2863, "step": 3673 }, { "epoch": 0.5, "grad_norm": 5.068741947519447, "learning_rate": 8.78516721014384e-06, "loss": 0.3343, "step": 3674 }, { "epoch": 0.5, "grad_norm": 4.082667754914286, "learning_rate": 8.78444739821794e-06, "loss": 0.3055, "step": 3675 }, { "epoch": 0.5, "grad_norm": 3.156346762319341, "learning_rate": 8.783727402611491e-06, "loss": 0.2743, "step": 3676 }, { "epoch": 0.5, "grad_norm": 3.602834037471883, "learning_rate": 8.78300722335944e-06, "loss": 0.3362, "step": 3677 }, { "epoch": 0.5, "grad_norm": 3.9366407472244886, "learning_rate": 8.782286860496743e-06, "loss": 0.3406, "step": 3678 }, { "epoch": 0.5, "grad_norm": 3.712583922643276, "learning_rate": 8.781566314058361e-06, "loss": 0.2826, "step": 3679 }, { "epoch": 0.5, "grad_norm": 3.5002762329768182, "learning_rate": 8.780845584079265e-06, "loss": 0.338, "step": 3680 }, { "epoch": 0.5, "grad_norm": 3.169789247421087, "learning_rate": 8.78012467059444e-06, "loss": 0.2754, "step": 3681 }, { "epoch": 0.5, "grad_norm": 3.785009721576519, "learning_rate": 8.77940357363887e-06, "loss": 0.3552, "step": 3682 }, { "epoch": 0.5, "grad_norm": 3.2076827432172044, "learning_rate": 8.778682293247558e-06, "loss": 0.2633, "step": 3683 }, { "epoch": 0.5, "grad_norm": 3.0439619622375047, "learning_rate": 8.777960829455515e-06, "loss": 0.2431, "step": 3684 }, { "epoch": 0.5, "grad_norm": 3.252358597882353, "learning_rate": 8.77723918229775e-06, "loss": 0.3041, "step": 3685 }, { "epoch": 0.5, "grad_norm": 2.7187151310937643, "learning_rate": 8.776517351809293e-06, "loss": 0.2332, "step": 3686 }, { "epoch": 0.5, "grad_norm": 2.715887592564287, "learning_rate": 8.77579533802518e-06, "loss": 0.2584, "step": 3687 }, { "epoch": 0.5, "grad_norm": 2.988341084713504, "learning_rate": 8.775073140980453e-06, "loss": 0.2875, "step": 3688 }, { "epoch": 0.5, "grad_norm": 2.8003102863220093, "learning_rate": 8.774350760710161e-06, "loss": 0.2247, "step": 3689 }, { "epoch": 0.5, "grad_norm": 3.3149360588219663, "learning_rate": 8.77362819724937e-06, "loss": 0.3077, "step": 3690 }, { "epoch": 0.5, "grad_norm": 3.9980171972105043, "learning_rate": 8.772905450633147e-06, "loss": 0.3121, "step": 3691 }, { "epoch": 0.5, "grad_norm": 3.2125893396281997, "learning_rate": 8.772182520896573e-06, "loss": 0.2275, "step": 3692 }, { "epoch": 0.5, "grad_norm": 3.6555890270151785, "learning_rate": 8.771459408074734e-06, "loss": 0.3024, "step": 3693 }, { "epoch": 0.5, "grad_norm": 4.313663808852401, "learning_rate": 8.770736112202728e-06, "loss": 0.3323, "step": 3694 }, { "epoch": 0.5, "grad_norm": 3.9437599389533204, "learning_rate": 8.77001263331566e-06, "loss": 0.2525, "step": 3695 }, { "epoch": 0.5, "grad_norm": 3.795815525552101, "learning_rate": 8.769288971448645e-06, "loss": 0.3549, "step": 3696 }, { "epoch": 0.5, "grad_norm": 4.352950210498795, "learning_rate": 8.768565126636806e-06, "loss": 0.3374, "step": 3697 }, { "epoch": 0.5, "grad_norm": 3.4381751645854464, "learning_rate": 8.767841098915276e-06, "loss": 0.2706, "step": 3698 }, { "epoch": 0.5, "grad_norm": 3.5118026447257913, "learning_rate": 8.767116888319195e-06, "loss": 0.3295, "step": 3699 }, { "epoch": 0.5, "grad_norm": 3.638284953642143, "learning_rate": 8.766392494883713e-06, "loss": 0.2793, "step": 3700 }, { "epoch": 0.5, "grad_norm": 3.9229694668798567, "learning_rate": 8.76566791864399e-06, "loss": 0.2796, "step": 3701 }, { "epoch": 0.5, "grad_norm": 3.0558068661037807, "learning_rate": 8.764943159635193e-06, "loss": 0.2492, "step": 3702 }, { "epoch": 0.5, "grad_norm": 3.1523197343106375, "learning_rate": 8.7642182178925e-06, "loss": 0.236, "step": 3703 }, { "epoch": 0.5, "grad_norm": 5.688735808791103, "learning_rate": 8.763493093451093e-06, "loss": 0.2375, "step": 3704 }, { "epoch": 0.5, "grad_norm": 3.403118709275436, "learning_rate": 8.762767786346169e-06, "loss": 0.2625, "step": 3705 }, { "epoch": 0.5, "grad_norm": 3.010764829551988, "learning_rate": 8.762042296612934e-06, "loss": 0.2534, "step": 3706 }, { "epoch": 0.5, "grad_norm": 3.536684187269166, "learning_rate": 8.761316624286593e-06, "loss": 0.3142, "step": 3707 }, { "epoch": 0.5, "grad_norm": 4.036534240311782, "learning_rate": 8.760590769402372e-06, "loss": 0.2555, "step": 3708 }, { "epoch": 0.5, "grad_norm": 2.966695387229186, "learning_rate": 8.759864731995502e-06, "loss": 0.2424, "step": 3709 }, { "epoch": 0.5, "grad_norm": 3.5679171654830526, "learning_rate": 8.75913851210122e-06, "loss": 0.2896, "step": 3710 }, { "epoch": 0.5, "grad_norm": 3.180271629664057, "learning_rate": 8.75841210975477e-06, "loss": 0.2718, "step": 3711 }, { "epoch": 0.5, "grad_norm": 4.502148062984218, "learning_rate": 8.757685524991414e-06, "loss": 0.271, "step": 3712 }, { "epoch": 0.51, "grad_norm": 2.364362300507454, "learning_rate": 8.756958757846412e-06, "loss": 0.1888, "step": 3713 }, { "epoch": 0.51, "grad_norm": 3.3783564263654684, "learning_rate": 8.756231808355042e-06, "loss": 0.2629, "step": 3714 }, { "epoch": 0.51, "grad_norm": 3.3446184602797926, "learning_rate": 8.755504676552586e-06, "loss": 0.3198, "step": 3715 }, { "epoch": 0.51, "grad_norm": 3.421683346935398, "learning_rate": 8.754777362474338e-06, "loss": 0.3586, "step": 3716 }, { "epoch": 0.51, "grad_norm": 3.426012837910216, "learning_rate": 8.754049866155594e-06, "loss": 0.3285, "step": 3717 }, { "epoch": 0.51, "grad_norm": 3.531322422010145, "learning_rate": 8.753322187631665e-06, "loss": 0.3212, "step": 3718 }, { "epoch": 0.51, "grad_norm": 3.359283871408477, "learning_rate": 8.75259432693787e-06, "loss": 0.2377, "step": 3719 }, { "epoch": 0.51, "grad_norm": 3.7371202514357056, "learning_rate": 8.751866284109538e-06, "loss": 0.3116, "step": 3720 }, { "epoch": 0.51, "grad_norm": 3.1378160064191207, "learning_rate": 8.751138059182001e-06, "loss": 0.2733, "step": 3721 }, { "epoch": 0.51, "grad_norm": 3.233736948770276, "learning_rate": 8.750409652190609e-06, "loss": 0.2826, "step": 3722 }, { "epoch": 0.51, "grad_norm": 3.7340577909351076, "learning_rate": 8.74968106317071e-06, "loss": 0.2478, "step": 3723 }, { "epoch": 0.51, "grad_norm": 3.185415754964161, "learning_rate": 8.748952292157672e-06, "loss": 0.2463, "step": 3724 }, { "epoch": 0.51, "grad_norm": 3.968243269157046, "learning_rate": 8.748223339186861e-06, "loss": 0.3242, "step": 3725 }, { "epoch": 0.51, "grad_norm": 2.856452495897657, "learning_rate": 8.74749420429366e-06, "loss": 0.2553, "step": 3726 }, { "epoch": 0.51, "grad_norm": 4.046012833809931, "learning_rate": 8.74676488751346e-06, "loss": 0.2591, "step": 3727 }, { "epoch": 0.51, "grad_norm": 3.6458983573608794, "learning_rate": 8.746035388881655e-06, "loss": 0.2636, "step": 3728 }, { "epoch": 0.51, "grad_norm": 4.870922448531485, "learning_rate": 8.745305708433655e-06, "loss": 0.3245, "step": 3729 }, { "epoch": 0.51, "grad_norm": 3.4108696245140893, "learning_rate": 8.744575846204872e-06, "loss": 0.2921, "step": 3730 }, { "epoch": 0.51, "grad_norm": 4.272458143706965, "learning_rate": 8.743845802230734e-06, "loss": 0.2856, "step": 3731 }, { "epoch": 0.51, "grad_norm": 3.6038335771782455, "learning_rate": 8.743115576546672e-06, "loss": 0.2872, "step": 3732 }, { "epoch": 0.51, "grad_norm": 3.482174424849175, "learning_rate": 8.742385169188128e-06, "loss": 0.2667, "step": 3733 }, { "epoch": 0.51, "grad_norm": 3.2554102605336435, "learning_rate": 8.741654580190554e-06, "loss": 0.2932, "step": 3734 }, { "epoch": 0.51, "grad_norm": 3.0305084704414846, "learning_rate": 8.740923809589408e-06, "loss": 0.2977, "step": 3735 }, { "epoch": 0.51, "grad_norm": 3.251907270018336, "learning_rate": 8.74019285742016e-06, "loss": 0.2677, "step": 3736 }, { "epoch": 0.51, "grad_norm": 3.0126074914675165, "learning_rate": 8.739461723718286e-06, "loss": 0.2371, "step": 3737 }, { "epoch": 0.51, "grad_norm": 2.7700375759973346, "learning_rate": 8.738730408519271e-06, "loss": 0.2402, "step": 3738 }, { "epoch": 0.51, "grad_norm": 3.096866574291473, "learning_rate": 8.737998911858613e-06, "loss": 0.2544, "step": 3739 }, { "epoch": 0.51, "grad_norm": 2.968879373295531, "learning_rate": 8.737267233771813e-06, "loss": 0.2402, "step": 3740 }, { "epoch": 0.51, "grad_norm": 3.2052353563582585, "learning_rate": 8.736535374294385e-06, "loss": 0.2726, "step": 3741 }, { "epoch": 0.51, "grad_norm": 3.9447098668170946, "learning_rate": 8.73580333346185e-06, "loss": 0.3353, "step": 3742 }, { "epoch": 0.51, "grad_norm": 3.9011326590377213, "learning_rate": 8.735071111309736e-06, "loss": 0.3139, "step": 3743 }, { "epoch": 0.51, "grad_norm": 3.215162115841284, "learning_rate": 8.734338707873583e-06, "loss": 0.2591, "step": 3744 }, { "epoch": 0.51, "grad_norm": 3.155722155540406, "learning_rate": 8.733606123188942e-06, "loss": 0.2934, "step": 3745 }, { "epoch": 0.51, "grad_norm": 3.133203330368428, "learning_rate": 8.732873357291367e-06, "loss": 0.2847, "step": 3746 }, { "epoch": 0.51, "grad_norm": 3.8080768487251033, "learning_rate": 8.732140410216422e-06, "loss": 0.3051, "step": 3747 }, { "epoch": 0.51, "grad_norm": 3.147490722888383, "learning_rate": 8.73140728199968e-06, "loss": 0.2412, "step": 3748 }, { "epoch": 0.51, "grad_norm": 3.48960847065788, "learning_rate": 8.73067397267673e-06, "loss": 0.2768, "step": 3749 }, { "epoch": 0.51, "grad_norm": 4.11614273898044, "learning_rate": 8.729940482283156e-06, "loss": 0.3523, "step": 3750 }, { "epoch": 0.51, "grad_norm": 3.4766475060648334, "learning_rate": 8.729206810854563e-06, "loss": 0.2801, "step": 3751 }, { "epoch": 0.51, "grad_norm": 3.3713448458329736, "learning_rate": 8.72847295842656e-06, "loss": 0.2512, "step": 3752 }, { "epoch": 0.51, "grad_norm": 3.6576891444334887, "learning_rate": 8.727738925034761e-06, "loss": 0.2642, "step": 3753 }, { "epoch": 0.51, "grad_norm": 3.761659162823562, "learning_rate": 8.7270047107148e-06, "loss": 0.2739, "step": 3754 }, { "epoch": 0.51, "grad_norm": 3.4553239017501927, "learning_rate": 8.726270315502305e-06, "loss": 0.332, "step": 3755 }, { "epoch": 0.51, "grad_norm": 3.5100755330550015, "learning_rate": 8.725535739432926e-06, "loss": 0.3051, "step": 3756 }, { "epoch": 0.51, "grad_norm": 3.7981334210647204, "learning_rate": 8.724800982542313e-06, "loss": 0.3209, "step": 3757 }, { "epoch": 0.51, "grad_norm": 3.8153594407468026, "learning_rate": 8.724066044866128e-06, "loss": 0.2879, "step": 3758 }, { "epoch": 0.51, "grad_norm": 3.2067038260700964, "learning_rate": 8.723330926440045e-06, "loss": 0.256, "step": 3759 }, { "epoch": 0.51, "grad_norm": 3.368846965025368, "learning_rate": 8.72259562729974e-06, "loss": 0.2839, "step": 3760 }, { "epoch": 0.51, "grad_norm": 3.358012230965094, "learning_rate": 8.7218601474809e-06, "loss": 0.2726, "step": 3761 }, { "epoch": 0.51, "grad_norm": 3.2855586568967525, "learning_rate": 8.721124487019226e-06, "loss": 0.27, "step": 3762 }, { "epoch": 0.51, "grad_norm": 3.2371535434901784, "learning_rate": 8.72038864595042e-06, "loss": 0.293, "step": 3763 }, { "epoch": 0.51, "grad_norm": 3.309294959173422, "learning_rate": 8.7196526243102e-06, "loss": 0.3003, "step": 3764 }, { "epoch": 0.51, "grad_norm": 3.1230418491367495, "learning_rate": 8.718916422134286e-06, "loss": 0.2833, "step": 3765 }, { "epoch": 0.51, "grad_norm": 3.249601270689942, "learning_rate": 8.718180039458414e-06, "loss": 0.3343, "step": 3766 }, { "epoch": 0.51, "grad_norm": 3.280608622618923, "learning_rate": 8.717443476318322e-06, "loss": 0.2721, "step": 3767 }, { "epoch": 0.51, "grad_norm": 3.1753616827417566, "learning_rate": 8.716706732749758e-06, "loss": 0.2598, "step": 3768 }, { "epoch": 0.51, "grad_norm": 3.6948513777471557, "learning_rate": 8.715969808788484e-06, "loss": 0.1976, "step": 3769 }, { "epoch": 0.51, "grad_norm": 3.231175069940734, "learning_rate": 8.715232704470265e-06, "loss": 0.2516, "step": 3770 }, { "epoch": 0.51, "grad_norm": 3.7261027377255553, "learning_rate": 8.714495419830878e-06, "loss": 0.2968, "step": 3771 }, { "epoch": 0.51, "grad_norm": 3.3446250064403693, "learning_rate": 8.713757954906105e-06, "loss": 0.2785, "step": 3772 }, { "epoch": 0.51, "grad_norm": 3.669158842716648, "learning_rate": 8.713020309731742e-06, "loss": 0.2944, "step": 3773 }, { "epoch": 0.51, "grad_norm": 3.790447927639789, "learning_rate": 8.71228248434359e-06, "loss": 0.3366, "step": 3774 }, { "epoch": 0.51, "grad_norm": 3.866324592287745, "learning_rate": 8.71154447877746e-06, "loss": 0.3461, "step": 3775 }, { "epoch": 0.51, "grad_norm": 3.5012417382608643, "learning_rate": 8.710806293069172e-06, "loss": 0.2972, "step": 3776 }, { "epoch": 0.51, "grad_norm": 3.8078160871653663, "learning_rate": 8.710067927254555e-06, "loss": 0.2901, "step": 3777 }, { "epoch": 0.51, "grad_norm": 3.5136433431381753, "learning_rate": 8.709329381369443e-06, "loss": 0.2617, "step": 3778 }, { "epoch": 0.51, "grad_norm": 3.3187813918573634, "learning_rate": 8.708590655449684e-06, "loss": 0.2894, "step": 3779 }, { "epoch": 0.51, "grad_norm": 4.127816566710691, "learning_rate": 8.707851749531133e-06, "loss": 0.326, "step": 3780 }, { "epoch": 0.51, "grad_norm": 3.503980543876884, "learning_rate": 8.707112663649653e-06, "loss": 0.2412, "step": 3781 }, { "epoch": 0.51, "grad_norm": 3.2394674804076566, "learning_rate": 8.706373397841114e-06, "loss": 0.2903, "step": 3782 }, { "epoch": 0.51, "grad_norm": 3.1659032566706165, "learning_rate": 8.705633952141398e-06, "loss": 0.3343, "step": 3783 }, { "epoch": 0.51, "grad_norm": 3.46616084237562, "learning_rate": 8.704894326586398e-06, "loss": 0.2995, "step": 3784 }, { "epoch": 0.51, "grad_norm": 3.397767433101572, "learning_rate": 8.704154521212007e-06, "loss": 0.2513, "step": 3785 }, { "epoch": 0.51, "grad_norm": 3.2807450896944355, "learning_rate": 8.703414536054132e-06, "loss": 0.2707, "step": 3786 }, { "epoch": 0.52, "grad_norm": 3.793975635630126, "learning_rate": 8.702674371148692e-06, "loss": 0.2762, "step": 3787 }, { "epoch": 0.52, "grad_norm": 3.3222995338746317, "learning_rate": 8.70193402653161e-06, "loss": 0.2288, "step": 3788 }, { "epoch": 0.52, "grad_norm": 3.65938349497644, "learning_rate": 8.701193502238818e-06, "loss": 0.3161, "step": 3789 }, { "epoch": 0.52, "grad_norm": 3.240410862800501, "learning_rate": 8.70045279830626e-06, "loss": 0.244, "step": 3790 }, { "epoch": 0.52, "grad_norm": 3.480613172628878, "learning_rate": 8.699711914769885e-06, "loss": 0.2725, "step": 3791 }, { "epoch": 0.52, "grad_norm": 3.899312028776182, "learning_rate": 8.698970851665652e-06, "loss": 0.2945, "step": 3792 }, { "epoch": 0.52, "grad_norm": 3.234570694474455, "learning_rate": 8.69822960902953e-06, "loss": 0.3095, "step": 3793 }, { "epoch": 0.52, "grad_norm": 3.5074505466181023, "learning_rate": 8.697488186897496e-06, "loss": 0.3154, "step": 3794 }, { "epoch": 0.52, "grad_norm": 3.0542357259470565, "learning_rate": 8.696746585305533e-06, "loss": 0.2769, "step": 3795 }, { "epoch": 0.52, "grad_norm": 3.3383478437422367, "learning_rate": 8.696004804289637e-06, "loss": 0.2933, "step": 3796 }, { "epoch": 0.52, "grad_norm": 3.5461915778752946, "learning_rate": 8.695262843885812e-06, "loss": 0.3578, "step": 3797 }, { "epoch": 0.52, "grad_norm": 3.302020133540848, "learning_rate": 8.694520704130067e-06, "loss": 0.2664, "step": 3798 }, { "epoch": 0.52, "grad_norm": 3.4883695480518417, "learning_rate": 8.693778385058423e-06, "loss": 0.3013, "step": 3799 }, { "epoch": 0.52, "grad_norm": 3.2160130825349165, "learning_rate": 8.693035886706909e-06, "loss": 0.2629, "step": 3800 }, { "epoch": 0.52, "grad_norm": 3.1903084950692087, "learning_rate": 8.692293209111563e-06, "loss": 0.2783, "step": 3801 }, { "epoch": 0.52, "grad_norm": 3.5499466857910917, "learning_rate": 8.691550352308431e-06, "loss": 0.2834, "step": 3802 }, { "epoch": 0.52, "grad_norm": 3.3930632107285, "learning_rate": 8.690807316333569e-06, "loss": 0.3068, "step": 3803 }, { "epoch": 0.52, "grad_norm": 3.5242933220263453, "learning_rate": 8.69006410122304e-06, "loss": 0.2826, "step": 3804 }, { "epoch": 0.52, "grad_norm": 2.9022815630191947, "learning_rate": 8.689320707012914e-06, "loss": 0.2678, "step": 3805 }, { "epoch": 0.52, "grad_norm": 4.021006541387549, "learning_rate": 8.688577133739276e-06, "loss": 0.3296, "step": 3806 }, { "epoch": 0.52, "grad_norm": 3.037006700477201, "learning_rate": 8.687833381438215e-06, "loss": 0.2341, "step": 3807 }, { "epoch": 0.52, "grad_norm": 3.0052574633271396, "learning_rate": 8.687089450145828e-06, "loss": 0.2561, "step": 3808 }, { "epoch": 0.52, "grad_norm": 3.1553015869874175, "learning_rate": 8.686345339898221e-06, "loss": 0.2603, "step": 3809 }, { "epoch": 0.52, "grad_norm": 3.495906012422733, "learning_rate": 8.685601050731515e-06, "loss": 0.3129, "step": 3810 }, { "epoch": 0.52, "grad_norm": 3.5481845193516834, "learning_rate": 8.684856582681828e-06, "loss": 0.3279, "step": 3811 }, { "epoch": 0.52, "grad_norm": 2.9346690945758667, "learning_rate": 8.684111935785299e-06, "loss": 0.259, "step": 3812 }, { "epoch": 0.52, "grad_norm": 3.6895892682497546, "learning_rate": 8.683367110078068e-06, "loss": 0.3164, "step": 3813 }, { "epoch": 0.52, "grad_norm": 3.2089325371197592, "learning_rate": 8.682622105596281e-06, "loss": 0.2636, "step": 3814 }, { "epoch": 0.52, "grad_norm": 3.8209698088586053, "learning_rate": 8.681876922376104e-06, "loss": 0.2879, "step": 3815 }, { "epoch": 0.52, "grad_norm": 2.869522017191074, "learning_rate": 8.681131560453702e-06, "loss": 0.2809, "step": 3816 }, { "epoch": 0.52, "grad_norm": 3.1711437961974305, "learning_rate": 8.680386019865253e-06, "loss": 0.2731, "step": 3817 }, { "epoch": 0.52, "grad_norm": 3.2582648854119634, "learning_rate": 8.679640300646938e-06, "loss": 0.2729, "step": 3818 }, { "epoch": 0.52, "grad_norm": 4.153996366833352, "learning_rate": 8.678894402834957e-06, "loss": 0.3385, "step": 3819 }, { "epoch": 0.52, "grad_norm": 3.305078043200578, "learning_rate": 8.678148326465509e-06, "loss": 0.2705, "step": 3820 }, { "epoch": 0.52, "grad_norm": 3.5420237388976035, "learning_rate": 8.677402071574806e-06, "loss": 0.2559, "step": 3821 }, { "epoch": 0.52, "grad_norm": 3.568219622480378, "learning_rate": 8.676655638199068e-06, "loss": 0.2752, "step": 3822 }, { "epoch": 0.52, "grad_norm": 2.7770715932304126, "learning_rate": 8.675909026374521e-06, "loss": 0.2522, "step": 3823 }, { "epoch": 0.52, "grad_norm": 3.3624538342740244, "learning_rate": 8.675162236137406e-06, "loss": 0.2712, "step": 3824 }, { "epoch": 0.52, "grad_norm": 3.3526504839941214, "learning_rate": 8.67441526752397e-06, "loss": 0.2393, "step": 3825 }, { "epoch": 0.52, "grad_norm": 3.7101885125039513, "learning_rate": 8.673668120570465e-06, "loss": 0.3416, "step": 3826 }, { "epoch": 0.52, "grad_norm": 3.4784891127056805, "learning_rate": 8.67292079531315e-06, "loss": 0.2711, "step": 3827 }, { "epoch": 0.52, "grad_norm": 4.020402592107121, "learning_rate": 8.672173291788306e-06, "loss": 0.2754, "step": 3828 }, { "epoch": 0.52, "grad_norm": 4.133435808418224, "learning_rate": 8.671425610032208e-06, "loss": 0.3678, "step": 3829 }, { "epoch": 0.52, "grad_norm": 3.908098128962925, "learning_rate": 8.670677750081146e-06, "loss": 0.2586, "step": 3830 }, { "epoch": 0.52, "grad_norm": 2.92139206252748, "learning_rate": 8.669929711971419e-06, "loss": 0.2953, "step": 3831 }, { "epoch": 0.52, "grad_norm": 3.4027169339946433, "learning_rate": 8.669181495739332e-06, "loss": 0.3333, "step": 3832 }, { "epoch": 0.52, "grad_norm": 3.462184761324405, "learning_rate": 8.6684331014212e-06, "loss": 0.2673, "step": 3833 }, { "epoch": 0.52, "grad_norm": 3.871465655784022, "learning_rate": 8.667684529053348e-06, "loss": 0.3086, "step": 3834 }, { "epoch": 0.52, "grad_norm": 3.285150154701466, "learning_rate": 8.666935778672107e-06, "loss": 0.2534, "step": 3835 }, { "epoch": 0.52, "grad_norm": 3.6256893175796803, "learning_rate": 8.66618685031382e-06, "loss": 0.2917, "step": 3836 }, { "epoch": 0.52, "grad_norm": 3.3486813475532786, "learning_rate": 8.665437744014838e-06, "loss": 0.3167, "step": 3837 }, { "epoch": 0.52, "grad_norm": 3.484588928011699, "learning_rate": 8.664688459811514e-06, "loss": 0.2238, "step": 3838 }, { "epoch": 0.52, "grad_norm": 3.0012830592161337, "learning_rate": 8.66393899774022e-06, "loss": 0.2352, "step": 3839 }, { "epoch": 0.52, "grad_norm": 3.476901693244236, "learning_rate": 8.663189357837327e-06, "loss": 0.2637, "step": 3840 }, { "epoch": 0.52, "grad_norm": 3.3362823900773053, "learning_rate": 8.662439540139224e-06, "loss": 0.3025, "step": 3841 }, { "epoch": 0.52, "grad_norm": 2.6548621288971668, "learning_rate": 8.661689544682301e-06, "loss": 0.2554, "step": 3842 }, { "epoch": 0.52, "grad_norm": 3.384656473018084, "learning_rate": 8.66093937150296e-06, "loss": 0.2455, "step": 3843 }, { "epoch": 0.52, "grad_norm": 3.5132002519288745, "learning_rate": 8.660189020637614e-06, "loss": 0.3041, "step": 3844 }, { "epoch": 0.52, "grad_norm": 3.687141978382175, "learning_rate": 8.659438492122677e-06, "loss": 0.303, "step": 3845 }, { "epoch": 0.52, "grad_norm": 2.8541411532075456, "learning_rate": 8.658687785994579e-06, "loss": 0.2753, "step": 3846 }, { "epoch": 0.52, "grad_norm": 3.191355700022973, "learning_rate": 8.657936902289756e-06, "loss": 0.2487, "step": 3847 }, { "epoch": 0.52, "grad_norm": 3.6751845708546456, "learning_rate": 8.657185841044653e-06, "loss": 0.2835, "step": 3848 }, { "epoch": 0.52, "grad_norm": 3.455117587269959, "learning_rate": 8.65643460229572e-06, "loss": 0.3059, "step": 3849 }, { "epoch": 0.52, "grad_norm": 3.1296532389432143, "learning_rate": 8.655683186079423e-06, "loss": 0.2222, "step": 3850 }, { "epoch": 0.52, "grad_norm": 3.37813956520511, "learning_rate": 8.65493159243223e-06, "loss": 0.2291, "step": 3851 }, { "epoch": 0.52, "grad_norm": 3.239467061889867, "learning_rate": 8.65417982139062e-06, "loss": 0.2812, "step": 3852 }, { "epoch": 0.52, "grad_norm": 3.5297704634436666, "learning_rate": 8.653427872991085e-06, "loss": 0.2921, "step": 3853 }, { "epoch": 0.52, "grad_norm": 3.4973890085716604, "learning_rate": 8.652675747270114e-06, "loss": 0.3152, "step": 3854 }, { "epoch": 0.52, "grad_norm": 3.565603233068732, "learning_rate": 8.651923444264218e-06, "loss": 0.2483, "step": 3855 }, { "epoch": 0.52, "grad_norm": 4.411085631587661, "learning_rate": 8.651170964009907e-06, "loss": 0.315, "step": 3856 }, { "epoch": 0.52, "grad_norm": 3.778369050401026, "learning_rate": 8.650418306543704e-06, "loss": 0.282, "step": 3857 }, { "epoch": 0.52, "grad_norm": 3.9490473164656166, "learning_rate": 8.649665471902142e-06, "loss": 0.3111, "step": 3858 }, { "epoch": 0.52, "grad_norm": 3.389475686904579, "learning_rate": 8.648912460121757e-06, "loss": 0.2584, "step": 3859 }, { "epoch": 0.53, "grad_norm": 3.972175494425493, "learning_rate": 8.648159271239097e-06, "loss": 0.307, "step": 3860 }, { "epoch": 0.53, "grad_norm": 3.2869797124212816, "learning_rate": 8.647405905290721e-06, "loss": 0.2576, "step": 3861 }, { "epoch": 0.53, "grad_norm": 3.8619213516197823, "learning_rate": 8.646652362313193e-06, "loss": 0.3047, "step": 3862 }, { "epoch": 0.53, "grad_norm": 3.231304361808912, "learning_rate": 8.645898642343086e-06, "loss": 0.2608, "step": 3863 }, { "epoch": 0.53, "grad_norm": 3.257944788213985, "learning_rate": 8.645144745416983e-06, "loss": 0.2762, "step": 3864 }, { "epoch": 0.53, "grad_norm": 8.084357199207217, "learning_rate": 8.644390671571474e-06, "loss": 0.3536, "step": 3865 }, { "epoch": 0.53, "grad_norm": 3.491068111333431, "learning_rate": 8.64363642084316e-06, "loss": 0.3525, "step": 3866 }, { "epoch": 0.53, "grad_norm": 3.8379925445942082, "learning_rate": 8.642881993268647e-06, "loss": 0.2045, "step": 3867 }, { "epoch": 0.53, "grad_norm": 4.291004236397803, "learning_rate": 8.642127388884554e-06, "loss": 0.3373, "step": 3868 }, { "epoch": 0.53, "grad_norm": 3.350210176472467, "learning_rate": 8.641372607727503e-06, "loss": 0.3012, "step": 3869 }, { "epoch": 0.53, "grad_norm": 2.7843938263068058, "learning_rate": 8.640617649834133e-06, "loss": 0.2131, "step": 3870 }, { "epoch": 0.53, "grad_norm": 3.2360091443062733, "learning_rate": 8.63986251524108e-06, "loss": 0.2772, "step": 3871 }, { "epoch": 0.53, "grad_norm": 3.8313464375317854, "learning_rate": 8.639107203985e-06, "loss": 0.2939, "step": 3872 }, { "epoch": 0.53, "grad_norm": 3.2840512785655873, "learning_rate": 8.63835171610255e-06, "loss": 0.2708, "step": 3873 }, { "epoch": 0.53, "grad_norm": 3.0061017586597507, "learning_rate": 8.637596051630398e-06, "loss": 0.2765, "step": 3874 }, { "epoch": 0.53, "grad_norm": 3.464953507538354, "learning_rate": 8.636840210605221e-06, "loss": 0.2817, "step": 3875 }, { "epoch": 0.53, "grad_norm": 2.9338024174662323, "learning_rate": 8.636084193063704e-06, "loss": 0.2648, "step": 3876 }, { "epoch": 0.53, "grad_norm": 3.4896086798241206, "learning_rate": 8.635327999042543e-06, "loss": 0.2813, "step": 3877 }, { "epoch": 0.53, "grad_norm": 3.9328125773006803, "learning_rate": 8.634571628578438e-06, "loss": 0.2821, "step": 3878 }, { "epoch": 0.53, "grad_norm": 3.6854572643525554, "learning_rate": 8.633815081708101e-06, "loss": 0.2523, "step": 3879 }, { "epoch": 0.53, "grad_norm": 3.7340618222376754, "learning_rate": 8.633058358468249e-06, "loss": 0.3025, "step": 3880 }, { "epoch": 0.53, "grad_norm": 3.31810269463759, "learning_rate": 8.632301458895615e-06, "loss": 0.2521, "step": 3881 }, { "epoch": 0.53, "grad_norm": 2.997184260446072, "learning_rate": 8.63154438302693e-06, "loss": 0.275, "step": 3882 }, { "epoch": 0.53, "grad_norm": 3.7588622478445513, "learning_rate": 8.630787130898943e-06, "loss": 0.2854, "step": 3883 }, { "epoch": 0.53, "grad_norm": 3.4004807050884627, "learning_rate": 8.630029702548406e-06, "loss": 0.298, "step": 3884 }, { "epoch": 0.53, "grad_norm": 3.357782569133061, "learning_rate": 8.629272098012082e-06, "loss": 0.2854, "step": 3885 }, { "epoch": 0.53, "grad_norm": 2.9657330248867404, "learning_rate": 8.628514317326744e-06, "loss": 0.2381, "step": 3886 }, { "epoch": 0.53, "grad_norm": 3.5456603780049174, "learning_rate": 8.627756360529166e-06, "loss": 0.2816, "step": 3887 }, { "epoch": 0.53, "grad_norm": 3.7256269869147443, "learning_rate": 8.62699822765614e-06, "loss": 0.2674, "step": 3888 }, { "epoch": 0.53, "grad_norm": 2.8460741253640993, "learning_rate": 8.626239918744461e-06, "loss": 0.2339, "step": 3889 }, { "epoch": 0.53, "grad_norm": 3.0305655072110755, "learning_rate": 8.625481433830933e-06, "loss": 0.2293, "step": 3890 }, { "epoch": 0.53, "grad_norm": 3.2830529250924645, "learning_rate": 8.624722772952373e-06, "loss": 0.2932, "step": 3891 }, { "epoch": 0.53, "grad_norm": 4.846352533104853, "learning_rate": 8.6239639361456e-06, "loss": 0.3255, "step": 3892 }, { "epoch": 0.53, "grad_norm": 4.09269163106944, "learning_rate": 8.623204923447445e-06, "loss": 0.2968, "step": 3893 }, { "epoch": 0.53, "grad_norm": 2.9382392989505384, "learning_rate": 8.62244573489475e-06, "loss": 0.2329, "step": 3894 }, { "epoch": 0.53, "grad_norm": 3.244821180445158, "learning_rate": 8.621686370524359e-06, "loss": 0.2466, "step": 3895 }, { "epoch": 0.53, "grad_norm": 3.871465276093707, "learning_rate": 8.620926830373129e-06, "loss": 0.4204, "step": 3896 }, { "epoch": 0.53, "grad_norm": 3.7602221055156257, "learning_rate": 8.620167114477926e-06, "loss": 0.2928, "step": 3897 }, { "epoch": 0.53, "grad_norm": 3.2203809990302226, "learning_rate": 8.619407222875624e-06, "loss": 0.3191, "step": 3898 }, { "epoch": 0.53, "grad_norm": 3.229026274923726, "learning_rate": 8.618647155603103e-06, "loss": 0.28, "step": 3899 }, { "epoch": 0.53, "grad_norm": 3.9401612701121342, "learning_rate": 8.617886912697253e-06, "loss": 0.3068, "step": 3900 }, { "epoch": 0.53, "grad_norm": 3.849730300296844, "learning_rate": 8.617126494194973e-06, "loss": 0.3363, "step": 3901 }, { "epoch": 0.53, "grad_norm": 3.3143376184644957, "learning_rate": 8.616365900133175e-06, "loss": 0.2771, "step": 3902 }, { "epoch": 0.53, "grad_norm": 3.3340393629195186, "learning_rate": 8.615605130548768e-06, "loss": 0.337, "step": 3903 }, { "epoch": 0.53, "grad_norm": 2.873519197404006, "learning_rate": 8.614844185478678e-06, "loss": 0.3039, "step": 3904 }, { "epoch": 0.53, "grad_norm": 3.544814573125414, "learning_rate": 8.614083064959842e-06, "loss": 0.3398, "step": 3905 }, { "epoch": 0.53, "grad_norm": 4.4120588740655675, "learning_rate": 8.613321769029199e-06, "loss": 0.2611, "step": 3906 }, { "epoch": 0.53, "grad_norm": 3.2506734531016432, "learning_rate": 8.612560297723697e-06, "loss": 0.2259, "step": 3907 }, { "epoch": 0.53, "grad_norm": 3.6143181210826145, "learning_rate": 8.611798651080298e-06, "loss": 0.3416, "step": 3908 }, { "epoch": 0.53, "grad_norm": 3.356559006053063, "learning_rate": 8.611036829135967e-06, "loss": 0.2607, "step": 3909 }, { "epoch": 0.53, "grad_norm": 3.4607771953374593, "learning_rate": 8.61027483192768e-06, "loss": 0.3032, "step": 3910 }, { "epoch": 0.53, "grad_norm": 2.91467821617915, "learning_rate": 8.60951265949242e-06, "loss": 0.2569, "step": 3911 }, { "epoch": 0.53, "grad_norm": 3.3098779074755584, "learning_rate": 8.608750311867182e-06, "loss": 0.3028, "step": 3912 }, { "epoch": 0.53, "grad_norm": 2.9516105710913485, "learning_rate": 8.607987789088966e-06, "loss": 0.2369, "step": 3913 }, { "epoch": 0.53, "grad_norm": 4.161016983570317, "learning_rate": 8.60722509119478e-06, "loss": 0.319, "step": 3914 }, { "epoch": 0.53, "grad_norm": 3.766389204527959, "learning_rate": 8.606462218221643e-06, "loss": 0.2656, "step": 3915 }, { "epoch": 0.53, "grad_norm": 3.2395345546654197, "learning_rate": 8.60569917020658e-06, "loss": 0.2623, "step": 3916 }, { "epoch": 0.53, "grad_norm": 4.462674090850865, "learning_rate": 8.60493594718663e-06, "loss": 0.2345, "step": 3917 }, { "epoch": 0.53, "grad_norm": 12.545810997398805, "learning_rate": 8.604172549198832e-06, "loss": 0.3127, "step": 3918 }, { "epoch": 0.53, "grad_norm": 3.0887952378210786, "learning_rate": 8.603408976280243e-06, "loss": 0.2242, "step": 3919 }, { "epoch": 0.53, "grad_norm": 3.1151450349295966, "learning_rate": 8.602645228467918e-06, "loss": 0.2311, "step": 3920 }, { "epoch": 0.53, "grad_norm": 3.653248525283675, "learning_rate": 8.601881305798929e-06, "loss": 0.2757, "step": 3921 }, { "epoch": 0.53, "grad_norm": 3.0968581638137427, "learning_rate": 8.601117208310351e-06, "loss": 0.2575, "step": 3922 }, { "epoch": 0.53, "grad_norm": 3.7567267329270133, "learning_rate": 8.600352936039276e-06, "loss": 0.3227, "step": 3923 }, { "epoch": 0.53, "grad_norm": 3.506406665315962, "learning_rate": 8.599588489022792e-06, "loss": 0.3181, "step": 3924 }, { "epoch": 0.53, "grad_norm": 3.3794581945024036, "learning_rate": 8.598823867298005e-06, "loss": 0.3012, "step": 3925 }, { "epoch": 0.53, "grad_norm": 3.3066136662377654, "learning_rate": 8.598059070902026e-06, "loss": 0.2459, "step": 3926 }, { "epoch": 0.53, "grad_norm": 3.418445172810002, "learning_rate": 8.597294099871974e-06, "loss": 0.3101, "step": 3927 }, { "epoch": 0.53, "grad_norm": 3.6624825720582885, "learning_rate": 8.596528954244978e-06, "loss": 0.2417, "step": 3928 }, { "epoch": 0.53, "grad_norm": 3.1602160335149225, "learning_rate": 8.595763634058177e-06, "loss": 0.2674, "step": 3929 }, { "epoch": 0.53, "grad_norm": 3.4198504598891857, "learning_rate": 8.594998139348712e-06, "loss": 0.2746, "step": 3930 }, { "epoch": 0.53, "grad_norm": 3.6378935382198145, "learning_rate": 8.594232470153739e-06, "loss": 0.2705, "step": 3931 }, { "epoch": 0.53, "grad_norm": 3.1834734741064654, "learning_rate": 8.59346662651042e-06, "loss": 0.2806, "step": 3932 }, { "epoch": 0.53, "grad_norm": 4.37161603401894, "learning_rate": 8.592700608455924e-06, "loss": 0.3375, "step": 3933 }, { "epoch": 0.54, "grad_norm": 3.6981140972262687, "learning_rate": 8.591934416027435e-06, "loss": 0.3199, "step": 3934 }, { "epoch": 0.54, "grad_norm": 3.3517181232373043, "learning_rate": 8.591168049262138e-06, "loss": 0.2289, "step": 3935 }, { "epoch": 0.54, "grad_norm": 3.8308772729315503, "learning_rate": 8.590401508197226e-06, "loss": 0.2615, "step": 3936 }, { "epoch": 0.54, "grad_norm": 3.558899578486801, "learning_rate": 8.589634792869908e-06, "loss": 0.2463, "step": 3937 }, { "epoch": 0.54, "grad_norm": 3.894560507693618, "learning_rate": 8.588867903317395e-06, "loss": 0.2627, "step": 3938 }, { "epoch": 0.54, "grad_norm": 4.321888274460711, "learning_rate": 8.588100839576908e-06, "loss": 0.2779, "step": 3939 }, { "epoch": 0.54, "grad_norm": 2.962138998815552, "learning_rate": 8.587333601685678e-06, "loss": 0.2132, "step": 3940 }, { "epoch": 0.54, "grad_norm": 3.25816197996941, "learning_rate": 8.586566189680943e-06, "loss": 0.3233, "step": 3941 }, { "epoch": 0.54, "grad_norm": 4.425045999763552, "learning_rate": 8.58579860359995e-06, "loss": 0.3669, "step": 3942 }, { "epoch": 0.54, "grad_norm": 3.8527844611068143, "learning_rate": 8.585030843479954e-06, "loss": 0.3339, "step": 3943 }, { "epoch": 0.54, "grad_norm": 3.412934739352286, "learning_rate": 8.58426290935822e-06, "loss": 0.2614, "step": 3944 }, { "epoch": 0.54, "grad_norm": 3.8229020328109513, "learning_rate": 8.583494801272018e-06, "loss": 0.3226, "step": 3945 }, { "epoch": 0.54, "grad_norm": 3.3626096778820798, "learning_rate": 8.58272651925863e-06, "loss": 0.2495, "step": 3946 }, { "epoch": 0.54, "grad_norm": 3.486538143214949, "learning_rate": 8.581958063355344e-06, "loss": 0.2972, "step": 3947 }, { "epoch": 0.54, "grad_norm": 3.450998732724129, "learning_rate": 8.581189433599459e-06, "loss": 0.2856, "step": 3948 }, { "epoch": 0.54, "grad_norm": 3.1981914087939995, "learning_rate": 8.58042063002828e-06, "loss": 0.3105, "step": 3949 }, { "epoch": 0.54, "grad_norm": 3.5662423763740834, "learning_rate": 8.579651652679121e-06, "loss": 0.3009, "step": 3950 }, { "epoch": 0.54, "grad_norm": 3.502066573559423, "learning_rate": 8.578882501589305e-06, "loss": 0.2825, "step": 3951 }, { "epoch": 0.54, "grad_norm": 2.8842403076816234, "learning_rate": 8.578113176796165e-06, "loss": 0.2674, "step": 3952 }, { "epoch": 0.54, "grad_norm": 3.8571257964524, "learning_rate": 8.577343678337038e-06, "loss": 0.3395, "step": 3953 }, { "epoch": 0.54, "grad_norm": 3.554757343075477, "learning_rate": 8.576574006249274e-06, "loss": 0.2893, "step": 3954 }, { "epoch": 0.54, "grad_norm": 3.5481146743660523, "learning_rate": 8.57580416057023e-06, "loss": 0.2779, "step": 3955 }, { "epoch": 0.54, "grad_norm": 4.46063436561201, "learning_rate": 8.575034141337266e-06, "loss": 0.3554, "step": 3956 }, { "epoch": 0.54, "grad_norm": 3.058038455785746, "learning_rate": 8.574263948587762e-06, "loss": 0.284, "step": 3957 }, { "epoch": 0.54, "grad_norm": 3.8890092601730757, "learning_rate": 8.573493582359096e-06, "loss": 0.2766, "step": 3958 }, { "epoch": 0.54, "grad_norm": 3.189737600352257, "learning_rate": 8.57272304268866e-06, "loss": 0.2085, "step": 3959 }, { "epoch": 0.54, "grad_norm": 3.9475868945603576, "learning_rate": 8.571952329613851e-06, "loss": 0.2994, "step": 3960 }, { "epoch": 0.54, "grad_norm": 3.4050903287008167, "learning_rate": 8.571181443172078e-06, "loss": 0.2985, "step": 3961 }, { "epoch": 0.54, "grad_norm": 3.4006954293822207, "learning_rate": 8.570410383400754e-06, "loss": 0.2587, "step": 3962 }, { "epoch": 0.54, "grad_norm": 3.0680635690283133, "learning_rate": 8.569639150337303e-06, "loss": 0.3022, "step": 3963 }, { "epoch": 0.54, "grad_norm": 4.525005981827703, "learning_rate": 8.56886774401916e-06, "loss": 0.2825, "step": 3964 }, { "epoch": 0.54, "grad_norm": 4.18778967545735, "learning_rate": 8.568096164483765e-06, "loss": 0.2946, "step": 3965 }, { "epoch": 0.54, "grad_norm": 3.0372598591740085, "learning_rate": 8.567324411768564e-06, "loss": 0.2251, "step": 3966 }, { "epoch": 0.54, "grad_norm": 3.735343061179575, "learning_rate": 8.56655248591102e-06, "loss": 0.2855, "step": 3967 }, { "epoch": 0.54, "grad_norm": 3.5970696416876744, "learning_rate": 8.565780386948593e-06, "loss": 0.262, "step": 3968 }, { "epoch": 0.54, "grad_norm": 2.924848011991321, "learning_rate": 8.56500811491876e-06, "loss": 0.2141, "step": 3969 }, { "epoch": 0.54, "grad_norm": 3.4119640142805348, "learning_rate": 8.564235669859004e-06, "loss": 0.2928, "step": 3970 }, { "epoch": 0.54, "grad_norm": 4.019417361259612, "learning_rate": 8.563463051806816e-06, "loss": 0.2586, "step": 3971 }, { "epoch": 0.54, "grad_norm": 3.876324460978952, "learning_rate": 8.562690260799696e-06, "loss": 0.2996, "step": 3972 }, { "epoch": 0.54, "grad_norm": 4.029105598804776, "learning_rate": 8.56191729687515e-06, "loss": 0.2458, "step": 3973 }, { "epoch": 0.54, "grad_norm": 4.014342256831054, "learning_rate": 8.561144160070696e-06, "loss": 0.3075, "step": 3974 }, { "epoch": 0.54, "grad_norm": 3.5111917191641915, "learning_rate": 8.560370850423859e-06, "loss": 0.3042, "step": 3975 }, { "epoch": 0.54, "grad_norm": 4.235222279361554, "learning_rate": 8.559597367972168e-06, "loss": 0.2547, "step": 3976 }, { "epoch": 0.54, "grad_norm": 3.7919021006077975, "learning_rate": 8.558823712753171e-06, "loss": 0.2995, "step": 3977 }, { "epoch": 0.54, "grad_norm": 3.330329244284114, "learning_rate": 8.558049884804414e-06, "loss": 0.2319, "step": 3978 }, { "epoch": 0.54, "grad_norm": 3.816643340184783, "learning_rate": 8.557275884163455e-06, "loss": 0.3057, "step": 3979 }, { "epoch": 0.54, "grad_norm": 4.15697691369403, "learning_rate": 8.556501710867862e-06, "loss": 0.292, "step": 3980 }, { "epoch": 0.54, "grad_norm": 4.040945994079908, "learning_rate": 8.555727364955208e-06, "loss": 0.3186, "step": 3981 }, { "epoch": 0.54, "grad_norm": 3.548423074453824, "learning_rate": 8.554952846463081e-06, "loss": 0.28, "step": 3982 }, { "epoch": 0.54, "grad_norm": 3.1406539872616968, "learning_rate": 8.554178155429065e-06, "loss": 0.2848, "step": 3983 }, { "epoch": 0.54, "grad_norm": 2.7857684807515626, "learning_rate": 8.553403291890767e-06, "loss": 0.2401, "step": 3984 }, { "epoch": 0.54, "grad_norm": 3.87380026682458, "learning_rate": 8.552628255885794e-06, "loss": 0.3039, "step": 3985 }, { "epoch": 0.54, "grad_norm": 3.479523957877495, "learning_rate": 8.55185304745176e-06, "loss": 0.3042, "step": 3986 }, { "epoch": 0.54, "grad_norm": 3.361678990563115, "learning_rate": 8.551077666626292e-06, "loss": 0.2615, "step": 3987 }, { "epoch": 0.54, "grad_norm": 3.5420056520184398, "learning_rate": 8.550302113447026e-06, "loss": 0.2724, "step": 3988 }, { "epoch": 0.54, "grad_norm": 3.2220747164316568, "learning_rate": 8.549526387951601e-06, "loss": 0.2512, "step": 3989 }, { "epoch": 0.54, "grad_norm": 4.111846901262551, "learning_rate": 8.548750490177667e-06, "loss": 0.3106, "step": 3990 }, { "epoch": 0.54, "grad_norm": 3.461438996896162, "learning_rate": 8.547974420162885e-06, "loss": 0.2706, "step": 3991 }, { "epoch": 0.54, "grad_norm": 3.199916070358576, "learning_rate": 8.54719817794492e-06, "loss": 0.2277, "step": 3992 }, { "epoch": 0.54, "grad_norm": 3.4495977849729753, "learning_rate": 8.546421763561448e-06, "loss": 0.2975, "step": 3993 }, { "epoch": 0.54, "grad_norm": 3.2524441711436656, "learning_rate": 8.545645177050154e-06, "loss": 0.2634, "step": 3994 }, { "epoch": 0.54, "grad_norm": 4.760930849551091, "learning_rate": 8.544868418448729e-06, "loss": 0.3389, "step": 3995 }, { "epoch": 0.54, "grad_norm": 4.175718078065642, "learning_rate": 8.544091487794872e-06, "loss": 0.2373, "step": 3996 }, { "epoch": 0.54, "grad_norm": 3.631139459763576, "learning_rate": 8.543314385126296e-06, "loss": 0.256, "step": 3997 }, { "epoch": 0.54, "grad_norm": 3.0909702853057666, "learning_rate": 8.542537110480712e-06, "loss": 0.2325, "step": 3998 }, { "epoch": 0.54, "grad_norm": 3.8127172717507944, "learning_rate": 8.541759663895852e-06, "loss": 0.2855, "step": 3999 }, { "epoch": 0.54, "grad_norm": 4.431075331154463, "learning_rate": 8.540982045409446e-06, "loss": 0.3647, "step": 4000 }, { "epoch": 0.54, "grad_norm": 3.3906857394525103, "learning_rate": 8.540204255059236e-06, "loss": 0.2769, "step": 4001 }, { "epoch": 0.54, "grad_norm": 3.2618985938130827, "learning_rate": 8.539426292882976e-06, "loss": 0.2385, "step": 4002 }, { "epoch": 0.54, "grad_norm": 3.477811843010108, "learning_rate": 8.53864815891842e-06, "loss": 0.2642, "step": 4003 }, { "epoch": 0.54, "grad_norm": 2.662270870552394, "learning_rate": 8.53786985320334e-06, "loss": 0.1963, "step": 4004 }, { "epoch": 0.54, "grad_norm": 3.6870677106025, "learning_rate": 8.537091375775508e-06, "loss": 0.3028, "step": 4005 }, { "epoch": 0.54, "grad_norm": 3.2594099272442567, "learning_rate": 8.53631272667271e-06, "loss": 0.2384, "step": 4006 }, { "epoch": 0.55, "grad_norm": 3.792059166254736, "learning_rate": 8.535533905932739e-06, "loss": 0.3012, "step": 4007 }, { "epoch": 0.55, "grad_norm": 3.235305082236057, "learning_rate": 8.534754913593393e-06, "loss": 0.3013, "step": 4008 }, { "epoch": 0.55, "grad_norm": 3.629640592230768, "learning_rate": 8.53397574969248e-06, "loss": 0.2674, "step": 4009 }, { "epoch": 0.55, "grad_norm": 3.9865245664433093, "learning_rate": 8.53319641426782e-06, "loss": 0.2392, "step": 4010 }, { "epoch": 0.55, "grad_norm": 3.872054230162423, "learning_rate": 8.53241690735724e-06, "loss": 0.2735, "step": 4011 }, { "epoch": 0.55, "grad_norm": 2.6821207646117298, "learning_rate": 8.531637228998569e-06, "loss": 0.2152, "step": 4012 }, { "epoch": 0.55, "grad_norm": 3.0281670787800676, "learning_rate": 8.530857379229651e-06, "loss": 0.2663, "step": 4013 }, { "epoch": 0.55, "grad_norm": 4.083642558302089, "learning_rate": 8.530077358088339e-06, "loss": 0.2864, "step": 4014 }, { "epoch": 0.55, "grad_norm": 3.806588547162481, "learning_rate": 8.529297165612488e-06, "loss": 0.2627, "step": 4015 }, { "epoch": 0.55, "grad_norm": 3.5293448787089896, "learning_rate": 8.52851680183997e-06, "loss": 0.2735, "step": 4016 }, { "epoch": 0.55, "grad_norm": 4.114349105545965, "learning_rate": 8.527736266808658e-06, "loss": 0.3012, "step": 4017 }, { "epoch": 0.55, "grad_norm": 3.299982739870908, "learning_rate": 8.526955560556433e-06, "loss": 0.2269, "step": 4018 }, { "epoch": 0.55, "grad_norm": 4.073279425819165, "learning_rate": 8.52617468312119e-06, "loss": 0.344, "step": 4019 }, { "epoch": 0.55, "grad_norm": 3.466587363665344, "learning_rate": 8.525393634540828e-06, "loss": 0.27, "step": 4020 }, { "epoch": 0.55, "grad_norm": 4.165187389691914, "learning_rate": 8.524612414853259e-06, "loss": 0.2781, "step": 4021 }, { "epoch": 0.55, "grad_norm": 3.9980478005584787, "learning_rate": 8.523831024096396e-06, "loss": 0.2425, "step": 4022 }, { "epoch": 0.55, "grad_norm": 3.7824053585806285, "learning_rate": 8.523049462308165e-06, "loss": 0.3063, "step": 4023 }, { "epoch": 0.55, "grad_norm": 3.719877690161142, "learning_rate": 8.5222677295265e-06, "loss": 0.2798, "step": 4024 }, { "epoch": 0.55, "grad_norm": 3.70232713189555, "learning_rate": 8.521485825789346e-06, "loss": 0.2772, "step": 4025 }, { "epoch": 0.55, "grad_norm": 3.076812906904743, "learning_rate": 8.52070375113465e-06, "loss": 0.2407, "step": 4026 }, { "epoch": 0.55, "grad_norm": 3.547281619874765, "learning_rate": 8.519921505600368e-06, "loss": 0.303, "step": 4027 }, { "epoch": 0.55, "grad_norm": 3.584075663145823, "learning_rate": 8.519139089224473e-06, "loss": 0.2868, "step": 4028 }, { "epoch": 0.55, "grad_norm": 3.948394751821968, "learning_rate": 8.518356502044935e-06, "loss": 0.3342, "step": 4029 }, { "epoch": 0.55, "grad_norm": 3.383246259353292, "learning_rate": 8.51757374409974e-06, "loss": 0.2252, "step": 4030 }, { "epoch": 0.55, "grad_norm": 3.028711186892707, "learning_rate": 8.516790815426879e-06, "loss": 0.2515, "step": 4031 }, { "epoch": 0.55, "grad_norm": 3.420344722125317, "learning_rate": 8.516007716064352e-06, "loss": 0.2472, "step": 4032 }, { "epoch": 0.55, "grad_norm": 3.397310458190045, "learning_rate": 8.515224446050166e-06, "loss": 0.3137, "step": 4033 }, { "epoch": 0.55, "grad_norm": 3.9711146943353226, "learning_rate": 8.51444100542234e-06, "loss": 0.2943, "step": 4034 }, { "epoch": 0.55, "grad_norm": 3.872476549165155, "learning_rate": 8.513657394218897e-06, "loss": 0.3078, "step": 4035 }, { "epoch": 0.55, "grad_norm": 3.544140630033937, "learning_rate": 8.51287361247787e-06, "loss": 0.3113, "step": 4036 }, { "epoch": 0.55, "grad_norm": 3.3124689635343554, "learning_rate": 8.5120896602373e-06, "loss": 0.2437, "step": 4037 }, { "epoch": 0.55, "grad_norm": 3.945549943644282, "learning_rate": 8.511305537535238e-06, "loss": 0.3209, "step": 4038 }, { "epoch": 0.55, "grad_norm": 3.5661396790806954, "learning_rate": 8.510521244409739e-06, "loss": 0.2757, "step": 4039 }, { "epoch": 0.55, "grad_norm": 3.0829083530166073, "learning_rate": 8.509736780898874e-06, "loss": 0.2807, "step": 4040 }, { "epoch": 0.55, "grad_norm": 3.420486133387409, "learning_rate": 8.508952147040715e-06, "loss": 0.2744, "step": 4041 }, { "epoch": 0.55, "grad_norm": 3.6437603240412075, "learning_rate": 8.508167342873342e-06, "loss": 0.2973, "step": 4042 }, { "epoch": 0.55, "grad_norm": 3.8095880116494008, "learning_rate": 8.507382368434851e-06, "loss": 0.3255, "step": 4043 }, { "epoch": 0.55, "grad_norm": 4.019484203004427, "learning_rate": 8.506597223763337e-06, "loss": 0.2431, "step": 4044 }, { "epoch": 0.55, "grad_norm": 3.440019554240605, "learning_rate": 8.505811908896911e-06, "loss": 0.3346, "step": 4045 }, { "epoch": 0.55, "grad_norm": 3.321484692956812, "learning_rate": 8.505026423873686e-06, "loss": 0.242, "step": 4046 }, { "epoch": 0.55, "grad_norm": 3.487042293281889, "learning_rate": 8.504240768731787e-06, "loss": 0.2837, "step": 4047 }, { "epoch": 0.55, "grad_norm": 3.3615744734999, "learning_rate": 8.503454943509346e-06, "loss": 0.2939, "step": 4048 }, { "epoch": 0.55, "grad_norm": 3.6116376705501687, "learning_rate": 8.502668948244504e-06, "loss": 0.2449, "step": 4049 }, { "epoch": 0.55, "grad_norm": 3.7058198402670235, "learning_rate": 8.50188278297541e-06, "loss": 0.2195, "step": 4050 }, { "epoch": 0.55, "grad_norm": 3.104298976268481, "learning_rate": 8.50109644774022e-06, "loss": 0.219, "step": 4051 }, { "epoch": 0.55, "grad_norm": 5.31881637233317, "learning_rate": 8.500309942577098e-06, "loss": 0.3201, "step": 4052 }, { "epoch": 0.55, "grad_norm": 4.007666691586651, "learning_rate": 8.499523267524223e-06, "loss": 0.2961, "step": 4053 }, { "epoch": 0.55, "grad_norm": 3.648625599180769, "learning_rate": 8.498736422619771e-06, "loss": 0.2995, "step": 4054 }, { "epoch": 0.55, "grad_norm": 3.815962296121392, "learning_rate": 8.497949407901936e-06, "loss": 0.3727, "step": 4055 }, { "epoch": 0.55, "grad_norm": 3.379855291647901, "learning_rate": 8.49716222340891e-06, "loss": 0.2602, "step": 4056 }, { "epoch": 0.55, "grad_norm": 3.6514583866330015, "learning_rate": 8.496374869178908e-06, "loss": 0.2702, "step": 4057 }, { "epoch": 0.55, "grad_norm": 2.947252800025874, "learning_rate": 8.49558734525014e-06, "loss": 0.2615, "step": 4058 }, { "epoch": 0.55, "grad_norm": 3.524392065550873, "learning_rate": 8.494799651660828e-06, "loss": 0.2999, "step": 4059 }, { "epoch": 0.55, "grad_norm": 4.2741554850061005, "learning_rate": 8.494011788449208e-06, "loss": 0.2473, "step": 4060 }, { "epoch": 0.55, "grad_norm": 3.5132603668931828, "learning_rate": 8.493223755653513e-06, "loss": 0.2958, "step": 4061 }, { "epoch": 0.55, "grad_norm": 3.384294033089384, "learning_rate": 8.492435553311995e-06, "loss": 0.2705, "step": 4062 }, { "epoch": 0.55, "grad_norm": 3.189414123666831, "learning_rate": 8.49164718146291e-06, "loss": 0.2547, "step": 4063 }, { "epoch": 0.55, "grad_norm": 3.1228488820760005, "learning_rate": 8.490858640144518e-06, "loss": 0.2786, "step": 4064 }, { "epoch": 0.55, "grad_norm": 5.069683458412716, "learning_rate": 8.490069929395098e-06, "loss": 0.3054, "step": 4065 }, { "epoch": 0.55, "grad_norm": 3.0320828521165133, "learning_rate": 8.489281049252926e-06, "loss": 0.2629, "step": 4066 }, { "epoch": 0.55, "grad_norm": 3.1164813980212975, "learning_rate": 8.48849199975629e-06, "loss": 0.28, "step": 4067 }, { "epoch": 0.55, "grad_norm": 3.632756188003151, "learning_rate": 8.48770278094349e-06, "loss": 0.2181, "step": 4068 }, { "epoch": 0.55, "grad_norm": 3.4554547501370636, "learning_rate": 8.48691339285283e-06, "loss": 0.2884, "step": 4069 }, { "epoch": 0.55, "grad_norm": 3.5705928069002377, "learning_rate": 8.486123835522623e-06, "loss": 0.3012, "step": 4070 }, { "epoch": 0.55, "grad_norm": 3.7239219544667934, "learning_rate": 8.48533410899119e-06, "loss": 0.2921, "step": 4071 }, { "epoch": 0.55, "grad_norm": 3.570585392595172, "learning_rate": 8.484544213296864e-06, "loss": 0.3418, "step": 4072 }, { "epoch": 0.55, "grad_norm": 5.922733668529267, "learning_rate": 8.48375414847798e-06, "loss": 0.3702, "step": 4073 }, { "epoch": 0.55, "grad_norm": 3.293234907150142, "learning_rate": 8.482963914572885e-06, "loss": 0.2574, "step": 4074 }, { "epoch": 0.55, "grad_norm": 3.812819543716674, "learning_rate": 8.482173511619934e-06, "loss": 0.2094, "step": 4075 }, { "epoch": 0.55, "grad_norm": 3.207475119925824, "learning_rate": 8.48138293965749e-06, "loss": 0.2296, "step": 4076 }, { "epoch": 0.55, "grad_norm": 4.020652584196159, "learning_rate": 8.480592198723922e-06, "loss": 0.2931, "step": 4077 }, { "epoch": 0.55, "grad_norm": 4.081560165297491, "learning_rate": 8.479801288857613e-06, "loss": 0.2923, "step": 4078 }, { "epoch": 0.55, "grad_norm": 2.9528547299785846, "learning_rate": 8.479010210096945e-06, "loss": 0.2242, "step": 4079 }, { "epoch": 0.55, "grad_norm": 3.1319706601036037, "learning_rate": 8.478218962480318e-06, "loss": 0.2879, "step": 4080 }, { "epoch": 0.56, "grad_norm": 3.7057562248032037, "learning_rate": 8.477427546046134e-06, "loss": 0.2878, "step": 4081 }, { "epoch": 0.56, "grad_norm": 2.9888453063004743, "learning_rate": 8.476635960832805e-06, "loss": 0.2862, "step": 4082 }, { "epoch": 0.56, "grad_norm": 4.051822893828658, "learning_rate": 8.47584420687875e-06, "loss": 0.3132, "step": 4083 }, { "epoch": 0.56, "grad_norm": 3.695189644963137, "learning_rate": 8.4750522842224e-06, "loss": 0.2666, "step": 4084 }, { "epoch": 0.56, "grad_norm": 3.404856922483896, "learning_rate": 8.474260192902189e-06, "loss": 0.2616, "step": 4085 }, { "epoch": 0.56, "grad_norm": 3.679360240313485, "learning_rate": 8.473467932956561e-06, "loss": 0.2655, "step": 4086 }, { "epoch": 0.56, "grad_norm": 3.2553330314359443, "learning_rate": 8.472675504423972e-06, "loss": 0.2518, "step": 4087 }, { "epoch": 0.56, "grad_norm": 3.2085940416418137, "learning_rate": 8.47188290734288e-06, "loss": 0.2529, "step": 4088 }, { "epoch": 0.56, "grad_norm": 3.5276084126636795, "learning_rate": 8.471090141751758e-06, "loss": 0.2836, "step": 4089 }, { "epoch": 0.56, "grad_norm": 3.4369746093110423, "learning_rate": 8.47029720768908e-06, "loss": 0.2565, "step": 4090 }, { "epoch": 0.56, "grad_norm": 3.711632426562642, "learning_rate": 8.469504105193332e-06, "loss": 0.2662, "step": 4091 }, { "epoch": 0.56, "grad_norm": 3.3787143289033605, "learning_rate": 8.468710834303007e-06, "loss": 0.2676, "step": 4092 }, { "epoch": 0.56, "grad_norm": 3.523150092218902, "learning_rate": 8.46791739505661e-06, "loss": 0.2854, "step": 4093 }, { "epoch": 0.56, "grad_norm": 4.467627124551692, "learning_rate": 8.467123787492647e-06, "loss": 0.3638, "step": 4094 }, { "epoch": 0.56, "grad_norm": 3.63328590841133, "learning_rate": 8.46633001164964e-06, "loss": 0.2438, "step": 4095 }, { "epoch": 0.56, "grad_norm": 3.1473963124855473, "learning_rate": 8.465536067566113e-06, "loss": 0.2341, "step": 4096 }, { "epoch": 0.56, "grad_norm": 4.044664292779911, "learning_rate": 8.464741955280603e-06, "loss": 0.3115, "step": 4097 }, { "epoch": 0.56, "grad_norm": 3.6742074675384657, "learning_rate": 8.46394767483165e-06, "loss": 0.277, "step": 4098 }, { "epoch": 0.56, "grad_norm": 3.8178190977520226, "learning_rate": 8.463153226257805e-06, "loss": 0.3121, "step": 4099 }, { "epoch": 0.56, "grad_norm": 3.2576894349000596, "learning_rate": 8.462358609597629e-06, "loss": 0.2244, "step": 4100 }, { "epoch": 0.56, "grad_norm": 3.6202110895043713, "learning_rate": 8.46156382488969e-06, "loss": 0.2715, "step": 4101 }, { "epoch": 0.56, "grad_norm": 3.341090735584136, "learning_rate": 8.460768872172558e-06, "loss": 0.3259, "step": 4102 }, { "epoch": 0.56, "grad_norm": 3.6309172458519927, "learning_rate": 8.459973751484822e-06, "loss": 0.1967, "step": 4103 }, { "epoch": 0.56, "grad_norm": 3.8331575279465713, "learning_rate": 8.459178462865072e-06, "loss": 0.324, "step": 4104 }, { "epoch": 0.56, "grad_norm": 3.6017568799354143, "learning_rate": 8.458383006351909e-06, "loss": 0.2588, "step": 4105 }, { "epoch": 0.56, "grad_norm": 3.251704267127131, "learning_rate": 8.457587381983939e-06, "loss": 0.3427, "step": 4106 }, { "epoch": 0.56, "grad_norm": 3.5759745241091103, "learning_rate": 8.456791589799777e-06, "loss": 0.3016, "step": 4107 }, { "epoch": 0.56, "grad_norm": 3.119665431395647, "learning_rate": 8.455995629838051e-06, "loss": 0.2017, "step": 4108 }, { "epoch": 0.56, "grad_norm": 3.6369714797159753, "learning_rate": 8.455199502137393e-06, "loss": 0.2493, "step": 4109 }, { "epoch": 0.56, "grad_norm": 3.374608695488695, "learning_rate": 8.45440320673644e-06, "loss": 0.2236, "step": 4110 }, { "epoch": 0.56, "grad_norm": 3.5244904312112775, "learning_rate": 8.453606743673843e-06, "loss": 0.2404, "step": 4111 }, { "epoch": 0.56, "grad_norm": 2.892942034587793, "learning_rate": 8.45281011298826e-06, "loss": 0.2403, "step": 4112 }, { "epoch": 0.56, "grad_norm": 3.6256123576119084, "learning_rate": 8.452013314718353e-06, "loss": 0.2907, "step": 4113 }, { "epoch": 0.56, "grad_norm": 3.8052527947077555, "learning_rate": 8.451216348902798e-06, "loss": 0.2802, "step": 4114 }, { "epoch": 0.56, "grad_norm": 3.3456120191869156, "learning_rate": 8.450419215580273e-06, "loss": 0.2617, "step": 4115 }, { "epoch": 0.56, "grad_norm": 4.351420924137729, "learning_rate": 8.449621914789471e-06, "loss": 0.3002, "step": 4116 }, { "epoch": 0.56, "grad_norm": 3.6983201840209303, "learning_rate": 8.448824446569087e-06, "loss": 0.3295, "step": 4117 }, { "epoch": 0.56, "grad_norm": 3.6565140645743877, "learning_rate": 8.448026810957828e-06, "loss": 0.285, "step": 4118 }, { "epoch": 0.56, "grad_norm": 3.482410152411, "learning_rate": 8.447229007994407e-06, "loss": 0.2639, "step": 4119 }, { "epoch": 0.56, "grad_norm": 3.389190748105736, "learning_rate": 8.446431037717546e-06, "loss": 0.2082, "step": 4120 }, { "epoch": 0.56, "grad_norm": 3.099187706871054, "learning_rate": 8.445632900165976e-06, "loss": 0.217, "step": 4121 }, { "epoch": 0.56, "grad_norm": 3.558674057289893, "learning_rate": 8.444834595378434e-06, "loss": 0.2442, "step": 4122 }, { "epoch": 0.56, "grad_norm": 2.953274579205508, "learning_rate": 8.444036123393667e-06, "loss": 0.2172, "step": 4123 }, { "epoch": 0.56, "grad_norm": 3.091663421270548, "learning_rate": 8.443237484250427e-06, "loss": 0.2712, "step": 4124 }, { "epoch": 0.56, "grad_norm": 3.2008924123519127, "learning_rate": 8.44243867798748e-06, "loss": 0.2562, "step": 4125 }, { "epoch": 0.56, "grad_norm": 3.8465277586847835, "learning_rate": 8.441639704643592e-06, "loss": 0.261, "step": 4126 }, { "epoch": 0.56, "grad_norm": 3.7472266838665034, "learning_rate": 8.440840564257547e-06, "loss": 0.2632, "step": 4127 }, { "epoch": 0.56, "grad_norm": 3.107980303005477, "learning_rate": 8.440041256868128e-06, "loss": 0.1973, "step": 4128 }, { "epoch": 0.56, "grad_norm": 3.3912429292292816, "learning_rate": 8.439241782514132e-06, "loss": 0.2326, "step": 4129 }, { "epoch": 0.56, "grad_norm": 4.0223818441413455, "learning_rate": 8.43844214123436e-06, "loss": 0.3699, "step": 4130 }, { "epoch": 0.56, "grad_norm": 3.7691213998945536, "learning_rate": 8.437642333067626e-06, "loss": 0.321, "step": 4131 }, { "epoch": 0.56, "grad_norm": 4.115867282421882, "learning_rate": 8.436842358052746e-06, "loss": 0.3103, "step": 4132 }, { "epoch": 0.56, "grad_norm": 4.4126127056342686, "learning_rate": 8.43604221622855e-06, "loss": 0.3133, "step": 4133 }, { "epoch": 0.56, "grad_norm": 3.4081068320304047, "learning_rate": 8.435241907633871e-06, "loss": 0.3279, "step": 4134 }, { "epoch": 0.56, "grad_norm": 3.7340007768600385, "learning_rate": 8.434441432307555e-06, "loss": 0.3045, "step": 4135 }, { "epoch": 0.56, "grad_norm": 4.161164226500416, "learning_rate": 8.43364079028845e-06, "loss": 0.3091, "step": 4136 }, { "epoch": 0.56, "grad_norm": 3.3581381063250437, "learning_rate": 8.432839981615419e-06, "loss": 0.2614, "step": 4137 }, { "epoch": 0.56, "grad_norm": 4.118569161470857, "learning_rate": 8.432039006327328e-06, "loss": 0.2668, "step": 4138 }, { "epoch": 0.56, "grad_norm": 3.560087943378511, "learning_rate": 8.431237864463055e-06, "loss": 0.2534, "step": 4139 }, { "epoch": 0.56, "grad_norm": 3.772652341940007, "learning_rate": 8.430436556061482e-06, "loss": 0.2672, "step": 4140 }, { "epoch": 0.56, "grad_norm": 3.577260431585988, "learning_rate": 8.4296350811615e-06, "loss": 0.2145, "step": 4141 }, { "epoch": 0.56, "grad_norm": 4.407123742374353, "learning_rate": 8.428833439802012e-06, "loss": 0.2799, "step": 4142 }, { "epoch": 0.56, "grad_norm": 3.2809390073382816, "learning_rate": 8.428031632021922e-06, "loss": 0.2559, "step": 4143 }, { "epoch": 0.56, "grad_norm": 3.287756933109961, "learning_rate": 8.427229657860152e-06, "loss": 0.3078, "step": 4144 }, { "epoch": 0.56, "grad_norm": 3.0527021475585636, "learning_rate": 8.42642751735562e-06, "loss": 0.2291, "step": 4145 }, { "epoch": 0.56, "grad_norm": 3.7642237630617448, "learning_rate": 8.425625210547265e-06, "loss": 0.2855, "step": 4146 }, { "epoch": 0.56, "grad_norm": 3.564580098684969, "learning_rate": 8.424822737474023e-06, "loss": 0.2755, "step": 4147 }, { "epoch": 0.56, "grad_norm": 5.635108232374167, "learning_rate": 8.424020098174843e-06, "loss": 0.2257, "step": 4148 }, { "epoch": 0.56, "grad_norm": 2.884786942427414, "learning_rate": 8.423217292688683e-06, "loss": 0.22, "step": 4149 }, { "epoch": 0.56, "grad_norm": 3.333991020672686, "learning_rate": 8.422414321054506e-06, "loss": 0.2533, "step": 4150 }, { "epoch": 0.56, "grad_norm": 3.0186715776953847, "learning_rate": 8.421611183311285e-06, "loss": 0.2544, "step": 4151 }, { "epoch": 0.56, "grad_norm": 3.9128108514979183, "learning_rate": 8.420807879498002e-06, "loss": 0.2606, "step": 4152 }, { "epoch": 0.56, "grad_norm": 4.22627464029893, "learning_rate": 8.420004409653648e-06, "loss": 0.276, "step": 4153 }, { "epoch": 0.57, "grad_norm": 3.3036480051074424, "learning_rate": 8.419200773817214e-06, "loss": 0.2305, "step": 4154 }, { "epoch": 0.57, "grad_norm": 3.600163374258635, "learning_rate": 8.41839697202771e-06, "loss": 0.318, "step": 4155 }, { "epoch": 0.57, "grad_norm": 3.527231557753674, "learning_rate": 8.417593004324145e-06, "loss": 0.2976, "step": 4156 }, { "epoch": 0.57, "grad_norm": 2.820635512198256, "learning_rate": 8.416788870745544e-06, "loss": 0.2587, "step": 4157 }, { "epoch": 0.57, "grad_norm": 4.49371735617251, "learning_rate": 8.415984571330933e-06, "loss": 0.3207, "step": 4158 }, { "epoch": 0.57, "grad_norm": 3.3082296692840703, "learning_rate": 8.415180106119351e-06, "loss": 0.2221, "step": 4159 }, { "epoch": 0.57, "grad_norm": 3.5869029476293814, "learning_rate": 8.414375475149844e-06, "loss": 0.282, "step": 4160 }, { "epoch": 0.57, "grad_norm": 4.0668622279735125, "learning_rate": 8.413570678461463e-06, "loss": 0.2743, "step": 4161 }, { "epoch": 0.57, "grad_norm": 2.8691815598761656, "learning_rate": 8.412765716093273e-06, "loss": 0.1806, "step": 4162 }, { "epoch": 0.57, "grad_norm": 3.0124770675927692, "learning_rate": 8.411960588084337e-06, "loss": 0.2979, "step": 4163 }, { "epoch": 0.57, "grad_norm": 3.18676406291354, "learning_rate": 8.411155294473737e-06, "loss": 0.2513, "step": 4164 }, { "epoch": 0.57, "grad_norm": 3.421226143705718, "learning_rate": 8.41034983530056e-06, "loss": 0.2606, "step": 4165 }, { "epoch": 0.57, "grad_norm": 3.817565874997899, "learning_rate": 8.409544210603895e-06, "loss": 0.2831, "step": 4166 }, { "epoch": 0.57, "grad_norm": 4.265508566847657, "learning_rate": 8.408738420422847e-06, "loss": 0.263, "step": 4167 }, { "epoch": 0.57, "grad_norm": 3.559778236760247, "learning_rate": 8.407932464796521e-06, "loss": 0.3176, "step": 4168 }, { "epoch": 0.57, "grad_norm": 3.8993809042082352, "learning_rate": 8.40712634376404e-06, "loss": 0.254, "step": 4169 }, { "epoch": 0.57, "grad_norm": 2.899747776251811, "learning_rate": 8.406320057364528e-06, "loss": 0.2087, "step": 4170 }, { "epoch": 0.57, "grad_norm": 3.7531562899128295, "learning_rate": 8.405513605637117e-06, "loss": 0.2873, "step": 4171 }, { "epoch": 0.57, "grad_norm": 3.462716356027992, "learning_rate": 8.40470698862095e-06, "loss": 0.277, "step": 4172 }, { "epoch": 0.57, "grad_norm": 3.365620366524209, "learning_rate": 8.403900206355174e-06, "loss": 0.2864, "step": 4173 }, { "epoch": 0.57, "grad_norm": 4.080529318197073, "learning_rate": 8.40309325887895e-06, "loss": 0.3062, "step": 4174 }, { "epoch": 0.57, "grad_norm": 3.2897194865232504, "learning_rate": 8.402286146231444e-06, "loss": 0.2013, "step": 4175 }, { "epoch": 0.57, "grad_norm": 3.8865096186871897, "learning_rate": 8.401478868451828e-06, "loss": 0.3622, "step": 4176 }, { "epoch": 0.57, "grad_norm": 3.146682616770062, "learning_rate": 8.400671425579283e-06, "loss": 0.2712, "step": 4177 }, { "epoch": 0.57, "grad_norm": 3.5922435595686015, "learning_rate": 8.399863817653e-06, "loss": 0.2898, "step": 4178 }, { "epoch": 0.57, "grad_norm": 3.351686870194176, "learning_rate": 8.399056044712178e-06, "loss": 0.3125, "step": 4179 }, { "epoch": 0.57, "grad_norm": 4.020343017669448, "learning_rate": 8.39824810679602e-06, "loss": 0.2988, "step": 4180 }, { "epoch": 0.57, "grad_norm": 3.5715755401551217, "learning_rate": 8.397440003943741e-06, "loss": 0.2624, "step": 4181 }, { "epoch": 0.57, "grad_norm": 3.504336668284907, "learning_rate": 8.396631736194563e-06, "loss": 0.3189, "step": 4182 }, { "epoch": 0.57, "grad_norm": 4.0617112534270605, "learning_rate": 8.395823303587716e-06, "loss": 0.2727, "step": 4183 }, { "epoch": 0.57, "grad_norm": 3.471138753467579, "learning_rate": 8.395014706162437e-06, "loss": 0.2526, "step": 4184 }, { "epoch": 0.57, "grad_norm": 3.835803750463455, "learning_rate": 8.394205943957974e-06, "loss": 0.294, "step": 4185 }, { "epoch": 0.57, "grad_norm": 3.582914518231128, "learning_rate": 8.393397017013577e-06, "loss": 0.2499, "step": 4186 }, { "epoch": 0.57, "grad_norm": 3.9666029353746177, "learning_rate": 8.39258792536851e-06, "loss": 0.2644, "step": 4187 }, { "epoch": 0.57, "grad_norm": 3.134538650300774, "learning_rate": 8.391778669062045e-06, "loss": 0.2871, "step": 4188 }, { "epoch": 0.57, "grad_norm": 3.634602209984865, "learning_rate": 8.390969248133455e-06, "loss": 0.2321, "step": 4189 }, { "epoch": 0.57, "grad_norm": 3.313080861682536, "learning_rate": 8.390159662622029e-06, "loss": 0.2456, "step": 4190 }, { "epoch": 0.57, "grad_norm": 5.046540165530516, "learning_rate": 8.389349912567058e-06, "loss": 0.2301, "step": 4191 }, { "epoch": 0.57, "grad_norm": 4.098454573623363, "learning_rate": 8.388539998007847e-06, "loss": 0.2622, "step": 4192 }, { "epoch": 0.57, "grad_norm": 3.225289057943615, "learning_rate": 8.387729918983706e-06, "loss": 0.2905, "step": 4193 }, { "epoch": 0.57, "grad_norm": 3.4854770629346956, "learning_rate": 8.386919675533949e-06, "loss": 0.2561, "step": 4194 }, { "epoch": 0.57, "grad_norm": 3.3430009035516077, "learning_rate": 8.386109267697904e-06, "loss": 0.3424, "step": 4195 }, { "epoch": 0.57, "grad_norm": 3.099026610282706, "learning_rate": 8.385298695514905e-06, "loss": 0.2642, "step": 4196 }, { "epoch": 0.57, "grad_norm": 3.531019849459976, "learning_rate": 8.384487959024293e-06, "loss": 0.2755, "step": 4197 }, { "epoch": 0.57, "grad_norm": 3.4385992876138736, "learning_rate": 8.383677058265416e-06, "loss": 0.2783, "step": 4198 }, { "epoch": 0.57, "grad_norm": 3.1896093852325174, "learning_rate": 8.382865993277634e-06, "loss": 0.2409, "step": 4199 }, { "epoch": 0.57, "grad_norm": 3.3675239243486437, "learning_rate": 8.382054764100312e-06, "loss": 0.2092, "step": 4200 }, { "epoch": 0.57, "grad_norm": 3.8070253272272443, "learning_rate": 8.381243370772823e-06, "loss": 0.2436, "step": 4201 }, { "epoch": 0.57, "grad_norm": 3.739748596641592, "learning_rate": 8.380431813334548e-06, "loss": 0.2898, "step": 4202 }, { "epoch": 0.57, "grad_norm": 3.4232709299232553, "learning_rate": 8.37962009182488e-06, "loss": 0.2763, "step": 4203 }, { "epoch": 0.57, "grad_norm": 3.3174163564715604, "learning_rate": 8.378808206283211e-06, "loss": 0.2544, "step": 4204 }, { "epoch": 0.57, "grad_norm": 3.0818656087382803, "learning_rate": 8.377996156748948e-06, "loss": 0.233, "step": 4205 }, { "epoch": 0.57, "grad_norm": 3.176905881830272, "learning_rate": 8.37718394326151e-06, "loss": 0.3313, "step": 4206 }, { "epoch": 0.57, "grad_norm": 3.0686118414462182, "learning_rate": 8.37637156586031e-06, "loss": 0.2288, "step": 4207 }, { "epoch": 0.57, "grad_norm": 4.456788247822261, "learning_rate": 8.37555902458478e-06, "loss": 0.3157, "step": 4208 }, { "epoch": 0.57, "grad_norm": 3.398517474545285, "learning_rate": 8.37474631947436e-06, "loss": 0.3197, "step": 4209 }, { "epoch": 0.57, "grad_norm": 3.1521473095707013, "learning_rate": 8.373933450568493e-06, "loss": 0.2968, "step": 4210 }, { "epoch": 0.57, "grad_norm": 3.47501463708798, "learning_rate": 8.373120417906631e-06, "loss": 0.2572, "step": 4211 }, { "epoch": 0.57, "grad_norm": 3.4705767883376955, "learning_rate": 8.372307221528239e-06, "loss": 0.28, "step": 4212 }, { "epoch": 0.57, "grad_norm": 3.366829353470803, "learning_rate": 8.37149386147278e-06, "loss": 0.2954, "step": 4213 }, { "epoch": 0.57, "grad_norm": 4.05695107875136, "learning_rate": 8.370680337779737e-06, "loss": 0.2656, "step": 4214 }, { "epoch": 0.57, "grad_norm": 3.6057496445108974, "learning_rate": 8.369866650488592e-06, "loss": 0.3163, "step": 4215 }, { "epoch": 0.57, "grad_norm": 3.740608618077043, "learning_rate": 8.369052799638838e-06, "loss": 0.2939, "step": 4216 }, { "epoch": 0.57, "grad_norm": 3.04588486652247, "learning_rate": 8.368238785269976e-06, "loss": 0.2647, "step": 4217 }, { "epoch": 0.57, "grad_norm": 2.872007884398736, "learning_rate": 8.367424607421512e-06, "loss": 0.2084, "step": 4218 }, { "epoch": 0.57, "grad_norm": 3.6977730024417412, "learning_rate": 8.366610266132967e-06, "loss": 0.3152, "step": 4219 }, { "epoch": 0.57, "grad_norm": 3.990022607926224, "learning_rate": 8.365795761443865e-06, "loss": 0.2612, "step": 4220 }, { "epoch": 0.57, "grad_norm": 2.8230911227773428, "learning_rate": 8.364981093393734e-06, "loss": 0.2226, "step": 4221 }, { "epoch": 0.57, "grad_norm": 3.914551020550052, "learning_rate": 8.36416626202212e-06, "loss": 0.2976, "step": 4222 }, { "epoch": 0.57, "grad_norm": 3.0480398215514746, "learning_rate": 8.363351267368568e-06, "loss": 0.2581, "step": 4223 }, { "epoch": 0.57, "grad_norm": 4.530703486701039, "learning_rate": 8.362536109472637e-06, "loss": 0.3187, "step": 4224 }, { "epoch": 0.57, "grad_norm": 3.1466654366748092, "learning_rate": 8.361720788373887e-06, "loss": 0.2654, "step": 4225 }, { "epoch": 0.57, "grad_norm": 3.622798638326905, "learning_rate": 8.360905304111894e-06, "loss": 0.2989, "step": 4226 }, { "epoch": 0.57, "grad_norm": 4.663232537916062, "learning_rate": 8.360089656726238e-06, "loss": 0.3272, "step": 4227 }, { "epoch": 0.58, "grad_norm": 3.914526814072208, "learning_rate": 8.359273846256505e-06, "loss": 0.303, "step": 4228 }, { "epoch": 0.58, "grad_norm": 2.8247106912243223, "learning_rate": 8.35845787274229e-06, "loss": 0.2075, "step": 4229 }, { "epoch": 0.58, "grad_norm": 3.382688893762994, "learning_rate": 8.357641736223199e-06, "loss": 0.2041, "step": 4230 }, { "epoch": 0.58, "grad_norm": 4.008339956348646, "learning_rate": 8.356825436738842e-06, "loss": 0.3139, "step": 4231 }, { "epoch": 0.58, "grad_norm": 4.014625815324273, "learning_rate": 8.356008974328843e-06, "loss": 0.2682, "step": 4232 }, { "epoch": 0.58, "grad_norm": 3.1544185094473836, "learning_rate": 8.355192349032824e-06, "loss": 0.2898, "step": 4233 }, { "epoch": 0.58, "grad_norm": 3.999322270708509, "learning_rate": 8.35437556089042e-06, "loss": 0.3097, "step": 4234 }, { "epoch": 0.58, "grad_norm": 4.131834630214244, "learning_rate": 8.35355860994128e-06, "loss": 0.3485, "step": 4235 }, { "epoch": 0.58, "grad_norm": 3.1035176239756748, "learning_rate": 8.352741496225053e-06, "loss": 0.2324, "step": 4236 }, { "epoch": 0.58, "grad_norm": 4.485997387599121, "learning_rate": 8.351924219781393e-06, "loss": 0.2284, "step": 4237 }, { "epoch": 0.58, "grad_norm": 3.2032048210610684, "learning_rate": 8.351106780649974e-06, "loss": 0.2361, "step": 4238 }, { "epoch": 0.58, "grad_norm": 3.3381402646351988, "learning_rate": 8.350289178870468e-06, "loss": 0.2602, "step": 4239 }, { "epoch": 0.58, "grad_norm": 4.286846116903057, "learning_rate": 8.349471414482556e-06, "loss": 0.2591, "step": 4240 }, { "epoch": 0.58, "grad_norm": 4.74351212761802, "learning_rate": 8.348653487525935e-06, "loss": 0.2787, "step": 4241 }, { "epoch": 0.58, "grad_norm": 3.6814354675472654, "learning_rate": 8.347835398040297e-06, "loss": 0.3426, "step": 4242 }, { "epoch": 0.58, "grad_norm": 3.4215721157858336, "learning_rate": 8.347017146065351e-06, "loss": 0.2623, "step": 4243 }, { "epoch": 0.58, "grad_norm": 3.7324470427922614, "learning_rate": 8.346198731640811e-06, "loss": 0.2615, "step": 4244 }, { "epoch": 0.58, "grad_norm": 3.8961822588642874, "learning_rate": 8.3453801548064e-06, "loss": 0.263, "step": 4245 }, { "epoch": 0.58, "grad_norm": 3.329380111887092, "learning_rate": 8.344561415601847e-06, "loss": 0.2354, "step": 4246 }, { "epoch": 0.58, "grad_norm": 3.904901236017731, "learning_rate": 8.34374251406689e-06, "loss": 0.2592, "step": 4247 }, { "epoch": 0.58, "grad_norm": 2.951501308400415, "learning_rate": 8.342923450241275e-06, "loss": 0.215, "step": 4248 }, { "epoch": 0.58, "grad_norm": 2.975625523698402, "learning_rate": 8.342104224164759e-06, "loss": 0.2599, "step": 4249 }, { "epoch": 0.58, "grad_norm": 3.2378884245305555, "learning_rate": 8.341284835877099e-06, "loss": 0.2319, "step": 4250 }, { "epoch": 0.58, "grad_norm": 4.125880897411674, "learning_rate": 8.340465285418067e-06, "loss": 0.2988, "step": 4251 }, { "epoch": 0.58, "grad_norm": 3.916515748129375, "learning_rate": 8.339645572827439e-06, "loss": 0.255, "step": 4252 }, { "epoch": 0.58, "grad_norm": 3.761009281378374, "learning_rate": 8.338825698145003e-06, "loss": 0.2787, "step": 4253 }, { "epoch": 0.58, "grad_norm": 3.6635285223122307, "learning_rate": 8.338005661410548e-06, "loss": 0.2865, "step": 4254 }, { "epoch": 0.58, "grad_norm": 4.04320887963207, "learning_rate": 8.33718546266388e-06, "loss": 0.2502, "step": 4255 }, { "epoch": 0.58, "grad_norm": 3.6233234124308087, "learning_rate": 8.336365101944804e-06, "loss": 0.2789, "step": 4256 }, { "epoch": 0.58, "grad_norm": 3.138102807465223, "learning_rate": 8.335544579293138e-06, "loss": 0.2258, "step": 4257 }, { "epoch": 0.58, "grad_norm": 3.991908830591191, "learning_rate": 8.334723894748705e-06, "loss": 0.3061, "step": 4258 }, { "epoch": 0.58, "grad_norm": 3.313348312144305, "learning_rate": 8.33390304835134e-06, "loss": 0.2511, "step": 4259 }, { "epoch": 0.58, "grad_norm": 4.194858793486396, "learning_rate": 8.333082040140884e-06, "loss": 0.271, "step": 4260 }, { "epoch": 0.58, "grad_norm": 3.49712606173413, "learning_rate": 8.332260870157181e-06, "loss": 0.2536, "step": 4261 }, { "epoch": 0.58, "grad_norm": 3.854391465324948, "learning_rate": 8.331439538440089e-06, "loss": 0.2452, "step": 4262 }, { "epoch": 0.58, "grad_norm": 4.01850294736097, "learning_rate": 8.330618045029473e-06, "loss": 0.2329, "step": 4263 }, { "epoch": 0.58, "grad_norm": 3.5632687793770046, "learning_rate": 8.329796389965204e-06, "loss": 0.32, "step": 4264 }, { "epoch": 0.58, "grad_norm": 3.0685647059115384, "learning_rate": 8.328974573287162e-06, "loss": 0.2189, "step": 4265 }, { "epoch": 0.58, "grad_norm": 4.016758034098606, "learning_rate": 8.328152595035234e-06, "loss": 0.3033, "step": 4266 }, { "epoch": 0.58, "grad_norm": 3.389919295291306, "learning_rate": 8.327330455249316e-06, "loss": 0.2564, "step": 4267 }, { "epoch": 0.58, "grad_norm": 4.398120604036799, "learning_rate": 8.326508153969308e-06, "loss": 0.2997, "step": 4268 }, { "epoch": 0.58, "grad_norm": 4.5611461476017405, "learning_rate": 8.325685691235124e-06, "loss": 0.3233, "step": 4269 }, { "epoch": 0.58, "grad_norm": 4.092778846233072, "learning_rate": 8.32486306708668e-06, "loss": 0.303, "step": 4270 }, { "epoch": 0.58, "grad_norm": 3.3062091162040894, "learning_rate": 8.324040281563907e-06, "loss": 0.2552, "step": 4271 }, { "epoch": 0.58, "grad_norm": 3.9524740476669336, "learning_rate": 8.323217334706736e-06, "loss": 0.3078, "step": 4272 }, { "epoch": 0.58, "grad_norm": 4.234773244253735, "learning_rate": 8.32239422655511e-06, "loss": 0.2847, "step": 4273 }, { "epoch": 0.58, "grad_norm": 3.2635267155938803, "learning_rate": 8.321570957148982e-06, "loss": 0.2504, "step": 4274 }, { "epoch": 0.58, "grad_norm": 3.3715672986002105, "learning_rate": 8.320747526528303e-06, "loss": 0.3234, "step": 4275 }, { "epoch": 0.58, "grad_norm": 4.442317177207381, "learning_rate": 8.319923934733044e-06, "loss": 0.2885, "step": 4276 }, { "epoch": 0.58, "grad_norm": 3.507298736087409, "learning_rate": 8.319100181803177e-06, "loss": 0.2733, "step": 4277 }, { "epoch": 0.58, "grad_norm": 3.262833687180574, "learning_rate": 8.318276267778684e-06, "loss": 0.207, "step": 4278 }, { "epoch": 0.58, "grad_norm": 3.3245002052432175, "learning_rate": 8.317452192699555e-06, "loss": 0.2707, "step": 4279 }, { "epoch": 0.58, "grad_norm": 4.169225643143287, "learning_rate": 8.316627956605785e-06, "loss": 0.2707, "step": 4280 }, { "epoch": 0.58, "grad_norm": 2.956323525441474, "learning_rate": 8.315803559537379e-06, "loss": 0.258, "step": 4281 }, { "epoch": 0.58, "grad_norm": 2.851344113282629, "learning_rate": 8.314979001534351e-06, "loss": 0.2273, "step": 4282 }, { "epoch": 0.58, "grad_norm": 3.6143841778472496, "learning_rate": 8.314154282636722e-06, "loss": 0.2992, "step": 4283 }, { "epoch": 0.58, "grad_norm": 3.696147449161409, "learning_rate": 8.313329402884516e-06, "loss": 0.3166, "step": 4284 }, { "epoch": 0.58, "grad_norm": 2.698676888926633, "learning_rate": 8.312504362317775e-06, "loss": 0.2474, "step": 4285 }, { "epoch": 0.58, "grad_norm": 3.191965592767239, "learning_rate": 8.31167916097654e-06, "loss": 0.2483, "step": 4286 }, { "epoch": 0.58, "grad_norm": 3.6368626344769295, "learning_rate": 8.310853798900861e-06, "loss": 0.2625, "step": 4287 }, { "epoch": 0.58, "grad_norm": 4.217355770885107, "learning_rate": 8.3100282761308e-06, "loss": 0.266, "step": 4288 }, { "epoch": 0.58, "grad_norm": 3.617164777348001, "learning_rate": 8.309202592706423e-06, "loss": 0.2969, "step": 4289 }, { "epoch": 0.58, "grad_norm": 3.3118873134042452, "learning_rate": 8.308376748667807e-06, "loss": 0.2553, "step": 4290 }, { "epoch": 0.58, "grad_norm": 3.834073360019426, "learning_rate": 8.30755074405503e-06, "loss": 0.2899, "step": 4291 }, { "epoch": 0.58, "grad_norm": 3.5305931972994644, "learning_rate": 8.306724578908187e-06, "loss": 0.3007, "step": 4292 }, { "epoch": 0.58, "grad_norm": 3.594795567689558, "learning_rate": 8.305898253267377e-06, "loss": 0.2646, "step": 4293 }, { "epoch": 0.58, "grad_norm": 3.62771769511745, "learning_rate": 8.305071767172704e-06, "loss": 0.2229, "step": 4294 }, { "epoch": 0.58, "grad_norm": 3.536024177026334, "learning_rate": 8.304245120664282e-06, "loss": 0.2708, "step": 4295 }, { "epoch": 0.58, "grad_norm": 3.753878603308491, "learning_rate": 8.303418313782234e-06, "loss": 0.2816, "step": 4296 }, { "epoch": 0.58, "grad_norm": 3.9129335186409087, "learning_rate": 8.302591346566691e-06, "loss": 0.2656, "step": 4297 }, { "epoch": 0.58, "grad_norm": 6.711558380876592, "learning_rate": 8.301764219057786e-06, "loss": 0.2949, "step": 4298 }, { "epoch": 0.58, "grad_norm": 3.3697927356367003, "learning_rate": 8.300936931295667e-06, "loss": 0.2359, "step": 4299 }, { "epoch": 0.58, "grad_norm": 3.2552994450290598, "learning_rate": 8.300109483320488e-06, "loss": 0.2468, "step": 4300 }, { "epoch": 0.59, "grad_norm": 3.6226254271825153, "learning_rate": 8.299281875172407e-06, "loss": 0.2662, "step": 4301 }, { "epoch": 0.59, "grad_norm": 3.482046520555934, "learning_rate": 8.298454106891593e-06, "loss": 0.2475, "step": 4302 }, { "epoch": 0.59, "grad_norm": 3.456198383936231, "learning_rate": 8.297626178518224e-06, "loss": 0.2703, "step": 4303 }, { "epoch": 0.59, "grad_norm": 4.129211201161081, "learning_rate": 8.296798090092482e-06, "loss": 0.319, "step": 4304 }, { "epoch": 0.59, "grad_norm": 3.596432466738805, "learning_rate": 8.295969841654563e-06, "loss": 0.2559, "step": 4305 }, { "epoch": 0.59, "grad_norm": 3.8327746195973833, "learning_rate": 8.29514143324466e-06, "loss": 0.3345, "step": 4306 }, { "epoch": 0.59, "grad_norm": 3.350026254897178, "learning_rate": 8.294312864902985e-06, "loss": 0.2383, "step": 4307 }, { "epoch": 0.59, "grad_norm": 3.299789597553567, "learning_rate": 8.293484136669751e-06, "loss": 0.2945, "step": 4308 }, { "epoch": 0.59, "grad_norm": 3.921958945496777, "learning_rate": 8.292655248585184e-06, "loss": 0.3694, "step": 4309 }, { "epoch": 0.59, "grad_norm": 3.211515447922096, "learning_rate": 8.29182620068951e-06, "loss": 0.2433, "step": 4310 }, { "epoch": 0.59, "grad_norm": 3.6147401719456482, "learning_rate": 8.29099699302297e-06, "loss": 0.2722, "step": 4311 }, { "epoch": 0.59, "grad_norm": 3.7703646779306577, "learning_rate": 8.290167625625811e-06, "loss": 0.2612, "step": 4312 }, { "epoch": 0.59, "grad_norm": 3.426474778169555, "learning_rate": 8.289338098538285e-06, "loss": 0.2806, "step": 4313 }, { "epoch": 0.59, "grad_norm": 3.5809733433898474, "learning_rate": 8.288508411800653e-06, "loss": 0.2813, "step": 4314 }, { "epoch": 0.59, "grad_norm": 3.5046129156978805, "learning_rate": 8.287678565453187e-06, "loss": 0.2873, "step": 4315 }, { "epoch": 0.59, "grad_norm": 4.214970972833488, "learning_rate": 8.286848559536161e-06, "loss": 0.2517, "step": 4316 }, { "epoch": 0.59, "grad_norm": 3.558633947553824, "learning_rate": 8.286018394089864e-06, "loss": 0.2644, "step": 4317 }, { "epoch": 0.59, "grad_norm": 4.076168459465584, "learning_rate": 8.285188069154584e-06, "loss": 0.285, "step": 4318 }, { "epoch": 0.59, "grad_norm": 3.2469392648217776, "learning_rate": 8.284357584770626e-06, "loss": 0.296, "step": 4319 }, { "epoch": 0.59, "grad_norm": 4.358844250094569, "learning_rate": 8.283526940978292e-06, "loss": 0.2602, "step": 4320 }, { "epoch": 0.59, "grad_norm": 3.1313465733712658, "learning_rate": 8.282696137817905e-06, "loss": 0.2616, "step": 4321 }, { "epoch": 0.59, "grad_norm": 3.7740033285843566, "learning_rate": 8.281865175329783e-06, "loss": 0.2936, "step": 4322 }, { "epoch": 0.59, "grad_norm": 3.5023581467885996, "learning_rate": 8.28103405355426e-06, "loss": 0.2916, "step": 4323 }, { "epoch": 0.59, "grad_norm": 3.669610296985965, "learning_rate": 8.280202772531674e-06, "loss": 0.2614, "step": 4324 }, { "epoch": 0.59, "grad_norm": 3.8077663793141494, "learning_rate": 8.279371332302373e-06, "loss": 0.2511, "step": 4325 }, { "epoch": 0.59, "grad_norm": 3.848901543116737, "learning_rate": 8.27853973290671e-06, "loss": 0.27, "step": 4326 }, { "epoch": 0.59, "grad_norm": 2.838944114396536, "learning_rate": 8.277707974385047e-06, "loss": 0.2188, "step": 4327 }, { "epoch": 0.59, "grad_norm": 3.362443865941836, "learning_rate": 8.276876056777755e-06, "loss": 0.2718, "step": 4328 }, { "epoch": 0.59, "grad_norm": 3.6226116525693994, "learning_rate": 8.276043980125213e-06, "loss": 0.2306, "step": 4329 }, { "epoch": 0.59, "grad_norm": 3.5230423962092803, "learning_rate": 8.275211744467801e-06, "loss": 0.3233, "step": 4330 }, { "epoch": 0.59, "grad_norm": 3.6633020146123965, "learning_rate": 8.274379349845918e-06, "loss": 0.3049, "step": 4331 }, { "epoch": 0.59, "grad_norm": 2.9098991310371263, "learning_rate": 8.273546796299962e-06, "loss": 0.2318, "step": 4332 }, { "epoch": 0.59, "grad_norm": 2.93241299258365, "learning_rate": 8.272714083870344e-06, "loss": 0.2407, "step": 4333 }, { "epoch": 0.59, "grad_norm": 3.633418026310484, "learning_rate": 8.271881212597476e-06, "loss": 0.2795, "step": 4334 }, { "epoch": 0.59, "grad_norm": 3.753342907789223, "learning_rate": 8.271048182521786e-06, "loss": 0.2795, "step": 4335 }, { "epoch": 0.59, "grad_norm": 3.35202166506505, "learning_rate": 8.270214993683703e-06, "loss": 0.241, "step": 4336 }, { "epoch": 0.59, "grad_norm": 3.857327923083195, "learning_rate": 8.269381646123666e-06, "loss": 0.2797, "step": 4337 }, { "epoch": 0.59, "grad_norm": 5.013299869049046, "learning_rate": 8.268548139882126e-06, "loss": 0.3077, "step": 4338 }, { "epoch": 0.59, "grad_norm": 3.24950844346804, "learning_rate": 8.267714474999533e-06, "loss": 0.2795, "step": 4339 }, { "epoch": 0.59, "grad_norm": 3.9837170404159634, "learning_rate": 8.266880651516353e-06, "loss": 0.2876, "step": 4340 }, { "epoch": 0.59, "grad_norm": 3.8215869216362672, "learning_rate": 8.266046669473053e-06, "loss": 0.2336, "step": 4341 }, { "epoch": 0.59, "grad_norm": 3.303470641103438, "learning_rate": 8.265212528910113e-06, "loss": 0.2898, "step": 4342 }, { "epoch": 0.59, "grad_norm": 3.6381420322942923, "learning_rate": 8.26437822986802e-06, "loss": 0.2668, "step": 4343 }, { "epoch": 0.59, "grad_norm": 3.1297851709662408, "learning_rate": 8.263543772387261e-06, "loss": 0.2524, "step": 4344 }, { "epoch": 0.59, "grad_norm": 3.036248897010743, "learning_rate": 8.262709156508345e-06, "loss": 0.231, "step": 4345 }, { "epoch": 0.59, "grad_norm": 2.830946445498571, "learning_rate": 8.261874382271775e-06, "loss": 0.2787, "step": 4346 }, { "epoch": 0.59, "grad_norm": 3.3884397740403602, "learning_rate": 8.261039449718068e-06, "loss": 0.2444, "step": 4347 }, { "epoch": 0.59, "grad_norm": 3.992205251224511, "learning_rate": 8.260204358887753e-06, "loss": 0.2832, "step": 4348 }, { "epoch": 0.59, "grad_norm": 3.3595544779678113, "learning_rate": 8.259369109821357e-06, "loss": 0.2702, "step": 4349 }, { "epoch": 0.59, "grad_norm": 3.844010952351389, "learning_rate": 8.258533702559417e-06, "loss": 0.2943, "step": 4350 }, { "epoch": 0.59, "grad_norm": 3.2897326019427324, "learning_rate": 8.257698137142485e-06, "loss": 0.2703, "step": 4351 }, { "epoch": 0.59, "grad_norm": 4.372268389228694, "learning_rate": 8.256862413611113e-06, "loss": 0.286, "step": 4352 }, { "epoch": 0.59, "grad_norm": 3.1411671527156044, "learning_rate": 8.256026532005863e-06, "loss": 0.2702, "step": 4353 }, { "epoch": 0.59, "grad_norm": 3.292687553548399, "learning_rate": 8.255190492367309e-06, "loss": 0.2449, "step": 4354 }, { "epoch": 0.59, "grad_norm": 3.6001674827178958, "learning_rate": 8.254354294736023e-06, "loss": 0.3055, "step": 4355 }, { "epoch": 0.59, "grad_norm": 3.1262763236985647, "learning_rate": 8.253517939152596e-06, "loss": 0.2369, "step": 4356 }, { "epoch": 0.59, "grad_norm": 3.3792916878216834, "learning_rate": 8.252681425657617e-06, "loss": 0.2718, "step": 4357 }, { "epoch": 0.59, "grad_norm": 4.077203731927472, "learning_rate": 8.251844754291688e-06, "loss": 0.3127, "step": 4358 }, { "epoch": 0.59, "grad_norm": 3.466631429130733, "learning_rate": 8.251007925095416e-06, "loss": 0.246, "step": 4359 }, { "epoch": 0.59, "grad_norm": 3.9988473281000116, "learning_rate": 8.25017093810942e-06, "loss": 0.2711, "step": 4360 }, { "epoch": 0.59, "grad_norm": 3.9890108750142397, "learning_rate": 8.249333793374324e-06, "loss": 0.2986, "step": 4361 }, { "epoch": 0.59, "grad_norm": 3.310358513790767, "learning_rate": 8.248496490930753e-06, "loss": 0.3018, "step": 4362 }, { "epoch": 0.59, "grad_norm": 3.3065096703111503, "learning_rate": 8.247659030819353e-06, "loss": 0.293, "step": 4363 }, { "epoch": 0.59, "grad_norm": 3.5009232727997888, "learning_rate": 8.24682141308077e-06, "loss": 0.2954, "step": 4364 }, { "epoch": 0.59, "grad_norm": 3.7537711661301483, "learning_rate": 8.245983637755655e-06, "loss": 0.2682, "step": 4365 }, { "epoch": 0.59, "grad_norm": 3.790310446760243, "learning_rate": 8.245145704884669e-06, "loss": 0.2834, "step": 4366 }, { "epoch": 0.59, "grad_norm": 3.2769850143433956, "learning_rate": 8.244307614508487e-06, "loss": 0.2582, "step": 4367 }, { "epoch": 0.59, "grad_norm": 3.7961814170052164, "learning_rate": 8.243469366667783e-06, "loss": 0.2205, "step": 4368 }, { "epoch": 0.59, "grad_norm": 5.331611468474943, "learning_rate": 8.242630961403242e-06, "loss": 0.2832, "step": 4369 }, { "epoch": 0.59, "grad_norm": 3.755128776952831, "learning_rate": 8.241792398755555e-06, "loss": 0.3461, "step": 4370 }, { "epoch": 0.59, "grad_norm": 3.6556969000352457, "learning_rate": 8.240953678765425e-06, "loss": 0.2662, "step": 4371 }, { "epoch": 0.59, "grad_norm": 3.280082088251302, "learning_rate": 8.240114801473558e-06, "loss": 0.2491, "step": 4372 }, { "epoch": 0.59, "grad_norm": 3.2161014022992096, "learning_rate": 8.239275766920672e-06, "loss": 0.2709, "step": 4373 }, { "epoch": 0.59, "grad_norm": 3.312438666157512, "learning_rate": 8.238436575147486e-06, "loss": 0.2792, "step": 4374 }, { "epoch": 0.6, "grad_norm": 3.440620239928278, "learning_rate": 8.237597226194733e-06, "loss": 0.2833, "step": 4375 }, { "epoch": 0.6, "grad_norm": 3.1916178843192244, "learning_rate": 8.236757720103154e-06, "loss": 0.2255, "step": 4376 }, { "epoch": 0.6, "grad_norm": 3.362819202175802, "learning_rate": 8.23591805691349e-06, "loss": 0.3136, "step": 4377 }, { "epoch": 0.6, "grad_norm": 3.55922250346872, "learning_rate": 8.235078236666496e-06, "loss": 0.2419, "step": 4378 }, { "epoch": 0.6, "grad_norm": 3.1636254725014092, "learning_rate": 8.234238259402936e-06, "loss": 0.2783, "step": 4379 }, { "epoch": 0.6, "grad_norm": 3.3136342838690225, "learning_rate": 8.233398125163574e-06, "loss": 0.2818, "step": 4380 }, { "epoch": 0.6, "grad_norm": 3.5351434841752583, "learning_rate": 8.232557833989193e-06, "loss": 0.2435, "step": 4381 }, { "epoch": 0.6, "grad_norm": 3.6785640156694934, "learning_rate": 8.23171738592057e-06, "loss": 0.271, "step": 4382 }, { "epoch": 0.6, "grad_norm": 3.528884314734583, "learning_rate": 8.2308767809985e-06, "loss": 0.2684, "step": 4383 }, { "epoch": 0.6, "grad_norm": 3.011027374410594, "learning_rate": 8.230036019263783e-06, "loss": 0.2867, "step": 4384 }, { "epoch": 0.6, "grad_norm": 3.9821167353795985, "learning_rate": 8.229195100757226e-06, "loss": 0.2681, "step": 4385 }, { "epoch": 0.6, "grad_norm": 3.5700461118568865, "learning_rate": 8.228354025519642e-06, "loss": 0.2557, "step": 4386 }, { "epoch": 0.6, "grad_norm": 4.678176485223439, "learning_rate": 8.227512793591855e-06, "loss": 0.2944, "step": 4387 }, { "epoch": 0.6, "grad_norm": 3.313885894075238, "learning_rate": 8.226671405014692e-06, "loss": 0.2357, "step": 4388 }, { "epoch": 0.6, "grad_norm": 3.036311859842429, "learning_rate": 8.225829859828994e-06, "loss": 0.2165, "step": 4389 }, { "epoch": 0.6, "grad_norm": 3.600543038463837, "learning_rate": 8.224988158075602e-06, "loss": 0.2759, "step": 4390 }, { "epoch": 0.6, "grad_norm": 3.776487676671449, "learning_rate": 8.22414629979537e-06, "loss": 0.2934, "step": 4391 }, { "epoch": 0.6, "grad_norm": 3.6576917555427113, "learning_rate": 8.223304285029159e-06, "loss": 0.2238, "step": 4392 }, { "epoch": 0.6, "grad_norm": 3.3248763181820387, "learning_rate": 8.222462113817835e-06, "loss": 0.2552, "step": 4393 }, { "epoch": 0.6, "grad_norm": 3.8230648889122723, "learning_rate": 8.221619786202275e-06, "loss": 0.2535, "step": 4394 }, { "epoch": 0.6, "grad_norm": 3.981630373455406, "learning_rate": 8.220777302223363e-06, "loss": 0.3195, "step": 4395 }, { "epoch": 0.6, "grad_norm": 3.8631540807958147, "learning_rate": 8.219934661921986e-06, "loss": 0.2681, "step": 4396 }, { "epoch": 0.6, "grad_norm": 3.207029799233959, "learning_rate": 8.219091865339045e-06, "loss": 0.2253, "step": 4397 }, { "epoch": 0.6, "grad_norm": 2.969204103360666, "learning_rate": 8.218248912515443e-06, "loss": 0.2478, "step": 4398 }, { "epoch": 0.6, "grad_norm": 3.973335691655764, "learning_rate": 8.217405803492095e-06, "loss": 0.2819, "step": 4399 }, { "epoch": 0.6, "grad_norm": 3.554152979048042, "learning_rate": 8.216562538309924e-06, "loss": 0.2749, "step": 4400 }, { "epoch": 0.6, "grad_norm": 3.7864941597920403, "learning_rate": 8.215719117009855e-06, "loss": 0.2997, "step": 4401 }, { "epoch": 0.6, "grad_norm": 3.5400259320419996, "learning_rate": 8.214875539632825e-06, "loss": 0.2755, "step": 4402 }, { "epoch": 0.6, "grad_norm": 4.412269468565807, "learning_rate": 8.214031806219777e-06, "loss": 0.2733, "step": 4403 }, { "epoch": 0.6, "grad_norm": 3.795194651715882, "learning_rate": 8.213187916811663e-06, "loss": 0.2574, "step": 4404 }, { "epoch": 0.6, "grad_norm": 3.896564732955062, "learning_rate": 8.212343871449441e-06, "loss": 0.2441, "step": 4405 }, { "epoch": 0.6, "grad_norm": 4.171223970170487, "learning_rate": 8.211499670174079e-06, "loss": 0.2508, "step": 4406 }, { "epoch": 0.6, "grad_norm": 2.9800842147087985, "learning_rate": 8.21065531302655e-06, "loss": 0.198, "step": 4407 }, { "epoch": 0.6, "grad_norm": 4.094610448445488, "learning_rate": 8.209810800047833e-06, "loss": 0.2666, "step": 4408 }, { "epoch": 0.6, "grad_norm": 3.6753753576431074, "learning_rate": 8.20896613127892e-06, "loss": 0.2896, "step": 4409 }, { "epoch": 0.6, "grad_norm": 3.9240016491139222, "learning_rate": 8.208121306760806e-06, "loss": 0.277, "step": 4410 }, { "epoch": 0.6, "grad_norm": 4.440504011905315, "learning_rate": 8.207276326534497e-06, "loss": 0.2796, "step": 4411 }, { "epoch": 0.6, "grad_norm": 3.639820634913097, "learning_rate": 8.206431190641002e-06, "loss": 0.2809, "step": 4412 }, { "epoch": 0.6, "grad_norm": 3.4656359585634564, "learning_rate": 8.20558589912134e-06, "loss": 0.2267, "step": 4413 }, { "epoch": 0.6, "grad_norm": 3.0104657193887148, "learning_rate": 8.204740452016543e-06, "loss": 0.235, "step": 4414 }, { "epoch": 0.6, "grad_norm": 3.503096389050954, "learning_rate": 8.20389484936764e-06, "loss": 0.2662, "step": 4415 }, { "epoch": 0.6, "grad_norm": 3.9288395805154326, "learning_rate": 8.203049091215672e-06, "loss": 0.2567, "step": 4416 }, { "epoch": 0.6, "grad_norm": 3.267514201230857, "learning_rate": 8.202203177601693e-06, "loss": 0.3013, "step": 4417 }, { "epoch": 0.6, "grad_norm": 3.307845612800748, "learning_rate": 8.201357108566757e-06, "loss": 0.2931, "step": 4418 }, { "epoch": 0.6, "grad_norm": 4.2196610492905, "learning_rate": 8.20051088415193e-06, "loss": 0.2405, "step": 4419 }, { "epoch": 0.6, "grad_norm": 3.6201131701602356, "learning_rate": 8.199664504398281e-06, "loss": 0.2546, "step": 4420 }, { "epoch": 0.6, "grad_norm": 3.6858938513003032, "learning_rate": 8.198817969346895e-06, "loss": 0.35, "step": 4421 }, { "epoch": 0.6, "grad_norm": 3.4533584866837677, "learning_rate": 8.197971279038854e-06, "loss": 0.2838, "step": 4422 }, { "epoch": 0.6, "grad_norm": 3.5548946092639966, "learning_rate": 8.197124433515256e-06, "loss": 0.2711, "step": 4423 }, { "epoch": 0.6, "grad_norm": 3.568593071760516, "learning_rate": 8.1962774328172e-06, "loss": 0.2936, "step": 4424 }, { "epoch": 0.6, "grad_norm": 3.2761120332638956, "learning_rate": 8.195430276985798e-06, "loss": 0.2822, "step": 4425 }, { "epoch": 0.6, "grad_norm": 3.4098604501129124, "learning_rate": 8.194582966062168e-06, "loss": 0.2413, "step": 4426 }, { "epoch": 0.6, "grad_norm": 4.361093741137064, "learning_rate": 8.193735500087432e-06, "loss": 0.2936, "step": 4427 }, { "epoch": 0.6, "grad_norm": 3.3571234805568846, "learning_rate": 8.192887879102726e-06, "loss": 0.2442, "step": 4428 }, { "epoch": 0.6, "grad_norm": 3.1263821763842867, "learning_rate": 8.192040103149185e-06, "loss": 0.298, "step": 4429 }, { "epoch": 0.6, "grad_norm": 5.683579829712689, "learning_rate": 8.191192172267962e-06, "loss": 0.2655, "step": 4430 }, { "epoch": 0.6, "grad_norm": 3.120041035453433, "learning_rate": 8.190344086500206e-06, "loss": 0.2899, "step": 4431 }, { "epoch": 0.6, "grad_norm": 3.133909152028711, "learning_rate": 8.189495845887083e-06, "loss": 0.2576, "step": 4432 }, { "epoch": 0.6, "grad_norm": 4.900005093414994, "learning_rate": 8.188647450469762e-06, "loss": 0.2246, "step": 4433 }, { "epoch": 0.6, "grad_norm": 3.2644860674153136, "learning_rate": 8.187798900289421e-06, "loss": 0.2348, "step": 4434 }, { "epoch": 0.6, "grad_norm": 3.4818597056157103, "learning_rate": 8.186950195387244e-06, "loss": 0.248, "step": 4435 }, { "epoch": 0.6, "grad_norm": 3.249853471505576, "learning_rate": 8.186101335804424e-06, "loss": 0.2826, "step": 4436 }, { "epoch": 0.6, "grad_norm": 3.272835255959788, "learning_rate": 8.185252321582162e-06, "loss": 0.2584, "step": 4437 }, { "epoch": 0.6, "grad_norm": 3.6085776578505193, "learning_rate": 8.184403152761663e-06, "loss": 0.2921, "step": 4438 }, { "epoch": 0.6, "grad_norm": 3.6712046005394727, "learning_rate": 8.183553829384144e-06, "loss": 0.2771, "step": 4439 }, { "epoch": 0.6, "grad_norm": 2.615014512147407, "learning_rate": 8.182704351490824e-06, "loss": 0.211, "step": 4440 }, { "epoch": 0.6, "grad_norm": 3.5605158927684566, "learning_rate": 8.181854719122938e-06, "loss": 0.2537, "step": 4441 }, { "epoch": 0.6, "grad_norm": 2.9799185052887665, "learning_rate": 8.18100493232172e-06, "loss": 0.2048, "step": 4442 }, { "epoch": 0.6, "grad_norm": 4.127611804977173, "learning_rate": 8.180154991128418e-06, "loss": 0.2737, "step": 4443 }, { "epoch": 0.6, "grad_norm": 2.991012017940089, "learning_rate": 8.179304895584282e-06, "loss": 0.2623, "step": 4444 }, { "epoch": 0.6, "grad_norm": 3.7826804108892014, "learning_rate": 8.17845464573057e-06, "loss": 0.333, "step": 4445 }, { "epoch": 0.6, "grad_norm": 3.674196407631066, "learning_rate": 8.177604241608554e-06, "loss": 0.209, "step": 4446 }, { "epoch": 0.6, "grad_norm": 3.683655646631497, "learning_rate": 8.176753683259506e-06, "loss": 0.2756, "step": 4447 }, { "epoch": 0.61, "grad_norm": 3.5980592492067363, "learning_rate": 8.175902970724707e-06, "loss": 0.2704, "step": 4448 }, { "epoch": 0.61, "grad_norm": 3.1511250956162926, "learning_rate": 8.17505210404545e-06, "loss": 0.2083, "step": 4449 }, { "epoch": 0.61, "grad_norm": 3.554853274634849, "learning_rate": 8.174201083263034e-06, "loss": 0.2892, "step": 4450 }, { "epoch": 0.61, "grad_norm": 3.948190545170249, "learning_rate": 8.173349908418759e-06, "loss": 0.244, "step": 4451 }, { "epoch": 0.61, "grad_norm": 3.073135707289009, "learning_rate": 8.172498579553939e-06, "loss": 0.2417, "step": 4452 }, { "epoch": 0.61, "grad_norm": 3.355004754946159, "learning_rate": 8.171647096709894e-06, "loss": 0.2724, "step": 4453 }, { "epoch": 0.61, "grad_norm": 3.7807916853785937, "learning_rate": 8.170795459927951e-06, "loss": 0.2707, "step": 4454 }, { "epoch": 0.61, "grad_norm": 4.886411701475539, "learning_rate": 8.169943669249446e-06, "loss": 0.2348, "step": 4455 }, { "epoch": 0.61, "grad_norm": 3.6389684428025686, "learning_rate": 8.169091724715721e-06, "loss": 0.2634, "step": 4456 }, { "epoch": 0.61, "grad_norm": 5.633902433705274, "learning_rate": 8.168239626368126e-06, "loss": 0.2651, "step": 4457 }, { "epoch": 0.61, "grad_norm": 3.487302417015625, "learning_rate": 8.167387374248015e-06, "loss": 0.2583, "step": 4458 }, { "epoch": 0.61, "grad_norm": 3.5803477912758352, "learning_rate": 8.166534968396755e-06, "loss": 0.3098, "step": 4459 }, { "epoch": 0.61, "grad_norm": 3.2141129524810728, "learning_rate": 8.16568240885572e-06, "loss": 0.2143, "step": 4460 }, { "epoch": 0.61, "grad_norm": 3.3686210815671096, "learning_rate": 8.164829695666285e-06, "loss": 0.2609, "step": 4461 }, { "epoch": 0.61, "grad_norm": 3.6440976080655454, "learning_rate": 8.16397682886984e-06, "loss": 0.2504, "step": 4462 }, { "epoch": 0.61, "grad_norm": 4.2843360110033775, "learning_rate": 8.163123808507777e-06, "loss": 0.3061, "step": 4463 }, { "epoch": 0.61, "grad_norm": 3.5795954624295625, "learning_rate": 8.162270634621503e-06, "loss": 0.242, "step": 4464 }, { "epoch": 0.61, "grad_norm": 7.13533485118604, "learning_rate": 8.161417307252422e-06, "loss": 0.29, "step": 4465 }, { "epoch": 0.61, "grad_norm": 10.628032334837084, "learning_rate": 8.160563826441954e-06, "loss": 0.2656, "step": 4466 }, { "epoch": 0.61, "grad_norm": 3.521915331983961, "learning_rate": 8.15971019223152e-06, "loss": 0.258, "step": 4467 }, { "epoch": 0.61, "grad_norm": 3.5981563916045407, "learning_rate": 8.158856404662556e-06, "loss": 0.2049, "step": 4468 }, { "epoch": 0.61, "grad_norm": 3.866176792877818, "learning_rate": 8.158002463776498e-06, "loss": 0.2827, "step": 4469 }, { "epoch": 0.61, "grad_norm": 3.8415888644025635, "learning_rate": 8.157148369614792e-06, "loss": 0.2704, "step": 4470 }, { "epoch": 0.61, "grad_norm": 3.486685702765085, "learning_rate": 8.156294122218896e-06, "loss": 0.2284, "step": 4471 }, { "epoch": 0.61, "grad_norm": 3.669716183939139, "learning_rate": 8.155439721630265e-06, "loss": 0.2815, "step": 4472 }, { "epoch": 0.61, "grad_norm": 3.298598865975544, "learning_rate": 8.154585167890374e-06, "loss": 0.2507, "step": 4473 }, { "epoch": 0.61, "grad_norm": 3.5490818615605777, "learning_rate": 8.153730461040697e-06, "loss": 0.2737, "step": 4474 }, { "epoch": 0.61, "grad_norm": 3.4938667821963545, "learning_rate": 8.152875601122716e-06, "loss": 0.2874, "step": 4475 }, { "epoch": 0.61, "grad_norm": 3.0716833213322214, "learning_rate": 8.152020588177924e-06, "loss": 0.2447, "step": 4476 }, { "epoch": 0.61, "grad_norm": 3.5859166422128, "learning_rate": 8.151165422247822e-06, "loss": 0.3689, "step": 4477 }, { "epoch": 0.61, "grad_norm": 3.0888811392750184, "learning_rate": 8.150310103373911e-06, "loss": 0.2546, "step": 4478 }, { "epoch": 0.61, "grad_norm": 3.4526720405014215, "learning_rate": 8.149454631597708e-06, "loss": 0.2298, "step": 4479 }, { "epoch": 0.61, "grad_norm": 3.1753000150692814, "learning_rate": 8.148599006960733e-06, "loss": 0.2526, "step": 4480 }, { "epoch": 0.61, "grad_norm": 3.044391006748875, "learning_rate": 8.147743229504513e-06, "loss": 0.2555, "step": 4481 }, { "epoch": 0.61, "grad_norm": 2.9963802322918975, "learning_rate": 8.146887299270585e-06, "loss": 0.2458, "step": 4482 }, { "epoch": 0.61, "grad_norm": 3.75705254513346, "learning_rate": 8.146031216300493e-06, "loss": 0.2911, "step": 4483 }, { "epoch": 0.61, "grad_norm": 3.7168256056259388, "learning_rate": 8.145174980635786e-06, "loss": 0.2731, "step": 4484 }, { "epoch": 0.61, "grad_norm": 3.290152849356111, "learning_rate": 8.14431859231802e-06, "loss": 0.2435, "step": 4485 }, { "epoch": 0.61, "grad_norm": 3.462672592770933, "learning_rate": 8.143462051388766e-06, "loss": 0.3103, "step": 4486 }, { "epoch": 0.61, "grad_norm": 2.778846065368353, "learning_rate": 8.142605357889592e-06, "loss": 0.2365, "step": 4487 }, { "epoch": 0.61, "grad_norm": 3.7082753811306284, "learning_rate": 8.14174851186208e-06, "loss": 0.2856, "step": 4488 }, { "epoch": 0.61, "grad_norm": 4.074635212145501, "learning_rate": 8.140891513347817e-06, "loss": 0.257, "step": 4489 }, { "epoch": 0.61, "grad_norm": 3.650373672917617, "learning_rate": 8.140034362388398e-06, "loss": 0.2247, "step": 4490 }, { "epoch": 0.61, "grad_norm": 3.088011105961298, "learning_rate": 8.139177059025425e-06, "loss": 0.2008, "step": 4491 }, { "epoch": 0.61, "grad_norm": 3.274955626259443, "learning_rate": 8.13831960330051e-06, "loss": 0.2439, "step": 4492 }, { "epoch": 0.61, "grad_norm": 3.6906389150981527, "learning_rate": 8.137461995255268e-06, "loss": 0.3049, "step": 4493 }, { "epoch": 0.61, "grad_norm": 3.8828212737892005, "learning_rate": 8.13660423493132e-06, "loss": 0.3369, "step": 4494 }, { "epoch": 0.61, "grad_norm": 4.044084354788373, "learning_rate": 8.135746322370306e-06, "loss": 0.2508, "step": 4495 }, { "epoch": 0.61, "grad_norm": 4.012756432974991, "learning_rate": 8.13488825761386e-06, "loss": 0.2878, "step": 4496 }, { "epoch": 0.61, "grad_norm": 3.545560679425645, "learning_rate": 8.13403004070363e-06, "loss": 0.2529, "step": 4497 }, { "epoch": 0.61, "grad_norm": 2.9360084036150953, "learning_rate": 8.13317167168127e-06, "loss": 0.2174, "step": 4498 }, { "epoch": 0.61, "grad_norm": 3.7660953095195953, "learning_rate": 8.132313150588442e-06, "loss": 0.2916, "step": 4499 }, { "epoch": 0.61, "grad_norm": 3.8107547510299726, "learning_rate": 8.131454477466815e-06, "loss": 0.2922, "step": 4500 }, { "epoch": 0.61, "grad_norm": 3.8042372303930376, "learning_rate": 8.130595652358065e-06, "loss": 0.3011, "step": 4501 }, { "epoch": 0.61, "grad_norm": 3.8118664302068304, "learning_rate": 8.129736675303873e-06, "loss": 0.2245, "step": 4502 }, { "epoch": 0.61, "grad_norm": 3.6156781515843646, "learning_rate": 8.128877546345932e-06, "loss": 0.1932, "step": 4503 }, { "epoch": 0.61, "grad_norm": 3.88396277940488, "learning_rate": 8.128018265525944e-06, "loss": 0.3038, "step": 4504 }, { "epoch": 0.61, "grad_norm": 3.533303002017001, "learning_rate": 8.127158832885609e-06, "loss": 0.276, "step": 4505 }, { "epoch": 0.61, "grad_norm": 4.165555713276088, "learning_rate": 8.126299248466642e-06, "loss": 0.2183, "step": 4506 }, { "epoch": 0.61, "grad_norm": 3.469562069380132, "learning_rate": 8.125439512310765e-06, "loss": 0.2663, "step": 4507 }, { "epoch": 0.61, "grad_norm": 3.1472275607551805, "learning_rate": 8.124579624459705e-06, "loss": 0.2939, "step": 4508 }, { "epoch": 0.61, "grad_norm": 3.485620265539875, "learning_rate": 8.123719584955195e-06, "loss": 0.2362, "step": 4509 }, { "epoch": 0.61, "grad_norm": 3.645250984558887, "learning_rate": 8.122859393838983e-06, "loss": 0.2563, "step": 4510 }, { "epoch": 0.61, "grad_norm": 4.15254824267404, "learning_rate": 8.121999051152811e-06, "loss": 0.287, "step": 4511 }, { "epoch": 0.61, "grad_norm": 3.137162570789958, "learning_rate": 8.121138556938444e-06, "loss": 0.2249, "step": 4512 }, { "epoch": 0.61, "grad_norm": 3.5868467011809098, "learning_rate": 8.120277911237641e-06, "loss": 0.3088, "step": 4513 }, { "epoch": 0.61, "grad_norm": 3.196115804575088, "learning_rate": 8.119417114092179e-06, "loss": 0.2754, "step": 4514 }, { "epoch": 0.61, "grad_norm": 3.882824436473958, "learning_rate": 8.118556165543833e-06, "loss": 0.2486, "step": 4515 }, { "epoch": 0.61, "grad_norm": 5.139546685644648, "learning_rate": 8.117695065634393e-06, "loss": 0.2929, "step": 4516 }, { "epoch": 0.61, "grad_norm": 4.142892205127521, "learning_rate": 8.116833814405648e-06, "loss": 0.2742, "step": 4517 }, { "epoch": 0.61, "grad_norm": 3.428980303331549, "learning_rate": 8.115972411899407e-06, "loss": 0.2299, "step": 4518 }, { "epoch": 0.61, "grad_norm": 4.056246822911087, "learning_rate": 8.115110858157472e-06, "loss": 0.2749, "step": 4519 }, { "epoch": 0.61, "grad_norm": 3.598572554195962, "learning_rate": 8.114249153221664e-06, "loss": 0.2892, "step": 4520 }, { "epoch": 0.61, "grad_norm": 3.8120559734903816, "learning_rate": 8.1133872971338e-06, "loss": 0.2593, "step": 4521 }, { "epoch": 0.62, "grad_norm": 3.2306164840972533, "learning_rate": 8.112525289935716e-06, "loss": 0.2783, "step": 4522 }, { "epoch": 0.62, "grad_norm": 4.274646665119707, "learning_rate": 8.11166313166925e-06, "loss": 0.2733, "step": 4523 }, { "epoch": 0.62, "grad_norm": 4.8871922604143725, "learning_rate": 8.110800822376245e-06, "loss": 0.3345, "step": 4524 }, { "epoch": 0.62, "grad_norm": 5.028835475473851, "learning_rate": 8.109938362098556e-06, "loss": 0.2594, "step": 4525 }, { "epoch": 0.62, "grad_norm": 3.4473163886207367, "learning_rate": 8.10907575087804e-06, "loss": 0.2576, "step": 4526 }, { "epoch": 0.62, "grad_norm": 4.3574027041785985, "learning_rate": 8.108212988756568e-06, "loss": 0.2568, "step": 4527 }, { "epoch": 0.62, "grad_norm": 3.631600558014914, "learning_rate": 8.107350075776013e-06, "loss": 0.3262, "step": 4528 }, { "epoch": 0.62, "grad_norm": 3.6365185424654047, "learning_rate": 8.106487011978254e-06, "loss": 0.2911, "step": 4529 }, { "epoch": 0.62, "grad_norm": 3.403082188294506, "learning_rate": 8.105623797405187e-06, "loss": 0.2724, "step": 4530 }, { "epoch": 0.62, "grad_norm": 2.960447802903521, "learning_rate": 8.104760432098704e-06, "loss": 0.2077, "step": 4531 }, { "epoch": 0.62, "grad_norm": 2.891140463742949, "learning_rate": 8.10389691610071e-06, "loss": 0.2685, "step": 4532 }, { "epoch": 0.62, "grad_norm": 4.5447716619635905, "learning_rate": 8.103033249453116e-06, "loss": 0.2697, "step": 4533 }, { "epoch": 0.62, "grad_norm": 3.860371891112774, "learning_rate": 8.102169432197842e-06, "loss": 0.2524, "step": 4534 }, { "epoch": 0.62, "grad_norm": 3.672737147742819, "learning_rate": 8.10130546437681e-06, "loss": 0.2444, "step": 4535 }, { "epoch": 0.62, "grad_norm": 3.8189386360282693, "learning_rate": 8.100441346031958e-06, "loss": 0.27, "step": 4536 }, { "epoch": 0.62, "grad_norm": 3.175575302093552, "learning_rate": 8.099577077205225e-06, "loss": 0.26, "step": 4537 }, { "epoch": 0.62, "grad_norm": 3.4505375740653577, "learning_rate": 8.098712657938558e-06, "loss": 0.2378, "step": 4538 }, { "epoch": 0.62, "grad_norm": 3.846054343553348, "learning_rate": 8.097848088273912e-06, "loss": 0.2999, "step": 4539 }, { "epoch": 0.62, "grad_norm": 3.883776644860052, "learning_rate": 8.096983368253251e-06, "loss": 0.2462, "step": 4540 }, { "epoch": 0.62, "grad_norm": 3.520358953582849, "learning_rate": 8.096118497918544e-06, "loss": 0.3232, "step": 4541 }, { "epoch": 0.62, "grad_norm": 4.1869209346485965, "learning_rate": 8.095253477311765e-06, "loss": 0.2676, "step": 4542 }, { "epoch": 0.62, "grad_norm": 3.5536718391948976, "learning_rate": 8.094388306474903e-06, "loss": 0.235, "step": 4543 }, { "epoch": 0.62, "grad_norm": 2.9393598971345365, "learning_rate": 8.093522985449948e-06, "loss": 0.247, "step": 4544 }, { "epoch": 0.62, "grad_norm": 3.3693947206785104, "learning_rate": 8.092657514278899e-06, "loss": 0.2616, "step": 4545 }, { "epoch": 0.62, "grad_norm": 3.2281446683797643, "learning_rate": 8.09179189300376e-06, "loss": 0.2907, "step": 4546 }, { "epoch": 0.62, "grad_norm": 3.312854346556017, "learning_rate": 8.090926121666547e-06, "loss": 0.2393, "step": 4547 }, { "epoch": 0.62, "grad_norm": 3.4074448871231144, "learning_rate": 8.090060200309282e-06, "loss": 0.2295, "step": 4548 }, { "epoch": 0.62, "grad_norm": 3.058062203435577, "learning_rate": 8.089194128973989e-06, "loss": 0.2202, "step": 4549 }, { "epoch": 0.62, "grad_norm": 3.5725208731421323, "learning_rate": 8.088327907702706e-06, "loss": 0.2872, "step": 4550 }, { "epoch": 0.62, "grad_norm": 3.034006509516386, "learning_rate": 8.087461536537472e-06, "loss": 0.2108, "step": 4551 }, { "epoch": 0.62, "grad_norm": 3.620864088680639, "learning_rate": 8.086595015520345e-06, "loss": 0.2402, "step": 4552 }, { "epoch": 0.62, "grad_norm": 4.473705479816099, "learning_rate": 8.085728344693374e-06, "loss": 0.2733, "step": 4553 }, { "epoch": 0.62, "grad_norm": 3.9637945000705788, "learning_rate": 8.084861524098627e-06, "loss": 0.1896, "step": 4554 }, { "epoch": 0.62, "grad_norm": 4.307849183640954, "learning_rate": 8.083994553778175e-06, "loss": 0.2809, "step": 4555 }, { "epoch": 0.62, "grad_norm": 3.4230250037896393, "learning_rate": 8.083127433774096e-06, "loss": 0.2367, "step": 4556 }, { "epoch": 0.62, "grad_norm": 3.2797282084833412, "learning_rate": 8.08226016412848e-06, "loss": 0.2325, "step": 4557 }, { "epoch": 0.62, "grad_norm": 3.839943471887736, "learning_rate": 8.081392744883415e-06, "loss": 0.2743, "step": 4558 }, { "epoch": 0.62, "grad_norm": 3.5787479582944277, "learning_rate": 8.080525176081006e-06, "loss": 0.2405, "step": 4559 }, { "epoch": 0.62, "grad_norm": 3.3756101758673367, "learning_rate": 8.07965745776336e-06, "loss": 0.2886, "step": 4560 }, { "epoch": 0.62, "grad_norm": 3.410311802140068, "learning_rate": 8.07878958997259e-06, "loss": 0.2099, "step": 4561 }, { "epoch": 0.62, "grad_norm": 3.808257271318912, "learning_rate": 8.07792157275082e-06, "loss": 0.2682, "step": 4562 }, { "epoch": 0.62, "grad_norm": 4.129797100175062, "learning_rate": 8.077053406140181e-06, "loss": 0.2466, "step": 4563 }, { "epoch": 0.62, "grad_norm": 3.6966391989527616, "learning_rate": 8.076185090182808e-06, "loss": 0.234, "step": 4564 }, { "epoch": 0.62, "grad_norm": 3.443350929150025, "learning_rate": 8.075316624920848e-06, "loss": 0.2413, "step": 4565 }, { "epoch": 0.62, "grad_norm": 3.3897401191078766, "learning_rate": 8.07444801039645e-06, "loss": 0.2831, "step": 4566 }, { "epoch": 0.62, "grad_norm": 3.818183646429799, "learning_rate": 8.073579246651775e-06, "loss": 0.2751, "step": 4567 }, { "epoch": 0.62, "grad_norm": 3.6098652471017605, "learning_rate": 8.072710333728986e-06, "loss": 0.2884, "step": 4568 }, { "epoch": 0.62, "grad_norm": 3.335780217638707, "learning_rate": 8.071841271670258e-06, "loss": 0.2616, "step": 4569 }, { "epoch": 0.62, "grad_norm": 3.5635607404757743, "learning_rate": 8.070972060517771e-06, "loss": 0.2513, "step": 4570 }, { "epoch": 0.62, "grad_norm": 3.152724649100599, "learning_rate": 8.070102700313713e-06, "loss": 0.2445, "step": 4571 }, { "epoch": 0.62, "grad_norm": 4.294892459699333, "learning_rate": 8.069233191100278e-06, "loss": 0.2905, "step": 4572 }, { "epoch": 0.62, "grad_norm": 4.036151520597074, "learning_rate": 8.068363532919672e-06, "loss": 0.2817, "step": 4573 }, { "epoch": 0.62, "grad_norm": 3.838152722403744, "learning_rate": 8.067493725814099e-06, "loss": 0.2722, "step": 4574 }, { "epoch": 0.62, "grad_norm": 3.7000030457424007, "learning_rate": 8.066623769825778e-06, "loss": 0.2883, "step": 4575 }, { "epoch": 0.62, "grad_norm": 3.342252220449437, "learning_rate": 8.065753664996935e-06, "loss": 0.2355, "step": 4576 }, { "epoch": 0.62, "grad_norm": 3.5407270165490354, "learning_rate": 8.064883411369799e-06, "loss": 0.2908, "step": 4577 }, { "epoch": 0.62, "grad_norm": 3.6121082992397566, "learning_rate": 8.064013008986606e-06, "loss": 0.2734, "step": 4578 }, { "epoch": 0.62, "grad_norm": 3.817078852247829, "learning_rate": 8.063142457889607e-06, "loss": 0.2502, "step": 4579 }, { "epoch": 0.62, "grad_norm": 3.63691881221396, "learning_rate": 8.06227175812105e-06, "loss": 0.2023, "step": 4580 }, { "epoch": 0.62, "grad_norm": 3.64769345844629, "learning_rate": 8.061400909723198e-06, "loss": 0.2154, "step": 4581 }, { "epoch": 0.62, "grad_norm": 3.595531058232526, "learning_rate": 8.060529912738316e-06, "loss": 0.2572, "step": 4582 }, { "epoch": 0.62, "grad_norm": 3.2789244914155917, "learning_rate": 8.059658767208678e-06, "loss": 0.2746, "step": 4583 }, { "epoch": 0.62, "grad_norm": 3.6232160939044804, "learning_rate": 8.058787473176568e-06, "loss": 0.2717, "step": 4584 }, { "epoch": 0.62, "grad_norm": 3.56859151302084, "learning_rate": 8.057916030684273e-06, "loss": 0.2112, "step": 4585 }, { "epoch": 0.62, "grad_norm": 3.5616134236396184, "learning_rate": 8.05704443977409e-06, "loss": 0.2603, "step": 4586 }, { "epoch": 0.62, "grad_norm": 3.549484723567333, "learning_rate": 8.056172700488324e-06, "loss": 0.3051, "step": 4587 }, { "epoch": 0.62, "grad_norm": 4.285196027864316, "learning_rate": 8.05530081286928e-06, "loss": 0.2248, "step": 4588 }, { "epoch": 0.62, "grad_norm": 2.9332926237372234, "learning_rate": 8.05442877695928e-06, "loss": 0.2489, "step": 4589 }, { "epoch": 0.62, "grad_norm": 3.3470382652157733, "learning_rate": 8.053556592800648e-06, "loss": 0.2556, "step": 4590 }, { "epoch": 0.62, "grad_norm": 3.6983273957969525, "learning_rate": 8.052684260435716e-06, "loss": 0.2359, "step": 4591 }, { "epoch": 0.62, "grad_norm": 3.763527645104573, "learning_rate": 8.051811779906823e-06, "loss": 0.2582, "step": 4592 }, { "epoch": 0.62, "grad_norm": 3.3881945910900404, "learning_rate": 8.050939151256314e-06, "loss": 0.2297, "step": 4593 }, { "epoch": 0.62, "grad_norm": 3.115886052459789, "learning_rate": 8.050066374526546e-06, "loss": 0.2325, "step": 4594 }, { "epoch": 0.63, "grad_norm": 3.1503146599578122, "learning_rate": 8.049193449759875e-06, "loss": 0.2309, "step": 4595 }, { "epoch": 0.63, "grad_norm": 3.6259985696283654, "learning_rate": 8.048320376998675e-06, "loss": 0.2943, "step": 4596 }, { "epoch": 0.63, "grad_norm": 3.0528340424538105, "learning_rate": 8.047447156285314e-06, "loss": 0.2495, "step": 4597 }, { "epoch": 0.63, "grad_norm": 4.005013023825806, "learning_rate": 8.04657378766218e-06, "loss": 0.2287, "step": 4598 }, { "epoch": 0.63, "grad_norm": 4.755484476156937, "learning_rate": 8.045700271171662e-06, "loss": 0.3184, "step": 4599 }, { "epoch": 0.63, "grad_norm": 3.564820546956161, "learning_rate": 8.044826606856153e-06, "loss": 0.3071, "step": 4600 }, { "epoch": 0.63, "grad_norm": 3.6234185757036554, "learning_rate": 8.04395279475806e-06, "loss": 0.2924, "step": 4601 }, { "epoch": 0.63, "grad_norm": 3.380740278752955, "learning_rate": 8.043078834919792e-06, "loss": 0.2658, "step": 4602 }, { "epoch": 0.63, "grad_norm": 4.3151276013640585, "learning_rate": 8.042204727383772e-06, "loss": 0.2582, "step": 4603 }, { "epoch": 0.63, "grad_norm": 3.285559753252493, "learning_rate": 8.041330472192417e-06, "loss": 0.2563, "step": 4604 }, { "epoch": 0.63, "grad_norm": 3.786856604305315, "learning_rate": 8.040456069388167e-06, "loss": 0.3021, "step": 4605 }, { "epoch": 0.63, "grad_norm": 3.5977355743230146, "learning_rate": 8.039581519013458e-06, "loss": 0.2945, "step": 4606 }, { "epoch": 0.63, "grad_norm": 3.2677549761114824, "learning_rate": 8.038706821110738e-06, "loss": 0.2445, "step": 4607 }, { "epoch": 0.63, "grad_norm": 3.342088243794068, "learning_rate": 8.03783197572246e-06, "loss": 0.2262, "step": 4608 }, { "epoch": 0.63, "grad_norm": 3.942367276763178, "learning_rate": 8.036956982891088e-06, "loss": 0.247, "step": 4609 }, { "epoch": 0.63, "grad_norm": 2.878331062926818, "learning_rate": 8.036081842659087e-06, "loss": 0.2243, "step": 4610 }, { "epoch": 0.63, "grad_norm": 3.2228104409520317, "learning_rate": 8.035206555068932e-06, "loss": 0.2418, "step": 4611 }, { "epoch": 0.63, "grad_norm": 3.305688333594395, "learning_rate": 8.03433112016311e-06, "loss": 0.2618, "step": 4612 }, { "epoch": 0.63, "grad_norm": 3.3370688178520487, "learning_rate": 8.033455537984107e-06, "loss": 0.2768, "step": 4613 }, { "epoch": 0.63, "grad_norm": 3.598544477764351, "learning_rate": 8.032579808574423e-06, "loss": 0.2967, "step": 4614 }, { "epoch": 0.63, "grad_norm": 3.401518898957322, "learning_rate": 8.031703931976558e-06, "loss": 0.2667, "step": 4615 }, { "epoch": 0.63, "grad_norm": 3.1018492725125664, "learning_rate": 8.030827908233026e-06, "loss": 0.27, "step": 4616 }, { "epoch": 0.63, "grad_norm": 3.48694993775802, "learning_rate": 8.029951737386345e-06, "loss": 0.262, "step": 4617 }, { "epoch": 0.63, "grad_norm": 3.4347137220869275, "learning_rate": 8.029075419479041e-06, "loss": 0.2305, "step": 4618 }, { "epoch": 0.63, "grad_norm": 3.6553320872038544, "learning_rate": 8.028198954553646e-06, "loss": 0.2357, "step": 4619 }, { "epoch": 0.63, "grad_norm": 4.074818477461548, "learning_rate": 8.027322342652698e-06, "loss": 0.3422, "step": 4620 }, { "epoch": 0.63, "grad_norm": 4.497925245301903, "learning_rate": 8.026445583818748e-06, "loss": 0.2579, "step": 4621 }, { "epoch": 0.63, "grad_norm": 4.088007745787896, "learning_rate": 8.025568678094346e-06, "loss": 0.2698, "step": 4622 }, { "epoch": 0.63, "grad_norm": 4.4691166715539055, "learning_rate": 8.024691625522054e-06, "loss": 0.266, "step": 4623 }, { "epoch": 0.63, "grad_norm": 3.431851935217984, "learning_rate": 8.023814426144443e-06, "loss": 0.2672, "step": 4624 }, { "epoch": 0.63, "grad_norm": 3.408603604251559, "learning_rate": 8.022937080004085e-06, "loss": 0.286, "step": 4625 }, { "epoch": 0.63, "grad_norm": 3.1651589618327725, "learning_rate": 8.022059587143568e-06, "loss": 0.2103, "step": 4626 }, { "epoch": 0.63, "grad_norm": 4.880984370980163, "learning_rate": 8.021181947605474e-06, "loss": 0.2474, "step": 4627 }, { "epoch": 0.63, "grad_norm": 3.749166650987808, "learning_rate": 8.020304161432404e-06, "loss": 0.2552, "step": 4628 }, { "epoch": 0.63, "grad_norm": 5.1181995059136876, "learning_rate": 8.019426228666962e-06, "loss": 0.2596, "step": 4629 }, { "epoch": 0.63, "grad_norm": 4.007993948590178, "learning_rate": 8.018548149351759e-06, "loss": 0.2823, "step": 4630 }, { "epoch": 0.63, "grad_norm": 3.368598470256399, "learning_rate": 8.017669923529412e-06, "loss": 0.2759, "step": 4631 }, { "epoch": 0.63, "grad_norm": 3.58878651841689, "learning_rate": 8.016791551242548e-06, "loss": 0.2629, "step": 4632 }, { "epoch": 0.63, "grad_norm": 4.159789069150756, "learning_rate": 8.015913032533798e-06, "loss": 0.2365, "step": 4633 }, { "epoch": 0.63, "grad_norm": 2.889950693388996, "learning_rate": 8.0150343674458e-06, "loss": 0.2302, "step": 4634 }, { "epoch": 0.63, "grad_norm": 3.2559560358623494, "learning_rate": 8.014155556021204e-06, "loss": 0.2668, "step": 4635 }, { "epoch": 0.63, "grad_norm": 3.518611322380662, "learning_rate": 8.013276598302662e-06, "loss": 0.2546, "step": 4636 }, { "epoch": 0.63, "grad_norm": 2.734175770453403, "learning_rate": 8.012397494332832e-06, "loss": 0.2562, "step": 4637 }, { "epoch": 0.63, "grad_norm": 3.687506756563826, "learning_rate": 8.011518244154388e-06, "loss": 0.2096, "step": 4638 }, { "epoch": 0.63, "grad_norm": 2.7518051131986847, "learning_rate": 8.01063884781e-06, "loss": 0.2186, "step": 4639 }, { "epoch": 0.63, "grad_norm": 4.356006342875004, "learning_rate": 8.009759305342352e-06, "loss": 0.2863, "step": 4640 }, { "epoch": 0.63, "grad_norm": 2.893556282248555, "learning_rate": 8.008879616794133e-06, "loss": 0.232, "step": 4641 }, { "epoch": 0.63, "grad_norm": 3.029162295085205, "learning_rate": 8.00799978220804e-06, "loss": 0.2012, "step": 4642 }, { "epoch": 0.63, "grad_norm": 3.2272888330015053, "learning_rate": 8.007119801626772e-06, "loss": 0.2457, "step": 4643 }, { "epoch": 0.63, "grad_norm": 3.2850449047257104, "learning_rate": 8.006239675093046e-06, "loss": 0.2235, "step": 4644 }, { "epoch": 0.63, "grad_norm": 3.8656638524669886, "learning_rate": 8.005359402649576e-06, "loss": 0.2742, "step": 4645 }, { "epoch": 0.63, "grad_norm": 3.8398727161286526, "learning_rate": 8.004478984339086e-06, "loss": 0.2898, "step": 4646 }, { "epoch": 0.63, "grad_norm": 3.191728063140781, "learning_rate": 8.003598420204307e-06, "loss": 0.2354, "step": 4647 }, { "epoch": 0.63, "grad_norm": 3.8147766598017787, "learning_rate": 8.002717710287982e-06, "loss": 0.31, "step": 4648 }, { "epoch": 0.63, "grad_norm": 4.439001477553326, "learning_rate": 8.001836854632852e-06, "loss": 0.275, "step": 4649 }, { "epoch": 0.63, "grad_norm": 4.179682076652388, "learning_rate": 8.000955853281672e-06, "loss": 0.224, "step": 4650 }, { "epoch": 0.63, "grad_norm": 3.522047889725096, "learning_rate": 8.000074706277204e-06, "loss": 0.2703, "step": 4651 }, { "epoch": 0.63, "grad_norm": 3.6268302237595362, "learning_rate": 7.99919341366221e-06, "loss": 0.3147, "step": 4652 }, { "epoch": 0.63, "grad_norm": 4.439719671778674, "learning_rate": 7.998311975479469e-06, "loss": 0.2661, "step": 4653 }, { "epoch": 0.63, "grad_norm": 3.599425163901322, "learning_rate": 7.997430391771759e-06, "loss": 0.2354, "step": 4654 }, { "epoch": 0.63, "grad_norm": 3.6307868650974364, "learning_rate": 7.99654866258187e-06, "loss": 0.2732, "step": 4655 }, { "epoch": 0.63, "grad_norm": 3.7166824196930155, "learning_rate": 7.995666787952596e-06, "loss": 0.2639, "step": 4656 }, { "epoch": 0.63, "grad_norm": 3.7051760925097033, "learning_rate": 7.994784767926743e-06, "loss": 0.3106, "step": 4657 }, { "epoch": 0.63, "grad_norm": 4.614291423241149, "learning_rate": 7.993902602547113e-06, "loss": 0.261, "step": 4658 }, { "epoch": 0.63, "grad_norm": 3.1451297518585504, "learning_rate": 7.99302029185653e-06, "loss": 0.2196, "step": 4659 }, { "epoch": 0.63, "grad_norm": 3.6902925809548695, "learning_rate": 7.992137835897814e-06, "loss": 0.2602, "step": 4660 }, { "epoch": 0.63, "grad_norm": 4.402483125779187, "learning_rate": 7.991255234713795e-06, "loss": 0.3225, "step": 4661 }, { "epoch": 0.63, "grad_norm": 2.9909630993203598, "learning_rate": 7.99037248834731e-06, "loss": 0.2664, "step": 4662 }, { "epoch": 0.63, "grad_norm": 3.3734233398145483, "learning_rate": 7.989489596841207e-06, "loss": 0.2452, "step": 4663 }, { "epoch": 0.63, "grad_norm": 4.115470659398125, "learning_rate": 7.988606560238337e-06, "loss": 0.3547, "step": 4664 }, { "epoch": 0.63, "grad_norm": 3.7974025232268134, "learning_rate": 7.987723378581556e-06, "loss": 0.2118, "step": 4665 }, { "epoch": 0.63, "grad_norm": 3.684295080622686, "learning_rate": 7.986840051913732e-06, "loss": 0.2323, "step": 4666 }, { "epoch": 0.63, "grad_norm": 3.437367741711112, "learning_rate": 7.985956580277738e-06, "loss": 0.2508, "step": 4667 }, { "epoch": 0.63, "grad_norm": 3.3464472073371794, "learning_rate": 7.985072963716451e-06, "loss": 0.255, "step": 4668 }, { "epoch": 0.64, "grad_norm": 3.0829987736043565, "learning_rate": 7.984189202272763e-06, "loss": 0.2541, "step": 4669 }, { "epoch": 0.64, "grad_norm": 4.117097667430533, "learning_rate": 7.983305295989562e-06, "loss": 0.3446, "step": 4670 }, { "epoch": 0.64, "grad_norm": 2.83343656681648, "learning_rate": 7.982421244909752e-06, "loss": 0.1598, "step": 4671 }, { "epoch": 0.64, "grad_norm": 8.437142496084602, "learning_rate": 7.981537049076243e-06, "loss": 0.2442, "step": 4672 }, { "epoch": 0.64, "grad_norm": 3.7902054397326728, "learning_rate": 7.980652708531946e-06, "loss": 0.2813, "step": 4673 }, { "epoch": 0.64, "grad_norm": 3.7757661220861825, "learning_rate": 7.979768223319786e-06, "loss": 0.2821, "step": 4674 }, { "epoch": 0.64, "grad_norm": 3.5749809807888546, "learning_rate": 7.97888359348269e-06, "loss": 0.2858, "step": 4675 }, { "epoch": 0.64, "grad_norm": 3.587125821378373, "learning_rate": 7.977998819063596e-06, "loss": 0.2964, "step": 4676 }, { "epoch": 0.64, "grad_norm": 3.048257979202859, "learning_rate": 7.977113900105444e-06, "loss": 0.2082, "step": 4677 }, { "epoch": 0.64, "grad_norm": 3.6035008550090435, "learning_rate": 7.976228836651188e-06, "loss": 0.2151, "step": 4678 }, { "epoch": 0.64, "grad_norm": 4.034153371883403, "learning_rate": 7.975343628743784e-06, "loss": 0.2987, "step": 4679 }, { "epoch": 0.64, "grad_norm": 3.497141514560119, "learning_rate": 7.974458276426193e-06, "loss": 0.2311, "step": 4680 }, { "epoch": 0.64, "grad_norm": 3.5699870830788005, "learning_rate": 7.97357277974139e-06, "loss": 0.2631, "step": 4681 }, { "epoch": 0.64, "grad_norm": 3.9218335929939494, "learning_rate": 7.972687138732352e-06, "loss": 0.28, "step": 4682 }, { "epoch": 0.64, "grad_norm": 3.35829234664875, "learning_rate": 7.971801353442063e-06, "loss": 0.2339, "step": 4683 }, { "epoch": 0.64, "grad_norm": 3.644633666262191, "learning_rate": 7.970915423913516e-06, "loss": 0.2652, "step": 4684 }, { "epoch": 0.64, "grad_norm": 3.6583332568110754, "learning_rate": 7.97002935018971e-06, "loss": 0.2186, "step": 4685 }, { "epoch": 0.64, "grad_norm": 4.022413063084137, "learning_rate": 7.969143132313652e-06, "loss": 0.2459, "step": 4686 }, { "epoch": 0.64, "grad_norm": 3.5362878830889337, "learning_rate": 7.968256770328353e-06, "loss": 0.2844, "step": 4687 }, { "epoch": 0.64, "grad_norm": 4.0543017260945025, "learning_rate": 7.967370264276836e-06, "loss": 0.2782, "step": 4688 }, { "epoch": 0.64, "grad_norm": 3.082820646211465, "learning_rate": 7.966483614202127e-06, "loss": 0.2248, "step": 4689 }, { "epoch": 0.64, "grad_norm": 3.0393420195789496, "learning_rate": 7.965596820147259e-06, "loss": 0.2544, "step": 4690 }, { "epoch": 0.64, "grad_norm": 3.8061514183156846, "learning_rate": 7.964709882155273e-06, "loss": 0.298, "step": 4691 }, { "epoch": 0.64, "grad_norm": 3.6948811485252038, "learning_rate": 7.96382280026922e-06, "loss": 0.3083, "step": 4692 }, { "epoch": 0.64, "grad_norm": 3.425013684118277, "learning_rate": 7.962935574532152e-06, "loss": 0.2598, "step": 4693 }, { "epoch": 0.64, "grad_norm": 3.9027658765222757, "learning_rate": 7.962048204987132e-06, "loss": 0.2318, "step": 4694 }, { "epoch": 0.64, "grad_norm": 4.319039377200388, "learning_rate": 7.96116069167723e-06, "loss": 0.2745, "step": 4695 }, { "epoch": 0.64, "grad_norm": 4.353329485211722, "learning_rate": 7.96027303464552e-06, "loss": 0.2719, "step": 4696 }, { "epoch": 0.64, "grad_norm": 3.8071968311302022, "learning_rate": 7.959385233935087e-06, "loss": 0.2514, "step": 4697 }, { "epoch": 0.64, "grad_norm": 3.420136702320759, "learning_rate": 7.95849728958902e-06, "loss": 0.26, "step": 4698 }, { "epoch": 0.64, "grad_norm": 4.390101732719251, "learning_rate": 7.957609201650416e-06, "loss": 0.3289, "step": 4699 }, { "epoch": 0.64, "grad_norm": 3.3974345058562303, "learning_rate": 7.956720970162379e-06, "loss": 0.2894, "step": 4700 }, { "epoch": 0.64, "grad_norm": 3.6154319998794935, "learning_rate": 7.95583259516802e-06, "loss": 0.2328, "step": 4701 }, { "epoch": 0.64, "grad_norm": 3.8967372356988523, "learning_rate": 7.954944076710457e-06, "loss": 0.268, "step": 4702 }, { "epoch": 0.64, "grad_norm": 3.5990120229306473, "learning_rate": 7.954055414832815e-06, "loss": 0.2648, "step": 4703 }, { "epoch": 0.64, "grad_norm": 3.7001248097936434, "learning_rate": 7.953166609578225e-06, "loss": 0.2695, "step": 4704 }, { "epoch": 0.64, "grad_norm": 4.234904163194293, "learning_rate": 7.952277660989825e-06, "loss": 0.255, "step": 4705 }, { "epoch": 0.64, "grad_norm": 3.953453419510655, "learning_rate": 7.951388569110763e-06, "loss": 0.237, "step": 4706 }, { "epoch": 0.64, "grad_norm": 3.755383016973387, "learning_rate": 7.95049933398419e-06, "loss": 0.2455, "step": 4707 }, { "epoch": 0.64, "grad_norm": 3.7615645305039216, "learning_rate": 7.949609955653266e-06, "loss": 0.2572, "step": 4708 }, { "epoch": 0.64, "grad_norm": 3.4370093055015043, "learning_rate": 7.948720434161157e-06, "loss": 0.2623, "step": 4709 }, { "epoch": 0.64, "grad_norm": 3.2619289854173172, "learning_rate": 7.947830769551038e-06, "loss": 0.2641, "step": 4710 }, { "epoch": 0.64, "grad_norm": 3.530121571317201, "learning_rate": 7.946940961866088e-06, "loss": 0.2813, "step": 4711 }, { "epoch": 0.64, "grad_norm": 3.587518056950012, "learning_rate": 7.946051011149494e-06, "loss": 0.2457, "step": 4712 }, { "epoch": 0.64, "grad_norm": 4.322482165477575, "learning_rate": 7.945160917444452e-06, "loss": 0.2846, "step": 4713 }, { "epoch": 0.64, "grad_norm": 3.56411687350066, "learning_rate": 7.944270680794162e-06, "loss": 0.2378, "step": 4714 }, { "epoch": 0.64, "grad_norm": 3.6518767368130254, "learning_rate": 7.943380301241833e-06, "loss": 0.2733, "step": 4715 }, { "epoch": 0.64, "grad_norm": 3.853933424918806, "learning_rate": 7.942489778830678e-06, "loss": 0.3229, "step": 4716 }, { "epoch": 0.64, "grad_norm": 3.521763992829803, "learning_rate": 7.941599113603923e-06, "loss": 0.237, "step": 4717 }, { "epoch": 0.64, "grad_norm": 3.4754456174880777, "learning_rate": 7.940708305604793e-06, "loss": 0.2641, "step": 4718 }, { "epoch": 0.64, "grad_norm": 3.3556683023860243, "learning_rate": 7.939817354876528e-06, "loss": 0.2394, "step": 4719 }, { "epoch": 0.64, "grad_norm": 3.1513747297118053, "learning_rate": 7.938926261462366e-06, "loss": 0.2384, "step": 4720 }, { "epoch": 0.64, "grad_norm": 3.4334088226067614, "learning_rate": 7.938035025405562e-06, "loss": 0.2838, "step": 4721 }, { "epoch": 0.64, "grad_norm": 3.094100532330984, "learning_rate": 7.937143646749367e-06, "loss": 0.2264, "step": 4722 }, { "epoch": 0.64, "grad_norm": 3.340964746374121, "learning_rate": 7.93625212553705e-06, "loss": 0.2612, "step": 4723 }, { "epoch": 0.64, "grad_norm": 3.0696249410694247, "learning_rate": 7.93536046181188e-06, "loss": 0.1863, "step": 4724 }, { "epoch": 0.64, "grad_norm": 3.1975752737193655, "learning_rate": 7.93446865561713e-06, "loss": 0.2753, "step": 4725 }, { "epoch": 0.64, "grad_norm": 3.760738154503221, "learning_rate": 7.933576706996091e-06, "loss": 0.3444, "step": 4726 }, { "epoch": 0.64, "grad_norm": 3.5783378989535155, "learning_rate": 7.93268461599205e-06, "loss": 0.2045, "step": 4727 }, { "epoch": 0.64, "grad_norm": 3.4868369940604875, "learning_rate": 7.931792382648307e-06, "loss": 0.2866, "step": 4728 }, { "epoch": 0.64, "grad_norm": 3.569863104355557, "learning_rate": 7.930900007008168e-06, "loss": 0.1931, "step": 4729 }, { "epoch": 0.64, "grad_norm": 3.07469568509077, "learning_rate": 7.930007489114941e-06, "loss": 0.2264, "step": 4730 }, { "epoch": 0.64, "grad_norm": 3.6218912838154034, "learning_rate": 7.92911482901195e-06, "loss": 0.2557, "step": 4731 }, { "epoch": 0.64, "grad_norm": 2.8796847204093514, "learning_rate": 7.928222026742517e-06, "loss": 0.2676, "step": 4732 }, { "epoch": 0.64, "grad_norm": 4.179214887379512, "learning_rate": 7.927329082349977e-06, "loss": 0.3408, "step": 4733 }, { "epoch": 0.64, "grad_norm": 3.5384666075837288, "learning_rate": 7.926435995877668e-06, "loss": 0.3022, "step": 4734 }, { "epoch": 0.64, "grad_norm": 3.6184026902650688, "learning_rate": 7.925542767368938e-06, "loss": 0.342, "step": 4735 }, { "epoch": 0.64, "grad_norm": 3.9927048218110985, "learning_rate": 7.92464939686714e-06, "loss": 0.293, "step": 4736 }, { "epoch": 0.64, "grad_norm": 4.224268637243144, "learning_rate": 7.923755884415634e-06, "loss": 0.2724, "step": 4737 }, { "epoch": 0.64, "grad_norm": 2.9867858840966046, "learning_rate": 7.922862230057787e-06, "loss": 0.2135, "step": 4738 }, { "epoch": 0.64, "grad_norm": 3.138145232682331, "learning_rate": 7.921968433836973e-06, "loss": 0.2026, "step": 4739 }, { "epoch": 0.64, "grad_norm": 3.3563335037533997, "learning_rate": 7.921074495796572e-06, "loss": 0.247, "step": 4740 }, { "epoch": 0.64, "grad_norm": 3.6680887971801055, "learning_rate": 7.920180415979974e-06, "loss": 0.2694, "step": 4741 }, { "epoch": 0.65, "grad_norm": 3.8182601448225393, "learning_rate": 7.919286194430573e-06, "loss": 0.2782, "step": 4742 }, { "epoch": 0.65, "grad_norm": 3.0334768172640576, "learning_rate": 7.918391831191772e-06, "loss": 0.2399, "step": 4743 }, { "epoch": 0.65, "grad_norm": 3.3159552753540007, "learning_rate": 7.917497326306975e-06, "loss": 0.2568, "step": 4744 }, { "epoch": 0.65, "grad_norm": 3.2263741707177016, "learning_rate": 7.916602679819602e-06, "loss": 0.2289, "step": 4745 }, { "epoch": 0.65, "grad_norm": 3.000567130991973, "learning_rate": 7.915707891773074e-06, "loss": 0.2263, "step": 4746 }, { "epoch": 0.65, "grad_norm": 3.7526800888609246, "learning_rate": 7.914812962210819e-06, "loss": 0.286, "step": 4747 }, { "epoch": 0.65, "grad_norm": 2.7562786949297036, "learning_rate": 7.913917891176273e-06, "loss": 0.2114, "step": 4748 }, { "epoch": 0.65, "grad_norm": 3.659374356747787, "learning_rate": 7.91302267871288e-06, "loss": 0.2826, "step": 4749 }, { "epoch": 0.65, "grad_norm": 3.1838005304672707, "learning_rate": 7.912127324864088e-06, "loss": 0.2369, "step": 4750 }, { "epoch": 0.65, "grad_norm": 3.040402399855172, "learning_rate": 7.911231829673356e-06, "loss": 0.23, "step": 4751 }, { "epoch": 0.65, "grad_norm": 3.072895141929435, "learning_rate": 7.910336193184146e-06, "loss": 0.1716, "step": 4752 }, { "epoch": 0.65, "grad_norm": 3.3325966013253248, "learning_rate": 7.909440415439928e-06, "loss": 0.2463, "step": 4753 }, { "epoch": 0.65, "grad_norm": 2.9090509568890712, "learning_rate": 7.90854449648418e-06, "loss": 0.2265, "step": 4754 }, { "epoch": 0.65, "grad_norm": 3.665614422652213, "learning_rate": 7.907648436360385e-06, "loss": 0.2699, "step": 4755 }, { "epoch": 0.65, "grad_norm": 3.236252164658972, "learning_rate": 7.906752235112034e-06, "loss": 0.2772, "step": 4756 }, { "epoch": 0.65, "grad_norm": 3.682024361823986, "learning_rate": 7.905855892782625e-06, "loss": 0.2409, "step": 4757 }, { "epoch": 0.65, "grad_norm": 3.6543952953070376, "learning_rate": 7.904959409415665e-06, "loss": 0.2339, "step": 4758 }, { "epoch": 0.65, "grad_norm": 3.4114773487231926, "learning_rate": 7.904062785054661e-06, "loss": 0.2356, "step": 4759 }, { "epoch": 0.65, "grad_norm": 4.1670150588997315, "learning_rate": 7.903166019743133e-06, "loss": 0.2871, "step": 4760 }, { "epoch": 0.65, "grad_norm": 3.5349051089251518, "learning_rate": 7.902269113524606e-06, "loss": 0.2648, "step": 4761 }, { "epoch": 0.65, "grad_norm": 3.6823743355188623, "learning_rate": 7.901372066442615e-06, "loss": 0.2275, "step": 4762 }, { "epoch": 0.65, "grad_norm": 3.7724182373533415, "learning_rate": 7.900474878540691e-06, "loss": 0.318, "step": 4763 }, { "epoch": 0.65, "grad_norm": 3.361339752279313, "learning_rate": 7.899577549862388e-06, "loss": 0.2825, "step": 4764 }, { "epoch": 0.65, "grad_norm": 2.9663516218379313, "learning_rate": 7.898680080451255e-06, "loss": 0.2286, "step": 4765 }, { "epoch": 0.65, "grad_norm": 3.4466776484061623, "learning_rate": 7.89778247035085e-06, "loss": 0.2705, "step": 4766 }, { "epoch": 0.65, "grad_norm": 3.7495490163261507, "learning_rate": 7.89688471960474e-06, "loss": 0.2535, "step": 4767 }, { "epoch": 0.65, "grad_norm": 3.5627778756083863, "learning_rate": 7.8959868282565e-06, "loss": 0.2779, "step": 4768 }, { "epoch": 0.65, "grad_norm": 3.1689029121235635, "learning_rate": 7.895088796349706e-06, "loss": 0.2408, "step": 4769 }, { "epoch": 0.65, "grad_norm": 3.70589258806499, "learning_rate": 7.89419062392795e-06, "loss": 0.3036, "step": 4770 }, { "epoch": 0.65, "grad_norm": 3.6219230388814996, "learning_rate": 7.893292311034817e-06, "loss": 0.2292, "step": 4771 }, { "epoch": 0.65, "grad_norm": 3.725883910256554, "learning_rate": 7.892393857713914e-06, "loss": 0.2855, "step": 4772 }, { "epoch": 0.65, "grad_norm": 2.943946173851206, "learning_rate": 7.891495264008847e-06, "loss": 0.2503, "step": 4773 }, { "epoch": 0.65, "grad_norm": 3.5923938311675774, "learning_rate": 7.890596529963229e-06, "loss": 0.2952, "step": 4774 }, { "epoch": 0.65, "grad_norm": 3.6112243480435486, "learning_rate": 7.88969765562068e-06, "loss": 0.2343, "step": 4775 }, { "epoch": 0.65, "grad_norm": 2.970859850044112, "learning_rate": 7.888798641024825e-06, "loss": 0.2226, "step": 4776 }, { "epoch": 0.65, "grad_norm": 2.943082972695109, "learning_rate": 7.887899486219304e-06, "loss": 0.2579, "step": 4777 }, { "epoch": 0.65, "grad_norm": 3.57342738999483, "learning_rate": 7.887000191247753e-06, "loss": 0.2816, "step": 4778 }, { "epoch": 0.65, "grad_norm": 3.7020396573888688, "learning_rate": 7.886100756153826e-06, "loss": 0.2881, "step": 4779 }, { "epoch": 0.65, "grad_norm": 5.154644191983087, "learning_rate": 7.885201180981169e-06, "loss": 0.253, "step": 4780 }, { "epoch": 0.65, "grad_norm": 4.069592188175726, "learning_rate": 7.884301465773451e-06, "loss": 0.2515, "step": 4781 }, { "epoch": 0.65, "grad_norm": 3.6939807283609776, "learning_rate": 7.883401610574338e-06, "loss": 0.2774, "step": 4782 }, { "epoch": 0.65, "grad_norm": 3.46031248079321, "learning_rate": 7.882501615427501e-06, "loss": 0.2559, "step": 4783 }, { "epoch": 0.65, "grad_norm": 3.2892093977044192, "learning_rate": 7.881601480376627e-06, "loss": 0.2003, "step": 4784 }, { "epoch": 0.65, "grad_norm": 4.033034699192059, "learning_rate": 7.880701205465404e-06, "loss": 0.3373, "step": 4785 }, { "epoch": 0.65, "grad_norm": 3.7070436335774346, "learning_rate": 7.879800790737525e-06, "loss": 0.2758, "step": 4786 }, { "epoch": 0.65, "grad_norm": 3.306277516723827, "learning_rate": 7.878900236236693e-06, "loss": 0.2068, "step": 4787 }, { "epoch": 0.65, "grad_norm": 3.497384704238173, "learning_rate": 7.877999542006619e-06, "loss": 0.2538, "step": 4788 }, { "epoch": 0.65, "grad_norm": 3.1501963481080795, "learning_rate": 7.877098708091017e-06, "loss": 0.2756, "step": 4789 }, { "epoch": 0.65, "grad_norm": 3.1408041660813155, "learning_rate": 7.87619773453361e-06, "loss": 0.2235, "step": 4790 }, { "epoch": 0.65, "grad_norm": 3.6003512522929264, "learning_rate": 7.875296621378125e-06, "loss": 0.2247, "step": 4791 }, { "epoch": 0.65, "grad_norm": 3.872525409703566, "learning_rate": 7.874395368668302e-06, "loss": 0.2548, "step": 4792 }, { "epoch": 0.65, "grad_norm": 3.772929298142986, "learning_rate": 7.873493976447882e-06, "loss": 0.2467, "step": 4793 }, { "epoch": 0.65, "grad_norm": 4.231697372639225, "learning_rate": 7.872592444760613e-06, "loss": 0.3252, "step": 4794 }, { "epoch": 0.65, "grad_norm": 4.255240550931266, "learning_rate": 7.871690773650255e-06, "loss": 0.3175, "step": 4795 }, { "epoch": 0.65, "grad_norm": 3.2583708464958514, "learning_rate": 7.870788963160568e-06, "loss": 0.2173, "step": 4796 }, { "epoch": 0.65, "grad_norm": 3.6077790623668404, "learning_rate": 7.869887013335324e-06, "loss": 0.2763, "step": 4797 }, { "epoch": 0.65, "grad_norm": 3.8249874556103713, "learning_rate": 7.8689849242183e-06, "loss": 0.291, "step": 4798 }, { "epoch": 0.65, "grad_norm": 3.5240430505492704, "learning_rate": 7.868082695853276e-06, "loss": 0.3061, "step": 4799 }, { "epoch": 0.65, "grad_norm": 3.4507090842009873, "learning_rate": 7.867180328284044e-06, "loss": 0.2404, "step": 4800 }, { "epoch": 0.65, "grad_norm": 3.9856392458830894, "learning_rate": 7.866277821554405e-06, "loss": 0.3189, "step": 4801 }, { "epoch": 0.65, "grad_norm": 3.273965324738724, "learning_rate": 7.865375175708158e-06, "loss": 0.2349, "step": 4802 }, { "epoch": 0.65, "grad_norm": 3.6639293575209284, "learning_rate": 7.864472390789114e-06, "loss": 0.2687, "step": 4803 }, { "epoch": 0.65, "grad_norm": 4.704016268664282, "learning_rate": 7.863569466841091e-06, "loss": 0.2536, "step": 4804 }, { "epoch": 0.65, "grad_norm": 4.048056028512515, "learning_rate": 7.862666403907914e-06, "loss": 0.307, "step": 4805 }, { "epoch": 0.65, "grad_norm": 3.516473683868722, "learning_rate": 7.861763202033411e-06, "loss": 0.2362, "step": 4806 }, { "epoch": 0.65, "grad_norm": 3.6997918821250453, "learning_rate": 7.860859861261423e-06, "loss": 0.2664, "step": 4807 }, { "epoch": 0.65, "grad_norm": 4.079953967567098, "learning_rate": 7.85995638163579e-06, "loss": 0.2694, "step": 4808 }, { "epoch": 0.65, "grad_norm": 3.520769338806808, "learning_rate": 7.85905276320037e-06, "loss": 0.2219, "step": 4809 }, { "epoch": 0.65, "grad_norm": 3.6552245975963213, "learning_rate": 7.858149005999012e-06, "loss": 0.2671, "step": 4810 }, { "epoch": 0.65, "grad_norm": 4.3520135640037445, "learning_rate": 7.857245110075587e-06, "loss": 0.2203, "step": 4811 }, { "epoch": 0.65, "grad_norm": 3.5461866907239186, "learning_rate": 7.856341075473963e-06, "loss": 0.2605, "step": 4812 }, { "epoch": 0.65, "grad_norm": 4.828208921522304, "learning_rate": 7.855436902238018e-06, "loss": 0.3052, "step": 4813 }, { "epoch": 0.65, "grad_norm": 3.400642742102151, "learning_rate": 7.854532590411636e-06, "loss": 0.1962, "step": 4814 }, { "epoch": 0.65, "grad_norm": 4.26062586356901, "learning_rate": 7.85362814003871e-06, "loss": 0.3201, "step": 4815 }, { "epoch": 0.66, "grad_norm": 3.775478013582339, "learning_rate": 7.85272355116314e-06, "loss": 0.266, "step": 4816 }, { "epoch": 0.66, "grad_norm": 3.26084946565975, "learning_rate": 7.851818823828828e-06, "loss": 0.2689, "step": 4817 }, { "epoch": 0.66, "grad_norm": 3.614733644221338, "learning_rate": 7.850913958079685e-06, "loss": 0.2584, "step": 4818 }, { "epoch": 0.66, "grad_norm": 3.2678832781127585, "learning_rate": 7.85000895395963e-06, "loss": 0.2395, "step": 4819 }, { "epoch": 0.66, "grad_norm": 3.229352340442626, "learning_rate": 7.849103811512587e-06, "loss": 0.2543, "step": 4820 }, { "epoch": 0.66, "grad_norm": 3.302479901827721, "learning_rate": 7.84819853078249e-06, "loss": 0.2816, "step": 4821 }, { "epoch": 0.66, "grad_norm": 3.509828980110413, "learning_rate": 7.847293111813276e-06, "loss": 0.2236, "step": 4822 }, { "epoch": 0.66, "grad_norm": 3.0596940878488526, "learning_rate": 7.84638755464889e-06, "loss": 0.2263, "step": 4823 }, { "epoch": 0.66, "grad_norm": 3.373845366197361, "learning_rate": 7.845481859333283e-06, "loss": 0.2591, "step": 4824 }, { "epoch": 0.66, "grad_norm": 3.5388855547299944, "learning_rate": 7.844576025910418e-06, "loss": 0.2185, "step": 4825 }, { "epoch": 0.66, "grad_norm": 3.6330137751328135, "learning_rate": 7.843670054424254e-06, "loss": 0.2745, "step": 4826 }, { "epoch": 0.66, "grad_norm": 3.5024402421793095, "learning_rate": 7.842763944918766e-06, "loss": 0.2524, "step": 4827 }, { "epoch": 0.66, "grad_norm": 3.715580495765165, "learning_rate": 7.841857697437931e-06, "loss": 0.3244, "step": 4828 }, { "epoch": 0.66, "grad_norm": 3.361243705470568, "learning_rate": 7.84095131202574e-06, "loss": 0.2093, "step": 4829 }, { "epoch": 0.66, "grad_norm": 3.471427653500617, "learning_rate": 7.840044788726175e-06, "loss": 0.2223, "step": 4830 }, { "epoch": 0.66, "grad_norm": 3.851542675483228, "learning_rate": 7.839138127583243e-06, "loss": 0.3234, "step": 4831 }, { "epoch": 0.66, "grad_norm": 3.6518566870414433, "learning_rate": 7.838231328640945e-06, "loss": 0.2592, "step": 4832 }, { "epoch": 0.66, "grad_norm": 3.160161491064152, "learning_rate": 7.837324391943296e-06, "loss": 0.2328, "step": 4833 }, { "epoch": 0.66, "grad_norm": 3.687427133802324, "learning_rate": 7.836417317534314e-06, "loss": 0.2527, "step": 4834 }, { "epoch": 0.66, "grad_norm": 4.833773172241298, "learning_rate": 7.835510105458022e-06, "loss": 0.2685, "step": 4835 }, { "epoch": 0.66, "grad_norm": 4.388152409715583, "learning_rate": 7.834602755758454e-06, "loss": 0.2839, "step": 4836 }, { "epoch": 0.66, "grad_norm": 3.2904510222624106, "learning_rate": 7.83369526847965e-06, "loss": 0.2888, "step": 4837 }, { "epoch": 0.66, "grad_norm": 3.9952124970336182, "learning_rate": 7.832787643665654e-06, "loss": 0.2048, "step": 4838 }, { "epoch": 0.66, "grad_norm": 4.042552039569976, "learning_rate": 7.831879881360516e-06, "loss": 0.2625, "step": 4839 }, { "epoch": 0.66, "grad_norm": 5.011646160682478, "learning_rate": 7.8309719816083e-06, "loss": 0.3607, "step": 4840 }, { "epoch": 0.66, "grad_norm": 3.4862784160376834, "learning_rate": 7.830063944453066e-06, "loss": 0.224, "step": 4841 }, { "epoch": 0.66, "grad_norm": 3.3219302104935657, "learning_rate": 7.82915576993889e-06, "loss": 0.2388, "step": 4842 }, { "epoch": 0.66, "grad_norm": 3.217238061233746, "learning_rate": 7.82824745810985e-06, "loss": 0.2615, "step": 4843 }, { "epoch": 0.66, "grad_norm": 3.923551547640028, "learning_rate": 7.82733900901003e-06, "loss": 0.3287, "step": 4844 }, { "epoch": 0.66, "grad_norm": 3.2405586355246014, "learning_rate": 7.826430422683524e-06, "loss": 0.2714, "step": 4845 }, { "epoch": 0.66, "grad_norm": 3.352073109245812, "learning_rate": 7.82552169917443e-06, "loss": 0.2484, "step": 4846 }, { "epoch": 0.66, "grad_norm": 3.799035421481808, "learning_rate": 7.824612838526853e-06, "loss": 0.2332, "step": 4847 }, { "epoch": 0.66, "grad_norm": 4.119138320921092, "learning_rate": 7.823703840784906e-06, "loss": 0.2473, "step": 4848 }, { "epoch": 0.66, "grad_norm": 3.693693196580737, "learning_rate": 7.822794705992708e-06, "loss": 0.2938, "step": 4849 }, { "epoch": 0.66, "grad_norm": 3.1371267296078065, "learning_rate": 7.821885434194382e-06, "loss": 0.2176, "step": 4850 }, { "epoch": 0.66, "grad_norm": 3.218960432461163, "learning_rate": 7.820976025434064e-06, "loss": 0.3099, "step": 4851 }, { "epoch": 0.66, "grad_norm": 3.6686153669576815, "learning_rate": 7.82006647975589e-06, "loss": 0.2396, "step": 4852 }, { "epoch": 0.66, "grad_norm": 3.6747119284108924, "learning_rate": 7.819156797204006e-06, "loss": 0.2915, "step": 4853 }, { "epoch": 0.66, "grad_norm": 4.124296399442915, "learning_rate": 7.818246977822563e-06, "loss": 0.2726, "step": 4854 }, { "epoch": 0.66, "grad_norm": 3.552497483437483, "learning_rate": 7.817337021655722e-06, "loss": 0.3164, "step": 4855 }, { "epoch": 0.66, "grad_norm": 3.480151469858184, "learning_rate": 7.816426928747647e-06, "loss": 0.1974, "step": 4856 }, { "epoch": 0.66, "grad_norm": 3.568615818935624, "learning_rate": 7.81551669914251e-06, "loss": 0.2989, "step": 4857 }, { "epoch": 0.66, "grad_norm": 3.7433702230667514, "learning_rate": 7.81460633288449e-06, "loss": 0.2438, "step": 4858 }, { "epoch": 0.66, "grad_norm": 2.840529817522256, "learning_rate": 7.813695830017769e-06, "loss": 0.1803, "step": 4859 }, { "epoch": 0.66, "grad_norm": 3.8620390492712855, "learning_rate": 7.812785190586543e-06, "loss": 0.2834, "step": 4860 }, { "epoch": 0.66, "grad_norm": 3.907089457236755, "learning_rate": 7.81187441463501e-06, "loss": 0.3044, "step": 4861 }, { "epoch": 0.66, "grad_norm": 3.1802422069340035, "learning_rate": 7.810963502207373e-06, "loss": 0.2368, "step": 4862 }, { "epoch": 0.66, "grad_norm": 4.740379942637626, "learning_rate": 7.810052453347844e-06, "loss": 0.29, "step": 4863 }, { "epoch": 0.66, "grad_norm": 2.967699312620518, "learning_rate": 7.809141268100644e-06, "loss": 0.2158, "step": 4864 }, { "epoch": 0.66, "grad_norm": 3.7790209465719458, "learning_rate": 7.808229946509996e-06, "loss": 0.2871, "step": 4865 }, { "epoch": 0.66, "grad_norm": 4.02406388693073, "learning_rate": 7.80731848862013e-06, "loss": 0.3005, "step": 4866 }, { "epoch": 0.66, "grad_norm": 3.669401084120441, "learning_rate": 7.806406894475286e-06, "loss": 0.2195, "step": 4867 }, { "epoch": 0.66, "grad_norm": 3.5319154848492635, "learning_rate": 7.805495164119711e-06, "loss": 0.2716, "step": 4868 }, { "epoch": 0.66, "grad_norm": 3.784847249445409, "learning_rate": 7.804583297597653e-06, "loss": 0.2406, "step": 4869 }, { "epoch": 0.66, "grad_norm": 3.505323422811605, "learning_rate": 7.80367129495337e-06, "loss": 0.2318, "step": 4870 }, { "epoch": 0.66, "grad_norm": 3.6367120193158358, "learning_rate": 7.80275915623113e-06, "loss": 0.3171, "step": 4871 }, { "epoch": 0.66, "grad_norm": 3.46826206855173, "learning_rate": 7.801846881475199e-06, "loss": 0.2111, "step": 4872 }, { "epoch": 0.66, "grad_norm": 3.298426318509398, "learning_rate": 7.80093447072986e-06, "loss": 0.2724, "step": 4873 }, { "epoch": 0.66, "grad_norm": 3.2369375475925097, "learning_rate": 7.800021924039393e-06, "loss": 0.2501, "step": 4874 }, { "epoch": 0.66, "grad_norm": 4.492249612004869, "learning_rate": 7.799109241448091e-06, "loss": 0.2532, "step": 4875 }, { "epoch": 0.66, "grad_norm": 2.7629137013890936, "learning_rate": 7.798196423000255e-06, "loss": 0.2238, "step": 4876 }, { "epoch": 0.66, "grad_norm": 3.713112455485913, "learning_rate": 7.797283468740184e-06, "loss": 0.256, "step": 4877 }, { "epoch": 0.66, "grad_norm": 4.808962223375151, "learning_rate": 7.79637037871219e-06, "loss": 0.2599, "step": 4878 }, { "epoch": 0.66, "grad_norm": 3.725401941950786, "learning_rate": 7.795457152960592e-06, "loss": 0.2302, "step": 4879 }, { "epoch": 0.66, "grad_norm": 3.6027827053138344, "learning_rate": 7.794543791529715e-06, "loss": 0.2495, "step": 4880 }, { "epoch": 0.66, "grad_norm": 3.348592185298723, "learning_rate": 7.793630294463885e-06, "loss": 0.2642, "step": 4881 }, { "epoch": 0.66, "grad_norm": 4.362754881766281, "learning_rate": 7.792716661807443e-06, "loss": 0.2627, "step": 4882 }, { "epoch": 0.66, "grad_norm": 3.669357156905326, "learning_rate": 7.791802893604732e-06, "loss": 0.2741, "step": 4883 }, { "epoch": 0.66, "grad_norm": 3.040530718790551, "learning_rate": 7.790888989900103e-06, "loss": 0.2651, "step": 4884 }, { "epoch": 0.66, "grad_norm": 3.536219892561898, "learning_rate": 7.78997495073791e-06, "loss": 0.23, "step": 4885 }, { "epoch": 0.66, "grad_norm": 4.669213817646569, "learning_rate": 7.78906077616252e-06, "loss": 0.3112, "step": 4886 }, { "epoch": 0.66, "grad_norm": 3.276556094903754, "learning_rate": 7.788146466218301e-06, "loss": 0.244, "step": 4887 }, { "epoch": 0.66, "grad_norm": 3.187351766543437, "learning_rate": 7.787232020949632e-06, "loss": 0.246, "step": 4888 }, { "epoch": 0.67, "grad_norm": 3.206754270483333, "learning_rate": 7.78631744040089e-06, "loss": 0.2556, "step": 4889 }, { "epoch": 0.67, "grad_norm": 3.288852090723043, "learning_rate": 7.785402724616472e-06, "loss": 0.2299, "step": 4890 }, { "epoch": 0.67, "grad_norm": 3.2543829163650004, "learning_rate": 7.784487873640772e-06, "loss": 0.2464, "step": 4891 }, { "epoch": 0.67, "grad_norm": 3.4405194818042566, "learning_rate": 7.78357288751819e-06, "loss": 0.2513, "step": 4892 }, { "epoch": 0.67, "grad_norm": 3.2392174962292706, "learning_rate": 7.782657766293138e-06, "loss": 0.2517, "step": 4893 }, { "epoch": 0.67, "grad_norm": 3.425950746588361, "learning_rate": 7.78174251001003e-06, "loss": 0.2578, "step": 4894 }, { "epoch": 0.67, "grad_norm": 3.9868327886453723, "learning_rate": 7.780827118713295e-06, "loss": 0.2734, "step": 4895 }, { "epoch": 0.67, "grad_norm": 3.4830275483217914, "learning_rate": 7.779911592447354e-06, "loss": 0.2271, "step": 4896 }, { "epoch": 0.67, "grad_norm": 3.8524095350527894, "learning_rate": 7.778995931256646e-06, "loss": 0.2418, "step": 4897 }, { "epoch": 0.67, "grad_norm": 3.6690745456132747, "learning_rate": 7.778080135185613e-06, "loss": 0.2531, "step": 4898 }, { "epoch": 0.67, "grad_norm": 3.2047779211148857, "learning_rate": 7.777164204278703e-06, "loss": 0.2406, "step": 4899 }, { "epoch": 0.67, "grad_norm": 3.2778423431097994, "learning_rate": 7.776248138580374e-06, "loss": 0.2585, "step": 4900 }, { "epoch": 0.67, "grad_norm": 3.4246882852897444, "learning_rate": 7.775331938135084e-06, "loss": 0.26, "step": 4901 }, { "epoch": 0.67, "grad_norm": 3.270788898923075, "learning_rate": 7.774415602987304e-06, "loss": 0.2177, "step": 4902 }, { "epoch": 0.67, "grad_norm": 3.3806615423288253, "learning_rate": 7.773499133181508e-06, "loss": 0.2284, "step": 4903 }, { "epoch": 0.67, "grad_norm": 3.2290036128564403, "learning_rate": 7.772582528762179e-06, "loss": 0.2731, "step": 4904 }, { "epoch": 0.67, "grad_norm": 3.360930892358299, "learning_rate": 7.771665789773803e-06, "loss": 0.2552, "step": 4905 }, { "epoch": 0.67, "grad_norm": 3.791021744331155, "learning_rate": 7.770748916260875e-06, "loss": 0.2592, "step": 4906 }, { "epoch": 0.67, "grad_norm": 3.485911793080131, "learning_rate": 7.769831908267896e-06, "loss": 0.27, "step": 4907 }, { "epoch": 0.67, "grad_norm": 5.078626432409736, "learning_rate": 7.768914765839372e-06, "loss": 0.2552, "step": 4908 }, { "epoch": 0.67, "grad_norm": 3.79290576662922, "learning_rate": 7.767997489019823e-06, "loss": 0.2933, "step": 4909 }, { "epoch": 0.67, "grad_norm": 3.1416067320028676, "learning_rate": 7.767080077853762e-06, "loss": 0.2722, "step": 4910 }, { "epoch": 0.67, "grad_norm": 3.928738147560708, "learning_rate": 7.76616253238572e-06, "loss": 0.3178, "step": 4911 }, { "epoch": 0.67, "grad_norm": 3.594945706644922, "learning_rate": 7.765244852660233e-06, "loss": 0.2192, "step": 4912 }, { "epoch": 0.67, "grad_norm": 11.281778717610964, "learning_rate": 7.764327038721835e-06, "loss": 0.2923, "step": 4913 }, { "epoch": 0.67, "grad_norm": 3.4218394603188758, "learning_rate": 7.763409090615077e-06, "loss": 0.2349, "step": 4914 }, { "epoch": 0.67, "grad_norm": 3.609892027000992, "learning_rate": 7.76249100838451e-06, "loss": 0.2671, "step": 4915 }, { "epoch": 0.67, "grad_norm": 3.3120756055241687, "learning_rate": 7.761572792074697e-06, "loss": 0.2276, "step": 4916 }, { "epoch": 0.67, "grad_norm": 3.423467682718137, "learning_rate": 7.760654441730202e-06, "loss": 0.2712, "step": 4917 }, { "epoch": 0.67, "grad_norm": 3.9178283762921002, "learning_rate": 7.759735957395596e-06, "loss": 0.3006, "step": 4918 }, { "epoch": 0.67, "grad_norm": 4.295027106785292, "learning_rate": 7.758817339115462e-06, "loss": 0.3121, "step": 4919 }, { "epoch": 0.67, "grad_norm": 3.4827441551787905, "learning_rate": 7.757898586934383e-06, "loss": 0.2547, "step": 4920 }, { "epoch": 0.67, "grad_norm": 3.5870072847176337, "learning_rate": 7.756979700896951e-06, "loss": 0.2356, "step": 4921 }, { "epoch": 0.67, "grad_norm": 3.701086896639113, "learning_rate": 7.756060681047769e-06, "loss": 0.295, "step": 4922 }, { "epoch": 0.67, "grad_norm": 3.943237329148478, "learning_rate": 7.755141527431436e-06, "loss": 0.2302, "step": 4923 }, { "epoch": 0.67, "grad_norm": 3.6050507364180375, "learning_rate": 7.754222240092568e-06, "loss": 0.2485, "step": 4924 }, { "epoch": 0.67, "grad_norm": 3.788738934490393, "learning_rate": 7.75330281907578e-06, "loss": 0.2442, "step": 4925 }, { "epoch": 0.67, "grad_norm": 3.1837411734425576, "learning_rate": 7.752383264425701e-06, "loss": 0.2216, "step": 4926 }, { "epoch": 0.67, "grad_norm": 3.4149211181898127, "learning_rate": 7.751463576186957e-06, "loss": 0.2489, "step": 4927 }, { "epoch": 0.67, "grad_norm": 3.38419075432107, "learning_rate": 7.750543754404191e-06, "loss": 0.2437, "step": 4928 }, { "epoch": 0.67, "grad_norm": 3.525602957456965, "learning_rate": 7.749623799122042e-06, "loss": 0.2875, "step": 4929 }, { "epoch": 0.67, "grad_norm": 3.1914863623944303, "learning_rate": 7.748703710385165e-06, "loss": 0.2196, "step": 4930 }, { "epoch": 0.67, "grad_norm": 3.150824920355125, "learning_rate": 7.747783488238214e-06, "loss": 0.2044, "step": 4931 }, { "epoch": 0.67, "grad_norm": 4.094092300666747, "learning_rate": 7.746863132725856e-06, "loss": 0.2825, "step": 4932 }, { "epoch": 0.67, "grad_norm": 3.8955996473689876, "learning_rate": 7.745942643892755e-06, "loss": 0.2676, "step": 4933 }, { "epoch": 0.67, "grad_norm": 4.0376035227398095, "learning_rate": 7.745022021783595e-06, "loss": 0.2934, "step": 4934 }, { "epoch": 0.67, "grad_norm": 3.597812577481398, "learning_rate": 7.744101266443052e-06, "loss": 0.2607, "step": 4935 }, { "epoch": 0.67, "grad_norm": 4.090092657309023, "learning_rate": 7.74318037791582e-06, "loss": 0.2798, "step": 4936 }, { "epoch": 0.67, "grad_norm": 4.119932047312922, "learning_rate": 7.742259356246594e-06, "loss": 0.2419, "step": 4937 }, { "epoch": 0.67, "grad_norm": 3.1588689565696058, "learning_rate": 7.741338201480075e-06, "loss": 0.2175, "step": 4938 }, { "epoch": 0.67, "grad_norm": 3.3246013303401365, "learning_rate": 7.740416913660974e-06, "loss": 0.2207, "step": 4939 }, { "epoch": 0.67, "grad_norm": 3.728713906232755, "learning_rate": 7.739495492834005e-06, "loss": 0.2573, "step": 4940 }, { "epoch": 0.67, "grad_norm": 3.1199522389335224, "learning_rate": 7.73857393904389e-06, "loss": 0.2479, "step": 4941 }, { "epoch": 0.67, "grad_norm": 2.8011265864599038, "learning_rate": 7.737652252335356e-06, "loss": 0.235, "step": 4942 }, { "epoch": 0.67, "grad_norm": 2.851327260628669, "learning_rate": 7.73673043275314e-06, "loss": 0.2182, "step": 4943 }, { "epoch": 0.67, "grad_norm": 3.88108848684224, "learning_rate": 7.73580848034198e-06, "loss": 0.2786, "step": 4944 }, { "epoch": 0.67, "grad_norm": 3.6492037787496927, "learning_rate": 7.734886395146626e-06, "loss": 0.2912, "step": 4945 }, { "epoch": 0.67, "grad_norm": 4.3707734659596245, "learning_rate": 7.733964177211831e-06, "loss": 0.2632, "step": 4946 }, { "epoch": 0.67, "grad_norm": 2.7672411084719193, "learning_rate": 7.733041826582357e-06, "loss": 0.2552, "step": 4947 }, { "epoch": 0.67, "grad_norm": 3.379552344541445, "learning_rate": 7.732119343302969e-06, "loss": 0.2114, "step": 4948 }, { "epoch": 0.67, "grad_norm": 3.446119531347157, "learning_rate": 7.731196727418441e-06, "loss": 0.2892, "step": 4949 }, { "epoch": 0.67, "grad_norm": 3.746995162022322, "learning_rate": 7.730273978973552e-06, "loss": 0.2375, "step": 4950 }, { "epoch": 0.67, "grad_norm": 3.6637575485592473, "learning_rate": 7.72935109801309e-06, "loss": 0.2554, "step": 4951 }, { "epoch": 0.67, "grad_norm": 3.4051230844915086, "learning_rate": 7.728428084581844e-06, "loss": 0.2464, "step": 4952 }, { "epoch": 0.67, "grad_norm": 3.5921102298786565, "learning_rate": 7.727504938724619e-06, "loss": 0.2544, "step": 4953 }, { "epoch": 0.67, "grad_norm": 3.686495158742624, "learning_rate": 7.726581660486214e-06, "loss": 0.2648, "step": 4954 }, { "epoch": 0.67, "grad_norm": 3.3419145480080616, "learning_rate": 7.725658249911445e-06, "loss": 0.2589, "step": 4955 }, { "epoch": 0.67, "grad_norm": 3.017708396166579, "learning_rate": 7.724734707045129e-06, "loss": 0.2258, "step": 4956 }, { "epoch": 0.67, "grad_norm": 4.139389958756124, "learning_rate": 7.72381103193209e-06, "loss": 0.3138, "step": 4957 }, { "epoch": 0.67, "grad_norm": 3.6921011586772816, "learning_rate": 7.722887224617162e-06, "loss": 0.2754, "step": 4958 }, { "epoch": 0.67, "grad_norm": 4.048221747421249, "learning_rate": 7.72196328514518e-06, "loss": 0.2859, "step": 4959 }, { "epoch": 0.67, "grad_norm": 3.6532979810805926, "learning_rate": 7.721039213560988e-06, "loss": 0.2513, "step": 4960 }, { "epoch": 0.67, "grad_norm": 3.2822661304542713, "learning_rate": 7.720115009909437e-06, "loss": 0.257, "step": 4961 }, { "epoch": 0.67, "grad_norm": 3.5071483204638976, "learning_rate": 7.719190674235383e-06, "loss": 0.2499, "step": 4962 }, { "epoch": 0.68, "grad_norm": 3.8769172288094857, "learning_rate": 7.718266206583693e-06, "loss": 0.2446, "step": 4963 }, { "epoch": 0.68, "grad_norm": 3.441168781618758, "learning_rate": 7.717341606999231e-06, "loss": 0.2826, "step": 4964 }, { "epoch": 0.68, "grad_norm": 5.276722593904416, "learning_rate": 7.716416875526876e-06, "loss": 0.2185, "step": 4965 }, { "epoch": 0.68, "grad_norm": 3.556692773216399, "learning_rate": 7.715492012211512e-06, "loss": 0.2665, "step": 4966 }, { "epoch": 0.68, "grad_norm": 2.8804561011559864, "learning_rate": 7.714567017098023e-06, "loss": 0.2111, "step": 4967 }, { "epoch": 0.68, "grad_norm": 3.5682375437509144, "learning_rate": 7.71364189023131e-06, "loss": 0.309, "step": 4968 }, { "epoch": 0.68, "grad_norm": 3.328559397075914, "learning_rate": 7.712716631656269e-06, "loss": 0.2813, "step": 4969 }, { "epoch": 0.68, "grad_norm": 3.6611699663873085, "learning_rate": 7.711791241417812e-06, "loss": 0.279, "step": 4970 }, { "epoch": 0.68, "grad_norm": 3.118919852534863, "learning_rate": 7.710865719560854e-06, "loss": 0.2196, "step": 4971 }, { "epoch": 0.68, "grad_norm": 3.641895673019239, "learning_rate": 7.709940066130312e-06, "loss": 0.2301, "step": 4972 }, { "epoch": 0.68, "grad_norm": 3.250827354809198, "learning_rate": 7.709014281171115e-06, "loss": 0.2329, "step": 4973 }, { "epoch": 0.68, "grad_norm": 3.9450126232142373, "learning_rate": 7.708088364728196e-06, "loss": 0.2498, "step": 4974 }, { "epoch": 0.68, "grad_norm": 3.9400841267157576, "learning_rate": 7.707162316846497e-06, "loss": 0.2595, "step": 4975 }, { "epoch": 0.68, "grad_norm": 3.507055130549129, "learning_rate": 7.706236137570964e-06, "loss": 0.2572, "step": 4976 }, { "epoch": 0.68, "grad_norm": 3.758940219680002, "learning_rate": 7.705309826946547e-06, "loss": 0.2964, "step": 4977 }, { "epoch": 0.68, "grad_norm": 3.780953301201634, "learning_rate": 7.704383385018207e-06, "loss": 0.286, "step": 4978 }, { "epoch": 0.68, "grad_norm": 4.25721759522616, "learning_rate": 7.70345681183091e-06, "loss": 0.3059, "step": 4979 }, { "epoch": 0.68, "grad_norm": 3.6724066902450985, "learning_rate": 7.702530107429627e-06, "loss": 0.2755, "step": 4980 }, { "epoch": 0.68, "grad_norm": 3.667711732817709, "learning_rate": 7.701603271859335e-06, "loss": 0.2447, "step": 4981 }, { "epoch": 0.68, "grad_norm": 3.5771858584057887, "learning_rate": 7.70067630516502e-06, "loss": 0.2837, "step": 4982 }, { "epoch": 0.68, "grad_norm": 3.9417235730630065, "learning_rate": 7.699749207391673e-06, "loss": 0.2672, "step": 4983 }, { "epoch": 0.68, "grad_norm": 3.0588288181559395, "learning_rate": 7.69882197858429e-06, "loss": 0.2008, "step": 4984 }, { "epoch": 0.68, "grad_norm": 2.992020002698623, "learning_rate": 7.697894618787878e-06, "loss": 0.2177, "step": 4985 }, { "epoch": 0.68, "grad_norm": 3.1685273715578344, "learning_rate": 7.696967128047442e-06, "loss": 0.211, "step": 4986 }, { "epoch": 0.68, "grad_norm": 3.3801129016282814, "learning_rate": 7.696039506408001e-06, "loss": 0.2645, "step": 4987 }, { "epoch": 0.68, "grad_norm": 3.6469620268654532, "learning_rate": 7.69511175391458e-06, "loss": 0.2529, "step": 4988 }, { "epoch": 0.68, "grad_norm": 3.6575862956420253, "learning_rate": 7.694183870612203e-06, "loss": 0.2302, "step": 4989 }, { "epoch": 0.68, "grad_norm": 3.5216463174129236, "learning_rate": 7.693255856545912e-06, "loss": 0.2341, "step": 4990 }, { "epoch": 0.68, "grad_norm": 3.1678965885835693, "learning_rate": 7.69232771176074e-06, "loss": 0.196, "step": 4991 }, { "epoch": 0.68, "grad_norm": 3.1706799151413185, "learning_rate": 7.691399436301743e-06, "loss": 0.2641, "step": 4992 }, { "epoch": 0.68, "grad_norm": 3.3050889669754153, "learning_rate": 7.690471030213971e-06, "loss": 0.2285, "step": 4993 }, { "epoch": 0.68, "grad_norm": 3.8704697460007256, "learning_rate": 7.689542493542486e-06, "loss": 0.1968, "step": 4994 }, { "epoch": 0.68, "grad_norm": 3.140938137186307, "learning_rate": 7.688613826332357e-06, "loss": 0.2344, "step": 4995 }, { "epoch": 0.68, "grad_norm": 3.1881509542301445, "learning_rate": 7.687685028628653e-06, "loss": 0.2409, "step": 4996 }, { "epoch": 0.68, "grad_norm": 3.9647311851309563, "learning_rate": 7.686756100476458e-06, "loss": 0.2816, "step": 4997 }, { "epoch": 0.68, "grad_norm": 3.621256832578531, "learning_rate": 7.685827041920857e-06, "loss": 0.2719, "step": 4998 }, { "epoch": 0.68, "grad_norm": 3.8574012881050463, "learning_rate": 7.68489785300694e-06, "loss": 0.2473, "step": 4999 }, { "epoch": 0.68, "grad_norm": 3.753036604005725, "learning_rate": 7.683968533779811e-06, "loss": 0.2824, "step": 5000 }, { "epoch": 0.68, "grad_norm": 3.334872312727345, "learning_rate": 7.68303908428457e-06, "loss": 0.2424, "step": 5001 }, { "epoch": 0.68, "grad_norm": 12.340254706904096, "learning_rate": 7.68210950456633e-06, "loss": 0.2396, "step": 5002 }, { "epoch": 0.68, "grad_norm": 4.327734167156794, "learning_rate": 7.681179794670208e-06, "loss": 0.2627, "step": 5003 }, { "epoch": 0.68, "grad_norm": 3.2302175217763467, "learning_rate": 7.680249954641333e-06, "loss": 0.1776, "step": 5004 }, { "epoch": 0.68, "grad_norm": 3.650717111006307, "learning_rate": 7.679319984524828e-06, "loss": 0.283, "step": 5005 }, { "epoch": 0.68, "grad_norm": 2.5185015762632625, "learning_rate": 7.678389884365834e-06, "loss": 0.2062, "step": 5006 }, { "epoch": 0.68, "grad_norm": 3.8240588391107218, "learning_rate": 7.677459654209493e-06, "loss": 0.3244, "step": 5007 }, { "epoch": 0.68, "grad_norm": 3.502042663459711, "learning_rate": 7.676529294100957e-06, "loss": 0.284, "step": 5008 }, { "epoch": 0.68, "grad_norm": 3.1191591178491858, "learning_rate": 7.675598804085376e-06, "loss": 0.2322, "step": 5009 }, { "epoch": 0.68, "grad_norm": 3.7188471867855886, "learning_rate": 7.674668184207915e-06, "loss": 0.2647, "step": 5010 }, { "epoch": 0.68, "grad_norm": 3.5978373925739344, "learning_rate": 7.673737434513743e-06, "loss": 0.2726, "step": 5011 }, { "epoch": 0.68, "grad_norm": 3.4850181658827912, "learning_rate": 7.672806555048034e-06, "loss": 0.2917, "step": 5012 }, { "epoch": 0.68, "grad_norm": 3.617815269090424, "learning_rate": 7.67187554585597e-06, "loss": 0.2593, "step": 5013 }, { "epoch": 0.68, "grad_norm": 4.0442029554822705, "learning_rate": 7.670944406982734e-06, "loss": 0.3639, "step": 5014 }, { "epoch": 0.68, "grad_norm": 3.406518130986882, "learning_rate": 7.670013138473524e-06, "loss": 0.2624, "step": 5015 }, { "epoch": 0.68, "grad_norm": 3.581612312593685, "learning_rate": 7.669081740373536e-06, "loss": 0.271, "step": 5016 }, { "epoch": 0.68, "grad_norm": 3.0589327444835503, "learning_rate": 7.66815021272798e-06, "loss": 0.2688, "step": 5017 }, { "epoch": 0.68, "grad_norm": 2.9739876329157333, "learning_rate": 7.667218555582065e-06, "loss": 0.1834, "step": 5018 }, { "epoch": 0.68, "grad_norm": 3.637029740200408, "learning_rate": 7.666286768981012e-06, "loss": 0.2856, "step": 5019 }, { "epoch": 0.68, "grad_norm": 3.5416662920890363, "learning_rate": 7.665354852970044e-06, "loss": 0.3115, "step": 5020 }, { "epoch": 0.68, "grad_norm": 4.218126262317571, "learning_rate": 7.664422807594391e-06, "loss": 0.2394, "step": 5021 }, { "epoch": 0.68, "grad_norm": 3.6527139776368833, "learning_rate": 7.663490632899293e-06, "loss": 0.2365, "step": 5022 }, { "epoch": 0.68, "grad_norm": 3.363286514315342, "learning_rate": 7.662558328929994e-06, "loss": 0.2498, "step": 5023 }, { "epoch": 0.68, "grad_norm": 3.4501248034513994, "learning_rate": 7.66162589573174e-06, "loss": 0.2302, "step": 5024 }, { "epoch": 0.68, "grad_norm": 3.396890002395603, "learning_rate": 7.660693333349793e-06, "loss": 0.3292, "step": 5025 }, { "epoch": 0.68, "grad_norm": 3.4074069104791977, "learning_rate": 7.65976064182941e-06, "loss": 0.2173, "step": 5026 }, { "epoch": 0.68, "grad_norm": 3.850561928318104, "learning_rate": 7.658827821215863e-06, "loss": 0.2676, "step": 5027 }, { "epoch": 0.68, "grad_norm": 3.4843276660923146, "learning_rate": 7.657894871554428e-06, "loss": 0.2287, "step": 5028 }, { "epoch": 0.68, "grad_norm": 3.7687715099769936, "learning_rate": 7.656961792890382e-06, "loss": 0.2607, "step": 5029 }, { "epoch": 0.68, "grad_norm": 3.314278988805602, "learning_rate": 7.656028585269017e-06, "loss": 0.2138, "step": 5030 }, { "epoch": 0.68, "grad_norm": 4.379293968455376, "learning_rate": 7.655095248735625e-06, "loss": 0.2547, "step": 5031 }, { "epoch": 0.68, "grad_norm": 3.2803588770833416, "learning_rate": 7.654161783335506e-06, "loss": 0.2287, "step": 5032 }, { "epoch": 0.68, "grad_norm": 4.994184576633876, "learning_rate": 7.653228189113966e-06, "loss": 0.2335, "step": 5033 }, { "epoch": 0.68, "grad_norm": 4.156127801956017, "learning_rate": 7.652294466116317e-06, "loss": 0.3008, "step": 5034 }, { "epoch": 0.68, "grad_norm": 4.1359136156299074, "learning_rate": 7.651360614387879e-06, "loss": 0.3075, "step": 5035 }, { "epoch": 0.69, "grad_norm": 3.582283719708978, "learning_rate": 7.650426633973978e-06, "loss": 0.2594, "step": 5036 }, { "epoch": 0.69, "grad_norm": 3.664078453068343, "learning_rate": 7.649492524919944e-06, "loss": 0.2868, "step": 5037 }, { "epoch": 0.69, "grad_norm": 4.006229071806009, "learning_rate": 7.648558287271114e-06, "loss": 0.2937, "step": 5038 }, { "epoch": 0.69, "grad_norm": 3.329098351761852, "learning_rate": 7.647623921072833e-06, "loss": 0.2851, "step": 5039 }, { "epoch": 0.69, "grad_norm": 3.3124091981213177, "learning_rate": 7.64668942637045e-06, "loss": 0.2448, "step": 5040 }, { "epoch": 0.69, "grad_norm": 3.41953975403275, "learning_rate": 7.645754803209324e-06, "loss": 0.2675, "step": 5041 }, { "epoch": 0.69, "grad_norm": 3.6245557101314363, "learning_rate": 7.644820051634813e-06, "loss": 0.2477, "step": 5042 }, { "epoch": 0.69, "grad_norm": 3.566402848081423, "learning_rate": 7.64388517169229e-06, "loss": 0.2791, "step": 5043 }, { "epoch": 0.69, "grad_norm": 4.315351546314023, "learning_rate": 7.642950163427127e-06, "loss": 0.2377, "step": 5044 }, { "epoch": 0.69, "grad_norm": 3.737839204784189, "learning_rate": 7.642015026884707e-06, "loss": 0.2087, "step": 5045 }, { "epoch": 0.69, "grad_norm": 3.9392362781290564, "learning_rate": 7.641079762110416e-06, "loss": 0.2566, "step": 5046 }, { "epoch": 0.69, "grad_norm": 4.723645393063607, "learning_rate": 7.64014436914965e-06, "loss": 0.2724, "step": 5047 }, { "epoch": 0.69, "grad_norm": 3.327345005223837, "learning_rate": 7.639208848047808e-06, "loss": 0.2611, "step": 5048 }, { "epoch": 0.69, "grad_norm": 2.9961646480669133, "learning_rate": 7.638273198850292e-06, "loss": 0.2349, "step": 5049 }, { "epoch": 0.69, "grad_norm": 3.54674954676532, "learning_rate": 7.637337421602521e-06, "loss": 0.2242, "step": 5050 }, { "epoch": 0.69, "grad_norm": 3.4793246171478676, "learning_rate": 7.636401516349909e-06, "loss": 0.2422, "step": 5051 }, { "epoch": 0.69, "grad_norm": 3.791221344746509, "learning_rate": 7.635465483137885e-06, "loss": 0.2725, "step": 5052 }, { "epoch": 0.69, "grad_norm": 3.452486076546291, "learning_rate": 7.634529322011874e-06, "loss": 0.2356, "step": 5053 }, { "epoch": 0.69, "grad_norm": 3.8403687003628724, "learning_rate": 7.633593033017317e-06, "loss": 0.2988, "step": 5054 }, { "epoch": 0.69, "grad_norm": 3.588976522015359, "learning_rate": 7.63265661619966e-06, "loss": 0.1975, "step": 5055 }, { "epoch": 0.69, "grad_norm": 3.529830275326187, "learning_rate": 7.631720071604343e-06, "loss": 0.293, "step": 5056 }, { "epoch": 0.69, "grad_norm": 3.0218738480172314, "learning_rate": 7.63078339927683e-06, "loss": 0.1815, "step": 5057 }, { "epoch": 0.69, "grad_norm": 3.123193127013042, "learning_rate": 7.629846599262585e-06, "loss": 0.2575, "step": 5058 }, { "epoch": 0.69, "grad_norm": 3.6662567697130397, "learning_rate": 7.6289096716070685e-06, "loss": 0.2574, "step": 5059 }, { "epoch": 0.69, "grad_norm": 4.038266711568902, "learning_rate": 7.62797261635576e-06, "loss": 0.2249, "step": 5060 }, { "epoch": 0.69, "grad_norm": 3.2666141691642, "learning_rate": 7.627035433554138e-06, "loss": 0.2609, "step": 5061 }, { "epoch": 0.69, "grad_norm": 3.4626122479381123, "learning_rate": 7.626098123247691e-06, "loss": 0.1873, "step": 5062 }, { "epoch": 0.69, "grad_norm": 3.659356071427061, "learning_rate": 7.625160685481909e-06, "loss": 0.2709, "step": 5063 }, { "epoch": 0.69, "grad_norm": 4.136522629847989, "learning_rate": 7.624223120302294e-06, "loss": 0.2497, "step": 5064 }, { "epoch": 0.69, "grad_norm": 3.7261176319081426, "learning_rate": 7.62328542775435e-06, "loss": 0.2635, "step": 5065 }, { "epoch": 0.69, "grad_norm": 3.3180123054029984, "learning_rate": 7.62234760788359e-06, "loss": 0.2538, "step": 5066 }, { "epoch": 0.69, "grad_norm": 3.6557456481732293, "learning_rate": 7.621409660735531e-06, "loss": 0.2718, "step": 5067 }, { "epoch": 0.69, "grad_norm": 3.2688366941182254, "learning_rate": 7.620471586355693e-06, "loss": 0.2103, "step": 5068 }, { "epoch": 0.69, "grad_norm": 3.6318289989363275, "learning_rate": 7.619533384789612e-06, "loss": 0.2883, "step": 5069 }, { "epoch": 0.69, "grad_norm": 4.06950180880985, "learning_rate": 7.618595056082823e-06, "loss": 0.2725, "step": 5070 }, { "epoch": 0.69, "grad_norm": 3.784137786562821, "learning_rate": 7.6176566002808635e-06, "loss": 0.243, "step": 5071 }, { "epoch": 0.69, "grad_norm": 3.4788853697117625, "learning_rate": 7.616718017429288e-06, "loss": 0.2471, "step": 5072 }, { "epoch": 0.69, "grad_norm": 3.908568485714622, "learning_rate": 7.6157793075736476e-06, "loss": 0.2461, "step": 5073 }, { "epoch": 0.69, "grad_norm": 3.3662857568050737, "learning_rate": 7.614840470759505e-06, "loss": 0.2722, "step": 5074 }, { "epoch": 0.69, "grad_norm": 2.760418971027706, "learning_rate": 7.613901507032425e-06, "loss": 0.1999, "step": 5075 }, { "epoch": 0.69, "grad_norm": 2.8765146768265972, "learning_rate": 7.612962416437984e-06, "loss": 0.1848, "step": 5076 }, { "epoch": 0.69, "grad_norm": 4.076161483911498, "learning_rate": 7.612023199021759e-06, "loss": 0.2854, "step": 5077 }, { "epoch": 0.69, "grad_norm": 3.4550064040983317, "learning_rate": 7.611083854829337e-06, "loss": 0.2069, "step": 5078 }, { "epoch": 0.69, "grad_norm": 3.5522501716402313, "learning_rate": 7.610144383906308e-06, "loss": 0.2279, "step": 5079 }, { "epoch": 0.69, "grad_norm": 3.3795789906660443, "learning_rate": 7.60920478629827e-06, "loss": 0.2328, "step": 5080 }, { "epoch": 0.69, "grad_norm": 3.992913044350287, "learning_rate": 7.608265062050829e-06, "loss": 0.2119, "step": 5081 }, { "epoch": 0.69, "grad_norm": 3.566002009521646, "learning_rate": 7.607325211209593e-06, "loss": 0.2447, "step": 5082 }, { "epoch": 0.69, "grad_norm": 3.242778125078426, "learning_rate": 7.606385233820179e-06, "loss": 0.1966, "step": 5083 }, { "epoch": 0.69, "grad_norm": 3.544411158372572, "learning_rate": 7.605445129928211e-06, "loss": 0.28, "step": 5084 }, { "epoch": 0.69, "grad_norm": 4.329564948704599, "learning_rate": 7.6045048995793145e-06, "loss": 0.2261, "step": 5085 }, { "epoch": 0.69, "grad_norm": 3.4688409905823328, "learning_rate": 7.6035645428191264e-06, "loss": 0.2746, "step": 5086 }, { "epoch": 0.69, "grad_norm": 3.405407312231137, "learning_rate": 7.6026240596932854e-06, "loss": 0.2312, "step": 5087 }, { "epoch": 0.69, "grad_norm": 3.671926286715759, "learning_rate": 7.6016834502474415e-06, "loss": 0.2669, "step": 5088 }, { "epoch": 0.69, "grad_norm": 3.564334488158291, "learning_rate": 7.600742714527244e-06, "loss": 0.2996, "step": 5089 }, { "epoch": 0.69, "grad_norm": 3.8361215217759335, "learning_rate": 7.599801852578355e-06, "loss": 0.213, "step": 5090 }, { "epoch": 0.69, "grad_norm": 3.8376732684440324, "learning_rate": 7.59886086444644e-06, "loss": 0.2667, "step": 5091 }, { "epoch": 0.69, "grad_norm": 3.7344653191651163, "learning_rate": 7.597919750177168e-06, "loss": 0.2125, "step": 5092 }, { "epoch": 0.69, "grad_norm": 4.628462432319966, "learning_rate": 7.59697850981622e-06, "loss": 0.2611, "step": 5093 }, { "epoch": 0.69, "grad_norm": 3.839579648283617, "learning_rate": 7.5960371434092765e-06, "loss": 0.2127, "step": 5094 }, { "epoch": 0.69, "grad_norm": 3.2224110710700895, "learning_rate": 7.595095651002029e-06, "loss": 0.2363, "step": 5095 }, { "epoch": 0.69, "grad_norm": 3.2593295490113245, "learning_rate": 7.594154032640171e-06, "loss": 0.2734, "step": 5096 }, { "epoch": 0.69, "grad_norm": 3.5812320897436822, "learning_rate": 7.593212288369408e-06, "loss": 0.2403, "step": 5097 }, { "epoch": 0.69, "grad_norm": 3.721960590686198, "learning_rate": 7.592270418235448e-06, "loss": 0.2064, "step": 5098 }, { "epoch": 0.69, "grad_norm": 3.687192643750386, "learning_rate": 7.591328422284001e-06, "loss": 0.2284, "step": 5099 }, { "epoch": 0.69, "grad_norm": 3.6178920965078274, "learning_rate": 7.590386300560793e-06, "loss": 0.2771, "step": 5100 }, { "epoch": 0.69, "grad_norm": 3.464595565189973, "learning_rate": 7.589444053111547e-06, "loss": 0.2347, "step": 5101 }, { "epoch": 0.69, "grad_norm": 4.385355190551944, "learning_rate": 7.588501679981997e-06, "loss": 0.2745, "step": 5102 }, { "epoch": 0.69, "grad_norm": 2.891881352175296, "learning_rate": 7.587559181217878e-06, "loss": 0.2372, "step": 5103 }, { "epoch": 0.69, "grad_norm": 2.9804463861759087, "learning_rate": 7.58661655686494e-06, "loss": 0.2153, "step": 5104 }, { "epoch": 0.69, "grad_norm": 3.6655863112160567, "learning_rate": 7.585673806968932e-06, "loss": 0.3042, "step": 5105 }, { "epoch": 0.69, "grad_norm": 4.197389490447783, "learning_rate": 7.58473093157561e-06, "loss": 0.2443, "step": 5106 }, { "epoch": 0.69, "grad_norm": 3.9346739874685177, "learning_rate": 7.583787930730737e-06, "loss": 0.2226, "step": 5107 }, { "epoch": 0.69, "grad_norm": 3.772928727921326, "learning_rate": 7.582844804480085e-06, "loss": 0.2975, "step": 5108 }, { "epoch": 0.69, "grad_norm": 4.406031347614344, "learning_rate": 7.581901552869425e-06, "loss": 0.2775, "step": 5109 }, { "epoch": 0.7, "grad_norm": 3.797198381094265, "learning_rate": 7.580958175944542e-06, "loss": 0.25, "step": 5110 }, { "epoch": 0.7, "grad_norm": 3.3237408374835375, "learning_rate": 7.58001467375122e-06, "loss": 0.2437, "step": 5111 }, { "epoch": 0.7, "grad_norm": 2.896340622292104, "learning_rate": 7.579071046335256e-06, "loss": 0.2479, "step": 5112 }, { "epoch": 0.7, "grad_norm": 3.6030878796201895, "learning_rate": 7.578127293742446e-06, "loss": 0.2665, "step": 5113 }, { "epoch": 0.7, "grad_norm": 4.119251635157373, "learning_rate": 7.577183416018598e-06, "loss": 0.2845, "step": 5114 }, { "epoch": 0.7, "grad_norm": 3.796286324928937, "learning_rate": 7.576239413209523e-06, "loss": 0.2658, "step": 5115 }, { "epoch": 0.7, "grad_norm": 3.5007141088867093, "learning_rate": 7.575295285361039e-06, "loss": 0.2454, "step": 5116 }, { "epoch": 0.7, "grad_norm": 3.3795502841610845, "learning_rate": 7.57435103251897e-06, "loss": 0.2176, "step": 5117 }, { "epoch": 0.7, "grad_norm": 3.619607689636403, "learning_rate": 7.573406654729146e-06, "loss": 0.2278, "step": 5118 }, { "epoch": 0.7, "grad_norm": 3.4891149843589537, "learning_rate": 7.572462152037403e-06, "loss": 0.2567, "step": 5119 }, { "epoch": 0.7, "grad_norm": 3.3879470469889736, "learning_rate": 7.571517524489582e-06, "loss": 0.2539, "step": 5120 }, { "epoch": 0.7, "grad_norm": 3.3706866817088974, "learning_rate": 7.5705727721315306e-06, "loss": 0.2353, "step": 5121 }, { "epoch": 0.7, "grad_norm": 3.4101881577980886, "learning_rate": 7.569627895009104e-06, "loss": 0.2382, "step": 5122 }, { "epoch": 0.7, "grad_norm": 4.275860889828597, "learning_rate": 7.5686828931681646e-06, "loss": 0.314, "step": 5123 }, { "epoch": 0.7, "grad_norm": 3.2716436275185528, "learning_rate": 7.567737766654575e-06, "loss": 0.2446, "step": 5124 }, { "epoch": 0.7, "grad_norm": 3.393614941944725, "learning_rate": 7.566792515514209e-06, "loss": 0.2311, "step": 5125 }, { "epoch": 0.7, "grad_norm": 3.479910867354229, "learning_rate": 7.565847139792948e-06, "loss": 0.265, "step": 5126 }, { "epoch": 0.7, "grad_norm": 4.824827302479807, "learning_rate": 7.564901639536671e-06, "loss": 0.2765, "step": 5127 }, { "epoch": 0.7, "grad_norm": 3.4980474534030215, "learning_rate": 7.563956014791271e-06, "loss": 0.2406, "step": 5128 }, { "epoch": 0.7, "grad_norm": 4.209238995404274, "learning_rate": 7.563010265602644e-06, "loss": 0.3339, "step": 5129 }, { "epoch": 0.7, "grad_norm": 3.318796789605477, "learning_rate": 7.562064392016693e-06, "loss": 0.2764, "step": 5130 }, { "epoch": 0.7, "grad_norm": 3.4186079153199778, "learning_rate": 7.561118394079326e-06, "loss": 0.2385, "step": 5131 }, { "epoch": 0.7, "grad_norm": 3.4958835352373887, "learning_rate": 7.560172271836459e-06, "loss": 0.1977, "step": 5132 }, { "epoch": 0.7, "grad_norm": 3.750751171762038, "learning_rate": 7.5592260253340125e-06, "loss": 0.2745, "step": 5133 }, { "epoch": 0.7, "grad_norm": 3.2736583577739053, "learning_rate": 7.5582796546179125e-06, "loss": 0.2128, "step": 5134 }, { "epoch": 0.7, "grad_norm": 3.573178294973349, "learning_rate": 7.55733315973409e-06, "loss": 0.2213, "step": 5135 }, { "epoch": 0.7, "grad_norm": 3.4260177849577897, "learning_rate": 7.5563865407284865e-06, "loss": 0.2163, "step": 5136 }, { "epoch": 0.7, "grad_norm": 3.585206180145214, "learning_rate": 7.555439797647044e-06, "loss": 0.2618, "step": 5137 }, { "epoch": 0.7, "grad_norm": 3.7258271999854577, "learning_rate": 7.554492930535718e-06, "loss": 0.2642, "step": 5138 }, { "epoch": 0.7, "grad_norm": 4.139848079931412, "learning_rate": 7.5535459394404585e-06, "loss": 0.2223, "step": 5139 }, { "epoch": 0.7, "grad_norm": 3.8896215448235636, "learning_rate": 7.552598824407234e-06, "loss": 0.2293, "step": 5140 }, { "epoch": 0.7, "grad_norm": 3.8614891322258473, "learning_rate": 7.551651585482012e-06, "loss": 0.2945, "step": 5141 }, { "epoch": 0.7, "grad_norm": 3.18290108660376, "learning_rate": 7.5507042227107655e-06, "loss": 0.2174, "step": 5142 }, { "epoch": 0.7, "grad_norm": 4.308506013309377, "learning_rate": 7.5497567361394766e-06, "loss": 0.2376, "step": 5143 }, { "epoch": 0.7, "grad_norm": 3.3390908610358863, "learning_rate": 7.548809125814131e-06, "loss": 0.2645, "step": 5144 }, { "epoch": 0.7, "grad_norm": 3.8027543610375467, "learning_rate": 7.547861391780725e-06, "loss": 0.2673, "step": 5145 }, { "epoch": 0.7, "grad_norm": 3.789133649734664, "learning_rate": 7.546913534085254e-06, "loss": 0.3046, "step": 5146 }, { "epoch": 0.7, "grad_norm": 3.1765737756235413, "learning_rate": 7.545965552773724e-06, "loss": 0.1937, "step": 5147 }, { "epoch": 0.7, "grad_norm": 4.415918562491159, "learning_rate": 7.545017447892147e-06, "loss": 0.2632, "step": 5148 }, { "epoch": 0.7, "grad_norm": 4.145557115979417, "learning_rate": 7.544069219486537e-06, "loss": 0.2382, "step": 5149 }, { "epoch": 0.7, "grad_norm": 3.743167196036967, "learning_rate": 7.54312086760292e-06, "loss": 0.253, "step": 5150 }, { "epoch": 0.7, "grad_norm": 4.304018175977853, "learning_rate": 7.542172392287322e-06, "loss": 0.3042, "step": 5151 }, { "epoch": 0.7, "grad_norm": 4.0979100782531015, "learning_rate": 7.54122379358578e-06, "loss": 0.2802, "step": 5152 }, { "epoch": 0.7, "grad_norm": 4.504994480558751, "learning_rate": 7.540275071544334e-06, "loss": 0.2511, "step": 5153 }, { "epoch": 0.7, "grad_norm": 3.230292244178754, "learning_rate": 7.539326226209032e-06, "loss": 0.2204, "step": 5154 }, { "epoch": 0.7, "grad_norm": 2.9799438951802797, "learning_rate": 7.538377257625924e-06, "loss": 0.2503, "step": 5155 }, { "epoch": 0.7, "grad_norm": 3.6282218292952226, "learning_rate": 7.537428165841071e-06, "loss": 0.2852, "step": 5156 }, { "epoch": 0.7, "grad_norm": 3.5814339547649428, "learning_rate": 7.536478950900537e-06, "loss": 0.3007, "step": 5157 }, { "epoch": 0.7, "grad_norm": 3.6358809783169694, "learning_rate": 7.535529612850394e-06, "loss": 0.2419, "step": 5158 }, { "epoch": 0.7, "grad_norm": 3.867836238157382, "learning_rate": 7.534580151736718e-06, "loss": 0.3108, "step": 5159 }, { "epoch": 0.7, "grad_norm": 3.428169205776656, "learning_rate": 7.533630567605591e-06, "loss": 0.2475, "step": 5160 }, { "epoch": 0.7, "grad_norm": 3.4215239581482417, "learning_rate": 7.532680860503101e-06, "loss": 0.2639, "step": 5161 }, { "epoch": 0.7, "grad_norm": 3.489805614807558, "learning_rate": 7.531731030475345e-06, "loss": 0.219, "step": 5162 }, { "epoch": 0.7, "grad_norm": 3.746050654634962, "learning_rate": 7.530781077568423e-06, "loss": 0.2247, "step": 5163 }, { "epoch": 0.7, "grad_norm": 3.127388428524651, "learning_rate": 7.52983100182844e-06, "loss": 0.2213, "step": 5164 }, { "epoch": 0.7, "grad_norm": 3.30795133348984, "learning_rate": 7.528880803301511e-06, "loss": 0.2276, "step": 5165 }, { "epoch": 0.7, "grad_norm": 3.68393419936524, "learning_rate": 7.5279304820337515e-06, "loss": 0.283, "step": 5166 }, { "epoch": 0.7, "grad_norm": 3.8903733068631334, "learning_rate": 7.526980038071288e-06, "loss": 0.2798, "step": 5167 }, { "epoch": 0.7, "grad_norm": 3.094006960230041, "learning_rate": 7.5260294714602475e-06, "loss": 0.2369, "step": 5168 }, { "epoch": 0.7, "grad_norm": 3.71318995787377, "learning_rate": 7.525078782246773e-06, "loss": 0.2494, "step": 5169 }, { "epoch": 0.7, "grad_norm": 3.3931879661083753, "learning_rate": 7.524127970477002e-06, "loss": 0.2363, "step": 5170 }, { "epoch": 0.7, "grad_norm": 3.528221330421346, "learning_rate": 7.523177036197082e-06, "loss": 0.3018, "step": 5171 }, { "epoch": 0.7, "grad_norm": 3.2522325818675646, "learning_rate": 7.52222597945317e-06, "loss": 0.2218, "step": 5172 }, { "epoch": 0.7, "grad_norm": 3.2966532181996424, "learning_rate": 7.521274800291427e-06, "loss": 0.2235, "step": 5173 }, { "epoch": 0.7, "grad_norm": 3.3119584126115917, "learning_rate": 7.520323498758015e-06, "loss": 0.247, "step": 5174 }, { "epoch": 0.7, "grad_norm": 3.6846371092016867, "learning_rate": 7.519372074899107e-06, "loss": 0.2631, "step": 5175 }, { "epoch": 0.7, "grad_norm": 3.0147626710694744, "learning_rate": 7.518420528760884e-06, "loss": 0.2513, "step": 5176 }, { "epoch": 0.7, "grad_norm": 4.149306236963365, "learning_rate": 7.517468860389528e-06, "loss": 0.2046, "step": 5177 }, { "epoch": 0.7, "grad_norm": 3.671471369458357, "learning_rate": 7.51651706983123e-06, "loss": 0.2964, "step": 5178 }, { "epoch": 0.7, "grad_norm": 3.829359200645288, "learning_rate": 7.515565157132183e-06, "loss": 0.2382, "step": 5179 }, { "epoch": 0.7, "grad_norm": 3.6753918156641703, "learning_rate": 7.5146131223385895e-06, "loss": 0.3051, "step": 5180 }, { "epoch": 0.7, "grad_norm": 3.183634660985557, "learning_rate": 7.51366096549666e-06, "loss": 0.1916, "step": 5181 }, { "epoch": 0.7, "grad_norm": 3.4848000105374246, "learning_rate": 7.512708686652603e-06, "loss": 0.2027, "step": 5182 }, { "epoch": 0.71, "grad_norm": 3.3604306416343213, "learning_rate": 7.511756285852642e-06, "loss": 0.2424, "step": 5183 }, { "epoch": 0.71, "grad_norm": 3.786840357115308, "learning_rate": 7.510803763143003e-06, "loss": 0.2307, "step": 5184 }, { "epoch": 0.71, "grad_norm": 3.9066142563815736, "learning_rate": 7.509851118569915e-06, "loss": 0.3042, "step": 5185 }, { "epoch": 0.71, "grad_norm": 4.31027396314415, "learning_rate": 7.508898352179614e-06, "loss": 0.2908, "step": 5186 }, { "epoch": 0.71, "grad_norm": 4.120154746397927, "learning_rate": 7.507945464018347e-06, "loss": 0.3016, "step": 5187 }, { "epoch": 0.71, "grad_norm": 3.3357522550454073, "learning_rate": 7.50699245413236e-06, "loss": 0.259, "step": 5188 }, { "epoch": 0.71, "grad_norm": 4.024274672753112, "learning_rate": 7.506039322567908e-06, "loss": 0.2242, "step": 5189 }, { "epoch": 0.71, "grad_norm": 4.299590702593498, "learning_rate": 7.505086069371254e-06, "loss": 0.2571, "step": 5190 }, { "epoch": 0.71, "grad_norm": 3.9482936039965617, "learning_rate": 7.504132694588664e-06, "loss": 0.31, "step": 5191 }, { "epoch": 0.71, "grad_norm": 3.980818779384652, "learning_rate": 7.5031791982664085e-06, "loss": 0.2366, "step": 5192 }, { "epoch": 0.71, "grad_norm": 3.5952991277857183, "learning_rate": 7.502225580450768e-06, "loss": 0.2277, "step": 5193 }, { "epoch": 0.71, "grad_norm": 4.13364370848488, "learning_rate": 7.5012718411880255e-06, "loss": 0.2464, "step": 5194 }, { "epoch": 0.71, "grad_norm": 3.0458496409522824, "learning_rate": 7.500317980524474e-06, "loss": 0.2322, "step": 5195 }, { "epoch": 0.71, "grad_norm": 3.061342320258664, "learning_rate": 7.499363998506409e-06, "loss": 0.2486, "step": 5196 }, { "epoch": 0.71, "grad_norm": 3.502793365722764, "learning_rate": 7.49840989518013e-06, "loss": 0.2624, "step": 5197 }, { "epoch": 0.71, "grad_norm": 3.545414362831033, "learning_rate": 7.4974556705919466e-06, "loss": 0.229, "step": 5198 }, { "epoch": 0.71, "grad_norm": 3.32629845149808, "learning_rate": 7.496501324788175e-06, "loss": 0.2579, "step": 5199 }, { "epoch": 0.71, "grad_norm": 3.9590746547205824, "learning_rate": 7.49554685781513e-06, "loss": 0.2896, "step": 5200 }, { "epoch": 0.71, "grad_norm": 2.9613269264329496, "learning_rate": 7.494592269719141e-06, "loss": 0.1989, "step": 5201 }, { "epoch": 0.71, "grad_norm": 4.573140978511885, "learning_rate": 7.493637560546539e-06, "loss": 0.2101, "step": 5202 }, { "epoch": 0.71, "grad_norm": 3.846036299865754, "learning_rate": 7.492682730343661e-06, "loss": 0.2688, "step": 5203 }, { "epoch": 0.71, "grad_norm": 3.901663825494914, "learning_rate": 7.491727779156851e-06, "loss": 0.2521, "step": 5204 }, { "epoch": 0.71, "grad_norm": 3.9836773902444844, "learning_rate": 7.490772707032456e-06, "loss": 0.2871, "step": 5205 }, { "epoch": 0.71, "grad_norm": 3.0640632127835494, "learning_rate": 7.489817514016835e-06, "loss": 0.2154, "step": 5206 }, { "epoch": 0.71, "grad_norm": 3.000158016155318, "learning_rate": 7.488862200156345e-06, "loss": 0.2209, "step": 5207 }, { "epoch": 0.71, "grad_norm": 3.1218876385550542, "learning_rate": 7.487906765497352e-06, "loss": 0.2272, "step": 5208 }, { "epoch": 0.71, "grad_norm": 3.264891712013237, "learning_rate": 7.486951210086234e-06, "loss": 0.228, "step": 5209 }, { "epoch": 0.71, "grad_norm": 3.698700786645607, "learning_rate": 7.485995533969365e-06, "loss": 0.2433, "step": 5210 }, { "epoch": 0.71, "grad_norm": 3.1765003522048123, "learning_rate": 7.48503973719313e-06, "loss": 0.2447, "step": 5211 }, { "epoch": 0.71, "grad_norm": 3.4773992243927148, "learning_rate": 7.484083819803922e-06, "loss": 0.2521, "step": 5212 }, { "epoch": 0.71, "grad_norm": 3.338300101576678, "learning_rate": 7.483127781848133e-06, "loss": 0.2304, "step": 5213 }, { "epoch": 0.71, "grad_norm": 3.8704675361283556, "learning_rate": 7.482171623372168e-06, "loss": 0.2456, "step": 5214 }, { "epoch": 0.71, "grad_norm": 4.234743976164657, "learning_rate": 7.4812153444224325e-06, "loss": 0.3248, "step": 5215 }, { "epoch": 0.71, "grad_norm": 3.687304438475663, "learning_rate": 7.4802589450453415e-06, "loss": 0.2579, "step": 5216 }, { "epoch": 0.71, "grad_norm": 4.107067028277593, "learning_rate": 7.479302425287314e-06, "loss": 0.2307, "step": 5217 }, { "epoch": 0.71, "grad_norm": 3.6915364088228824, "learning_rate": 7.478345785194775e-06, "loss": 0.2823, "step": 5218 }, { "epoch": 0.71, "grad_norm": 3.4972082162307623, "learning_rate": 7.477389024814157e-06, "loss": 0.2512, "step": 5219 }, { "epoch": 0.71, "grad_norm": 3.479491757813318, "learning_rate": 7.476432144191895e-06, "loss": 0.2492, "step": 5220 }, { "epoch": 0.71, "grad_norm": 4.050244357610473, "learning_rate": 7.475475143374434e-06, "loss": 0.2906, "step": 5221 }, { "epoch": 0.71, "grad_norm": 3.08528045242571, "learning_rate": 7.4745180224082205e-06, "loss": 0.2293, "step": 5222 }, { "epoch": 0.71, "grad_norm": 3.429616682722749, "learning_rate": 7.47356078133971e-06, "loss": 0.2467, "step": 5223 }, { "epoch": 0.71, "grad_norm": 3.6950124865210783, "learning_rate": 7.472603420215363e-06, "loss": 0.2891, "step": 5224 }, { "epoch": 0.71, "grad_norm": 3.2767123428380605, "learning_rate": 7.471645939081645e-06, "loss": 0.2163, "step": 5225 }, { "epoch": 0.71, "grad_norm": 3.3243528569223235, "learning_rate": 7.470688337985029e-06, "loss": 0.2061, "step": 5226 }, { "epoch": 0.71, "grad_norm": 4.586941882684589, "learning_rate": 7.469730616971992e-06, "loss": 0.2877, "step": 5227 }, { "epoch": 0.71, "grad_norm": 3.864473378874353, "learning_rate": 7.468772776089017e-06, "loss": 0.2644, "step": 5228 }, { "epoch": 0.71, "grad_norm": 3.5830795132298143, "learning_rate": 7.4678148153825946e-06, "loss": 0.2397, "step": 5229 }, { "epoch": 0.71, "grad_norm": 4.322827448082992, "learning_rate": 7.466856734899219e-06, "loss": 0.254, "step": 5230 }, { "epoch": 0.71, "grad_norm": 3.792319706145689, "learning_rate": 7.465898534685393e-06, "loss": 0.2498, "step": 5231 }, { "epoch": 0.71, "grad_norm": 3.2683661997045315, "learning_rate": 7.464940214787622e-06, "loss": 0.2634, "step": 5232 }, { "epoch": 0.71, "grad_norm": 11.403475857205928, "learning_rate": 7.463981775252418e-06, "loss": 0.2875, "step": 5233 }, { "epoch": 0.71, "grad_norm": 4.689861679502126, "learning_rate": 7.463023216126299e-06, "loss": 0.2955, "step": 5234 }, { "epoch": 0.71, "grad_norm": 3.428838234343702, "learning_rate": 7.462064537455793e-06, "loss": 0.2516, "step": 5235 }, { "epoch": 0.71, "grad_norm": 3.6966234798618345, "learning_rate": 7.461105739287426e-06, "loss": 0.2746, "step": 5236 }, { "epoch": 0.71, "grad_norm": 3.5684286911922114, "learning_rate": 7.4601468216677375e-06, "loss": 0.2679, "step": 5237 }, { "epoch": 0.71, "grad_norm": 3.8434118903494716, "learning_rate": 7.459187784643266e-06, "loss": 0.3154, "step": 5238 }, { "epoch": 0.71, "grad_norm": 4.219741973517054, "learning_rate": 7.458228628260561e-06, "loss": 0.2424, "step": 5239 }, { "epoch": 0.71, "grad_norm": 3.246637745507687, "learning_rate": 7.457269352566174e-06, "loss": 0.2237, "step": 5240 }, { "epoch": 0.71, "grad_norm": 3.8445405682991485, "learning_rate": 7.456309957606664e-06, "loss": 0.29, "step": 5241 }, { "epoch": 0.71, "grad_norm": 3.861273453217729, "learning_rate": 7.455350443428598e-06, "loss": 0.2721, "step": 5242 }, { "epoch": 0.71, "grad_norm": 3.300519405889291, "learning_rate": 7.454390810078547e-06, "loss": 0.256, "step": 5243 }, { "epoch": 0.71, "grad_norm": 3.78901594021306, "learning_rate": 7.453431057603084e-06, "loss": 0.2668, "step": 5244 }, { "epoch": 0.71, "grad_norm": 3.3827504906824237, "learning_rate": 7.452471186048794e-06, "loss": 0.2181, "step": 5245 }, { "epoch": 0.71, "grad_norm": 3.5145311968975848, "learning_rate": 7.451511195462265e-06, "loss": 0.2384, "step": 5246 }, { "epoch": 0.71, "grad_norm": 3.649564641819501, "learning_rate": 7.450551085890087e-06, "loss": 0.2726, "step": 5247 }, { "epoch": 0.71, "grad_norm": 3.4911469697494577, "learning_rate": 7.449590857378863e-06, "loss": 0.2409, "step": 5248 }, { "epoch": 0.71, "grad_norm": 3.6776024636518874, "learning_rate": 7.448630509975199e-06, "loss": 0.2791, "step": 5249 }, { "epoch": 0.71, "grad_norm": 3.6238007104255017, "learning_rate": 7.4476700437257045e-06, "loss": 0.2179, "step": 5250 }, { "epoch": 0.71, "grad_norm": 3.625529381659866, "learning_rate": 7.446709458676996e-06, "loss": 0.232, "step": 5251 }, { "epoch": 0.71, "grad_norm": 3.518015474664962, "learning_rate": 7.445748754875697e-06, "loss": 0.2631, "step": 5252 }, { "epoch": 0.71, "grad_norm": 3.7648859751664836, "learning_rate": 7.444787932368436e-06, "loss": 0.2372, "step": 5253 }, { "epoch": 0.71, "grad_norm": 3.8192425545110296, "learning_rate": 7.443826991201845e-06, "loss": 0.2608, "step": 5254 }, { "epoch": 0.71, "grad_norm": 3.4286108633871617, "learning_rate": 7.4428659314225675e-06, "loss": 0.2611, "step": 5255 }, { "epoch": 0.71, "grad_norm": 5.191855862144869, "learning_rate": 7.441904753077246e-06, "loss": 0.2758, "step": 5256 }, { "epoch": 0.72, "grad_norm": 3.5360259887214625, "learning_rate": 7.440943456212533e-06, "loss": 0.2477, "step": 5257 }, { "epoch": 0.72, "grad_norm": 3.554058778040281, "learning_rate": 7.439982040875087e-06, "loss": 0.2104, "step": 5258 }, { "epoch": 0.72, "grad_norm": 3.907846984387277, "learning_rate": 7.439020507111568e-06, "loss": 0.2631, "step": 5259 }, { "epoch": 0.72, "grad_norm": 3.042701002866775, "learning_rate": 7.438058854968647e-06, "loss": 0.2383, "step": 5260 }, { "epoch": 0.72, "grad_norm": 3.7293458371506083, "learning_rate": 7.437097084492997e-06, "loss": 0.2519, "step": 5261 }, { "epoch": 0.72, "grad_norm": 3.639810383109034, "learning_rate": 7.4361351957313e-06, "loss": 0.2372, "step": 5262 }, { "epoch": 0.72, "grad_norm": 3.8004720175811086, "learning_rate": 7.43517318873024e-06, "loss": 0.279, "step": 5263 }, { "epoch": 0.72, "grad_norm": 3.6824115344427106, "learning_rate": 7.43421106353651e-06, "loss": 0.3007, "step": 5264 }, { "epoch": 0.72, "grad_norm": 3.353510417775388, "learning_rate": 7.4332488201968055e-06, "loss": 0.2992, "step": 5265 }, { "epoch": 0.72, "grad_norm": 3.3847730257318904, "learning_rate": 7.432286458757832e-06, "loss": 0.286, "step": 5266 }, { "epoch": 0.72, "grad_norm": 3.2850380711221434, "learning_rate": 7.431323979266296e-06, "loss": 0.2808, "step": 5267 }, { "epoch": 0.72, "grad_norm": 3.5549651694379363, "learning_rate": 7.430361381768912e-06, "loss": 0.1795, "step": 5268 }, { "epoch": 0.72, "grad_norm": 3.6941359382481767, "learning_rate": 7.429398666312404e-06, "loss": 0.2285, "step": 5269 }, { "epoch": 0.72, "grad_norm": 3.0350347730542566, "learning_rate": 7.428435832943495e-06, "loss": 0.2385, "step": 5270 }, { "epoch": 0.72, "grad_norm": 3.7841110827632667, "learning_rate": 7.427472881708917e-06, "loss": 0.213, "step": 5271 }, { "epoch": 0.72, "grad_norm": 3.2343044395413587, "learning_rate": 7.4265098126554065e-06, "loss": 0.232, "step": 5272 }, { "epoch": 0.72, "grad_norm": 3.6442513685783733, "learning_rate": 7.425546625829708e-06, "loss": 0.2559, "step": 5273 }, { "epoch": 0.72, "grad_norm": 3.8384443263113988, "learning_rate": 7.4245833212785705e-06, "loss": 0.2498, "step": 5274 }, { "epoch": 0.72, "grad_norm": 2.785877955002541, "learning_rate": 7.423619899048747e-06, "loss": 0.2259, "step": 5275 }, { "epoch": 0.72, "grad_norm": 3.1656700373656985, "learning_rate": 7.422656359187002e-06, "loss": 0.2374, "step": 5276 }, { "epoch": 0.72, "grad_norm": 4.235266209293637, "learning_rate": 7.421692701740095e-06, "loss": 0.1873, "step": 5277 }, { "epoch": 0.72, "grad_norm": 3.0346483332112886, "learning_rate": 7.420728926754803e-06, "loss": 0.2647, "step": 5278 }, { "epoch": 0.72, "grad_norm": 3.4350017401822255, "learning_rate": 7.419765034277903e-06, "loss": 0.2111, "step": 5279 }, { "epoch": 0.72, "grad_norm": 3.6085417114309495, "learning_rate": 7.4188010243561735e-06, "loss": 0.2413, "step": 5280 }, { "epoch": 0.72, "grad_norm": 2.9831070707129084, "learning_rate": 7.417836897036409e-06, "loss": 0.2457, "step": 5281 }, { "epoch": 0.72, "grad_norm": 3.9317160585588358, "learning_rate": 7.416872652365401e-06, "loss": 0.233, "step": 5282 }, { "epoch": 0.72, "grad_norm": 3.8005776714716863, "learning_rate": 7.415908290389952e-06, "loss": 0.2796, "step": 5283 }, { "epoch": 0.72, "grad_norm": 4.4742891182664986, "learning_rate": 7.414943811156865e-06, "loss": 0.2428, "step": 5284 }, { "epoch": 0.72, "grad_norm": 3.7305609523923917, "learning_rate": 7.413979214712952e-06, "loss": 0.208, "step": 5285 }, { "epoch": 0.72, "grad_norm": 3.7122258578515224, "learning_rate": 7.413014501105033e-06, "loss": 0.3029, "step": 5286 }, { "epoch": 0.72, "grad_norm": 4.107431987212594, "learning_rate": 7.412049670379927e-06, "loss": 0.3055, "step": 5287 }, { "epoch": 0.72, "grad_norm": 4.052703689958925, "learning_rate": 7.4110847225844675e-06, "loss": 0.2492, "step": 5288 }, { "epoch": 0.72, "grad_norm": 3.828764044224645, "learning_rate": 7.410119657765484e-06, "loss": 0.224, "step": 5289 }, { "epoch": 0.72, "grad_norm": 3.95851739640569, "learning_rate": 7.40915447596982e-06, "loss": 0.2176, "step": 5290 }, { "epoch": 0.72, "grad_norm": 3.5619151592943137, "learning_rate": 7.40818917724432e-06, "loss": 0.2717, "step": 5291 }, { "epoch": 0.72, "grad_norm": 3.630003214268967, "learning_rate": 7.4072237616358356e-06, "loss": 0.2842, "step": 5292 }, { "epoch": 0.72, "grad_norm": 3.135021003743689, "learning_rate": 7.406258229191223e-06, "loss": 0.2307, "step": 5293 }, { "epoch": 0.72, "grad_norm": 4.414742971637347, "learning_rate": 7.405292579957345e-06, "loss": 0.3069, "step": 5294 }, { "epoch": 0.72, "grad_norm": 4.04803171965673, "learning_rate": 7.404326813981071e-06, "loss": 0.3137, "step": 5295 }, { "epoch": 0.72, "grad_norm": 3.887207991928133, "learning_rate": 7.403360931309275e-06, "loss": 0.2355, "step": 5296 }, { "epoch": 0.72, "grad_norm": 3.1249355093390703, "learning_rate": 7.402394931988836e-06, "loss": 0.2079, "step": 5297 }, { "epoch": 0.72, "grad_norm": 3.1828263690845464, "learning_rate": 7.401428816066641e-06, "loss": 0.2187, "step": 5298 }, { "epoch": 0.72, "grad_norm": 4.312685534313485, "learning_rate": 7.400462583589578e-06, "loss": 0.3158, "step": 5299 }, { "epoch": 0.72, "grad_norm": 3.7515596979341455, "learning_rate": 7.3994962346045475e-06, "loss": 0.2484, "step": 5300 }, { "epoch": 0.72, "grad_norm": 3.5263062147803903, "learning_rate": 7.398529769158447e-06, "loss": 0.2051, "step": 5301 }, { "epoch": 0.72, "grad_norm": 3.6024400763429036, "learning_rate": 7.397563187298189e-06, "loss": 0.261, "step": 5302 }, { "epoch": 0.72, "grad_norm": 3.0863910746143306, "learning_rate": 7.396596489070686e-06, "loss": 0.2379, "step": 5303 }, { "epoch": 0.72, "grad_norm": 3.6258338551263067, "learning_rate": 7.395629674522856e-06, "loss": 0.2312, "step": 5304 }, { "epoch": 0.72, "grad_norm": 3.062430467537318, "learning_rate": 7.394662743701625e-06, "loss": 0.2471, "step": 5305 }, { "epoch": 0.72, "grad_norm": 4.079648338932271, "learning_rate": 7.393695696653924e-06, "loss": 0.2435, "step": 5306 }, { "epoch": 0.72, "grad_norm": 4.186887288489833, "learning_rate": 7.392728533426687e-06, "loss": 0.2409, "step": 5307 }, { "epoch": 0.72, "grad_norm": 3.5960918993425595, "learning_rate": 7.391761254066858e-06, "loss": 0.2831, "step": 5308 }, { "epoch": 0.72, "grad_norm": 3.1701425052138505, "learning_rate": 7.390793858621386e-06, "loss": 0.2421, "step": 5309 }, { "epoch": 0.72, "grad_norm": 3.707443214085218, "learning_rate": 7.389826347137221e-06, "loss": 0.1721, "step": 5310 }, { "epoch": 0.72, "grad_norm": 4.150784014733857, "learning_rate": 7.388858719661322e-06, "loss": 0.2775, "step": 5311 }, { "epoch": 0.72, "grad_norm": 3.4681959056718092, "learning_rate": 7.387890976240655e-06, "loss": 0.2255, "step": 5312 }, { "epoch": 0.72, "grad_norm": 4.495907463730625, "learning_rate": 7.386923116922189e-06, "loss": 0.2917, "step": 5313 }, { "epoch": 0.72, "grad_norm": 4.05548985694937, "learning_rate": 7.3859551417528995e-06, "loss": 0.2412, "step": 5314 }, { "epoch": 0.72, "grad_norm": 3.7673889579559456, "learning_rate": 7.384987050779769e-06, "loss": 0.338, "step": 5315 }, { "epoch": 0.72, "grad_norm": 3.770307303478587, "learning_rate": 7.384018844049785e-06, "loss": 0.2729, "step": 5316 }, { "epoch": 0.72, "grad_norm": 5.579739817647102, "learning_rate": 7.383050521609938e-06, "loss": 0.2346, "step": 5317 }, { "epoch": 0.72, "grad_norm": 3.1105163127083433, "learning_rate": 7.382082083507226e-06, "loss": 0.194, "step": 5318 }, { "epoch": 0.72, "grad_norm": 3.862772042691901, "learning_rate": 7.381113529788655e-06, "loss": 0.2997, "step": 5319 }, { "epoch": 0.72, "grad_norm": 3.821924602928668, "learning_rate": 7.3801448605012304e-06, "loss": 0.2295, "step": 5320 }, { "epoch": 0.72, "grad_norm": 4.2418307010229634, "learning_rate": 7.379176075691973e-06, "loss": 0.2688, "step": 5321 }, { "epoch": 0.72, "grad_norm": 4.086272319906812, "learning_rate": 7.378207175407899e-06, "loss": 0.2379, "step": 5322 }, { "epoch": 0.72, "grad_norm": 3.145227250952293, "learning_rate": 7.377238159696035e-06, "loss": 0.2546, "step": 5323 }, { "epoch": 0.72, "grad_norm": 3.2756214930911205, "learning_rate": 7.376269028603414e-06, "loss": 0.2481, "step": 5324 }, { "epoch": 0.72, "grad_norm": 3.7395090914204605, "learning_rate": 7.3752997821770746e-06, "loss": 0.2058, "step": 5325 }, { "epoch": 0.72, "grad_norm": 3.6240639848051077, "learning_rate": 7.374330420464057e-06, "loss": 0.28, "step": 5326 }, { "epoch": 0.72, "grad_norm": 3.5553041570159567, "learning_rate": 7.3733609435114096e-06, "loss": 0.2487, "step": 5327 }, { "epoch": 0.72, "grad_norm": 3.0330003106834362, "learning_rate": 7.372391351366188e-06, "loss": 0.1901, "step": 5328 }, { "epoch": 0.72, "grad_norm": 3.863457846431597, "learning_rate": 7.371421644075454e-06, "loss": 0.2389, "step": 5329 }, { "epoch": 0.73, "grad_norm": 4.107931554454362, "learning_rate": 7.37045182168627e-06, "loss": 0.259, "step": 5330 }, { "epoch": 0.73, "grad_norm": 3.4322445682970844, "learning_rate": 7.369481884245708e-06, "loss": 0.2511, "step": 5331 }, { "epoch": 0.73, "grad_norm": 3.1272287756103716, "learning_rate": 7.368511831800844e-06, "loss": 0.2384, "step": 5332 }, { "epoch": 0.73, "grad_norm": 3.7602852199395866, "learning_rate": 7.36754166439876e-06, "loss": 0.2476, "step": 5333 }, { "epoch": 0.73, "grad_norm": 3.755380810899194, "learning_rate": 7.3665713820865435e-06, "loss": 0.2458, "step": 5334 }, { "epoch": 0.73, "grad_norm": 4.533741270789649, "learning_rate": 7.365600984911289e-06, "loss": 0.2283, "step": 5335 }, { "epoch": 0.73, "grad_norm": 3.6028935958743693, "learning_rate": 7.364630472920096e-06, "loss": 0.2484, "step": 5336 }, { "epoch": 0.73, "grad_norm": 3.281436754039689, "learning_rate": 7.363659846160066e-06, "loss": 0.2485, "step": 5337 }, { "epoch": 0.73, "grad_norm": 3.518595243008555, "learning_rate": 7.362689104678311e-06, "loss": 0.2753, "step": 5338 }, { "epoch": 0.73, "grad_norm": 3.8425384210563407, "learning_rate": 7.361718248521947e-06, "loss": 0.2499, "step": 5339 }, { "epoch": 0.73, "grad_norm": 3.7899007737795283, "learning_rate": 7.360747277738094e-06, "loss": 0.2911, "step": 5340 }, { "epoch": 0.73, "grad_norm": 3.1613855046418537, "learning_rate": 7.359776192373877e-06, "loss": 0.207, "step": 5341 }, { "epoch": 0.73, "grad_norm": 3.9632223144877314, "learning_rate": 7.358804992476432e-06, "loss": 0.3237, "step": 5342 }, { "epoch": 0.73, "grad_norm": 3.438605404350528, "learning_rate": 7.357833678092895e-06, "loss": 0.2162, "step": 5343 }, { "epoch": 0.73, "grad_norm": 3.3883242052380265, "learning_rate": 7.356862249270409e-06, "loss": 0.2354, "step": 5344 }, { "epoch": 0.73, "grad_norm": 4.023747827753747, "learning_rate": 7.355890706056124e-06, "loss": 0.2654, "step": 5345 }, { "epoch": 0.73, "grad_norm": 3.4532963711456883, "learning_rate": 7.354919048497193e-06, "loss": 0.248, "step": 5346 }, { "epoch": 0.73, "grad_norm": 3.976446463718738, "learning_rate": 7.353947276640776e-06, "loss": 0.2418, "step": 5347 }, { "epoch": 0.73, "grad_norm": 3.6535724394782787, "learning_rate": 7.35297539053404e-06, "loss": 0.2263, "step": 5348 }, { "epoch": 0.73, "grad_norm": 4.090873976382015, "learning_rate": 7.352003390224157e-06, "loss": 0.2813, "step": 5349 }, { "epoch": 0.73, "grad_norm": 3.873593456413615, "learning_rate": 7.351031275758302e-06, "loss": 0.2387, "step": 5350 }, { "epoch": 0.73, "grad_norm": 4.708818449551483, "learning_rate": 7.350059047183655e-06, "loss": 0.2415, "step": 5351 }, { "epoch": 0.73, "grad_norm": 3.6785736822454917, "learning_rate": 7.349086704547408e-06, "loss": 0.2499, "step": 5352 }, { "epoch": 0.73, "grad_norm": 4.023997018734956, "learning_rate": 7.34811424789675e-06, "loss": 0.2996, "step": 5353 }, { "epoch": 0.73, "grad_norm": 3.7746485998770956, "learning_rate": 7.347141677278884e-06, "loss": 0.2528, "step": 5354 }, { "epoch": 0.73, "grad_norm": 3.6125655388455367, "learning_rate": 7.346168992741012e-06, "loss": 0.2802, "step": 5355 }, { "epoch": 0.73, "grad_norm": 3.5793105374180154, "learning_rate": 7.345196194330343e-06, "loss": 0.2801, "step": 5356 }, { "epoch": 0.73, "grad_norm": 3.445964546023373, "learning_rate": 7.344223282094095e-06, "loss": 0.2176, "step": 5357 }, { "epoch": 0.73, "grad_norm": 3.5080473924317457, "learning_rate": 7.343250256079487e-06, "loss": 0.2647, "step": 5358 }, { "epoch": 0.73, "grad_norm": 4.012450758456866, "learning_rate": 7.342277116333747e-06, "loss": 0.2748, "step": 5359 }, { "epoch": 0.73, "grad_norm": 3.958170004029711, "learning_rate": 7.341303862904104e-06, "loss": 0.2379, "step": 5360 }, { "epoch": 0.73, "grad_norm": 2.834021969140118, "learning_rate": 7.340330495837798e-06, "loss": 0.1865, "step": 5361 }, { "epoch": 0.73, "grad_norm": 3.8859472859766457, "learning_rate": 7.3393570151820714e-06, "loss": 0.2517, "step": 5362 }, { "epoch": 0.73, "grad_norm": 4.34816405693358, "learning_rate": 7.338383420984174e-06, "loss": 0.269, "step": 5363 }, { "epoch": 0.73, "grad_norm": 3.8349204699385027, "learning_rate": 7.337409713291357e-06, "loss": 0.285, "step": 5364 }, { "epoch": 0.73, "grad_norm": 3.2933053404660724, "learning_rate": 7.336435892150883e-06, "loss": 0.2763, "step": 5365 }, { "epoch": 0.73, "grad_norm": 5.16923976882372, "learning_rate": 7.335461957610014e-06, "loss": 0.2859, "step": 5366 }, { "epoch": 0.73, "grad_norm": 3.3093194540270563, "learning_rate": 7.334487909716021e-06, "loss": 0.2314, "step": 5367 }, { "epoch": 0.73, "grad_norm": 3.6794932448385262, "learning_rate": 7.333513748516183e-06, "loss": 0.2657, "step": 5368 }, { "epoch": 0.73, "grad_norm": 3.771650923915989, "learning_rate": 7.332539474057778e-06, "loss": 0.1997, "step": 5369 }, { "epoch": 0.73, "grad_norm": 3.20260971784398, "learning_rate": 7.331565086388096e-06, "loss": 0.2437, "step": 5370 }, { "epoch": 0.73, "grad_norm": 4.212020993974333, "learning_rate": 7.330590585554428e-06, "loss": 0.2484, "step": 5371 }, { "epoch": 0.73, "grad_norm": 3.946874378601571, "learning_rate": 7.329615971604071e-06, "loss": 0.2315, "step": 5372 }, { "epoch": 0.73, "grad_norm": 3.503973654136327, "learning_rate": 7.328641244584329e-06, "loss": 0.2672, "step": 5373 }, { "epoch": 0.73, "grad_norm": 4.011669935735193, "learning_rate": 7.327666404542514e-06, "loss": 0.2715, "step": 5374 }, { "epoch": 0.73, "grad_norm": 4.191326561819843, "learning_rate": 7.326691451525937e-06, "loss": 0.2658, "step": 5375 }, { "epoch": 0.73, "grad_norm": 4.511152819584283, "learning_rate": 7.325716385581918e-06, "loss": 0.2754, "step": 5376 }, { "epoch": 0.73, "grad_norm": 4.185052667761973, "learning_rate": 7.324741206757785e-06, "loss": 0.2841, "step": 5377 }, { "epoch": 0.73, "grad_norm": 2.9394211286045686, "learning_rate": 7.323765915100867e-06, "loss": 0.202, "step": 5378 }, { "epoch": 0.73, "grad_norm": 3.4643920399427963, "learning_rate": 7.3227905106585e-06, "loss": 0.2272, "step": 5379 }, { "epoch": 0.73, "grad_norm": 4.045489462394932, "learning_rate": 7.321814993478027e-06, "loss": 0.2403, "step": 5380 }, { "epoch": 0.73, "grad_norm": 3.6358068881838226, "learning_rate": 7.320839363606794e-06, "loss": 0.2245, "step": 5381 }, { "epoch": 0.73, "grad_norm": 4.092752836553236, "learning_rate": 7.3198636210921556e-06, "loss": 0.2086, "step": 5382 }, { "epoch": 0.73, "grad_norm": 3.183117454475703, "learning_rate": 7.3188877659814705e-06, "loss": 0.2002, "step": 5383 }, { "epoch": 0.73, "grad_norm": 3.981495809740444, "learning_rate": 7.3179117983221e-06, "loss": 0.2119, "step": 5384 }, { "epoch": 0.73, "grad_norm": 3.8818513956023537, "learning_rate": 7.316935718161415e-06, "loss": 0.2659, "step": 5385 }, { "epoch": 0.73, "grad_norm": 3.479498098617706, "learning_rate": 7.31595952554679e-06, "loss": 0.2608, "step": 5386 }, { "epoch": 0.73, "grad_norm": 4.875070034502991, "learning_rate": 7.314983220525604e-06, "loss": 0.2761, "step": 5387 }, { "epoch": 0.73, "grad_norm": 4.423926781975956, "learning_rate": 7.314006803145244e-06, "loss": 0.2648, "step": 5388 }, { "epoch": 0.73, "grad_norm": 3.915971487515331, "learning_rate": 7.313030273453101e-06, "loss": 0.2566, "step": 5389 }, { "epoch": 0.73, "grad_norm": 4.248421237229142, "learning_rate": 7.31205363149657e-06, "loss": 0.2681, "step": 5390 }, { "epoch": 0.73, "grad_norm": 3.666495082948392, "learning_rate": 7.311076877323054e-06, "loss": 0.2467, "step": 5391 }, { "epoch": 0.73, "grad_norm": 3.524561001398266, "learning_rate": 7.3101000109799616e-06, "loss": 0.2603, "step": 5392 }, { "epoch": 0.73, "grad_norm": 3.2784644749963756, "learning_rate": 7.309123032514702e-06, "loss": 0.2028, "step": 5393 }, { "epoch": 0.73, "grad_norm": 3.1955343675560886, "learning_rate": 7.308145941974697e-06, "loss": 0.2068, "step": 5394 }, { "epoch": 0.73, "grad_norm": 3.615894700308, "learning_rate": 7.307168739407369e-06, "loss": 0.2826, "step": 5395 }, { "epoch": 0.73, "grad_norm": 3.7335558445531545, "learning_rate": 7.306191424860147e-06, "loss": 0.2362, "step": 5396 }, { "epoch": 0.73, "grad_norm": 3.6191516678439912, "learning_rate": 7.305213998380466e-06, "loss": 0.2599, "step": 5397 }, { "epoch": 0.73, "grad_norm": 3.4834843546214755, "learning_rate": 7.304236460015766e-06, "loss": 0.2605, "step": 5398 }, { "epoch": 0.73, "grad_norm": 3.562353059191457, "learning_rate": 7.3032588098134904e-06, "loss": 0.2487, "step": 5399 }, { "epoch": 0.73, "grad_norm": 3.9132103543170285, "learning_rate": 7.302281047821093e-06, "loss": 0.2356, "step": 5400 }, { "epoch": 0.73, "grad_norm": 3.794652300564381, "learning_rate": 7.3013031740860295e-06, "loss": 0.2578, "step": 5401 }, { "epoch": 0.73, "grad_norm": 3.5824628857731255, "learning_rate": 7.300325188655762e-06, "loss": 0.2676, "step": 5402 }, { "epoch": 0.73, "grad_norm": 3.329183897259779, "learning_rate": 7.299347091577756e-06, "loss": 0.2968, "step": 5403 }, { "epoch": 0.74, "grad_norm": 3.1822602548476087, "learning_rate": 7.298368882899484e-06, "loss": 0.2466, "step": 5404 }, { "epoch": 0.74, "grad_norm": 5.903001713800573, "learning_rate": 7.297390562668425e-06, "loss": 0.3067, "step": 5405 }, { "epoch": 0.74, "grad_norm": 3.0904407676714865, "learning_rate": 7.296412130932062e-06, "loss": 0.2509, "step": 5406 }, { "epoch": 0.74, "grad_norm": 3.6677309577307478, "learning_rate": 7.295433587737885e-06, "loss": 0.2765, "step": 5407 }, { "epoch": 0.74, "grad_norm": 3.1599077228546006, "learning_rate": 7.294454933133387e-06, "loss": 0.2516, "step": 5408 }, { "epoch": 0.74, "grad_norm": 4.008433485273899, "learning_rate": 7.293476167166067e-06, "loss": 0.2369, "step": 5409 }, { "epoch": 0.74, "grad_norm": 3.0461253396349512, "learning_rate": 7.292497289883432e-06, "loss": 0.2497, "step": 5410 }, { "epoch": 0.74, "grad_norm": 3.0797374881161335, "learning_rate": 7.2915183013329904e-06, "loss": 0.2185, "step": 5411 }, { "epoch": 0.74, "grad_norm": 3.7470623829135845, "learning_rate": 7.29053920156226e-06, "loss": 0.2546, "step": 5412 }, { "epoch": 0.74, "grad_norm": 3.3002642742226618, "learning_rate": 7.28955999061876e-06, "loss": 0.239, "step": 5413 }, { "epoch": 0.74, "grad_norm": 4.157945418709543, "learning_rate": 7.288580668550017e-06, "loss": 0.2411, "step": 5414 }, { "epoch": 0.74, "grad_norm": 3.589658673509121, "learning_rate": 7.287601235403565e-06, "loss": 0.2427, "step": 5415 }, { "epoch": 0.74, "grad_norm": 3.078102666817852, "learning_rate": 7.28662169122694e-06, "loss": 0.2434, "step": 5416 }, { "epoch": 0.74, "grad_norm": 4.12458361308484, "learning_rate": 7.285642036067687e-06, "loss": 0.2466, "step": 5417 }, { "epoch": 0.74, "grad_norm": 3.7365867698473716, "learning_rate": 7.28466226997335e-06, "loss": 0.2004, "step": 5418 }, { "epoch": 0.74, "grad_norm": 3.43378353709868, "learning_rate": 7.283682392991487e-06, "loss": 0.2621, "step": 5419 }, { "epoch": 0.74, "grad_norm": 3.110076007900494, "learning_rate": 7.282702405169653e-06, "loss": 0.2148, "step": 5420 }, { "epoch": 0.74, "grad_norm": 3.107387490899534, "learning_rate": 7.281722306555415e-06, "loss": 0.2504, "step": 5421 }, { "epoch": 0.74, "grad_norm": 3.2501535815455473, "learning_rate": 7.280742097196342e-06, "loss": 0.2003, "step": 5422 }, { "epoch": 0.74, "grad_norm": 3.607073199461343, "learning_rate": 7.2797617771400085e-06, "loss": 0.2631, "step": 5423 }, { "epoch": 0.74, "grad_norm": 2.620244551106148, "learning_rate": 7.278781346433996e-06, "loss": 0.2093, "step": 5424 }, { "epoch": 0.74, "grad_norm": 3.7484825442804155, "learning_rate": 7.2778008051258906e-06, "loss": 0.2698, "step": 5425 }, { "epoch": 0.74, "grad_norm": 3.779777837942532, "learning_rate": 7.276820153263282e-06, "loss": 0.25, "step": 5426 }, { "epoch": 0.74, "grad_norm": 4.08924639754561, "learning_rate": 7.275839390893766e-06, "loss": 0.2215, "step": 5427 }, { "epoch": 0.74, "grad_norm": 4.03164242566486, "learning_rate": 7.274858518064946e-06, "loss": 0.2338, "step": 5428 }, { "epoch": 0.74, "grad_norm": 3.988305068369606, "learning_rate": 7.273877534824433e-06, "loss": 0.2412, "step": 5429 }, { "epoch": 0.74, "grad_norm": 2.958346266709583, "learning_rate": 7.272896441219833e-06, "loss": 0.2345, "step": 5430 }, { "epoch": 0.74, "grad_norm": 2.4596950018819115, "learning_rate": 7.271915237298767e-06, "loss": 0.2152, "step": 5431 }, { "epoch": 0.74, "grad_norm": 3.5498199688635728, "learning_rate": 7.270933923108857e-06, "loss": 0.2691, "step": 5432 }, { "epoch": 0.74, "grad_norm": 3.544710937692104, "learning_rate": 7.269952498697734e-06, "loss": 0.2696, "step": 5433 }, { "epoch": 0.74, "grad_norm": 3.418564212733988, "learning_rate": 7.268970964113033e-06, "loss": 0.2715, "step": 5434 }, { "epoch": 0.74, "grad_norm": 3.3061956285252334, "learning_rate": 7.2679893194023885e-06, "loss": 0.2142, "step": 5435 }, { "epoch": 0.74, "grad_norm": 3.7140868278373107, "learning_rate": 7.26700756461345e-06, "loss": 0.2511, "step": 5436 }, { "epoch": 0.74, "grad_norm": 3.3730227100598733, "learning_rate": 7.2660256997938635e-06, "loss": 0.2212, "step": 5437 }, { "epoch": 0.74, "grad_norm": 3.382823757601331, "learning_rate": 7.26504372499129e-06, "loss": 0.1928, "step": 5438 }, { "epoch": 0.74, "grad_norm": 4.265315893365066, "learning_rate": 7.264061640253383e-06, "loss": 0.2615, "step": 5439 }, { "epoch": 0.74, "grad_norm": 3.3093002395128144, "learning_rate": 7.263079445627814e-06, "loss": 0.2081, "step": 5440 }, { "epoch": 0.74, "grad_norm": 3.460768382705529, "learning_rate": 7.262097141162254e-06, "loss": 0.2318, "step": 5441 }, { "epoch": 0.74, "grad_norm": 3.5320588849728654, "learning_rate": 7.261114726904379e-06, "loss": 0.2721, "step": 5442 }, { "epoch": 0.74, "grad_norm": 3.49158791537203, "learning_rate": 7.26013220290187e-06, "loss": 0.2414, "step": 5443 }, { "epoch": 0.74, "grad_norm": 4.029802108380292, "learning_rate": 7.259149569202415e-06, "loss": 0.2409, "step": 5444 }, { "epoch": 0.74, "grad_norm": 3.196191505269043, "learning_rate": 7.258166825853707e-06, "loss": 0.1938, "step": 5445 }, { "epoch": 0.74, "grad_norm": 3.6971269667101634, "learning_rate": 7.257183972903443e-06, "loss": 0.2264, "step": 5446 }, { "epoch": 0.74, "grad_norm": 3.1630507725871078, "learning_rate": 7.25620101039933e-06, "loss": 0.206, "step": 5447 }, { "epoch": 0.74, "grad_norm": 3.291593261741443, "learning_rate": 7.255217938389072e-06, "loss": 0.2292, "step": 5448 }, { "epoch": 0.74, "grad_norm": 3.4763764700108934, "learning_rate": 7.254234756920386e-06, "loss": 0.2393, "step": 5449 }, { "epoch": 0.74, "grad_norm": 3.9737106520249013, "learning_rate": 7.253251466040992e-06, "loss": 0.2573, "step": 5450 }, { "epoch": 0.74, "grad_norm": 3.80628552625459, "learning_rate": 7.252268065798611e-06, "loss": 0.2595, "step": 5451 }, { "epoch": 0.74, "grad_norm": 3.6181640605220857, "learning_rate": 7.2512845562409764e-06, "loss": 0.2508, "step": 5452 }, { "epoch": 0.74, "grad_norm": 4.093921111935356, "learning_rate": 7.2503009374158205e-06, "loss": 0.2741, "step": 5453 }, { "epoch": 0.74, "grad_norm": 3.765542918320864, "learning_rate": 7.249317209370886e-06, "loss": 0.2609, "step": 5454 }, { "epoch": 0.74, "grad_norm": 3.911874548688036, "learning_rate": 7.2483333721539195e-06, "loss": 0.2227, "step": 5455 }, { "epoch": 0.74, "grad_norm": 3.9218499304477143, "learning_rate": 7.247349425812671e-06, "loss": 0.2295, "step": 5456 }, { "epoch": 0.74, "grad_norm": 3.337823352828477, "learning_rate": 7.246365370394896e-06, "loss": 0.2026, "step": 5457 }, { "epoch": 0.74, "grad_norm": 3.405312502140218, "learning_rate": 7.245381205948357e-06, "loss": 0.2503, "step": 5458 }, { "epoch": 0.74, "grad_norm": 3.701851818155378, "learning_rate": 7.244396932520823e-06, "loss": 0.249, "step": 5459 }, { "epoch": 0.74, "grad_norm": 3.558082293215266, "learning_rate": 7.243412550160061e-06, "loss": 0.2668, "step": 5460 }, { "epoch": 0.74, "grad_norm": 3.4112796552730837, "learning_rate": 7.242428058913855e-06, "loss": 0.2609, "step": 5461 }, { "epoch": 0.74, "grad_norm": 3.514775452635179, "learning_rate": 7.241443458829985e-06, "loss": 0.2226, "step": 5462 }, { "epoch": 0.74, "grad_norm": 4.074798477539717, "learning_rate": 7.240458749956238e-06, "loss": 0.323, "step": 5463 }, { "epoch": 0.74, "grad_norm": 3.7427497915455206, "learning_rate": 7.2394739323404105e-06, "loss": 0.2862, "step": 5464 }, { "epoch": 0.74, "grad_norm": 3.613713020690559, "learning_rate": 7.238489006030298e-06, "loss": 0.2723, "step": 5465 }, { "epoch": 0.74, "grad_norm": 3.860797668949846, "learning_rate": 7.237503971073707e-06, "loss": 0.2926, "step": 5466 }, { "epoch": 0.74, "grad_norm": 3.187821972133948, "learning_rate": 7.236518827518444e-06, "loss": 0.2313, "step": 5467 }, { "epoch": 0.74, "grad_norm": 4.30492381136328, "learning_rate": 7.235533575412327e-06, "loss": 0.2892, "step": 5468 }, { "epoch": 0.74, "grad_norm": 4.206789463177754, "learning_rate": 7.234548214803174e-06, "loss": 0.2667, "step": 5469 }, { "epoch": 0.74, "grad_norm": 3.6610612968858858, "learning_rate": 7.233562745738811e-06, "loss": 0.2762, "step": 5470 }, { "epoch": 0.74, "grad_norm": 3.5986757807170267, "learning_rate": 7.232577168267066e-06, "loss": 0.26, "step": 5471 }, { "epoch": 0.74, "grad_norm": 3.2426194582857386, "learning_rate": 7.231591482435777e-06, "loss": 0.233, "step": 5472 }, { "epoch": 0.74, "grad_norm": 4.790475282196458, "learning_rate": 7.230605688292785e-06, "loss": 0.206, "step": 5473 }, { "epoch": 0.74, "grad_norm": 4.021669287967819, "learning_rate": 7.229619785885933e-06, "loss": 0.2875, "step": 5474 }, { "epoch": 0.74, "grad_norm": 3.875393441548491, "learning_rate": 7.2286337752630765e-06, "loss": 0.1837, "step": 5475 }, { "epoch": 0.74, "grad_norm": 3.5400169378197575, "learning_rate": 7.22764765647207e-06, "loss": 0.1952, "step": 5476 }, { "epoch": 0.75, "grad_norm": 3.7621105727815234, "learning_rate": 7.226661429560776e-06, "loss": 0.2285, "step": 5477 }, { "epoch": 0.75, "grad_norm": 3.5094153238545416, "learning_rate": 7.225675094577061e-06, "loss": 0.2444, "step": 5478 }, { "epoch": 0.75, "grad_norm": 3.3672287352720085, "learning_rate": 7.224688651568797e-06, "loss": 0.2413, "step": 5479 }, { "epoch": 0.75, "grad_norm": 3.355996619222763, "learning_rate": 7.223702100583864e-06, "loss": 0.2905, "step": 5480 }, { "epoch": 0.75, "grad_norm": 3.3595795129872577, "learning_rate": 7.222715441670143e-06, "loss": 0.2174, "step": 5481 }, { "epoch": 0.75, "grad_norm": 2.830640829956155, "learning_rate": 7.221728674875522e-06, "loss": 0.2323, "step": 5482 }, { "epoch": 0.75, "grad_norm": 3.4077850180246303, "learning_rate": 7.2207418002478945e-06, "loss": 0.2258, "step": 5483 }, { "epoch": 0.75, "grad_norm": 3.7119587322667718, "learning_rate": 7.219754817835161e-06, "loss": 0.2635, "step": 5484 }, { "epoch": 0.75, "grad_norm": 3.8640986564994386, "learning_rate": 7.218767727685221e-06, "loss": 0.2617, "step": 5485 }, { "epoch": 0.75, "grad_norm": 3.815513851518198, "learning_rate": 7.217780529845987e-06, "loss": 0.273, "step": 5486 }, { "epoch": 0.75, "grad_norm": 3.608737724038586, "learning_rate": 7.216793224365373e-06, "loss": 0.1676, "step": 5487 }, { "epoch": 0.75, "grad_norm": 3.310169243428941, "learning_rate": 7.2158058112912985e-06, "loss": 0.2469, "step": 5488 }, { "epoch": 0.75, "grad_norm": 3.2768613213357356, "learning_rate": 7.214818290671687e-06, "loss": 0.2726, "step": 5489 }, { "epoch": 0.75, "grad_norm": 3.3339627920177386, "learning_rate": 7.213830662554469e-06, "loss": 0.2041, "step": 5490 }, { "epoch": 0.75, "grad_norm": 3.3506428034198708, "learning_rate": 7.212842926987581e-06, "loss": 0.2151, "step": 5491 }, { "epoch": 0.75, "grad_norm": 3.4891522438674416, "learning_rate": 7.2118550840189605e-06, "loss": 0.2554, "step": 5492 }, { "epoch": 0.75, "grad_norm": 4.585989109350837, "learning_rate": 7.210867133696555e-06, "loss": 0.2565, "step": 5493 }, { "epoch": 0.75, "grad_norm": 5.494022995525774, "learning_rate": 7.209879076068316e-06, "loss": 0.2366, "step": 5494 }, { "epoch": 0.75, "grad_norm": 4.7073573571665985, "learning_rate": 7.208890911182198e-06, "loss": 0.2418, "step": 5495 }, { "epoch": 0.75, "grad_norm": 3.9806058948795564, "learning_rate": 7.207902639086163e-06, "loss": 0.2264, "step": 5496 }, { "epoch": 0.75, "grad_norm": 2.873505479559965, "learning_rate": 7.206914259828177e-06, "loss": 0.2184, "step": 5497 }, { "epoch": 0.75, "grad_norm": 4.125425372657707, "learning_rate": 7.205925773456211e-06, "loss": 0.2274, "step": 5498 }, { "epoch": 0.75, "grad_norm": 3.576937251726278, "learning_rate": 7.204937180018243e-06, "loss": 0.2287, "step": 5499 }, { "epoch": 0.75, "grad_norm": 3.49493802613702, "learning_rate": 7.203948479562255e-06, "loss": 0.2505, "step": 5500 }, { "epoch": 0.75, "grad_norm": 3.441422417651655, "learning_rate": 7.202959672136234e-06, "loss": 0.1772, "step": 5501 }, { "epoch": 0.75, "grad_norm": 2.9501553620400998, "learning_rate": 7.201970757788172e-06, "loss": 0.203, "step": 5502 }, { "epoch": 0.75, "grad_norm": 2.914742077488631, "learning_rate": 7.200981736566066e-06, "loss": 0.197, "step": 5503 }, { "epoch": 0.75, "grad_norm": 3.3965530981377894, "learning_rate": 7.199992608517921e-06, "loss": 0.2464, "step": 5504 }, { "epoch": 0.75, "grad_norm": 3.599391199316584, "learning_rate": 7.199003373691743e-06, "loss": 0.277, "step": 5505 }, { "epoch": 0.75, "grad_norm": 3.34187772587252, "learning_rate": 7.198014032135544e-06, "loss": 0.2413, "step": 5506 }, { "epoch": 0.75, "grad_norm": 3.2967288353671234, "learning_rate": 7.197024583897345e-06, "loss": 0.1926, "step": 5507 }, { "epoch": 0.75, "grad_norm": 4.5883016047364835, "learning_rate": 7.196035029025169e-06, "loss": 0.2614, "step": 5508 }, { "epoch": 0.75, "grad_norm": 3.713638535084182, "learning_rate": 7.195045367567044e-06, "loss": 0.2378, "step": 5509 }, { "epoch": 0.75, "grad_norm": 3.089501010262968, "learning_rate": 7.194055599571005e-06, "loss": 0.2459, "step": 5510 }, { "epoch": 0.75, "grad_norm": 3.7576400941345702, "learning_rate": 7.1930657250850886e-06, "loss": 0.2563, "step": 5511 }, { "epoch": 0.75, "grad_norm": 4.028763119340169, "learning_rate": 7.19207574415734e-06, "loss": 0.2423, "step": 5512 }, { "epoch": 0.75, "grad_norm": 3.9193820575377756, "learning_rate": 7.191085656835809e-06, "loss": 0.2392, "step": 5513 }, { "epoch": 0.75, "grad_norm": 3.1863826522801104, "learning_rate": 7.19009546316855e-06, "loss": 0.2406, "step": 5514 }, { "epoch": 0.75, "grad_norm": 3.2317928245524077, "learning_rate": 7.189105163203625e-06, "loss": 0.2234, "step": 5515 }, { "epoch": 0.75, "grad_norm": 3.2698142631061167, "learning_rate": 7.188114756989094e-06, "loss": 0.241, "step": 5516 }, { "epoch": 0.75, "grad_norm": 3.8928701774905963, "learning_rate": 7.187124244573029e-06, "loss": 0.3022, "step": 5517 }, { "epoch": 0.75, "grad_norm": 4.637583595789733, "learning_rate": 7.186133626003506e-06, "loss": 0.3177, "step": 5518 }, { "epoch": 0.75, "grad_norm": 3.8510362120803476, "learning_rate": 7.185142901328607e-06, "loss": 0.1957, "step": 5519 }, { "epoch": 0.75, "grad_norm": 3.932460262065654, "learning_rate": 7.184152070596414e-06, "loss": 0.2372, "step": 5520 }, { "epoch": 0.75, "grad_norm": 3.3145039438548363, "learning_rate": 7.18316113385502e-06, "loss": 0.2411, "step": 5521 }, { "epoch": 0.75, "grad_norm": 3.610999970211563, "learning_rate": 7.182170091152518e-06, "loss": 0.1991, "step": 5522 }, { "epoch": 0.75, "grad_norm": 3.514445991418674, "learning_rate": 7.181178942537013e-06, "loss": 0.2678, "step": 5523 }, { "epoch": 0.75, "grad_norm": 3.721858667845633, "learning_rate": 7.180187688056607e-06, "loss": 0.2231, "step": 5524 }, { "epoch": 0.75, "grad_norm": 3.1857730651401774, "learning_rate": 7.179196327759412e-06, "loss": 0.2244, "step": 5525 }, { "epoch": 0.75, "grad_norm": 3.7391378934525363, "learning_rate": 7.178204861693546e-06, "loss": 0.2714, "step": 5526 }, { "epoch": 0.75, "grad_norm": 3.3963396437432096, "learning_rate": 7.17721328990713e-06, "loss": 0.1942, "step": 5527 }, { "epoch": 0.75, "grad_norm": 3.911268614151307, "learning_rate": 7.1762216124482905e-06, "loss": 0.2721, "step": 5528 }, { "epoch": 0.75, "grad_norm": 3.8563032688067116, "learning_rate": 7.175229829365158e-06, "loss": 0.3061, "step": 5529 }, { "epoch": 0.75, "grad_norm": 4.041380166030084, "learning_rate": 7.1742379407058705e-06, "loss": 0.2445, "step": 5530 }, { "epoch": 0.75, "grad_norm": 3.568555831311465, "learning_rate": 7.17324594651857e-06, "loss": 0.3015, "step": 5531 }, { "epoch": 0.75, "grad_norm": 3.871662541476074, "learning_rate": 7.1722538468514015e-06, "loss": 0.2467, "step": 5532 }, { "epoch": 0.75, "grad_norm": 3.540225041240485, "learning_rate": 7.1712616417525205e-06, "loss": 0.3024, "step": 5533 }, { "epoch": 0.75, "grad_norm": 3.147726273310862, "learning_rate": 7.170269331270082e-06, "loss": 0.2564, "step": 5534 }, { "epoch": 0.75, "grad_norm": 3.419621272020203, "learning_rate": 7.169276915452249e-06, "loss": 0.2173, "step": 5535 }, { "epoch": 0.75, "grad_norm": 3.6572121433234206, "learning_rate": 7.168284394347189e-06, "loss": 0.2317, "step": 5536 }, { "epoch": 0.75, "grad_norm": 4.429488559512734, "learning_rate": 7.167291768003075e-06, "loss": 0.301, "step": 5537 }, { "epoch": 0.75, "grad_norm": 3.739519425692206, "learning_rate": 7.1662990364680854e-06, "loss": 0.2973, "step": 5538 }, { "epoch": 0.75, "grad_norm": 3.577449596019056, "learning_rate": 7.1653061997904e-06, "loss": 0.2205, "step": 5539 }, { "epoch": 0.75, "grad_norm": 3.854428126551123, "learning_rate": 7.164313258018209e-06, "loss": 0.257, "step": 5540 }, { "epoch": 0.75, "grad_norm": 4.968525814886993, "learning_rate": 7.163320211199707e-06, "loss": 0.2558, "step": 5541 }, { "epoch": 0.75, "grad_norm": 3.6664587273020195, "learning_rate": 7.162327059383089e-06, "loss": 0.2525, "step": 5542 }, { "epoch": 0.75, "grad_norm": 3.344798151207875, "learning_rate": 7.161333802616561e-06, "loss": 0.2647, "step": 5543 }, { "epoch": 0.75, "grad_norm": 3.8587144531816886, "learning_rate": 7.1603404409483305e-06, "loss": 0.261, "step": 5544 }, { "epoch": 0.75, "grad_norm": 3.40529013689945, "learning_rate": 7.1593469744266096e-06, "loss": 0.2061, "step": 5545 }, { "epoch": 0.75, "grad_norm": 3.231789449507347, "learning_rate": 7.158353403099617e-06, "loss": 0.2212, "step": 5546 }, { "epoch": 0.75, "grad_norm": 3.933341756972281, "learning_rate": 7.157359727015578e-06, "loss": 0.2883, "step": 5547 }, { "epoch": 0.75, "grad_norm": 3.4821645332081994, "learning_rate": 7.156365946222721e-06, "loss": 0.2, "step": 5548 }, { "epoch": 0.75, "grad_norm": 3.526145870490163, "learning_rate": 7.155372060769278e-06, "loss": 0.1976, "step": 5549 }, { "epoch": 0.75, "grad_norm": 4.150012853513464, "learning_rate": 7.1543780707034904e-06, "loss": 0.2601, "step": 5550 }, { "epoch": 0.76, "grad_norm": 3.3657462255138544, "learning_rate": 7.153383976073599e-06, "loss": 0.205, "step": 5551 }, { "epoch": 0.76, "grad_norm": 4.463003598917536, "learning_rate": 7.152389776927855e-06, "loss": 0.255, "step": 5552 }, { "epoch": 0.76, "grad_norm": 5.014226465828631, "learning_rate": 7.151395473314512e-06, "loss": 0.2967, "step": 5553 }, { "epoch": 0.76, "grad_norm": 2.815857034682642, "learning_rate": 7.15040106528183e-06, "loss": 0.1915, "step": 5554 }, { "epoch": 0.76, "grad_norm": 3.1757286178084114, "learning_rate": 7.149406552878072e-06, "loss": 0.2393, "step": 5555 }, { "epoch": 0.76, "grad_norm": 3.835742629188371, "learning_rate": 7.148411936151507e-06, "loss": 0.2501, "step": 5556 }, { "epoch": 0.76, "grad_norm": 3.5956780160232813, "learning_rate": 7.147417215150411e-06, "loss": 0.2412, "step": 5557 }, { "epoch": 0.76, "grad_norm": 3.49431712117321, "learning_rate": 7.1464223899230615e-06, "loss": 0.2582, "step": 5558 }, { "epoch": 0.76, "grad_norm": 3.5011904174074315, "learning_rate": 7.145427460517744e-06, "loss": 0.2122, "step": 5559 }, { "epoch": 0.76, "grad_norm": 3.350131689502989, "learning_rate": 7.144432426982748e-06, "loss": 0.1834, "step": 5560 }, { "epoch": 0.76, "grad_norm": 3.6158306371437203, "learning_rate": 7.143437289366368e-06, "loss": 0.2967, "step": 5561 }, { "epoch": 0.76, "grad_norm": 3.5167491367711254, "learning_rate": 7.142442047716905e-06, "loss": 0.2448, "step": 5562 }, { "epoch": 0.76, "grad_norm": 3.362488528333752, "learning_rate": 7.141446702082661e-06, "loss": 0.2932, "step": 5563 }, { "epoch": 0.76, "grad_norm": 3.5851905587771995, "learning_rate": 7.140451252511949e-06, "loss": 0.2325, "step": 5564 }, { "epoch": 0.76, "grad_norm": 3.5543644610921548, "learning_rate": 7.13945569905308e-06, "loss": 0.2052, "step": 5565 }, { "epoch": 0.76, "grad_norm": 3.9470895917681013, "learning_rate": 7.138460041754375e-06, "loss": 0.243, "step": 5566 }, { "epoch": 0.76, "grad_norm": 3.703718361818386, "learning_rate": 7.13746428066416e-06, "loss": 0.3352, "step": 5567 }, { "epoch": 0.76, "grad_norm": 4.2612187583294885, "learning_rate": 7.136468415830766e-06, "loss": 0.2063, "step": 5568 }, { "epoch": 0.76, "grad_norm": 3.9053098048598276, "learning_rate": 7.135472447302525e-06, "loss": 0.2504, "step": 5569 }, { "epoch": 0.76, "grad_norm": 3.5813336230152877, "learning_rate": 7.134476375127779e-06, "loss": 0.2267, "step": 5570 }, { "epoch": 0.76, "grad_norm": 4.075810276243725, "learning_rate": 7.133480199354872e-06, "loss": 0.2552, "step": 5571 }, { "epoch": 0.76, "grad_norm": 3.3755207693717284, "learning_rate": 7.132483920032154e-06, "loss": 0.2361, "step": 5572 }, { "epoch": 0.76, "grad_norm": 3.339094358198023, "learning_rate": 7.131487537207981e-06, "loss": 0.2119, "step": 5573 }, { "epoch": 0.76, "grad_norm": 3.1726472402963477, "learning_rate": 7.1304910509307114e-06, "loss": 0.24, "step": 5574 }, { "epoch": 0.76, "grad_norm": 3.4159143477405967, "learning_rate": 7.129494461248713e-06, "loss": 0.2272, "step": 5575 }, { "epoch": 0.76, "grad_norm": 4.135799619794673, "learning_rate": 7.128497768210353e-06, "loss": 0.2468, "step": 5576 }, { "epoch": 0.76, "grad_norm": 3.4426307357455763, "learning_rate": 7.127500971864008e-06, "loss": 0.2401, "step": 5577 }, { "epoch": 0.76, "grad_norm": 3.4092682782670236, "learning_rate": 7.126504072258058e-06, "loss": 0.2477, "step": 5578 }, { "epoch": 0.76, "grad_norm": 3.243026925678752, "learning_rate": 7.1255070694408865e-06, "loss": 0.2361, "step": 5579 }, { "epoch": 0.76, "grad_norm": 3.169043629035146, "learning_rate": 7.124509963460888e-06, "loss": 0.24, "step": 5580 }, { "epoch": 0.76, "grad_norm": 4.625451744936279, "learning_rate": 7.123512754366454e-06, "loss": 0.2578, "step": 5581 }, { "epoch": 0.76, "grad_norm": 4.226292318777825, "learning_rate": 7.122515442205985e-06, "loss": 0.2593, "step": 5582 }, { "epoch": 0.76, "grad_norm": 4.146202060954784, "learning_rate": 7.1215180270278875e-06, "loss": 0.2953, "step": 5583 }, { "epoch": 0.76, "grad_norm": 4.090573368570592, "learning_rate": 7.1205205088805705e-06, "loss": 0.2963, "step": 5584 }, { "epoch": 0.76, "grad_norm": 3.7439669282251375, "learning_rate": 7.119522887812449e-06, "loss": 0.2876, "step": 5585 }, { "epoch": 0.76, "grad_norm": 3.4211805005411073, "learning_rate": 7.118525163871945e-06, "loss": 0.2741, "step": 5586 }, { "epoch": 0.76, "grad_norm": 3.6157879929193055, "learning_rate": 7.117527337107481e-06, "loss": 0.2341, "step": 5587 }, { "epoch": 0.76, "grad_norm": 3.8339021242630635, "learning_rate": 7.116529407567489e-06, "loss": 0.2071, "step": 5588 }, { "epoch": 0.76, "grad_norm": 3.5527667242755974, "learning_rate": 7.115531375300404e-06, "loss": 0.2289, "step": 5589 }, { "epoch": 0.76, "grad_norm": 3.9070594027210563, "learning_rate": 7.114533240354668e-06, "loss": 0.2618, "step": 5590 }, { "epoch": 0.76, "grad_norm": 3.4282845262326025, "learning_rate": 7.113535002778721e-06, "loss": 0.2229, "step": 5591 }, { "epoch": 0.76, "grad_norm": 3.143458646764904, "learning_rate": 7.112536662621017e-06, "loss": 0.2086, "step": 5592 }, { "epoch": 0.76, "grad_norm": 3.4037101780250385, "learning_rate": 7.111538219930009e-06, "loss": 0.1915, "step": 5593 }, { "epoch": 0.76, "grad_norm": 4.3985399979453765, "learning_rate": 7.11053967475416e-06, "loss": 0.258, "step": 5594 }, { "epoch": 0.76, "grad_norm": 4.086553896720943, "learning_rate": 7.109541027141933e-06, "loss": 0.2148, "step": 5595 }, { "epoch": 0.76, "grad_norm": 3.548338457050653, "learning_rate": 7.108542277141798e-06, "loss": 0.2398, "step": 5596 }, { "epoch": 0.76, "grad_norm": 3.4809365277358664, "learning_rate": 7.10754342480223e-06, "loss": 0.2057, "step": 5597 }, { "epoch": 0.76, "grad_norm": 3.392904153679417, "learning_rate": 7.106544470171708e-06, "loss": 0.2197, "step": 5598 }, { "epoch": 0.76, "grad_norm": 3.733322679141447, "learning_rate": 7.105545413298719e-06, "loss": 0.2701, "step": 5599 }, { "epoch": 0.76, "grad_norm": 3.154609775839445, "learning_rate": 7.104546254231752e-06, "loss": 0.253, "step": 5600 }, { "epoch": 0.76, "grad_norm": 3.590380502878714, "learning_rate": 7.103546993019302e-06, "loss": 0.2712, "step": 5601 }, { "epoch": 0.76, "grad_norm": 3.176388986631003, "learning_rate": 7.102547629709867e-06, "loss": 0.2265, "step": 5602 }, { "epoch": 0.76, "grad_norm": 3.315069869922474, "learning_rate": 7.101548164351954e-06, "loss": 0.1998, "step": 5603 }, { "epoch": 0.76, "grad_norm": 3.0634289721479213, "learning_rate": 7.1005485969940725e-06, "loss": 0.2445, "step": 5604 }, { "epoch": 0.76, "grad_norm": 3.0520350520773705, "learning_rate": 7.099548927684735e-06, "loss": 0.2375, "step": 5605 }, { "epoch": 0.76, "grad_norm": 2.8672009717657705, "learning_rate": 7.098549156472463e-06, "loss": 0.2122, "step": 5606 }, { "epoch": 0.76, "grad_norm": 5.554222387010535, "learning_rate": 7.097549283405782e-06, "loss": 0.2764, "step": 5607 }, { "epoch": 0.76, "grad_norm": 3.7043388535013464, "learning_rate": 7.0965493085332185e-06, "loss": 0.2911, "step": 5608 }, { "epoch": 0.76, "grad_norm": 3.178531108374086, "learning_rate": 7.0955492319033094e-06, "loss": 0.1982, "step": 5609 }, { "epoch": 0.76, "grad_norm": 3.4415777715862816, "learning_rate": 7.094549053564592e-06, "loss": 0.2999, "step": 5610 }, { "epoch": 0.76, "grad_norm": 3.107125438011015, "learning_rate": 7.093548773565613e-06, "loss": 0.2462, "step": 5611 }, { "epoch": 0.76, "grad_norm": 4.363346185391726, "learning_rate": 7.092548391954919e-06, "loss": 0.2576, "step": 5612 }, { "epoch": 0.76, "grad_norm": 3.0833979279064723, "learning_rate": 7.091547908781067e-06, "loss": 0.2483, "step": 5613 }, { "epoch": 0.76, "grad_norm": 3.8144277656393375, "learning_rate": 7.090547324092614e-06, "loss": 0.2777, "step": 5614 }, { "epoch": 0.76, "grad_norm": 3.9845346925693406, "learning_rate": 7.089546637938124e-06, "loss": 0.2627, "step": 5615 }, { "epoch": 0.76, "grad_norm": 3.4465300837328843, "learning_rate": 7.088545850366167e-06, "loss": 0.2953, "step": 5616 }, { "epoch": 0.76, "grad_norm": 3.9304663742772643, "learning_rate": 7.087544961425317e-06, "loss": 0.2442, "step": 5617 }, { "epoch": 0.76, "grad_norm": 3.5105730598325904, "learning_rate": 7.08654397116415e-06, "loss": 0.2333, "step": 5618 }, { "epoch": 0.76, "grad_norm": 2.910275774199327, "learning_rate": 7.085542879631253e-06, "loss": 0.2209, "step": 5619 }, { "epoch": 0.76, "grad_norm": 3.4444146286643598, "learning_rate": 7.084541686875215e-06, "loss": 0.2225, "step": 5620 }, { "epoch": 0.76, "grad_norm": 3.7028771093477633, "learning_rate": 7.083540392944626e-06, "loss": 0.225, "step": 5621 }, { "epoch": 0.76, "grad_norm": 3.297156746377306, "learning_rate": 7.082538997888087e-06, "loss": 0.218, "step": 5622 }, { "epoch": 0.76, "grad_norm": 3.5929918038707482, "learning_rate": 7.081537501754202e-06, "loss": 0.1958, "step": 5623 }, { "epoch": 0.77, "grad_norm": 3.335499840235683, "learning_rate": 7.080535904591576e-06, "loss": 0.2308, "step": 5624 }, { "epoch": 0.77, "grad_norm": 4.273522158937113, "learning_rate": 7.0795342064488245e-06, "loss": 0.2717, "step": 5625 }, { "epoch": 0.77, "grad_norm": 3.178440592195577, "learning_rate": 7.078532407374567e-06, "loss": 0.211, "step": 5626 }, { "epoch": 0.77, "grad_norm": 4.721317743613384, "learning_rate": 7.077530507417423e-06, "loss": 0.2865, "step": 5627 }, { "epoch": 0.77, "grad_norm": 3.7613708892664426, "learning_rate": 7.076528506626022e-06, "loss": 0.2648, "step": 5628 }, { "epoch": 0.77, "grad_norm": 3.1945005214606392, "learning_rate": 7.075526405048998e-06, "loss": 0.2513, "step": 5629 }, { "epoch": 0.77, "grad_norm": 3.884499286587952, "learning_rate": 7.074524202734987e-06, "loss": 0.2182, "step": 5630 }, { "epoch": 0.77, "grad_norm": 2.8511246064935585, "learning_rate": 7.073521899732632e-06, "loss": 0.2169, "step": 5631 }, { "epoch": 0.77, "grad_norm": 3.795440125529297, "learning_rate": 7.07251949609058e-06, "loss": 0.2452, "step": 5632 }, { "epoch": 0.77, "grad_norm": 3.505344918984923, "learning_rate": 7.071516991857483e-06, "loss": 0.2344, "step": 5633 }, { "epoch": 0.77, "grad_norm": 3.5825328848198303, "learning_rate": 7.070514387081999e-06, "loss": 0.2216, "step": 5634 }, { "epoch": 0.77, "grad_norm": 3.2422995388232865, "learning_rate": 7.069511681812791e-06, "loss": 0.2246, "step": 5635 }, { "epoch": 0.77, "grad_norm": 3.7757971471942744, "learning_rate": 7.068508876098523e-06, "loss": 0.2322, "step": 5636 }, { "epoch": 0.77, "grad_norm": 3.926100359625087, "learning_rate": 7.067505969987869e-06, "loss": 0.2051, "step": 5637 }, { "epoch": 0.77, "grad_norm": 3.0687677813751773, "learning_rate": 7.0665029635295056e-06, "loss": 0.2437, "step": 5638 }, { "epoch": 0.77, "grad_norm": 3.1004097409106826, "learning_rate": 7.0654998567721144e-06, "loss": 0.2056, "step": 5639 }, { "epoch": 0.77, "grad_norm": 3.277366601425859, "learning_rate": 7.064496649764381e-06, "loss": 0.208, "step": 5640 }, { "epoch": 0.77, "grad_norm": 2.7788342539283364, "learning_rate": 7.063493342554998e-06, "loss": 0.2045, "step": 5641 }, { "epoch": 0.77, "grad_norm": 3.49517253775559, "learning_rate": 7.06248993519266e-06, "loss": 0.2739, "step": 5642 }, { "epoch": 0.77, "grad_norm": 4.104215690010139, "learning_rate": 7.061486427726068e-06, "loss": 0.2841, "step": 5643 }, { "epoch": 0.77, "grad_norm": 3.06089792404109, "learning_rate": 7.060482820203929e-06, "loss": 0.1891, "step": 5644 }, { "epoch": 0.77, "grad_norm": 3.0230415908133588, "learning_rate": 7.059479112674954e-06, "loss": 0.2005, "step": 5645 }, { "epoch": 0.77, "grad_norm": 3.1024531717360007, "learning_rate": 7.0584753051878574e-06, "loss": 0.2107, "step": 5646 }, { "epoch": 0.77, "grad_norm": 3.0153429972457046, "learning_rate": 7.05747139779136e-06, "loss": 0.1961, "step": 5647 }, { "epoch": 0.77, "grad_norm": 2.9028593999019776, "learning_rate": 7.056467390534187e-06, "loss": 0.2708, "step": 5648 }, { "epoch": 0.77, "grad_norm": 3.491370822751452, "learning_rate": 7.055463283465071e-06, "loss": 0.1955, "step": 5649 }, { "epoch": 0.77, "grad_norm": 3.3748855729569143, "learning_rate": 7.054459076632742e-06, "loss": 0.2607, "step": 5650 }, { "epoch": 0.77, "grad_norm": 3.2272806493478274, "learning_rate": 7.053454770085942e-06, "loss": 0.287, "step": 5651 }, { "epoch": 0.77, "grad_norm": 3.937341322553069, "learning_rate": 7.0524503638734175e-06, "loss": 0.2583, "step": 5652 }, { "epoch": 0.77, "grad_norm": 3.4774395271264757, "learning_rate": 7.051445858043917e-06, "loss": 0.2408, "step": 5653 }, { "epoch": 0.77, "grad_norm": 3.1752193847481034, "learning_rate": 7.050441252646194e-06, "loss": 0.2243, "step": 5654 }, { "epoch": 0.77, "grad_norm": 4.262974261684303, "learning_rate": 7.0494365477290086e-06, "loss": 0.2369, "step": 5655 }, { "epoch": 0.77, "grad_norm": 4.079906078632096, "learning_rate": 7.0484317433411245e-06, "loss": 0.2181, "step": 5656 }, { "epoch": 0.77, "grad_norm": 2.932256818185178, "learning_rate": 7.047426839531308e-06, "loss": 0.2342, "step": 5657 }, { "epoch": 0.77, "grad_norm": 3.672995182732852, "learning_rate": 7.046421836348336e-06, "loss": 0.2415, "step": 5658 }, { "epoch": 0.77, "grad_norm": 2.806277271577128, "learning_rate": 7.045416733840988e-06, "loss": 0.2281, "step": 5659 }, { "epoch": 0.77, "grad_norm": 3.433876055072924, "learning_rate": 7.044411532058044e-06, "loss": 0.2524, "step": 5660 }, { "epoch": 0.77, "grad_norm": 4.033588464214429, "learning_rate": 7.043406231048293e-06, "loss": 0.2843, "step": 5661 }, { "epoch": 0.77, "grad_norm": 4.435539866194086, "learning_rate": 7.04240083086053e-06, "loss": 0.2277, "step": 5662 }, { "epoch": 0.77, "grad_norm": 4.5908892219688875, "learning_rate": 7.041395331543549e-06, "loss": 0.216, "step": 5663 }, { "epoch": 0.77, "grad_norm": 3.336895107870436, "learning_rate": 7.040389733146156e-06, "loss": 0.2161, "step": 5664 }, { "epoch": 0.77, "grad_norm": 4.249483044857246, "learning_rate": 7.039384035717156e-06, "loss": 0.25, "step": 5665 }, { "epoch": 0.77, "grad_norm": 3.618076612094413, "learning_rate": 7.038378239305363e-06, "loss": 0.2346, "step": 5666 }, { "epoch": 0.77, "grad_norm": 4.231451434125914, "learning_rate": 7.037372343959592e-06, "loss": 0.2898, "step": 5667 }, { "epoch": 0.77, "grad_norm": 3.836620212687049, "learning_rate": 7.036366349728667e-06, "loss": 0.2707, "step": 5668 }, { "epoch": 0.77, "grad_norm": 4.10070686891836, "learning_rate": 7.0353602566614136e-06, "loss": 0.2258, "step": 5669 }, { "epoch": 0.77, "grad_norm": 2.915598692592186, "learning_rate": 7.034354064806662e-06, "loss": 0.212, "step": 5670 }, { "epoch": 0.77, "grad_norm": 3.499287490859004, "learning_rate": 7.033347774213251e-06, "loss": 0.2102, "step": 5671 }, { "epoch": 0.77, "grad_norm": 3.045395731285692, "learning_rate": 7.032341384930018e-06, "loss": 0.2173, "step": 5672 }, { "epoch": 0.77, "grad_norm": 3.336923273205775, "learning_rate": 7.031334897005811e-06, "loss": 0.301, "step": 5673 }, { "epoch": 0.77, "grad_norm": 3.7093894763412387, "learning_rate": 7.0303283104894815e-06, "loss": 0.2764, "step": 5674 }, { "epoch": 0.77, "grad_norm": 3.518848283807237, "learning_rate": 7.0293216254298825e-06, "loss": 0.2807, "step": 5675 }, { "epoch": 0.77, "grad_norm": 4.44020446752065, "learning_rate": 7.028314841875875e-06, "loss": 0.2255, "step": 5676 }, { "epoch": 0.77, "grad_norm": 3.2594834077753814, "learning_rate": 7.0273079598763236e-06, "loss": 0.2712, "step": 5677 }, { "epoch": 0.77, "grad_norm": 4.378042665785945, "learning_rate": 7.0263009794800985e-06, "loss": 0.3004, "step": 5678 }, { "epoch": 0.77, "grad_norm": 3.9219558306785363, "learning_rate": 7.025293900736074e-06, "loss": 0.1991, "step": 5679 }, { "epoch": 0.77, "grad_norm": 3.422278077384525, "learning_rate": 7.02428672369313e-06, "loss": 0.239, "step": 5680 }, { "epoch": 0.77, "grad_norm": 3.5266654096371663, "learning_rate": 7.0232794484001495e-06, "loss": 0.2709, "step": 5681 }, { "epoch": 0.77, "grad_norm": 2.915968997273736, "learning_rate": 7.022272074906021e-06, "loss": 0.2024, "step": 5682 }, { "epoch": 0.77, "grad_norm": 3.8334737600849103, "learning_rate": 7.021264603259639e-06, "loss": 0.2711, "step": 5683 }, { "epoch": 0.77, "grad_norm": 3.789620139137543, "learning_rate": 7.020257033509901e-06, "loss": 0.2507, "step": 5684 }, { "epoch": 0.77, "grad_norm": 2.883070830261022, "learning_rate": 7.01924936570571e-06, "loss": 0.2226, "step": 5685 }, { "epoch": 0.77, "grad_norm": 2.77093003242866, "learning_rate": 7.018241599895974e-06, "loss": 0.1971, "step": 5686 }, { "epoch": 0.77, "grad_norm": 3.7603233166742545, "learning_rate": 7.017233736129606e-06, "loss": 0.2343, "step": 5687 }, { "epoch": 0.77, "grad_norm": 3.149351412090653, "learning_rate": 7.016225774455523e-06, "loss": 0.2347, "step": 5688 }, { "epoch": 0.77, "grad_norm": 7.113295954271745, "learning_rate": 7.015217714922647e-06, "loss": 0.2405, "step": 5689 }, { "epoch": 0.77, "grad_norm": 3.720904842531045, "learning_rate": 7.014209557579905e-06, "loss": 0.2159, "step": 5690 }, { "epoch": 0.77, "grad_norm": 3.5953514131135877, "learning_rate": 7.013201302476227e-06, "loss": 0.2833, "step": 5691 }, { "epoch": 0.77, "grad_norm": 3.3531001297765166, "learning_rate": 7.012192949660552e-06, "loss": 0.2139, "step": 5692 }, { "epoch": 0.77, "grad_norm": 3.417669929986654, "learning_rate": 7.011184499181819e-06, "loss": 0.2082, "step": 5693 }, { "epoch": 0.77, "grad_norm": 4.261480629219687, "learning_rate": 7.010175951088976e-06, "loss": 0.219, "step": 5694 }, { "epoch": 0.77, "grad_norm": 3.983464424246226, "learning_rate": 7.00916730543097e-06, "loss": 0.2872, "step": 5695 }, { "epoch": 0.77, "grad_norm": 3.620009833683469, "learning_rate": 7.00815856225676e-06, "loss": 0.2021, "step": 5696 }, { "epoch": 0.77, "grad_norm": 3.8674165888631458, "learning_rate": 7.007149721615303e-06, "loss": 0.2749, "step": 5697 }, { "epoch": 0.78, "grad_norm": 4.4368651891864115, "learning_rate": 7.006140783555564e-06, "loss": 0.2402, "step": 5698 }, { "epoch": 0.78, "grad_norm": 4.506659713021614, "learning_rate": 7.005131748126515e-06, "loss": 0.2763, "step": 5699 }, { "epoch": 0.78, "grad_norm": 3.5445747702136794, "learning_rate": 7.004122615377128e-06, "loss": 0.2292, "step": 5700 }, { "epoch": 0.78, "grad_norm": 3.521447765400461, "learning_rate": 7.003113385356384e-06, "loss": 0.2489, "step": 5701 }, { "epoch": 0.78, "grad_norm": 3.994049114709142, "learning_rate": 7.002104058113264e-06, "loss": 0.2474, "step": 5702 }, { "epoch": 0.78, "grad_norm": 3.5032119806373405, "learning_rate": 7.001094633696757e-06, "loss": 0.2261, "step": 5703 }, { "epoch": 0.78, "grad_norm": 3.0879728717894572, "learning_rate": 7.000085112155857e-06, "loss": 0.2082, "step": 5704 }, { "epoch": 0.78, "grad_norm": 3.2746858905133087, "learning_rate": 6.9990754935395595e-06, "loss": 0.1842, "step": 5705 }, { "epoch": 0.78, "grad_norm": 2.984805442515156, "learning_rate": 6.99806577789687e-06, "loss": 0.1856, "step": 5706 }, { "epoch": 0.78, "grad_norm": 2.9893659660219, "learning_rate": 6.997055965276796e-06, "loss": 0.148, "step": 5707 }, { "epoch": 0.78, "grad_norm": 2.9706279859856055, "learning_rate": 6.996046055728345e-06, "loss": 0.21, "step": 5708 }, { "epoch": 0.78, "grad_norm": 2.9683682763326695, "learning_rate": 6.995036049300539e-06, "loss": 0.275, "step": 5709 }, { "epoch": 0.78, "grad_norm": 3.5195613461637336, "learning_rate": 6.994025946042395e-06, "loss": 0.2409, "step": 5710 }, { "epoch": 0.78, "grad_norm": 3.555274250717056, "learning_rate": 6.993015746002942e-06, "loss": 0.238, "step": 5711 }, { "epoch": 0.78, "grad_norm": 4.53301191925554, "learning_rate": 6.9920054492312086e-06, "loss": 0.2589, "step": 5712 }, { "epoch": 0.78, "grad_norm": 4.502924269085089, "learning_rate": 6.99099505577623e-06, "loss": 0.2824, "step": 5713 }, { "epoch": 0.78, "grad_norm": 4.392996704477138, "learning_rate": 6.989984565687049e-06, "loss": 0.2729, "step": 5714 }, { "epoch": 0.78, "grad_norm": 3.6372696408108434, "learning_rate": 6.988973979012709e-06, "loss": 0.2474, "step": 5715 }, { "epoch": 0.78, "grad_norm": 4.286938582338388, "learning_rate": 6.987963295802258e-06, "loss": 0.2254, "step": 5716 }, { "epoch": 0.78, "grad_norm": 3.2816286123489165, "learning_rate": 6.98695251610475e-06, "loss": 0.2349, "step": 5717 }, { "epoch": 0.78, "grad_norm": 3.1785422732869613, "learning_rate": 6.985941639969247e-06, "loss": 0.2207, "step": 5718 }, { "epoch": 0.78, "grad_norm": 4.528095808915803, "learning_rate": 6.984930667444809e-06, "loss": 0.2989, "step": 5719 }, { "epoch": 0.78, "grad_norm": 3.8027292051552837, "learning_rate": 6.9839195985805076e-06, "loss": 0.3215, "step": 5720 }, { "epoch": 0.78, "grad_norm": 4.24436101862978, "learning_rate": 6.982908433425414e-06, "loss": 0.2546, "step": 5721 }, { "epoch": 0.78, "grad_norm": 3.236960415616055, "learning_rate": 6.981897172028605e-06, "loss": 0.2579, "step": 5722 }, { "epoch": 0.78, "grad_norm": 3.0523529259220776, "learning_rate": 6.980885814439162e-06, "loss": 0.233, "step": 5723 }, { "epoch": 0.78, "grad_norm": 3.5697540442905784, "learning_rate": 6.979874360706174e-06, "loss": 0.233, "step": 5724 }, { "epoch": 0.78, "grad_norm": 3.455240998171832, "learning_rate": 6.978862810878733e-06, "loss": 0.1561, "step": 5725 }, { "epoch": 0.78, "grad_norm": 3.782214349750155, "learning_rate": 6.977851165005933e-06, "loss": 0.2973, "step": 5726 }, { "epoch": 0.78, "grad_norm": 3.4385229258168324, "learning_rate": 6.9768394231368765e-06, "loss": 0.2437, "step": 5727 }, { "epoch": 0.78, "grad_norm": 3.1153096316978677, "learning_rate": 6.9758275853206695e-06, "loss": 0.2527, "step": 5728 }, { "epoch": 0.78, "grad_norm": 3.4717051998739366, "learning_rate": 6.9748156516064195e-06, "loss": 0.2506, "step": 5729 }, { "epoch": 0.78, "grad_norm": 3.907794344923477, "learning_rate": 6.9738036220432426e-06, "loss": 0.2022, "step": 5730 }, { "epoch": 0.78, "grad_norm": 3.014899277600153, "learning_rate": 6.972791496680259e-06, "loss": 0.1943, "step": 5731 }, { "epoch": 0.78, "grad_norm": 3.3839063899831165, "learning_rate": 6.971779275566593e-06, "loss": 0.2297, "step": 5732 }, { "epoch": 0.78, "grad_norm": 3.8159168180449186, "learning_rate": 6.970766958751374e-06, "loss": 0.2502, "step": 5733 }, { "epoch": 0.78, "grad_norm": 3.3283258949372323, "learning_rate": 6.969754546283734e-06, "loss": 0.2867, "step": 5734 }, { "epoch": 0.78, "grad_norm": 3.529656941152318, "learning_rate": 6.96874203821281e-06, "loss": 0.2857, "step": 5735 }, { "epoch": 0.78, "grad_norm": 3.542785848372722, "learning_rate": 6.967729434587747e-06, "loss": 0.2483, "step": 5736 }, { "epoch": 0.78, "grad_norm": 3.4891183367431986, "learning_rate": 6.96671673545769e-06, "loss": 0.2437, "step": 5737 }, { "epoch": 0.78, "grad_norm": 3.336590276601188, "learning_rate": 6.9657039408717955e-06, "loss": 0.1883, "step": 5738 }, { "epoch": 0.78, "grad_norm": 3.3435121710815805, "learning_rate": 6.964691050879215e-06, "loss": 0.2272, "step": 5739 }, { "epoch": 0.78, "grad_norm": 2.799641729089716, "learning_rate": 6.963678065529112e-06, "loss": 0.2335, "step": 5740 }, { "epoch": 0.78, "grad_norm": 3.6038146734965366, "learning_rate": 6.962664984870653e-06, "loss": 0.2486, "step": 5741 }, { "epoch": 0.78, "grad_norm": 3.6650810073159517, "learning_rate": 6.961651808953008e-06, "loss": 0.2032, "step": 5742 }, { "epoch": 0.78, "grad_norm": 3.597087147175215, "learning_rate": 6.960638537825352e-06, "loss": 0.2615, "step": 5743 }, { "epoch": 0.78, "grad_norm": 3.648195598914499, "learning_rate": 6.959625171536863e-06, "loss": 0.2541, "step": 5744 }, { "epoch": 0.78, "grad_norm": 3.8309261983503267, "learning_rate": 6.958611710136728e-06, "loss": 0.2211, "step": 5745 }, { "epoch": 0.78, "grad_norm": 3.741364535602143, "learning_rate": 6.957598153674135e-06, "loss": 0.204, "step": 5746 }, { "epoch": 0.78, "grad_norm": 3.499828211859166, "learning_rate": 6.956584502198278e-06, "loss": 0.247, "step": 5747 }, { "epoch": 0.78, "grad_norm": 4.410914928874167, "learning_rate": 6.955570755758355e-06, "loss": 0.2076, "step": 5748 }, { "epoch": 0.78, "grad_norm": 2.9231237651638615, "learning_rate": 6.954556914403569e-06, "loss": 0.2115, "step": 5749 }, { "epoch": 0.78, "grad_norm": 3.745473192161924, "learning_rate": 6.953542978183126e-06, "loss": 0.2406, "step": 5750 }, { "epoch": 0.78, "grad_norm": 3.6819737043533376, "learning_rate": 6.952528947146241e-06, "loss": 0.2173, "step": 5751 }, { "epoch": 0.78, "grad_norm": 3.830222713041455, "learning_rate": 6.9515148213421265e-06, "loss": 0.2618, "step": 5752 }, { "epoch": 0.78, "grad_norm": 3.5368187389918364, "learning_rate": 6.950500600820009e-06, "loss": 0.2246, "step": 5753 }, { "epoch": 0.78, "grad_norm": 3.312408977486961, "learning_rate": 6.949486285629108e-06, "loss": 0.282, "step": 5754 }, { "epoch": 0.78, "grad_norm": 3.8204956260652163, "learning_rate": 6.948471875818658e-06, "loss": 0.257, "step": 5755 }, { "epoch": 0.78, "grad_norm": 3.767003709034243, "learning_rate": 6.9474573714378945e-06, "loss": 0.2741, "step": 5756 }, { "epoch": 0.78, "grad_norm": 3.6215951993672912, "learning_rate": 6.946442772536055e-06, "loss": 0.2235, "step": 5757 }, { "epoch": 0.78, "grad_norm": 3.928559678241314, "learning_rate": 6.945428079162385e-06, "loss": 0.2314, "step": 5758 }, { "epoch": 0.78, "grad_norm": 3.2828252876857693, "learning_rate": 6.944413291366133e-06, "loss": 0.2299, "step": 5759 }, { "epoch": 0.78, "grad_norm": 3.8300047918952353, "learning_rate": 6.943398409196553e-06, "loss": 0.2181, "step": 5760 }, { "epoch": 0.78, "grad_norm": 3.069111601620853, "learning_rate": 6.942383432702902e-06, "loss": 0.2398, "step": 5761 }, { "epoch": 0.78, "grad_norm": 3.2738451082730773, "learning_rate": 6.941368361934442e-06, "loss": 0.2365, "step": 5762 }, { "epoch": 0.78, "grad_norm": 3.2694812407946103, "learning_rate": 6.940353196940439e-06, "loss": 0.2239, "step": 5763 }, { "epoch": 0.78, "grad_norm": 3.4458790518039852, "learning_rate": 6.9393379377701694e-06, "loss": 0.2344, "step": 5764 }, { "epoch": 0.78, "grad_norm": 3.4858173595089976, "learning_rate": 6.938322584472906e-06, "loss": 0.2357, "step": 5765 }, { "epoch": 0.78, "grad_norm": 3.130764969391255, "learning_rate": 6.937307137097931e-06, "loss": 0.1849, "step": 5766 }, { "epoch": 0.78, "grad_norm": 3.1433183618338596, "learning_rate": 6.9362915956945264e-06, "loss": 0.2583, "step": 5767 }, { "epoch": 0.78, "grad_norm": 2.9214546111895165, "learning_rate": 6.935275960311987e-06, "loss": 0.1968, "step": 5768 }, { "epoch": 0.78, "grad_norm": 3.0884872757122843, "learning_rate": 6.9342602309996045e-06, "loss": 0.2317, "step": 5769 }, { "epoch": 0.78, "grad_norm": 4.2254757895027835, "learning_rate": 6.933244407806677e-06, "loss": 0.2563, "step": 5770 }, { "epoch": 0.79, "grad_norm": 4.534845735127357, "learning_rate": 6.932228490782511e-06, "loss": 0.2454, "step": 5771 }, { "epoch": 0.79, "grad_norm": 3.554978937854599, "learning_rate": 6.931212479976413e-06, "loss": 0.2187, "step": 5772 }, { "epoch": 0.79, "grad_norm": 3.2861954204170245, "learning_rate": 6.930196375437697e-06, "loss": 0.2074, "step": 5773 }, { "epoch": 0.79, "grad_norm": 3.964732016458483, "learning_rate": 6.9291801772156775e-06, "loss": 0.2491, "step": 5774 }, { "epoch": 0.79, "grad_norm": 4.054443147474703, "learning_rate": 6.928163885359679e-06, "loss": 0.2855, "step": 5775 }, { "epoch": 0.79, "grad_norm": 3.4597478558332027, "learning_rate": 6.927147499919027e-06, "loss": 0.2815, "step": 5776 }, { "epoch": 0.79, "grad_norm": 3.4600152373567172, "learning_rate": 6.9261310209430525e-06, "loss": 0.2421, "step": 5777 }, { "epoch": 0.79, "grad_norm": 4.115321837429576, "learning_rate": 6.925114448481089e-06, "loss": 0.233, "step": 5778 }, { "epoch": 0.79, "grad_norm": 3.3240087765062207, "learning_rate": 6.924097782582481e-06, "loss": 0.2803, "step": 5779 }, { "epoch": 0.79, "grad_norm": 3.386512168929593, "learning_rate": 6.923081023296569e-06, "loss": 0.2422, "step": 5780 }, { "epoch": 0.79, "grad_norm": 3.0708314271668424, "learning_rate": 6.922064170672705e-06, "loss": 0.2521, "step": 5781 }, { "epoch": 0.79, "grad_norm": 3.821562079057293, "learning_rate": 6.921047224760239e-06, "loss": 0.2624, "step": 5782 }, { "epoch": 0.79, "grad_norm": 3.423866464730208, "learning_rate": 6.920030185608532e-06, "loss": 0.2184, "step": 5783 }, { "epoch": 0.79, "grad_norm": 3.7187815298723987, "learning_rate": 6.919013053266944e-06, "loss": 0.2732, "step": 5784 }, { "epoch": 0.79, "grad_norm": 2.9483467292616803, "learning_rate": 6.917995827784846e-06, "loss": 0.2412, "step": 5785 }, { "epoch": 0.79, "grad_norm": 3.000169582680279, "learning_rate": 6.916978509211607e-06, "loss": 0.19, "step": 5786 }, { "epoch": 0.79, "grad_norm": 3.0000922947241446, "learning_rate": 6.9159610975966044e-06, "loss": 0.2075, "step": 5787 }, { "epoch": 0.79, "grad_norm": 2.9465492282594257, "learning_rate": 6.914943592989217e-06, "loss": 0.1768, "step": 5788 }, { "epoch": 0.79, "grad_norm": 2.859275810452976, "learning_rate": 6.913925995438833e-06, "loss": 0.234, "step": 5789 }, { "epoch": 0.79, "grad_norm": 3.1314396840913123, "learning_rate": 6.9129083049948385e-06, "loss": 0.2231, "step": 5790 }, { "epoch": 0.79, "grad_norm": 3.2538746742732436, "learning_rate": 6.911890521706631e-06, "loss": 0.2278, "step": 5791 }, { "epoch": 0.79, "grad_norm": 3.330536904380468, "learning_rate": 6.910872645623608e-06, "loss": 0.2641, "step": 5792 }, { "epoch": 0.79, "grad_norm": 3.2750284852274896, "learning_rate": 6.9098546767951725e-06, "loss": 0.2439, "step": 5793 }, { "epoch": 0.79, "grad_norm": 3.5155360476698854, "learning_rate": 6.908836615270732e-06, "loss": 0.2502, "step": 5794 }, { "epoch": 0.79, "grad_norm": 2.5610774031320016, "learning_rate": 6.9078184610997e-06, "loss": 0.1791, "step": 5795 }, { "epoch": 0.79, "grad_norm": 3.3759935984458984, "learning_rate": 6.90680021433149e-06, "loss": 0.2295, "step": 5796 }, { "epoch": 0.79, "grad_norm": 2.9980480880843747, "learning_rate": 6.905781875015529e-06, "loss": 0.2342, "step": 5797 }, { "epoch": 0.79, "grad_norm": 3.6317773956259485, "learning_rate": 6.904763443201239e-06, "loss": 0.2573, "step": 5798 }, { "epoch": 0.79, "grad_norm": 3.241915988513646, "learning_rate": 6.903744918938049e-06, "loss": 0.252, "step": 5799 }, { "epoch": 0.79, "grad_norm": 3.9112732974208737, "learning_rate": 6.902726302275397e-06, "loss": 0.2679, "step": 5800 }, { "epoch": 0.79, "grad_norm": 3.5014629927991763, "learning_rate": 6.901707593262721e-06, "loss": 0.2241, "step": 5801 }, { "epoch": 0.79, "grad_norm": 4.312148228139937, "learning_rate": 6.900688791949463e-06, "loss": 0.2572, "step": 5802 }, { "epoch": 0.79, "grad_norm": 4.030026498330948, "learning_rate": 6.899669898385072e-06, "loss": 0.2463, "step": 5803 }, { "epoch": 0.79, "grad_norm": 2.9916436610449666, "learning_rate": 6.898650912619003e-06, "loss": 0.2112, "step": 5804 }, { "epoch": 0.79, "grad_norm": 3.3151198481416553, "learning_rate": 6.89763183470071e-06, "loss": 0.2633, "step": 5805 }, { "epoch": 0.79, "grad_norm": 4.171059360824842, "learning_rate": 6.896612664679656e-06, "loss": 0.2642, "step": 5806 }, { "epoch": 0.79, "grad_norm": 3.5548324954751225, "learning_rate": 6.895593402605308e-06, "loss": 0.2391, "step": 5807 }, { "epoch": 0.79, "grad_norm": 3.326745853361705, "learning_rate": 6.894574048527134e-06, "loss": 0.2293, "step": 5808 }, { "epoch": 0.79, "grad_norm": 3.3572568841944466, "learning_rate": 6.893554602494611e-06, "loss": 0.2187, "step": 5809 }, { "epoch": 0.79, "grad_norm": 3.4067205891313255, "learning_rate": 6.892535064557218e-06, "loss": 0.2177, "step": 5810 }, { "epoch": 0.79, "grad_norm": 4.474020844192248, "learning_rate": 6.891515434764439e-06, "loss": 0.2793, "step": 5811 }, { "epoch": 0.79, "grad_norm": 3.8213245975550687, "learning_rate": 6.890495713165761e-06, "loss": 0.231, "step": 5812 }, { "epoch": 0.79, "grad_norm": 2.916065830031357, "learning_rate": 6.889475899810679e-06, "loss": 0.2172, "step": 5813 }, { "epoch": 0.79, "grad_norm": 4.002874128204417, "learning_rate": 6.8884559947486895e-06, "loss": 0.2725, "step": 5814 }, { "epoch": 0.79, "grad_norm": 3.38375785860904, "learning_rate": 6.8874359980292945e-06, "loss": 0.2266, "step": 5815 }, { "epoch": 0.79, "grad_norm": 3.123569740496574, "learning_rate": 6.886415909701999e-06, "loss": 0.2161, "step": 5816 }, { "epoch": 0.79, "grad_norm": 3.8957405324026615, "learning_rate": 6.885395729816313e-06, "loss": 0.2228, "step": 5817 }, { "epoch": 0.79, "grad_norm": 3.3154832669384717, "learning_rate": 6.884375458421755e-06, "loss": 0.1767, "step": 5818 }, { "epoch": 0.79, "grad_norm": 3.862637238395483, "learning_rate": 6.883355095567843e-06, "loss": 0.243, "step": 5819 }, { "epoch": 0.79, "grad_norm": 4.558154305060502, "learning_rate": 6.882334641304098e-06, "loss": 0.2482, "step": 5820 }, { "epoch": 0.79, "grad_norm": 3.443369319260945, "learning_rate": 6.881314095680053e-06, "loss": 0.2441, "step": 5821 }, { "epoch": 0.79, "grad_norm": 3.164554442033822, "learning_rate": 6.880293458745237e-06, "loss": 0.2239, "step": 5822 }, { "epoch": 0.79, "grad_norm": 3.933741409494879, "learning_rate": 6.879272730549191e-06, "loss": 0.23, "step": 5823 }, { "epoch": 0.79, "grad_norm": 3.7678181458576123, "learning_rate": 6.8782519111414515e-06, "loss": 0.2419, "step": 5824 }, { "epoch": 0.79, "grad_norm": 3.2538313254611895, "learning_rate": 6.87723100057157e-06, "loss": 0.2277, "step": 5825 }, { "epoch": 0.79, "grad_norm": 3.291130625383987, "learning_rate": 6.876209998889094e-06, "loss": 0.2589, "step": 5826 }, { "epoch": 0.79, "grad_norm": 3.3649004682898678, "learning_rate": 6.87518890614358e-06, "loss": 0.2199, "step": 5827 }, { "epoch": 0.79, "grad_norm": 3.7724512673333126, "learning_rate": 6.874167722384586e-06, "loss": 0.2934, "step": 5828 }, { "epoch": 0.79, "grad_norm": 3.256884389791327, "learning_rate": 6.873146447661676e-06, "loss": 0.2242, "step": 5829 }, { "epoch": 0.79, "grad_norm": 3.3611137361519536, "learning_rate": 6.872125082024419e-06, "loss": 0.2364, "step": 5830 }, { "epoch": 0.79, "grad_norm": 3.0873738972804925, "learning_rate": 6.871103625522388e-06, "loss": 0.2009, "step": 5831 }, { "epoch": 0.79, "grad_norm": 5.280870015988352, "learning_rate": 6.870082078205158e-06, "loss": 0.2867, "step": 5832 }, { "epoch": 0.79, "grad_norm": 3.6352394729390767, "learning_rate": 6.869060440122313e-06, "loss": 0.2365, "step": 5833 }, { "epoch": 0.79, "grad_norm": 3.165052626768162, "learning_rate": 6.868038711323437e-06, "loss": 0.2218, "step": 5834 }, { "epoch": 0.79, "grad_norm": 4.191699218278761, "learning_rate": 6.867016891858121e-06, "loss": 0.2625, "step": 5835 }, { "epoch": 0.79, "grad_norm": 4.26748493659887, "learning_rate": 6.865994981775958e-06, "loss": 0.3416, "step": 5836 }, { "epoch": 0.79, "grad_norm": 3.234904443022996, "learning_rate": 6.86497298112655e-06, "loss": 0.174, "step": 5837 }, { "epoch": 0.79, "grad_norm": 3.225453295213018, "learning_rate": 6.863950889959498e-06, "loss": 0.2364, "step": 5838 }, { "epoch": 0.79, "grad_norm": 3.6165544759821753, "learning_rate": 6.8629287083244114e-06, "loss": 0.2197, "step": 5839 }, { "epoch": 0.79, "grad_norm": 4.213114620777071, "learning_rate": 6.861906436270902e-06, "loss": 0.2456, "step": 5840 }, { "epoch": 0.79, "grad_norm": 3.06553049345766, "learning_rate": 6.860884073848586e-06, "loss": 0.2408, "step": 5841 }, { "epoch": 0.79, "grad_norm": 3.694646822849758, "learning_rate": 6.859861621107084e-06, "loss": 0.2577, "step": 5842 }, { "epoch": 0.79, "grad_norm": 3.4661214589543206, "learning_rate": 6.858839078096022e-06, "loss": 0.2095, "step": 5843 }, { "epoch": 0.79, "grad_norm": 3.448565236073126, "learning_rate": 6.8578164448650304e-06, "loss": 0.2312, "step": 5844 }, { "epoch": 0.8, "grad_norm": 3.0805458104232026, "learning_rate": 6.8567937214637435e-06, "loss": 0.2378, "step": 5845 }, { "epoch": 0.8, "grad_norm": 3.733809156735885, "learning_rate": 6.855770907941799e-06, "loss": 0.2424, "step": 5846 }, { "epoch": 0.8, "grad_norm": 3.5060953218613835, "learning_rate": 6.85474800434884e-06, "loss": 0.2348, "step": 5847 }, { "epoch": 0.8, "grad_norm": 3.541159417680003, "learning_rate": 6.853725010734513e-06, "loss": 0.3012, "step": 5848 }, { "epoch": 0.8, "grad_norm": 3.9603571315728483, "learning_rate": 6.852701927148471e-06, "loss": 0.2432, "step": 5849 }, { "epoch": 0.8, "grad_norm": 4.244730994128298, "learning_rate": 6.851678753640367e-06, "loss": 0.2198, "step": 5850 }, { "epoch": 0.8, "grad_norm": 3.8145392359198618, "learning_rate": 6.850655490259867e-06, "loss": 0.2643, "step": 5851 }, { "epoch": 0.8, "grad_norm": 3.14149790040268, "learning_rate": 6.849632137056631e-06, "loss": 0.2366, "step": 5852 }, { "epoch": 0.8, "grad_norm": 3.617992171664531, "learning_rate": 6.84860869408033e-06, "loss": 0.2211, "step": 5853 }, { "epoch": 0.8, "grad_norm": 3.7225990440457006, "learning_rate": 6.8475851613806364e-06, "loss": 0.2065, "step": 5854 }, { "epoch": 0.8, "grad_norm": 3.074248970839079, "learning_rate": 6.846561539007228e-06, "loss": 0.194, "step": 5855 }, { "epoch": 0.8, "grad_norm": 4.649415214064653, "learning_rate": 6.8455378270097875e-06, "loss": 0.256, "step": 5856 }, { "epoch": 0.8, "grad_norm": 3.636478853270369, "learning_rate": 6.844514025438003e-06, "loss": 0.2057, "step": 5857 }, { "epoch": 0.8, "grad_norm": 4.1457800754021665, "learning_rate": 6.8434901343415624e-06, "loss": 0.2346, "step": 5858 }, { "epoch": 0.8, "grad_norm": 3.339470296054973, "learning_rate": 6.8424661537701635e-06, "loss": 0.2357, "step": 5859 }, { "epoch": 0.8, "grad_norm": 4.2183752686533715, "learning_rate": 6.841442083773504e-06, "loss": 0.28, "step": 5860 }, { "epoch": 0.8, "grad_norm": 3.536376550381716, "learning_rate": 6.8404179244012895e-06, "loss": 0.2758, "step": 5861 }, { "epoch": 0.8, "grad_norm": 3.5143819864272365, "learning_rate": 6.8393936757032255e-06, "loss": 0.2433, "step": 5862 }, { "epoch": 0.8, "grad_norm": 3.621231613854887, "learning_rate": 6.8383693377290265e-06, "loss": 0.2325, "step": 5863 }, { "epoch": 0.8, "grad_norm": 3.355505738290279, "learning_rate": 6.83734491052841e-06, "loss": 0.2852, "step": 5864 }, { "epoch": 0.8, "grad_norm": 4.377330796667787, "learning_rate": 6.836320394151096e-06, "loss": 0.2981, "step": 5865 }, { "epoch": 0.8, "grad_norm": 4.5941271314908105, "learning_rate": 6.8352957886468104e-06, "loss": 0.3066, "step": 5866 }, { "epoch": 0.8, "grad_norm": 3.562749122539145, "learning_rate": 6.834271094065284e-06, "loss": 0.2697, "step": 5867 }, { "epoch": 0.8, "grad_norm": 3.5745735678245385, "learning_rate": 6.833246310456249e-06, "loss": 0.2154, "step": 5868 }, { "epoch": 0.8, "grad_norm": 3.5670413088794497, "learning_rate": 6.832221437869444e-06, "loss": 0.211, "step": 5869 }, { "epoch": 0.8, "grad_norm": 4.014226269567473, "learning_rate": 6.831196476354615e-06, "loss": 0.2593, "step": 5870 }, { "epoch": 0.8, "grad_norm": 3.4972235027142515, "learning_rate": 6.830171425961506e-06, "loss": 0.2009, "step": 5871 }, { "epoch": 0.8, "grad_norm": 4.250569658462557, "learning_rate": 6.82914628673987e-06, "loss": 0.2533, "step": 5872 }, { "epoch": 0.8, "grad_norm": 3.3402928248354637, "learning_rate": 6.8281210587394605e-06, "loss": 0.2226, "step": 5873 }, { "epoch": 0.8, "grad_norm": 3.233506832742775, "learning_rate": 6.827095742010042e-06, "loss": 0.2896, "step": 5874 }, { "epoch": 0.8, "grad_norm": 3.4775074270332333, "learning_rate": 6.826070336601373e-06, "loss": 0.1873, "step": 5875 }, { "epoch": 0.8, "grad_norm": 3.6279133485004635, "learning_rate": 6.825044842563227e-06, "loss": 0.238, "step": 5876 }, { "epoch": 0.8, "grad_norm": 3.7250846702059106, "learning_rate": 6.824019259945376e-06, "loss": 0.2474, "step": 5877 }, { "epoch": 0.8, "grad_norm": 4.332454946334211, "learning_rate": 6.8229935887975965e-06, "loss": 0.2464, "step": 5878 }, { "epoch": 0.8, "grad_norm": 3.511572659877051, "learning_rate": 6.82196782916967e-06, "loss": 0.2118, "step": 5879 }, { "epoch": 0.8, "grad_norm": 3.1943736224083454, "learning_rate": 6.820941981111383e-06, "loss": 0.2577, "step": 5880 }, { "epoch": 0.8, "grad_norm": 3.605382896095952, "learning_rate": 6.819916044672525e-06, "loss": 0.2765, "step": 5881 }, { "epoch": 0.8, "grad_norm": 3.8798637435400227, "learning_rate": 6.818890019902891e-06, "loss": 0.2357, "step": 5882 }, { "epoch": 0.8, "grad_norm": 3.3387916013807644, "learning_rate": 6.817863906852279e-06, "loss": 0.2708, "step": 5883 }, { "epoch": 0.8, "grad_norm": 4.176560042965101, "learning_rate": 6.816837705570494e-06, "loss": 0.2895, "step": 5884 }, { "epoch": 0.8, "grad_norm": 3.13380704656777, "learning_rate": 6.815811416107341e-06, "loss": 0.2353, "step": 5885 }, { "epoch": 0.8, "grad_norm": 3.565676653970858, "learning_rate": 6.814785038512633e-06, "loss": 0.2379, "step": 5886 }, { "epoch": 0.8, "grad_norm": 4.106142119616324, "learning_rate": 6.813758572836187e-06, "loss": 0.3449, "step": 5887 }, { "epoch": 0.8, "grad_norm": 3.4979047114151096, "learning_rate": 6.812732019127819e-06, "loss": 0.2449, "step": 5888 }, { "epoch": 0.8, "grad_norm": 3.5700828460002767, "learning_rate": 6.811705377437357e-06, "loss": 0.2913, "step": 5889 }, { "epoch": 0.8, "grad_norm": 3.5372665517610558, "learning_rate": 6.810678647814629e-06, "loss": 0.2362, "step": 5890 }, { "epoch": 0.8, "grad_norm": 3.5258882150554207, "learning_rate": 6.809651830309467e-06, "loss": 0.224, "step": 5891 }, { "epoch": 0.8, "grad_norm": 3.387555802058097, "learning_rate": 6.808624924971711e-06, "loss": 0.2074, "step": 5892 }, { "epoch": 0.8, "grad_norm": 3.899789017677801, "learning_rate": 6.8075979318511996e-06, "loss": 0.2507, "step": 5893 }, { "epoch": 0.8, "grad_norm": 3.807111054283519, "learning_rate": 6.806570850997779e-06, "loss": 0.2591, "step": 5894 }, { "epoch": 0.8, "grad_norm": 3.5979624879041454, "learning_rate": 6.805543682461299e-06, "loss": 0.2142, "step": 5895 }, { "epoch": 0.8, "grad_norm": 4.211688166465731, "learning_rate": 6.804516426291616e-06, "loss": 0.2386, "step": 5896 }, { "epoch": 0.8, "grad_norm": 4.169516277484372, "learning_rate": 6.803489082538586e-06, "loss": 0.2542, "step": 5897 }, { "epoch": 0.8, "grad_norm": 4.271070590967503, "learning_rate": 6.802461651252073e-06, "loss": 0.2338, "step": 5898 }, { "epoch": 0.8, "grad_norm": 4.059089661851081, "learning_rate": 6.801434132481945e-06, "loss": 0.2582, "step": 5899 }, { "epoch": 0.8, "grad_norm": 2.834866063415304, "learning_rate": 6.800406526278072e-06, "loss": 0.1976, "step": 5900 }, { "epoch": 0.8, "grad_norm": 3.3934702418348524, "learning_rate": 6.799378832690328e-06, "loss": 0.2577, "step": 5901 }, { "epoch": 0.8, "grad_norm": 4.160362110496555, "learning_rate": 6.798351051768597e-06, "loss": 0.2089, "step": 5902 }, { "epoch": 0.8, "grad_norm": 3.625442452204044, "learning_rate": 6.7973231835627586e-06, "loss": 0.2335, "step": 5903 }, { "epoch": 0.8, "grad_norm": 3.645127807576446, "learning_rate": 6.796295228122703e-06, "loss": 0.2123, "step": 5904 }, { "epoch": 0.8, "grad_norm": 3.2141914164012424, "learning_rate": 6.795267185498323e-06, "loss": 0.1861, "step": 5905 }, { "epoch": 0.8, "grad_norm": 3.6525617544326954, "learning_rate": 6.794239055739516e-06, "loss": 0.2322, "step": 5906 }, { "epoch": 0.8, "grad_norm": 5.565720321533645, "learning_rate": 6.79321083889618e-06, "loss": 0.3033, "step": 5907 }, { "epoch": 0.8, "grad_norm": 3.986844865199951, "learning_rate": 6.792182535018222e-06, "loss": 0.238, "step": 5908 }, { "epoch": 0.8, "grad_norm": 3.5115545152382044, "learning_rate": 6.791154144155552e-06, "loss": 0.194, "step": 5909 }, { "epoch": 0.8, "grad_norm": 3.1553602510877266, "learning_rate": 6.7901256663580825e-06, "loss": 0.2515, "step": 5910 }, { "epoch": 0.8, "grad_norm": 3.5556544451080225, "learning_rate": 6.789097101675733e-06, "loss": 0.1776, "step": 5911 }, { "epoch": 0.8, "grad_norm": 3.839904784010018, "learning_rate": 6.788068450158422e-06, "loss": 0.2187, "step": 5912 }, { "epoch": 0.8, "grad_norm": 4.495995101269198, "learning_rate": 6.78703971185608e-06, "loss": 0.2475, "step": 5913 }, { "epoch": 0.8, "grad_norm": 3.3680731194727023, "learning_rate": 6.786010886818635e-06, "loss": 0.2046, "step": 5914 }, { "epoch": 0.8, "grad_norm": 3.399471452013802, "learning_rate": 6.784981975096021e-06, "loss": 0.2419, "step": 5915 }, { "epoch": 0.8, "grad_norm": 4.543570625799061, "learning_rate": 6.7839529767381785e-06, "loss": 0.2674, "step": 5916 }, { "epoch": 0.8, "grad_norm": 3.819267529453971, "learning_rate": 6.78292389179505e-06, "loss": 0.2279, "step": 5917 }, { "epoch": 0.81, "grad_norm": 3.5259635905735074, "learning_rate": 6.781894720316584e-06, "loss": 0.2797, "step": 5918 }, { "epoch": 0.81, "grad_norm": 3.4016514086657965, "learning_rate": 6.78086546235273e-06, "loss": 0.1828, "step": 5919 }, { "epoch": 0.81, "grad_norm": 3.1439558544079493, "learning_rate": 6.779836117953445e-06, "loss": 0.2115, "step": 5920 }, { "epoch": 0.81, "grad_norm": 3.113693600516526, "learning_rate": 6.778806687168689e-06, "loss": 0.2012, "step": 5921 }, { "epoch": 0.81, "grad_norm": 3.665982891090848, "learning_rate": 6.777777170048423e-06, "loss": 0.2202, "step": 5922 }, { "epoch": 0.81, "grad_norm": 2.7043912284574185, "learning_rate": 6.776747566642621e-06, "loss": 0.1873, "step": 5923 }, { "epoch": 0.81, "grad_norm": 3.63957107417053, "learning_rate": 6.77571787700125e-06, "loss": 0.2227, "step": 5924 }, { "epoch": 0.81, "grad_norm": 3.681394122763432, "learning_rate": 6.774688101174289e-06, "loss": 0.2203, "step": 5925 }, { "epoch": 0.81, "grad_norm": 5.104798209295471, "learning_rate": 6.77365823921172e-06, "loss": 0.2692, "step": 5926 }, { "epoch": 0.81, "grad_norm": 3.4966078961306004, "learning_rate": 6.772628291163527e-06, "loss": 0.217, "step": 5927 }, { "epoch": 0.81, "grad_norm": 4.083882302194696, "learning_rate": 6.771598257079698e-06, "loss": 0.2753, "step": 5928 }, { "epoch": 0.81, "grad_norm": 3.294399035132455, "learning_rate": 6.770568137010226e-06, "loss": 0.232, "step": 5929 }, { "epoch": 0.81, "grad_norm": 3.68382086118373, "learning_rate": 6.7695379310051125e-06, "loss": 0.2316, "step": 5930 }, { "epoch": 0.81, "grad_norm": 3.849410883736449, "learning_rate": 6.7685076391143545e-06, "loss": 0.2548, "step": 5931 }, { "epoch": 0.81, "grad_norm": 3.65206270850204, "learning_rate": 6.76747726138796e-06, "loss": 0.2247, "step": 5932 }, { "epoch": 0.81, "grad_norm": 3.7132220272050493, "learning_rate": 6.76644679787594e-06, "loss": 0.1807, "step": 5933 }, { "epoch": 0.81, "grad_norm": 3.0081956700755863, "learning_rate": 6.765416248628307e-06, "loss": 0.2114, "step": 5934 }, { "epoch": 0.81, "grad_norm": 3.8426699948904663, "learning_rate": 6.764385613695079e-06, "loss": 0.2551, "step": 5935 }, { "epoch": 0.81, "grad_norm": 3.8194955852282684, "learning_rate": 6.763354893126281e-06, "loss": 0.2039, "step": 5936 }, { "epoch": 0.81, "grad_norm": 3.215501934417748, "learning_rate": 6.762324086971936e-06, "loss": 0.1986, "step": 5937 }, { "epoch": 0.81, "grad_norm": 3.570356361436954, "learning_rate": 6.761293195282079e-06, "loss": 0.2631, "step": 5938 }, { "epoch": 0.81, "grad_norm": 3.5803804259831145, "learning_rate": 6.7602622181067425e-06, "loss": 0.1965, "step": 5939 }, { "epoch": 0.81, "grad_norm": 3.2106637545089973, "learning_rate": 6.759231155495967e-06, "loss": 0.2436, "step": 5940 }, { "epoch": 0.81, "grad_norm": 3.4891365027433414, "learning_rate": 6.758200007499793e-06, "loss": 0.2332, "step": 5941 }, { "epoch": 0.81, "grad_norm": 3.886732361109459, "learning_rate": 6.75716877416827e-06, "loss": 0.2878, "step": 5942 }, { "epoch": 0.81, "grad_norm": 3.1416981177276813, "learning_rate": 6.75613745555145e-06, "loss": 0.2445, "step": 5943 }, { "epoch": 0.81, "grad_norm": 3.9866821518110576, "learning_rate": 6.7551060516993874e-06, "loss": 0.266, "step": 5944 }, { "epoch": 0.81, "grad_norm": 3.3120677381307226, "learning_rate": 6.754074562662143e-06, "loss": 0.1909, "step": 5945 }, { "epoch": 0.81, "grad_norm": 3.8547124558232984, "learning_rate": 6.753042988489781e-06, "loss": 0.2238, "step": 5946 }, { "epoch": 0.81, "grad_norm": 3.655736922240509, "learning_rate": 6.752011329232369e-06, "loss": 0.2449, "step": 5947 }, { "epoch": 0.81, "grad_norm": 3.337926868166619, "learning_rate": 6.750979584939979e-06, "loss": 0.2182, "step": 5948 }, { "epoch": 0.81, "grad_norm": 7.492318057559236, "learning_rate": 6.749947755662686e-06, "loss": 0.195, "step": 5949 }, { "epoch": 0.81, "grad_norm": 3.759853756816093, "learning_rate": 6.748915841450574e-06, "loss": 0.2513, "step": 5950 }, { "epoch": 0.81, "grad_norm": 3.6976134504088245, "learning_rate": 6.747883842353725e-06, "loss": 0.28, "step": 5951 }, { "epoch": 0.81, "grad_norm": 3.8179573829969473, "learning_rate": 6.746851758422228e-06, "loss": 0.2265, "step": 5952 }, { "epoch": 0.81, "grad_norm": 3.4518381660596242, "learning_rate": 6.745819589706177e-06, "loss": 0.2326, "step": 5953 }, { "epoch": 0.81, "grad_norm": 3.9157630361666746, "learning_rate": 6.7447873362556684e-06, "loss": 0.2291, "step": 5954 }, { "epoch": 0.81, "grad_norm": 4.205066986501555, "learning_rate": 6.7437549981208014e-06, "loss": 0.2646, "step": 5955 }, { "epoch": 0.81, "grad_norm": 3.732197079096173, "learning_rate": 6.742722575351683e-06, "loss": 0.2438, "step": 5956 }, { "epoch": 0.81, "grad_norm": 4.0791624490569065, "learning_rate": 6.741690067998423e-06, "loss": 0.2591, "step": 5957 }, { "epoch": 0.81, "grad_norm": 3.447411823560548, "learning_rate": 6.740657476111136e-06, "loss": 0.2241, "step": 5958 }, { "epoch": 0.81, "grad_norm": 3.624332197662324, "learning_rate": 6.739624799739935e-06, "loss": 0.2129, "step": 5959 }, { "epoch": 0.81, "grad_norm": 3.4431406151180846, "learning_rate": 6.738592038934946e-06, "loss": 0.2197, "step": 5960 }, { "epoch": 0.81, "grad_norm": 4.539573661038029, "learning_rate": 6.737559193746291e-06, "loss": 0.2349, "step": 5961 }, { "epoch": 0.81, "grad_norm": 22.54263190038612, "learning_rate": 6.736526264224101e-06, "loss": 0.1808, "step": 5962 }, { "epoch": 0.81, "grad_norm": 2.9919611193310742, "learning_rate": 6.735493250418513e-06, "loss": 0.246, "step": 5963 }, { "epoch": 0.81, "grad_norm": 3.209489974723268, "learning_rate": 6.734460152379663e-06, "loss": 0.2031, "step": 5964 }, { "epoch": 0.81, "grad_norm": 3.1897965971631335, "learning_rate": 6.73342697015769e-06, "loss": 0.2331, "step": 5965 }, { "epoch": 0.81, "grad_norm": 3.119598919477796, "learning_rate": 6.732393703802746e-06, "loss": 0.1913, "step": 5966 }, { "epoch": 0.81, "grad_norm": 3.6278247445762437, "learning_rate": 6.731360353364975e-06, "loss": 0.2402, "step": 5967 }, { "epoch": 0.81, "grad_norm": 3.6063573452276825, "learning_rate": 6.7303269188945364e-06, "loss": 0.207, "step": 5968 }, { "epoch": 0.81, "grad_norm": 4.478064406325991, "learning_rate": 6.729293400441585e-06, "loss": 0.2578, "step": 5969 }, { "epoch": 0.81, "grad_norm": 5.1608760134170115, "learning_rate": 6.728259798056286e-06, "loss": 0.273, "step": 5970 }, { "epoch": 0.81, "grad_norm": 3.4652570193710304, "learning_rate": 6.727226111788805e-06, "loss": 0.2228, "step": 5971 }, { "epoch": 0.81, "grad_norm": 3.2474385475789065, "learning_rate": 6.726192341689311e-06, "loss": 0.241, "step": 5972 }, { "epoch": 0.81, "grad_norm": 4.007758044631031, "learning_rate": 6.725158487807981e-06, "loss": 0.315, "step": 5973 }, { "epoch": 0.81, "grad_norm": 4.2043378785429795, "learning_rate": 6.724124550194993e-06, "loss": 0.2426, "step": 5974 }, { "epoch": 0.81, "grad_norm": 3.1695572875097837, "learning_rate": 6.723090528900529e-06, "loss": 0.2376, "step": 5975 }, { "epoch": 0.81, "grad_norm": 3.817547432758709, "learning_rate": 6.722056423974777e-06, "loss": 0.2741, "step": 5976 }, { "epoch": 0.81, "grad_norm": 3.682570102657037, "learning_rate": 6.721022235467926e-06, "loss": 0.2442, "step": 5977 }, { "epoch": 0.81, "grad_norm": 3.352152283462939, "learning_rate": 6.719987963430175e-06, "loss": 0.2124, "step": 5978 }, { "epoch": 0.81, "grad_norm": 4.230737804668888, "learning_rate": 6.718953607911719e-06, "loss": 0.2687, "step": 5979 }, { "epoch": 0.81, "grad_norm": 4.339567629233029, "learning_rate": 6.717919168962763e-06, "loss": 0.2285, "step": 5980 }, { "epoch": 0.81, "grad_norm": 4.503161341087763, "learning_rate": 6.716884646633513e-06, "loss": 0.2368, "step": 5981 }, { "epoch": 0.81, "grad_norm": 3.766575931512485, "learning_rate": 6.7158500409741815e-06, "loss": 0.173, "step": 5982 }, { "epoch": 0.81, "grad_norm": 3.7485963353495806, "learning_rate": 6.714815352034983e-06, "loss": 0.2679, "step": 5983 }, { "epoch": 0.81, "grad_norm": 3.5602435720596404, "learning_rate": 6.713780579866137e-06, "loss": 0.2463, "step": 5984 }, { "epoch": 0.81, "grad_norm": 3.374059245961378, "learning_rate": 6.712745724517867e-06, "loss": 0.1989, "step": 5985 }, { "epoch": 0.81, "grad_norm": 3.751738168424271, "learning_rate": 6.7117107860404e-06, "loss": 0.2539, "step": 5986 }, { "epoch": 0.81, "grad_norm": 3.091192638295841, "learning_rate": 6.710675764483968e-06, "loss": 0.2313, "step": 5987 }, { "epoch": 0.81, "grad_norm": 3.835390118709665, "learning_rate": 6.7096406598988065e-06, "loss": 0.2817, "step": 5988 }, { "epoch": 0.81, "grad_norm": 3.8754184817422304, "learning_rate": 6.7086054723351514e-06, "loss": 0.2655, "step": 5989 }, { "epoch": 0.81, "grad_norm": 3.621517184082385, "learning_rate": 6.707570201843252e-06, "loss": 0.2157, "step": 5990 }, { "epoch": 0.81, "grad_norm": 3.3930758509975187, "learning_rate": 6.706534848473353e-06, "loss": 0.21, "step": 5991 }, { "epoch": 0.82, "grad_norm": 3.9658404512259544, "learning_rate": 6.7054994122757046e-06, "loss": 0.251, "step": 5992 }, { "epoch": 0.82, "grad_norm": 4.459616561872924, "learning_rate": 6.704463893300565e-06, "loss": 0.2973, "step": 5993 }, { "epoch": 0.82, "grad_norm": 2.9840998809677624, "learning_rate": 6.703428291598191e-06, "loss": 0.2343, "step": 5994 }, { "epoch": 0.82, "grad_norm": 2.8987216801856355, "learning_rate": 6.702392607218848e-06, "loss": 0.2246, "step": 5995 }, { "epoch": 0.82, "grad_norm": 3.7213355856896966, "learning_rate": 6.7013568402128035e-06, "loss": 0.1688, "step": 5996 }, { "epoch": 0.82, "grad_norm": 4.016767712533131, "learning_rate": 6.700320990630329e-06, "loss": 0.233, "step": 5997 }, { "epoch": 0.82, "grad_norm": 5.091551729787765, "learning_rate": 6.6992850585217005e-06, "loss": 0.2456, "step": 5998 }, { "epoch": 0.82, "grad_norm": 3.419862095382451, "learning_rate": 6.698249043937196e-06, "loss": 0.2067, "step": 5999 }, { "epoch": 0.82, "grad_norm": 3.42881637240229, "learning_rate": 6.697212946927101e-06, "loss": 0.2537, "step": 6000 }, { "epoch": 0.82, "grad_norm": 3.407291467562485, "learning_rate": 6.696176767541701e-06, "loss": 0.2178, "step": 6001 }, { "epoch": 0.82, "grad_norm": 3.2792280229243924, "learning_rate": 6.69514050583129e-06, "loss": 0.1726, "step": 6002 }, { "epoch": 0.82, "grad_norm": 4.089826462537082, "learning_rate": 6.694104161846162e-06, "loss": 0.267, "step": 6003 }, { "epoch": 0.82, "grad_norm": 5.325078674641659, "learning_rate": 6.693067735636618e-06, "loss": 0.2371, "step": 6004 }, { "epoch": 0.82, "grad_norm": 7.425575238842804, "learning_rate": 6.692031227252961e-06, "loss": 0.2249, "step": 6005 }, { "epoch": 0.82, "grad_norm": 3.7959470104784794, "learning_rate": 6.690994636745497e-06, "loss": 0.2291, "step": 6006 }, { "epoch": 0.82, "grad_norm": 3.4372988884285482, "learning_rate": 6.689957964164539e-06, "loss": 0.2349, "step": 6007 }, { "epoch": 0.82, "grad_norm": 4.036852477201136, "learning_rate": 6.6889212095604036e-06, "loss": 0.2199, "step": 6008 }, { "epoch": 0.82, "grad_norm": 4.281108955386541, "learning_rate": 6.687884372983408e-06, "loss": 0.2615, "step": 6009 }, { "epoch": 0.82, "grad_norm": 4.286404889136674, "learning_rate": 6.686847454483878e-06, "loss": 0.2663, "step": 6010 }, { "epoch": 0.82, "grad_norm": 4.3610807643073395, "learning_rate": 6.685810454112141e-06, "loss": 0.2872, "step": 6011 }, { "epoch": 0.82, "grad_norm": 3.2524443851295537, "learning_rate": 6.684773371918526e-06, "loss": 0.2219, "step": 6012 }, { "epoch": 0.82, "grad_norm": 3.856668049689157, "learning_rate": 6.683736207953371e-06, "loss": 0.2007, "step": 6013 }, { "epoch": 0.82, "grad_norm": 3.6002192611319566, "learning_rate": 6.682698962267013e-06, "loss": 0.1971, "step": 6014 }, { "epoch": 0.82, "grad_norm": 3.645957298669482, "learning_rate": 6.681661634909796e-06, "loss": 0.2561, "step": 6015 }, { "epoch": 0.82, "grad_norm": 3.582595134728873, "learning_rate": 6.68062422593207e-06, "loss": 0.2105, "step": 6016 }, { "epoch": 0.82, "grad_norm": 4.767532852429614, "learning_rate": 6.679586735384184e-06, "loss": 0.2249, "step": 6017 }, { "epoch": 0.82, "grad_norm": 4.230152473534246, "learning_rate": 6.678549163316493e-06, "loss": 0.275, "step": 6018 }, { "epoch": 0.82, "grad_norm": 3.7038271518521437, "learning_rate": 6.677511509779358e-06, "loss": 0.2743, "step": 6019 }, { "epoch": 0.82, "grad_norm": 3.5621934004088223, "learning_rate": 6.67647377482314e-06, "loss": 0.2425, "step": 6020 }, { "epoch": 0.82, "grad_norm": 4.267287090781665, "learning_rate": 6.6754359584982085e-06, "loss": 0.1833, "step": 6021 }, { "epoch": 0.82, "grad_norm": 4.1727418561169465, "learning_rate": 6.674398060854931e-06, "loss": 0.2336, "step": 6022 }, { "epoch": 0.82, "grad_norm": 3.4614742127518605, "learning_rate": 6.673360081943686e-06, "loss": 0.188, "step": 6023 }, { "epoch": 0.82, "grad_norm": 3.249979566207703, "learning_rate": 6.672322021814851e-06, "loss": 0.2159, "step": 6024 }, { "epoch": 0.82, "grad_norm": 3.1526451550753007, "learning_rate": 6.671283880518809e-06, "loss": 0.2033, "step": 6025 }, { "epoch": 0.82, "grad_norm": 2.888195515164189, "learning_rate": 6.670245658105948e-06, "loss": 0.2088, "step": 6026 }, { "epoch": 0.82, "grad_norm": 2.974402758631837, "learning_rate": 6.669207354626657e-06, "loss": 0.2454, "step": 6027 }, { "epoch": 0.82, "grad_norm": 4.262104801492896, "learning_rate": 6.668168970131332e-06, "loss": 0.2421, "step": 6028 }, { "epoch": 0.82, "grad_norm": 2.810708614070373, "learning_rate": 6.66713050467037e-06, "loss": 0.2289, "step": 6029 }, { "epoch": 0.82, "grad_norm": 4.30804767477289, "learning_rate": 6.6660919582941765e-06, "loss": 0.2489, "step": 6030 }, { "epoch": 0.82, "grad_norm": 3.1987619055731025, "learning_rate": 6.665053331053154e-06, "loss": 0.2562, "step": 6031 }, { "epoch": 0.82, "grad_norm": 4.682427556413552, "learning_rate": 6.664014622997717e-06, "loss": 0.2528, "step": 6032 }, { "epoch": 0.82, "grad_norm": 3.6464311652546866, "learning_rate": 6.6629758341782765e-06, "loss": 0.2224, "step": 6033 }, { "epoch": 0.82, "grad_norm": 3.3103176169505173, "learning_rate": 6.6619369646452514e-06, "loss": 0.2455, "step": 6034 }, { "epoch": 0.82, "grad_norm": 3.341182967055521, "learning_rate": 6.660898014449066e-06, "loss": 0.2135, "step": 6035 }, { "epoch": 0.82, "grad_norm": 3.1251796737365454, "learning_rate": 6.659858983640146e-06, "loss": 0.1865, "step": 6036 }, { "epoch": 0.82, "grad_norm": 3.7120670915003617, "learning_rate": 6.65881987226892e-06, "loss": 0.2613, "step": 6037 }, { "epoch": 0.82, "grad_norm": 3.0403777341972025, "learning_rate": 6.657780680385821e-06, "loss": 0.2051, "step": 6038 }, { "epoch": 0.82, "grad_norm": 3.4117190323187634, "learning_rate": 6.656741408041291e-06, "loss": 0.2206, "step": 6039 }, { "epoch": 0.82, "grad_norm": 2.928807286628159, "learning_rate": 6.655702055285768e-06, "loss": 0.2292, "step": 6040 }, { "epoch": 0.82, "grad_norm": 3.687405739626589, "learning_rate": 6.654662622169696e-06, "loss": 0.2466, "step": 6041 }, { "epoch": 0.82, "grad_norm": 4.143196735405495, "learning_rate": 6.65362310874353e-06, "loss": 0.2631, "step": 6042 }, { "epoch": 0.82, "grad_norm": 3.7332606222366254, "learning_rate": 6.65258351505772e-06, "loss": 0.2215, "step": 6043 }, { "epoch": 0.82, "grad_norm": 3.930321970945608, "learning_rate": 6.651543841162725e-06, "loss": 0.2807, "step": 6044 }, { "epoch": 0.82, "grad_norm": 3.451365740147286, "learning_rate": 6.650504087109004e-06, "loss": 0.195, "step": 6045 }, { "epoch": 0.82, "grad_norm": 2.7916361238417937, "learning_rate": 6.649464252947024e-06, "loss": 0.1775, "step": 6046 }, { "epoch": 0.82, "grad_norm": 3.1186583222394546, "learning_rate": 6.648424338727254e-06, "loss": 0.2291, "step": 6047 }, { "epoch": 0.82, "grad_norm": 3.296256776393866, "learning_rate": 6.647384344500164e-06, "loss": 0.2235, "step": 6048 }, { "epoch": 0.82, "grad_norm": 3.7295830686369635, "learning_rate": 6.646344270316235e-06, "loss": 0.1971, "step": 6049 }, { "epoch": 0.82, "grad_norm": 3.375463573271189, "learning_rate": 6.645304116225946e-06, "loss": 0.2746, "step": 6050 }, { "epoch": 0.82, "grad_norm": 3.2315653576523666, "learning_rate": 6.64426388227978e-06, "loss": 0.1865, "step": 6051 }, { "epoch": 0.82, "grad_norm": 3.452363046057658, "learning_rate": 6.643223568528228e-06, "loss": 0.216, "step": 6052 }, { "epoch": 0.82, "grad_norm": 2.846761504594373, "learning_rate": 6.642183175021779e-06, "loss": 0.1875, "step": 6053 }, { "epoch": 0.82, "grad_norm": 3.4125830182698995, "learning_rate": 6.641142701810932e-06, "loss": 0.2076, "step": 6054 }, { "epoch": 0.82, "grad_norm": 3.9584116434600816, "learning_rate": 6.640102148946186e-06, "loss": 0.2098, "step": 6055 }, { "epoch": 0.82, "grad_norm": 4.00368876904302, "learning_rate": 6.6390615164780445e-06, "loss": 0.2467, "step": 6056 }, { "epoch": 0.82, "grad_norm": 3.9124895995154585, "learning_rate": 6.638020804457017e-06, "loss": 0.2345, "step": 6057 }, { "epoch": 0.82, "grad_norm": 4.081249783031875, "learning_rate": 6.636980012933613e-06, "loss": 0.2909, "step": 6058 }, { "epoch": 0.82, "grad_norm": 4.291255865794301, "learning_rate": 6.635939141958351e-06, "loss": 0.2596, "step": 6059 }, { "epoch": 0.82, "grad_norm": 4.518903182920727, "learning_rate": 6.634898191581747e-06, "loss": 0.2593, "step": 6060 }, { "epoch": 0.82, "grad_norm": 4.010118574407686, "learning_rate": 6.633857161854324e-06, "loss": 0.2696, "step": 6061 }, { "epoch": 0.82, "grad_norm": 4.883602994315263, "learning_rate": 6.632816052826611e-06, "loss": 0.2546, "step": 6062 }, { "epoch": 0.82, "grad_norm": 3.9763018550813958, "learning_rate": 6.631774864549138e-06, "loss": 0.2492, "step": 6063 }, { "epoch": 0.82, "grad_norm": 3.403808874473121, "learning_rate": 6.6307335970724405e-06, "loss": 0.2443, "step": 6064 }, { "epoch": 0.83, "grad_norm": 3.548192470288723, "learning_rate": 6.629692250447057e-06, "loss": 0.2181, "step": 6065 }, { "epoch": 0.83, "grad_norm": 3.581016433252047, "learning_rate": 6.6286508247235305e-06, "loss": 0.2599, "step": 6066 }, { "epoch": 0.83, "grad_norm": 3.530176434047809, "learning_rate": 6.627609319952404e-06, "loss": 0.222, "step": 6067 }, { "epoch": 0.83, "grad_norm": 3.021336468343612, "learning_rate": 6.626567736184231e-06, "loss": 0.233, "step": 6068 }, { "epoch": 0.83, "grad_norm": 4.1973788642715375, "learning_rate": 6.625526073469563e-06, "loss": 0.2635, "step": 6069 }, { "epoch": 0.83, "grad_norm": 4.4244135177089845, "learning_rate": 6.6244843318589624e-06, "loss": 0.2568, "step": 6070 }, { "epoch": 0.83, "grad_norm": 4.925886065967633, "learning_rate": 6.623442511402986e-06, "loss": 0.2558, "step": 6071 }, { "epoch": 0.83, "grad_norm": 3.8308061336676476, "learning_rate": 6.622400612152199e-06, "loss": 0.3216, "step": 6072 }, { "epoch": 0.83, "grad_norm": 3.7592697705724554, "learning_rate": 6.621358634157173e-06, "loss": 0.2121, "step": 6073 }, { "epoch": 0.83, "grad_norm": 3.342074397924935, "learning_rate": 6.62031657746848e-06, "loss": 0.2353, "step": 6074 }, { "epoch": 0.83, "grad_norm": 4.3153029625245365, "learning_rate": 6.619274442136698e-06, "loss": 0.265, "step": 6075 }, { "epoch": 0.83, "grad_norm": 3.388587343758073, "learning_rate": 6.618232228212407e-06, "loss": 0.2056, "step": 6076 }, { "epoch": 0.83, "grad_norm": 2.785965755003335, "learning_rate": 6.617189935746191e-06, "loss": 0.1733, "step": 6077 }, { "epoch": 0.83, "grad_norm": 3.712255807943756, "learning_rate": 6.6161475647886396e-06, "loss": 0.2634, "step": 6078 }, { "epoch": 0.83, "grad_norm": 4.120464369524763, "learning_rate": 6.615105115390344e-06, "loss": 0.257, "step": 6079 }, { "epoch": 0.83, "grad_norm": 3.441930793901935, "learning_rate": 6.6140625876019e-06, "loss": 0.2192, "step": 6080 }, { "epoch": 0.83, "grad_norm": 3.174980003913414, "learning_rate": 6.613019981473906e-06, "loss": 0.2056, "step": 6081 }, { "epoch": 0.83, "grad_norm": 2.9841049088876286, "learning_rate": 6.6119772970569686e-06, "loss": 0.2353, "step": 6082 }, { "epoch": 0.83, "grad_norm": 3.4962696017539194, "learning_rate": 6.610934534401694e-06, "loss": 0.1985, "step": 6083 }, { "epoch": 0.83, "grad_norm": 3.4526892757576957, "learning_rate": 6.609891693558692e-06, "loss": 0.2449, "step": 6084 }, { "epoch": 0.83, "grad_norm": 3.483043127619135, "learning_rate": 6.6088487745785805e-06, "loss": 0.2411, "step": 6085 }, { "epoch": 0.83, "grad_norm": 3.5653360560810836, "learning_rate": 6.607805777511974e-06, "loss": 0.2375, "step": 6086 }, { "epoch": 0.83, "grad_norm": 3.2762744962752377, "learning_rate": 6.606762702409499e-06, "loss": 0.1928, "step": 6087 }, { "epoch": 0.83, "grad_norm": 3.2422924602467695, "learning_rate": 6.605719549321778e-06, "loss": 0.259, "step": 6088 }, { "epoch": 0.83, "grad_norm": 3.867329835015826, "learning_rate": 6.604676318299446e-06, "loss": 0.1969, "step": 6089 }, { "epoch": 0.83, "grad_norm": 4.124807779857838, "learning_rate": 6.603633009393132e-06, "loss": 0.218, "step": 6090 }, { "epoch": 0.83, "grad_norm": 4.703627175111084, "learning_rate": 6.6025896226534765e-06, "loss": 0.2229, "step": 6091 }, { "epoch": 0.83, "grad_norm": 3.971575264700906, "learning_rate": 6.60154615813112e-06, "loss": 0.2016, "step": 6092 }, { "epoch": 0.83, "grad_norm": 3.378732804757736, "learning_rate": 6.600502615876709e-06, "loss": 0.2078, "step": 6093 }, { "epoch": 0.83, "grad_norm": 3.8997351659108945, "learning_rate": 6.59945899594089e-06, "loss": 0.2282, "step": 6094 }, { "epoch": 0.83, "grad_norm": 4.33991129982221, "learning_rate": 6.598415298374318e-06, "loss": 0.2441, "step": 6095 }, { "epoch": 0.83, "grad_norm": 3.1723097685805004, "learning_rate": 6.5973715232276484e-06, "loss": 0.2381, "step": 6096 }, { "epoch": 0.83, "grad_norm": 3.883208371783473, "learning_rate": 6.596327670551541e-06, "loss": 0.2556, "step": 6097 }, { "epoch": 0.83, "grad_norm": 4.461027958977296, "learning_rate": 6.595283740396662e-06, "loss": 0.2381, "step": 6098 }, { "epoch": 0.83, "grad_norm": 4.763501158081531, "learning_rate": 6.594239732813678e-06, "loss": 0.2707, "step": 6099 }, { "epoch": 0.83, "grad_norm": 3.1513128166772906, "learning_rate": 6.5931956478532585e-06, "loss": 0.2246, "step": 6100 }, { "epoch": 0.83, "grad_norm": 4.512896580919544, "learning_rate": 6.592151485566081e-06, "loss": 0.2436, "step": 6101 }, { "epoch": 0.83, "grad_norm": 3.451268238098905, "learning_rate": 6.591107246002825e-06, "loss": 0.2025, "step": 6102 }, { "epoch": 0.83, "grad_norm": 3.581415592926513, "learning_rate": 6.5900629292141735e-06, "loss": 0.1977, "step": 6103 }, { "epoch": 0.83, "grad_norm": 3.4707319077339363, "learning_rate": 6.589018535250811e-06, "loss": 0.2068, "step": 6104 }, { "epoch": 0.83, "grad_norm": 4.262634502979414, "learning_rate": 6.587974064163429e-06, "loss": 0.2394, "step": 6105 }, { "epoch": 0.83, "grad_norm": 3.6532451718874412, "learning_rate": 6.586929516002723e-06, "loss": 0.2076, "step": 6106 }, { "epoch": 0.83, "grad_norm": 4.0761865137646325, "learning_rate": 6.585884890819388e-06, "loss": 0.245, "step": 6107 }, { "epoch": 0.83, "grad_norm": 2.7886901779377298, "learning_rate": 6.584840188664127e-06, "loss": 0.1941, "step": 6108 }, { "epoch": 0.83, "grad_norm": 3.5156661790138997, "learning_rate": 6.583795409587645e-06, "loss": 0.2396, "step": 6109 }, { "epoch": 0.83, "grad_norm": 3.9577842925111235, "learning_rate": 6.582750553640653e-06, "loss": 0.2432, "step": 6110 }, { "epoch": 0.83, "grad_norm": 3.6571338926932584, "learning_rate": 6.581705620873861e-06, "loss": 0.2591, "step": 6111 }, { "epoch": 0.83, "grad_norm": 4.931175710056868, "learning_rate": 6.5806606113379855e-06, "loss": 0.2459, "step": 6112 }, { "epoch": 0.83, "grad_norm": 3.055304075171273, "learning_rate": 6.579615525083748e-06, "loss": 0.1941, "step": 6113 }, { "epoch": 0.83, "grad_norm": 3.0563705552349947, "learning_rate": 6.578570362161872e-06, "loss": 0.1881, "step": 6114 }, { "epoch": 0.83, "grad_norm": 3.9309329163319626, "learning_rate": 6.5775251226230855e-06, "loss": 0.2626, "step": 6115 }, { "epoch": 0.83, "grad_norm": 3.4656857989761316, "learning_rate": 6.57647980651812e-06, "loss": 0.2316, "step": 6116 }, { "epoch": 0.83, "grad_norm": 3.3018599412332614, "learning_rate": 6.57543441389771e-06, "loss": 0.2149, "step": 6117 }, { "epoch": 0.83, "grad_norm": 4.151099310949423, "learning_rate": 6.574388944812594e-06, "loss": 0.2014, "step": 6118 }, { "epoch": 0.83, "grad_norm": 3.7442606599183827, "learning_rate": 6.573343399313515e-06, "loss": 0.2454, "step": 6119 }, { "epoch": 0.83, "grad_norm": 3.1809362487065833, "learning_rate": 6.5722977774512175e-06, "loss": 0.2607, "step": 6120 }, { "epoch": 0.83, "grad_norm": 3.302788823864341, "learning_rate": 6.571252079276456e-06, "loss": 0.2055, "step": 6121 }, { "epoch": 0.83, "grad_norm": 4.7682928952921095, "learning_rate": 6.570206304839979e-06, "loss": 0.302, "step": 6122 }, { "epoch": 0.83, "grad_norm": 2.6974020227224442, "learning_rate": 6.569160454192547e-06, "loss": 0.1766, "step": 6123 }, { "epoch": 0.83, "grad_norm": 3.110805421620253, "learning_rate": 6.568114527384921e-06, "loss": 0.1667, "step": 6124 }, { "epoch": 0.83, "grad_norm": 3.2562186626050065, "learning_rate": 6.567068524467864e-06, "loss": 0.1988, "step": 6125 }, { "epoch": 0.83, "grad_norm": 4.140321344972256, "learning_rate": 6.566022445492144e-06, "loss": 0.2876, "step": 6126 }, { "epoch": 0.83, "grad_norm": 3.50313509064066, "learning_rate": 6.564976290508535e-06, "loss": 0.2002, "step": 6127 }, { "epoch": 0.83, "grad_norm": 4.507544342887139, "learning_rate": 6.563930059567813e-06, "loss": 0.274, "step": 6128 }, { "epoch": 0.83, "grad_norm": 2.968208027785078, "learning_rate": 6.5628837527207545e-06, "loss": 0.2208, "step": 6129 }, { "epoch": 0.83, "grad_norm": 3.3800962491320337, "learning_rate": 6.561837370018147e-06, "loss": 0.184, "step": 6130 }, { "epoch": 0.83, "grad_norm": 3.8589384173241585, "learning_rate": 6.560790911510774e-06, "loss": 0.2405, "step": 6131 }, { "epoch": 0.83, "grad_norm": 4.164167245356595, "learning_rate": 6.559744377249426e-06, "loss": 0.2544, "step": 6132 }, { "epoch": 0.83, "grad_norm": 3.571226740210129, "learning_rate": 6.558697767284901e-06, "loss": 0.2636, "step": 6133 }, { "epoch": 0.83, "grad_norm": 4.573321327643927, "learning_rate": 6.55765108166799e-06, "loss": 0.2584, "step": 6134 }, { "epoch": 0.83, "grad_norm": 4.175832218561186, "learning_rate": 6.556604320449502e-06, "loss": 0.2075, "step": 6135 }, { "epoch": 0.83, "grad_norm": 2.9759782695642283, "learning_rate": 6.555557483680238e-06, "loss": 0.2256, "step": 6136 }, { "epoch": 0.83, "grad_norm": 5.070787586684564, "learning_rate": 6.554510571411009e-06, "loss": 0.2491, "step": 6137 }, { "epoch": 0.83, "grad_norm": 3.443808833235101, "learning_rate": 6.553463583692625e-06, "loss": 0.2497, "step": 6138 }, { "epoch": 0.84, "grad_norm": 4.133745286478319, "learning_rate": 6.5524165205759036e-06, "loss": 0.2848, "step": 6139 }, { "epoch": 0.84, "grad_norm": 3.795769895258605, "learning_rate": 6.551369382111665e-06, "loss": 0.2231, "step": 6140 }, { "epoch": 0.84, "grad_norm": 2.922680057945697, "learning_rate": 6.550322168350732e-06, "loss": 0.2177, "step": 6141 }, { "epoch": 0.84, "grad_norm": 3.0485385122575654, "learning_rate": 6.549274879343932e-06, "loss": 0.2126, "step": 6142 }, { "epoch": 0.84, "grad_norm": 2.802646892820064, "learning_rate": 6.548227515142097e-06, "loss": 0.2034, "step": 6143 }, { "epoch": 0.84, "grad_norm": 3.8391519575087205, "learning_rate": 6.54718007579606e-06, "loss": 0.2384, "step": 6144 }, { "epoch": 0.84, "grad_norm": 5.530874612664128, "learning_rate": 6.54613256135666e-06, "loss": 0.221, "step": 6145 }, { "epoch": 0.84, "grad_norm": 3.0323170741508685, "learning_rate": 6.545084971874738e-06, "loss": 0.2011, "step": 6146 }, { "epoch": 0.84, "grad_norm": 3.958267893557569, "learning_rate": 6.54403730740114e-06, "loss": 0.2889, "step": 6147 }, { "epoch": 0.84, "grad_norm": 3.263306951119682, "learning_rate": 6.542989567986713e-06, "loss": 0.2256, "step": 6148 }, { "epoch": 0.84, "grad_norm": 3.2887885161015205, "learning_rate": 6.541941753682314e-06, "loss": 0.231, "step": 6149 }, { "epoch": 0.84, "grad_norm": 3.1308980741369665, "learning_rate": 6.540893864538795e-06, "loss": 0.2582, "step": 6150 }, { "epoch": 0.84, "grad_norm": 3.633954294706777, "learning_rate": 6.53984590060702e-06, "loss": 0.1966, "step": 6151 }, { "epoch": 0.84, "grad_norm": 3.6824863971982453, "learning_rate": 6.53879786193785e-06, "loss": 0.2492, "step": 6152 }, { "epoch": 0.84, "grad_norm": 3.483159408602958, "learning_rate": 6.53774974858215e-06, "loss": 0.2522, "step": 6153 }, { "epoch": 0.84, "grad_norm": 2.937160962355774, "learning_rate": 6.536701560590797e-06, "loss": 0.2161, "step": 6154 }, { "epoch": 0.84, "grad_norm": 3.490799903082454, "learning_rate": 6.535653298014659e-06, "loss": 0.2223, "step": 6155 }, { "epoch": 0.84, "grad_norm": 3.866437288077878, "learning_rate": 6.534604960904619e-06, "loss": 0.2581, "step": 6156 }, { "epoch": 0.84, "grad_norm": 5.168782724770392, "learning_rate": 6.533556549311557e-06, "loss": 0.1854, "step": 6157 }, { "epoch": 0.84, "grad_norm": 4.112579056391784, "learning_rate": 6.532508063286357e-06, "loss": 0.2559, "step": 6158 }, { "epoch": 0.84, "grad_norm": 3.84974128332656, "learning_rate": 6.531459502879909e-06, "loss": 0.2177, "step": 6159 }, { "epoch": 0.84, "grad_norm": 3.719163927891572, "learning_rate": 6.530410868143105e-06, "loss": 0.1688, "step": 6160 }, { "epoch": 0.84, "grad_norm": 3.310361898554449, "learning_rate": 6.529362159126842e-06, "loss": 0.2292, "step": 6161 }, { "epoch": 0.84, "grad_norm": 3.392902772908039, "learning_rate": 6.52831337588202e-06, "loss": 0.2175, "step": 6162 }, { "epoch": 0.84, "grad_norm": 3.887610571909086, "learning_rate": 6.527264518459542e-06, "loss": 0.1917, "step": 6163 }, { "epoch": 0.84, "grad_norm": 4.336472792753965, "learning_rate": 6.526215586910315e-06, "loss": 0.2419, "step": 6164 }, { "epoch": 0.84, "grad_norm": 4.390564695533852, "learning_rate": 6.52516658128525e-06, "loss": 0.229, "step": 6165 }, { "epoch": 0.84, "grad_norm": 3.1854683158640085, "learning_rate": 6.524117501635258e-06, "loss": 0.189, "step": 6166 }, { "epoch": 0.84, "grad_norm": 4.607794756023584, "learning_rate": 6.52306834801126e-06, "loss": 0.2496, "step": 6167 }, { "epoch": 0.84, "grad_norm": 4.644109517118495, "learning_rate": 6.522019120464177e-06, "loss": 0.2677, "step": 6168 }, { "epoch": 0.84, "grad_norm": 3.470430692023889, "learning_rate": 6.520969819044933e-06, "loss": 0.2725, "step": 6169 }, { "epoch": 0.84, "grad_norm": 2.665458793541692, "learning_rate": 6.519920443804458e-06, "loss": 0.1318, "step": 6170 }, { "epoch": 0.84, "grad_norm": 3.879880824993248, "learning_rate": 6.5188709947936826e-06, "loss": 0.1817, "step": 6171 }, { "epoch": 0.84, "grad_norm": 3.1676118484268807, "learning_rate": 6.517821472063543e-06, "loss": 0.2073, "step": 6172 }, { "epoch": 0.84, "grad_norm": 3.5836968700363983, "learning_rate": 6.516771875664979e-06, "loss": 0.2124, "step": 6173 }, { "epoch": 0.84, "grad_norm": 4.847750796999602, "learning_rate": 6.515722205648931e-06, "loss": 0.2627, "step": 6174 }, { "epoch": 0.84, "grad_norm": 3.956032387181374, "learning_rate": 6.514672462066348e-06, "loss": 0.2381, "step": 6175 }, { "epoch": 0.84, "grad_norm": 3.820370829154166, "learning_rate": 6.513622644968181e-06, "loss": 0.1785, "step": 6176 }, { "epoch": 0.84, "grad_norm": 4.12246075738319, "learning_rate": 6.51257275440538e-06, "loss": 0.219, "step": 6177 }, { "epoch": 0.84, "grad_norm": 4.399263453126762, "learning_rate": 6.511522790428904e-06, "loss": 0.2416, "step": 6178 }, { "epoch": 0.84, "grad_norm": 4.11050758985172, "learning_rate": 6.510472753089714e-06, "loss": 0.2338, "step": 6179 }, { "epoch": 0.84, "grad_norm": 2.982313152539464, "learning_rate": 6.509422642438773e-06, "loss": 0.1722, "step": 6180 }, { "epoch": 0.84, "grad_norm": 3.649813536980381, "learning_rate": 6.508372458527049e-06, "loss": 0.2212, "step": 6181 }, { "epoch": 0.84, "grad_norm": 3.814648327951766, "learning_rate": 6.507322201405515e-06, "loss": 0.2487, "step": 6182 }, { "epoch": 0.84, "grad_norm": 4.039762049806979, "learning_rate": 6.506271871125145e-06, "loss": 0.2051, "step": 6183 }, { "epoch": 0.84, "grad_norm": 4.090986697788896, "learning_rate": 6.505221467736916e-06, "loss": 0.2681, "step": 6184 }, { "epoch": 0.84, "grad_norm": 3.459624024129312, "learning_rate": 6.504170991291812e-06, "loss": 0.2031, "step": 6185 }, { "epoch": 0.84, "grad_norm": 3.161759142760848, "learning_rate": 6.503120441840818e-06, "loss": 0.2214, "step": 6186 }, { "epoch": 0.84, "grad_norm": 4.088924586051309, "learning_rate": 6.502069819434921e-06, "loss": 0.2568, "step": 6187 }, { "epoch": 0.84, "grad_norm": 3.85896025219269, "learning_rate": 6.5010191241251165e-06, "loss": 0.2238, "step": 6188 }, { "epoch": 0.84, "grad_norm": 3.574318811720939, "learning_rate": 6.499968355962401e-06, "loss": 0.2367, "step": 6189 }, { "epoch": 0.84, "grad_norm": 3.4934777156385897, "learning_rate": 6.498917514997771e-06, "loss": 0.2666, "step": 6190 }, { "epoch": 0.84, "grad_norm": 3.4210906274045194, "learning_rate": 6.4978666012822334e-06, "loss": 0.2318, "step": 6191 }, { "epoch": 0.84, "grad_norm": 4.399319370512236, "learning_rate": 6.496815614866792e-06, "loss": 0.2688, "step": 6192 }, { "epoch": 0.84, "grad_norm": 4.6052689233414705, "learning_rate": 6.495764555802456e-06, "loss": 0.2485, "step": 6193 }, { "epoch": 0.84, "grad_norm": 3.837701515239734, "learning_rate": 6.494713424140244e-06, "loss": 0.2512, "step": 6194 }, { "epoch": 0.84, "grad_norm": 3.6985162741691187, "learning_rate": 6.493662219931169e-06, "loss": 0.2554, "step": 6195 }, { "epoch": 0.84, "grad_norm": 3.6548168766282467, "learning_rate": 6.492610943226255e-06, "loss": 0.2112, "step": 6196 }, { "epoch": 0.84, "grad_norm": 3.7326530590929594, "learning_rate": 6.491559594076526e-06, "loss": 0.2151, "step": 6197 }, { "epoch": 0.84, "grad_norm": 4.435586536780066, "learning_rate": 6.490508172533007e-06, "loss": 0.2059, "step": 6198 }, { "epoch": 0.84, "grad_norm": 4.472343925372764, "learning_rate": 6.489456678646732e-06, "loss": 0.2612, "step": 6199 }, { "epoch": 0.84, "grad_norm": 3.5733573408355173, "learning_rate": 6.488405112468733e-06, "loss": 0.2415, "step": 6200 }, { "epoch": 0.84, "grad_norm": 3.1211499257119955, "learning_rate": 6.487353474050053e-06, "loss": 0.1823, "step": 6201 }, { "epoch": 0.84, "grad_norm": 3.551751364139949, "learning_rate": 6.486301763441732e-06, "loss": 0.2318, "step": 6202 }, { "epoch": 0.84, "grad_norm": 3.9160919939536134, "learning_rate": 6.485249980694813e-06, "loss": 0.2253, "step": 6203 }, { "epoch": 0.84, "grad_norm": 3.233506801518247, "learning_rate": 6.484198125860348e-06, "loss": 0.1935, "step": 6204 }, { "epoch": 0.84, "grad_norm": 3.639994022036494, "learning_rate": 6.483146198989389e-06, "loss": 0.1992, "step": 6205 }, { "epoch": 0.84, "grad_norm": 3.0674490722667422, "learning_rate": 6.48209420013299e-06, "loss": 0.2155, "step": 6206 }, { "epoch": 0.84, "grad_norm": 3.9431509415786703, "learning_rate": 6.4810421293422124e-06, "loss": 0.2314, "step": 6207 }, { "epoch": 0.84, "grad_norm": 3.6651525371968283, "learning_rate": 6.479989986668118e-06, "loss": 0.2433, "step": 6208 }, { "epoch": 0.84, "grad_norm": 4.684260560782585, "learning_rate": 6.478937772161773e-06, "loss": 0.2745, "step": 6209 }, { "epoch": 0.84, "grad_norm": 3.6656514532444135, "learning_rate": 6.47788548587425e-06, "loss": 0.1989, "step": 6210 }, { "epoch": 0.84, "grad_norm": 4.097457766535136, "learning_rate": 6.476833127856619e-06, "loss": 0.2194, "step": 6211 }, { "epoch": 0.84, "grad_norm": 4.499660515940152, "learning_rate": 6.475780698159959e-06, "loss": 0.2442, "step": 6212 }, { "epoch": 0.85, "grad_norm": 2.896063517706738, "learning_rate": 6.474728196835348e-06, "loss": 0.1714, "step": 6213 }, { "epoch": 0.85, "grad_norm": 3.2910015575589635, "learning_rate": 6.473675623933873e-06, "loss": 0.2501, "step": 6214 }, { "epoch": 0.85, "grad_norm": 3.3601696343691736, "learning_rate": 6.472622979506619e-06, "loss": 0.2422, "step": 6215 }, { "epoch": 0.85, "grad_norm": 3.5050245539927514, "learning_rate": 6.471570263604678e-06, "loss": 0.2359, "step": 6216 }, { "epoch": 0.85, "grad_norm": 3.4316371822427847, "learning_rate": 6.470517476279143e-06, "loss": 0.232, "step": 6217 }, { "epoch": 0.85, "grad_norm": 3.892180489662563, "learning_rate": 6.469464617581112e-06, "loss": 0.2347, "step": 6218 }, { "epoch": 0.85, "grad_norm": 3.5180293955202266, "learning_rate": 6.468411687561687e-06, "loss": 0.1926, "step": 6219 }, { "epoch": 0.85, "grad_norm": 3.220953906969722, "learning_rate": 6.467358686271972e-06, "loss": 0.2206, "step": 6220 }, { "epoch": 0.85, "grad_norm": 3.709193696128476, "learning_rate": 6.466305613763075e-06, "loss": 0.2381, "step": 6221 }, { "epoch": 0.85, "grad_norm": 3.7971474150515827, "learning_rate": 6.465252470086109e-06, "loss": 0.1854, "step": 6222 }, { "epoch": 0.85, "grad_norm": 4.1418721963756155, "learning_rate": 6.464199255292186e-06, "loss": 0.255, "step": 6223 }, { "epoch": 0.85, "grad_norm": 3.2266059370604534, "learning_rate": 6.4631459694324275e-06, "loss": 0.2349, "step": 6224 }, { "epoch": 0.85, "grad_norm": 3.535323718208233, "learning_rate": 6.462092612557954e-06, "loss": 0.2084, "step": 6225 }, { "epoch": 0.85, "grad_norm": 3.9214412242875603, "learning_rate": 6.46103918471989e-06, "loss": 0.1952, "step": 6226 }, { "epoch": 0.85, "grad_norm": 3.4870949664070747, "learning_rate": 6.459985685969365e-06, "loss": 0.2045, "step": 6227 }, { "epoch": 0.85, "grad_norm": 4.089938536511139, "learning_rate": 6.4589321163575124e-06, "loss": 0.2238, "step": 6228 }, { "epoch": 0.85, "grad_norm": 4.100596018806647, "learning_rate": 6.457878475935467e-06, "loss": 0.2863, "step": 6229 }, { "epoch": 0.85, "grad_norm": 3.7547577952265705, "learning_rate": 6.456824764754368e-06, "loss": 0.1925, "step": 6230 }, { "epoch": 0.85, "grad_norm": 4.074629514530908, "learning_rate": 6.4557709828653594e-06, "loss": 0.2307, "step": 6231 }, { "epoch": 0.85, "grad_norm": 4.931279333706991, "learning_rate": 6.454717130319583e-06, "loss": 0.2895, "step": 6232 }, { "epoch": 0.85, "grad_norm": 3.855612164626369, "learning_rate": 6.45366320716819e-06, "loss": 0.231, "step": 6233 }, { "epoch": 0.85, "grad_norm": 3.7770381669778157, "learning_rate": 6.452609213462335e-06, "loss": 0.1907, "step": 6234 }, { "epoch": 0.85, "grad_norm": 4.244606031457142, "learning_rate": 6.451555149253174e-06, "loss": 0.2786, "step": 6235 }, { "epoch": 0.85, "grad_norm": 4.038567008006982, "learning_rate": 6.450501014591866e-06, "loss": 0.2164, "step": 6236 }, { "epoch": 0.85, "grad_norm": 3.593821900602644, "learning_rate": 6.449446809529573e-06, "loss": 0.2241, "step": 6237 }, { "epoch": 0.85, "grad_norm": 3.1889075965716365, "learning_rate": 6.4483925341174625e-06, "loss": 0.2074, "step": 6238 }, { "epoch": 0.85, "grad_norm": 3.2514907519708474, "learning_rate": 6.447338188406705e-06, "loss": 0.214, "step": 6239 }, { "epoch": 0.85, "grad_norm": 3.9136408516224312, "learning_rate": 6.446283772448471e-06, "loss": 0.2674, "step": 6240 }, { "epoch": 0.85, "grad_norm": 3.9800250105406523, "learning_rate": 6.445229286293943e-06, "loss": 0.2657, "step": 6241 }, { "epoch": 0.85, "grad_norm": 3.4735664713437293, "learning_rate": 6.444174729994295e-06, "loss": 0.2177, "step": 6242 }, { "epoch": 0.85, "grad_norm": 3.7744713040494715, "learning_rate": 6.443120103600715e-06, "loss": 0.2095, "step": 6243 }, { "epoch": 0.85, "grad_norm": 3.482237550744052, "learning_rate": 6.442065407164387e-06, "loss": 0.2019, "step": 6244 }, { "epoch": 0.85, "grad_norm": 2.7039204014058997, "learning_rate": 6.441010640736502e-06, "loss": 0.1656, "step": 6245 }, { "epoch": 0.85, "grad_norm": 3.548400622383901, "learning_rate": 6.4399558043682556e-06, "loss": 0.2361, "step": 6246 }, { "epoch": 0.85, "grad_norm": 3.4101089792550603, "learning_rate": 6.438900898110843e-06, "loss": 0.2048, "step": 6247 }, { "epoch": 0.85, "grad_norm": 3.863436689157889, "learning_rate": 6.437845922015465e-06, "loss": 0.228, "step": 6248 }, { "epoch": 0.85, "grad_norm": 3.6229316633740414, "learning_rate": 6.4367908761333275e-06, "loss": 0.2655, "step": 6249 }, { "epoch": 0.85, "grad_norm": 3.193152356772067, "learning_rate": 6.435735760515635e-06, "loss": 0.2562, "step": 6250 }, { "epoch": 0.85, "grad_norm": 3.5382910900212, "learning_rate": 6.4346805752136e-06, "loss": 0.2302, "step": 6251 }, { "epoch": 0.85, "grad_norm": 3.0797695244296324, "learning_rate": 6.433625320278435e-06, "loss": 0.2364, "step": 6252 }, { "epoch": 0.85, "grad_norm": 3.6911450347535193, "learning_rate": 6.432569995761359e-06, "loss": 0.2143, "step": 6253 }, { "epoch": 0.85, "grad_norm": 3.6354663136194945, "learning_rate": 6.431514601713592e-06, "loss": 0.186, "step": 6254 }, { "epoch": 0.85, "grad_norm": 5.456174956650405, "learning_rate": 6.430459138186359e-06, "loss": 0.2238, "step": 6255 }, { "epoch": 0.85, "grad_norm": 4.571978858027498, "learning_rate": 6.429403605230886e-06, "loss": 0.2385, "step": 6256 }, { "epoch": 0.85, "grad_norm": 2.8719562583673586, "learning_rate": 6.4283480028984065e-06, "loss": 0.1962, "step": 6257 }, { "epoch": 0.85, "grad_norm": 3.8371889960656715, "learning_rate": 6.427292331240154e-06, "loss": 0.2517, "step": 6258 }, { "epoch": 0.85, "grad_norm": 3.9133025937746044, "learning_rate": 6.426236590307364e-06, "loss": 0.2286, "step": 6259 }, { "epoch": 0.85, "grad_norm": 5.49961433873701, "learning_rate": 6.425180780151279e-06, "loss": 0.1937, "step": 6260 }, { "epoch": 0.85, "grad_norm": 3.416703922157321, "learning_rate": 6.424124900823144e-06, "loss": 0.2071, "step": 6261 }, { "epoch": 0.85, "grad_norm": 3.362578076230735, "learning_rate": 6.423068952374208e-06, "loss": 0.2323, "step": 6262 }, { "epoch": 0.85, "grad_norm": 4.621723545830709, "learning_rate": 6.422012934855718e-06, "loss": 0.2118, "step": 6263 }, { "epoch": 0.85, "grad_norm": 3.0899930897409664, "learning_rate": 6.420956848318932e-06, "loss": 0.1923, "step": 6264 }, { "epoch": 0.85, "grad_norm": 3.0004298011401063, "learning_rate": 6.419900692815107e-06, "loss": 0.1703, "step": 6265 }, { "epoch": 0.85, "grad_norm": 3.61652265738641, "learning_rate": 6.4188444683955045e-06, "loss": 0.2189, "step": 6266 }, { "epoch": 0.85, "grad_norm": 3.774722531038187, "learning_rate": 6.4177881751113854e-06, "loss": 0.1944, "step": 6267 }, { "epoch": 0.85, "grad_norm": 3.788014625095387, "learning_rate": 6.416731813014024e-06, "loss": 0.2543, "step": 6268 }, { "epoch": 0.85, "grad_norm": 3.1291849676281815, "learning_rate": 6.415675382154687e-06, "loss": 0.2314, "step": 6269 }, { "epoch": 0.85, "grad_norm": 3.558750073673958, "learning_rate": 6.41461888258465e-06, "loss": 0.2182, "step": 6270 }, { "epoch": 0.85, "grad_norm": 3.1997019499411565, "learning_rate": 6.4135623143551924e-06, "loss": 0.237, "step": 6271 }, { "epoch": 0.85, "grad_norm": 3.670099034851638, "learning_rate": 6.412505677517592e-06, "loss": 0.2768, "step": 6272 }, { "epoch": 0.85, "grad_norm": 3.2743693262706715, "learning_rate": 6.411448972123135e-06, "loss": 0.21, "step": 6273 }, { "epoch": 0.85, "grad_norm": 3.9990569273927576, "learning_rate": 6.410392198223112e-06, "loss": 0.2247, "step": 6274 }, { "epoch": 0.85, "grad_norm": 3.2740342189520373, "learning_rate": 6.4093353558688095e-06, "loss": 0.2242, "step": 6275 }, { "epoch": 0.85, "grad_norm": 3.5896021269753575, "learning_rate": 6.408278445111525e-06, "loss": 0.2037, "step": 6276 }, { "epoch": 0.85, "grad_norm": 4.1486879727890065, "learning_rate": 6.4072214660025555e-06, "loss": 0.2173, "step": 6277 }, { "epoch": 0.85, "grad_norm": 4.133453859169418, "learning_rate": 6.406164418593203e-06, "loss": 0.2364, "step": 6278 }, { "epoch": 0.85, "grad_norm": 3.7741799088636325, "learning_rate": 6.405107302934769e-06, "loss": 0.2263, "step": 6279 }, { "epoch": 0.85, "grad_norm": 3.154710933693885, "learning_rate": 6.404050119078565e-06, "loss": 0.2021, "step": 6280 }, { "epoch": 0.85, "grad_norm": 3.614171766159116, "learning_rate": 6.402992867075901e-06, "loss": 0.1963, "step": 6281 }, { "epoch": 0.85, "grad_norm": 4.002724795759052, "learning_rate": 6.401935546978091e-06, "loss": 0.2297, "step": 6282 }, { "epoch": 0.85, "grad_norm": 3.53737755573692, "learning_rate": 6.400878158836452e-06, "loss": 0.2649, "step": 6283 }, { "epoch": 0.85, "grad_norm": 3.834776668864288, "learning_rate": 6.3998207027023056e-06, "loss": 0.2232, "step": 6284 }, { "epoch": 0.85, "grad_norm": 3.524859717129629, "learning_rate": 6.398763178626977e-06, "loss": 0.2447, "step": 6285 }, { "epoch": 0.86, "grad_norm": 4.028739324626779, "learning_rate": 6.397705586661791e-06, "loss": 0.2201, "step": 6286 }, { "epoch": 0.86, "grad_norm": 3.290085845351364, "learning_rate": 6.396647926858082e-06, "loss": 0.2019, "step": 6287 }, { "epoch": 0.86, "grad_norm": 3.358901668407234, "learning_rate": 6.395590199267182e-06, "loss": 0.2261, "step": 6288 }, { "epoch": 0.86, "grad_norm": 3.820131181378966, "learning_rate": 6.39453240394043e-06, "loss": 0.2556, "step": 6289 }, { "epoch": 0.86, "grad_norm": 4.645973335986187, "learning_rate": 6.393474540929166e-06, "loss": 0.1811, "step": 6290 }, { "epoch": 0.86, "grad_norm": 3.70708155412316, "learning_rate": 6.3924166102847326e-06, "loss": 0.2508, "step": 6291 }, { "epoch": 0.86, "grad_norm": 3.046881683094606, "learning_rate": 6.391358612058479e-06, "loss": 0.2466, "step": 6292 }, { "epoch": 0.86, "grad_norm": 3.578242947231897, "learning_rate": 6.390300546301754e-06, "loss": 0.2271, "step": 6293 }, { "epoch": 0.86, "grad_norm": 2.88971158997138, "learning_rate": 6.389242413065915e-06, "loss": 0.1755, "step": 6294 }, { "epoch": 0.86, "grad_norm": 2.98810527930011, "learning_rate": 6.388184212402316e-06, "loss": 0.2059, "step": 6295 }, { "epoch": 0.86, "grad_norm": 3.988116170919707, "learning_rate": 6.387125944362319e-06, "loss": 0.2689, "step": 6296 }, { "epoch": 0.86, "grad_norm": 3.203476460855589, "learning_rate": 6.386067608997286e-06, "loss": 0.2301, "step": 6297 }, { "epoch": 0.86, "grad_norm": 5.443615597816084, "learning_rate": 6.385009206358587e-06, "loss": 0.2339, "step": 6298 }, { "epoch": 0.86, "grad_norm": 3.3169251592565456, "learning_rate": 6.383950736497588e-06, "loss": 0.2145, "step": 6299 }, { "epoch": 0.86, "grad_norm": 3.8086233402688072, "learning_rate": 6.382892199465665e-06, "loss": 0.235, "step": 6300 }, { "epoch": 0.86, "grad_norm": 4.008234326190306, "learning_rate": 6.3818335953141955e-06, "loss": 0.2107, "step": 6301 }, { "epoch": 0.86, "grad_norm": 3.316212988096248, "learning_rate": 6.3807749240945594e-06, "loss": 0.2573, "step": 6302 }, { "epoch": 0.86, "grad_norm": 3.0963904527569612, "learning_rate": 6.379716185858138e-06, "loss": 0.216, "step": 6303 }, { "epoch": 0.86, "grad_norm": 4.416106113140859, "learning_rate": 6.378657380656319e-06, "loss": 0.2428, "step": 6304 }, { "epoch": 0.86, "grad_norm": 4.272595989649147, "learning_rate": 6.377598508540492e-06, "loss": 0.2338, "step": 6305 }, { "epoch": 0.86, "grad_norm": 3.9453212792031533, "learning_rate": 6.37653956956205e-06, "loss": 0.2216, "step": 6306 }, { "epoch": 0.86, "grad_norm": 3.6938652641683576, "learning_rate": 6.375480563772391e-06, "loss": 0.2204, "step": 6307 }, { "epoch": 0.86, "grad_norm": 5.576086871645379, "learning_rate": 6.374421491222912e-06, "loss": 0.2772, "step": 6308 }, { "epoch": 0.86, "grad_norm": 3.288294116753507, "learning_rate": 6.373362351965016e-06, "loss": 0.2142, "step": 6309 }, { "epoch": 0.86, "grad_norm": 4.382252357656026, "learning_rate": 6.3723031460501115e-06, "loss": 0.2276, "step": 6310 }, { "epoch": 0.86, "grad_norm": 3.926541166222237, "learning_rate": 6.371243873529607e-06, "loss": 0.1874, "step": 6311 }, { "epoch": 0.86, "grad_norm": 4.387147855416779, "learning_rate": 6.3701845344549105e-06, "loss": 0.2242, "step": 6312 }, { "epoch": 0.86, "grad_norm": 3.4100166258679936, "learning_rate": 6.3691251288774425e-06, "loss": 0.1962, "step": 6313 }, { "epoch": 0.86, "grad_norm": 4.146699568292299, "learning_rate": 6.368065656848621e-06, "loss": 0.2185, "step": 6314 }, { "epoch": 0.86, "grad_norm": 4.065046790356425, "learning_rate": 6.367006118419867e-06, "loss": 0.2271, "step": 6315 }, { "epoch": 0.86, "grad_norm": 3.671789176653451, "learning_rate": 6.365946513642607e-06, "loss": 0.2733, "step": 6316 }, { "epoch": 0.86, "grad_norm": 3.8075093207660893, "learning_rate": 6.3648868425682695e-06, "loss": 0.2443, "step": 6317 }, { "epoch": 0.86, "grad_norm": 3.2266883425360176, "learning_rate": 6.363827105248286e-06, "loss": 0.22, "step": 6318 }, { "epoch": 0.86, "grad_norm": 3.408215218698312, "learning_rate": 6.36276730173409e-06, "loss": 0.2453, "step": 6319 }, { "epoch": 0.86, "grad_norm": 3.6251194598569496, "learning_rate": 6.361707432077122e-06, "loss": 0.2487, "step": 6320 }, { "epoch": 0.86, "grad_norm": 3.7994626072437923, "learning_rate": 6.3606474963288225e-06, "loss": 0.2538, "step": 6321 }, { "epoch": 0.86, "grad_norm": 4.209252772571734, "learning_rate": 6.359587494540638e-06, "loss": 0.2093, "step": 6322 }, { "epoch": 0.86, "grad_norm": 4.9583717964171115, "learning_rate": 6.358527426764013e-06, "loss": 0.2163, "step": 6323 }, { "epoch": 0.86, "grad_norm": 3.0346532744836865, "learning_rate": 6.3574672930504e-06, "loss": 0.2059, "step": 6324 }, { "epoch": 0.86, "grad_norm": 4.24051324167431, "learning_rate": 6.356407093451254e-06, "loss": 0.2127, "step": 6325 }, { "epoch": 0.86, "grad_norm": 4.168879300754164, "learning_rate": 6.355346828018032e-06, "loss": 0.2669, "step": 6326 }, { "epoch": 0.86, "grad_norm": 3.9065575767978657, "learning_rate": 6.354286496802195e-06, "loss": 0.2456, "step": 6327 }, { "epoch": 0.86, "grad_norm": 3.9332219327376734, "learning_rate": 6.3532260998552055e-06, "loss": 0.221, "step": 6328 }, { "epoch": 0.86, "grad_norm": 3.2723031290400146, "learning_rate": 6.352165637228531e-06, "loss": 0.1922, "step": 6329 }, { "epoch": 0.86, "grad_norm": 7.01317795905018, "learning_rate": 6.351105108973644e-06, "loss": 0.2651, "step": 6330 }, { "epoch": 0.86, "grad_norm": 3.5354937207679127, "learning_rate": 6.350044515142015e-06, "loss": 0.2267, "step": 6331 }, { "epoch": 0.86, "grad_norm": 3.6829392302140356, "learning_rate": 6.348983855785122e-06, "loss": 0.2114, "step": 6332 }, { "epoch": 0.86, "grad_norm": 3.391927247977577, "learning_rate": 6.347923130954444e-06, "loss": 0.2094, "step": 6333 }, { "epoch": 0.86, "grad_norm": 4.161414014670577, "learning_rate": 6.346862340701465e-06, "loss": 0.2115, "step": 6334 }, { "epoch": 0.86, "grad_norm": 4.072931741532394, "learning_rate": 6.345801485077671e-06, "loss": 0.2583, "step": 6335 }, { "epoch": 0.86, "grad_norm": 3.7344684501709247, "learning_rate": 6.344740564134551e-06, "loss": 0.2314, "step": 6336 }, { "epoch": 0.86, "grad_norm": 3.6948265153874256, "learning_rate": 6.343679577923596e-06, "loss": 0.2538, "step": 6337 }, { "epoch": 0.86, "grad_norm": 3.7988542539733716, "learning_rate": 6.342618526496303e-06, "loss": 0.2337, "step": 6338 }, { "epoch": 0.86, "grad_norm": 4.047297128615549, "learning_rate": 6.341557409904171e-06, "loss": 0.2432, "step": 6339 }, { "epoch": 0.86, "grad_norm": 2.9186720730918956, "learning_rate": 6.340496228198703e-06, "loss": 0.1743, "step": 6340 }, { "epoch": 0.86, "grad_norm": 2.868747051513251, "learning_rate": 6.3394349814314e-06, "loss": 0.1947, "step": 6341 }, { "epoch": 0.86, "grad_norm": 3.581354006199669, "learning_rate": 6.338373669653777e-06, "loss": 0.2113, "step": 6342 }, { "epoch": 0.86, "grad_norm": 6.538372338705542, "learning_rate": 6.337312292917339e-06, "loss": 0.2489, "step": 6343 }, { "epoch": 0.86, "grad_norm": 3.592049558330966, "learning_rate": 6.336250851273605e-06, "loss": 0.2791, "step": 6344 }, { "epoch": 0.86, "grad_norm": 3.3355803174592045, "learning_rate": 6.335189344774088e-06, "loss": 0.2359, "step": 6345 }, { "epoch": 0.86, "grad_norm": 3.464364697359595, "learning_rate": 6.334127773470313e-06, "loss": 0.2153, "step": 6346 }, { "epoch": 0.86, "grad_norm": 4.01696291785864, "learning_rate": 6.333066137413803e-06, "loss": 0.2223, "step": 6347 }, { "epoch": 0.86, "grad_norm": 3.242262851219089, "learning_rate": 6.332004436656085e-06, "loss": 0.2176, "step": 6348 }, { "epoch": 0.86, "grad_norm": 3.256344209673177, "learning_rate": 6.33094267124869e-06, "loss": 0.2031, "step": 6349 }, { "epoch": 0.86, "grad_norm": 3.896204551140596, "learning_rate": 6.329880841243151e-06, "loss": 0.2106, "step": 6350 }, { "epoch": 0.86, "grad_norm": 4.0844651596895805, "learning_rate": 6.328818946691005e-06, "loss": 0.2432, "step": 6351 }, { "epoch": 0.86, "grad_norm": 4.057704175071436, "learning_rate": 6.327756987643788e-06, "loss": 0.2435, "step": 6352 }, { "epoch": 0.86, "grad_norm": 4.237015790531594, "learning_rate": 6.3266949641530486e-06, "loss": 0.2368, "step": 6353 }, { "epoch": 0.86, "grad_norm": 11.777968936890868, "learning_rate": 6.325632876270329e-06, "loss": 0.2462, "step": 6354 }, { "epoch": 0.86, "grad_norm": 4.02382775496425, "learning_rate": 6.32457072404718e-06, "loss": 0.2601, "step": 6355 }, { "epoch": 0.86, "grad_norm": 3.2529648399578237, "learning_rate": 6.323508507535153e-06, "loss": 0.2104, "step": 6356 }, { "epoch": 0.86, "grad_norm": 3.264515119448179, "learning_rate": 6.322446226785803e-06, "loss": 0.2211, "step": 6357 }, { "epoch": 0.86, "grad_norm": 3.6138470217224787, "learning_rate": 6.321383881850689e-06, "loss": 0.2549, "step": 6358 }, { "epoch": 0.86, "grad_norm": 4.284429578593238, "learning_rate": 6.320321472781375e-06, "loss": 0.218, "step": 6359 }, { "epoch": 0.87, "grad_norm": 3.130516004043148, "learning_rate": 6.319258999629421e-06, "loss": 0.1881, "step": 6360 }, { "epoch": 0.87, "grad_norm": 3.5842057235350997, "learning_rate": 6.318196462446397e-06, "loss": 0.253, "step": 6361 }, { "epoch": 0.87, "grad_norm": 3.0729298133435496, "learning_rate": 6.317133861283876e-06, "loss": 0.1928, "step": 6362 }, { "epoch": 0.87, "grad_norm": 3.107022032247247, "learning_rate": 6.31607119619343e-06, "loss": 0.2334, "step": 6363 }, { "epoch": 0.87, "grad_norm": 3.1574319165755784, "learning_rate": 6.315008467226635e-06, "loss": 0.191, "step": 6364 }, { "epoch": 0.87, "grad_norm": 3.266987760333765, "learning_rate": 6.313945674435071e-06, "loss": 0.1952, "step": 6365 }, { "epoch": 0.87, "grad_norm": 4.107560830272779, "learning_rate": 6.312882817870325e-06, "loss": 0.2486, "step": 6366 }, { "epoch": 0.87, "grad_norm": 4.0250421514807675, "learning_rate": 6.311819897583981e-06, "loss": 0.2301, "step": 6367 }, { "epoch": 0.87, "grad_norm": 8.873939843767625, "learning_rate": 6.310756913627628e-06, "loss": 0.2151, "step": 6368 }, { "epoch": 0.87, "grad_norm": 3.2677539970256104, "learning_rate": 6.309693866052859e-06, "loss": 0.2008, "step": 6369 }, { "epoch": 0.87, "grad_norm": 3.8160229082452477, "learning_rate": 6.3086307549112715e-06, "loss": 0.2415, "step": 6370 }, { "epoch": 0.87, "grad_norm": 2.7543634191287705, "learning_rate": 6.3075675802544615e-06, "loss": 0.2094, "step": 6371 }, { "epoch": 0.87, "grad_norm": 3.1912248743560423, "learning_rate": 6.306504342134032e-06, "loss": 0.2392, "step": 6372 }, { "epoch": 0.87, "grad_norm": 3.2706506841636958, "learning_rate": 6.305441040601589e-06, "loss": 0.2274, "step": 6373 }, { "epoch": 0.87, "grad_norm": 3.1867996287982514, "learning_rate": 6.3043776757087385e-06, "loss": 0.2111, "step": 6374 }, { "epoch": 0.87, "grad_norm": 4.076212395640897, "learning_rate": 6.303314247507093e-06, "loss": 0.2941, "step": 6375 }, { "epoch": 0.87, "grad_norm": 2.954251641951496, "learning_rate": 6.302250756048267e-06, "loss": 0.2087, "step": 6376 }, { "epoch": 0.87, "grad_norm": 3.346140957267289, "learning_rate": 6.301187201383876e-06, "loss": 0.2328, "step": 6377 }, { "epoch": 0.87, "grad_norm": 4.098965228430852, "learning_rate": 6.3001235835655424e-06, "loss": 0.2592, "step": 6378 }, { "epoch": 0.87, "grad_norm": 3.4409679602539627, "learning_rate": 6.299059902644886e-06, "loss": 0.2226, "step": 6379 }, { "epoch": 0.87, "grad_norm": 3.9361331596928455, "learning_rate": 6.297996158673538e-06, "loss": 0.2246, "step": 6380 }, { "epoch": 0.87, "grad_norm": 4.12736392363275, "learning_rate": 6.296932351703125e-06, "loss": 0.2333, "step": 6381 }, { "epoch": 0.87, "grad_norm": 5.974071738147866, "learning_rate": 6.295868481785281e-06, "loss": 0.2592, "step": 6382 }, { "epoch": 0.87, "grad_norm": 3.856283062173541, "learning_rate": 6.29480454897164e-06, "loss": 0.2125, "step": 6383 }, { "epoch": 0.87, "grad_norm": 3.3369511886614345, "learning_rate": 6.293740553313841e-06, "loss": 0.2063, "step": 6384 }, { "epoch": 0.87, "grad_norm": 4.230800853554127, "learning_rate": 6.292676494863528e-06, "loss": 0.2943, "step": 6385 }, { "epoch": 0.87, "grad_norm": 3.644353374436022, "learning_rate": 6.291612373672341e-06, "loss": 0.2267, "step": 6386 }, { "epoch": 0.87, "grad_norm": 3.4011880341861582, "learning_rate": 6.290548189791932e-06, "loss": 0.1866, "step": 6387 }, { "epoch": 0.87, "grad_norm": 3.594532224285504, "learning_rate": 6.28948394327395e-06, "loss": 0.2303, "step": 6388 }, { "epoch": 0.87, "grad_norm": 3.6681460521650266, "learning_rate": 6.288419634170051e-06, "loss": 0.2569, "step": 6389 }, { "epoch": 0.87, "grad_norm": 3.145050891762851, "learning_rate": 6.287355262531889e-06, "loss": 0.2095, "step": 6390 }, { "epoch": 0.87, "grad_norm": 3.088194690993271, "learning_rate": 6.286290828411126e-06, "loss": 0.2269, "step": 6391 }, { "epoch": 0.87, "grad_norm": 3.287622697928698, "learning_rate": 6.285226331859423e-06, "loss": 0.2132, "step": 6392 }, { "epoch": 0.87, "grad_norm": 3.83020609738234, "learning_rate": 6.284161772928448e-06, "loss": 0.2383, "step": 6393 }, { "epoch": 0.87, "grad_norm": 3.868486744213251, "learning_rate": 6.283097151669869e-06, "loss": 0.2685, "step": 6394 }, { "epoch": 0.87, "grad_norm": 3.5791407716707906, "learning_rate": 6.282032468135358e-06, "loss": 0.2427, "step": 6395 }, { "epoch": 0.87, "grad_norm": 3.4270914252481663, "learning_rate": 6.280967722376592e-06, "loss": 0.2175, "step": 6396 }, { "epoch": 0.87, "grad_norm": 3.3805052122365145, "learning_rate": 6.279902914445246e-06, "loss": 0.3102, "step": 6397 }, { "epoch": 0.87, "grad_norm": 4.29997292433313, "learning_rate": 6.2788380443930015e-06, "loss": 0.2165, "step": 6398 }, { "epoch": 0.87, "grad_norm": 3.7179655358731867, "learning_rate": 6.277773112271545e-06, "loss": 0.2664, "step": 6399 }, { "epoch": 0.87, "grad_norm": 3.870496331386032, "learning_rate": 6.276708118132564e-06, "loss": 0.2726, "step": 6400 }, { "epoch": 0.87, "grad_norm": 3.1379046341493027, "learning_rate": 6.275643062027746e-06, "loss": 0.1855, "step": 6401 }, { "epoch": 0.87, "grad_norm": 3.47471908933929, "learning_rate": 6.274577944008785e-06, "loss": 0.221, "step": 6402 }, { "epoch": 0.87, "grad_norm": 2.946044311861358, "learning_rate": 6.273512764127377e-06, "loss": 0.248, "step": 6403 }, { "epoch": 0.87, "grad_norm": 4.5698923020173625, "learning_rate": 6.272447522435223e-06, "loss": 0.2811, "step": 6404 }, { "epoch": 0.87, "grad_norm": 2.996362724307452, "learning_rate": 6.271382218984021e-06, "loss": 0.2187, "step": 6405 }, { "epoch": 0.87, "grad_norm": 3.9955963350145596, "learning_rate": 6.270316853825482e-06, "loss": 0.2074, "step": 6406 }, { "epoch": 0.87, "grad_norm": 3.3411095234330306, "learning_rate": 6.26925142701131e-06, "loss": 0.2245, "step": 6407 }, { "epoch": 0.87, "grad_norm": 3.908978358472412, "learning_rate": 6.268185938593218e-06, "loss": 0.2634, "step": 6408 }, { "epoch": 0.87, "grad_norm": 3.1291502219796494, "learning_rate": 6.2671203886229186e-06, "loss": 0.232, "step": 6409 }, { "epoch": 0.87, "grad_norm": 3.138820562919338, "learning_rate": 6.266054777152131e-06, "loss": 0.2316, "step": 6410 }, { "epoch": 0.87, "grad_norm": 4.495193853707699, "learning_rate": 6.264989104232573e-06, "loss": 0.2173, "step": 6411 }, { "epoch": 0.87, "grad_norm": 3.62231674595275, "learning_rate": 6.263923369915968e-06, "loss": 0.2486, "step": 6412 }, { "epoch": 0.87, "grad_norm": 2.9599900102277403, "learning_rate": 6.262857574254044e-06, "loss": 0.2218, "step": 6413 }, { "epoch": 0.87, "grad_norm": 4.190379491475833, "learning_rate": 6.2617917172985285e-06, "loss": 0.2474, "step": 6414 }, { "epoch": 0.87, "grad_norm": 4.1347064601124455, "learning_rate": 6.260725799101156e-06, "loss": 0.2314, "step": 6415 }, { "epoch": 0.87, "grad_norm": 3.6448488967648047, "learning_rate": 6.259659819713658e-06, "loss": 0.2195, "step": 6416 }, { "epoch": 0.87, "grad_norm": 3.357941066820471, "learning_rate": 6.258593779187774e-06, "loss": 0.2301, "step": 6417 }, { "epoch": 0.87, "grad_norm": 3.6844676355086747, "learning_rate": 6.257527677575246e-06, "loss": 0.2513, "step": 6418 }, { "epoch": 0.87, "grad_norm": 3.2491899539599105, "learning_rate": 6.256461514927815e-06, "loss": 0.2127, "step": 6419 }, { "epoch": 0.87, "grad_norm": 3.730817762641752, "learning_rate": 6.255395291297232e-06, "loss": 0.2626, "step": 6420 }, { "epoch": 0.87, "grad_norm": 3.5559618164773985, "learning_rate": 6.254329006735242e-06, "loss": 0.2578, "step": 6421 }, { "epoch": 0.87, "grad_norm": 4.264205584057877, "learning_rate": 6.2532626612936035e-06, "loss": 0.2737, "step": 6422 }, { "epoch": 0.87, "grad_norm": 3.6154909153211783, "learning_rate": 6.252196255024068e-06, "loss": 0.223, "step": 6423 }, { "epoch": 0.87, "grad_norm": 3.294660866971795, "learning_rate": 6.251129787978395e-06, "loss": 0.1955, "step": 6424 }, { "epoch": 0.87, "grad_norm": 3.306091705209511, "learning_rate": 6.250063260208345e-06, "loss": 0.2098, "step": 6425 }, { "epoch": 0.87, "grad_norm": 3.8056924986101985, "learning_rate": 6.2489966717656866e-06, "loss": 0.2052, "step": 6426 }, { "epoch": 0.87, "grad_norm": 4.044805601771355, "learning_rate": 6.247930022702184e-06, "loss": 0.1868, "step": 6427 }, { "epoch": 0.87, "grad_norm": 3.3869949937929293, "learning_rate": 6.246863313069609e-06, "loss": 0.1852, "step": 6428 }, { "epoch": 0.87, "grad_norm": 3.766711637571515, "learning_rate": 6.2457965429197346e-06, "loss": 0.2453, "step": 6429 }, { "epoch": 0.87, "grad_norm": 3.5340184178470557, "learning_rate": 6.244729712304337e-06, "loss": 0.2716, "step": 6430 }, { "epoch": 0.87, "grad_norm": 3.3375129556612677, "learning_rate": 6.243662821275196e-06, "loss": 0.231, "step": 6431 }, { "epoch": 0.87, "grad_norm": 3.746934404040534, "learning_rate": 6.242595869884093e-06, "loss": 0.2463, "step": 6432 }, { "epoch": 0.88, "grad_norm": 3.598907095285613, "learning_rate": 6.241528858182814e-06, "loss": 0.2922, "step": 6433 }, { "epoch": 0.88, "grad_norm": 3.7429758356059875, "learning_rate": 6.240461786223146e-06, "loss": 0.1844, "step": 6434 }, { "epoch": 0.88, "grad_norm": 3.497715113696888, "learning_rate": 6.23939465405688e-06, "loss": 0.2252, "step": 6435 }, { "epoch": 0.88, "grad_norm": 3.1039341324185625, "learning_rate": 6.238327461735813e-06, "loss": 0.2353, "step": 6436 }, { "epoch": 0.88, "grad_norm": 3.2659117913501414, "learning_rate": 6.237260209311738e-06, "loss": 0.1931, "step": 6437 }, { "epoch": 0.88, "grad_norm": 3.1653655881965315, "learning_rate": 6.236192896836456e-06, "loss": 0.2288, "step": 6438 }, { "epoch": 0.88, "grad_norm": 3.7301442682792203, "learning_rate": 6.235125524361772e-06, "loss": 0.24, "step": 6439 }, { "epoch": 0.88, "grad_norm": 3.9763890770358135, "learning_rate": 6.234058091939488e-06, "loss": 0.2945, "step": 6440 }, { "epoch": 0.88, "grad_norm": 3.126381975343611, "learning_rate": 6.232990599621414e-06, "loss": 0.1853, "step": 6441 }, { "epoch": 0.88, "grad_norm": 5.007332127859017, "learning_rate": 6.231923047459362e-06, "loss": 0.346, "step": 6442 }, { "epoch": 0.88, "grad_norm": 3.6819304047434214, "learning_rate": 6.230855435505145e-06, "loss": 0.2465, "step": 6443 }, { "epoch": 0.88, "grad_norm": 3.5623073584531806, "learning_rate": 6.2297877638105815e-06, "loss": 0.1988, "step": 6444 }, { "epoch": 0.88, "grad_norm": 3.4917045010217915, "learning_rate": 6.228720032427491e-06, "loss": 0.2566, "step": 6445 }, { "epoch": 0.88, "grad_norm": 4.706213192485414, "learning_rate": 6.227652241407698e-06, "loss": 0.2222, "step": 6446 }, { "epoch": 0.88, "grad_norm": 4.366223999049163, "learning_rate": 6.2265843908030255e-06, "loss": 0.2581, "step": 6447 }, { "epoch": 0.88, "grad_norm": 3.195305806928291, "learning_rate": 6.225516480665306e-06, "loss": 0.2247, "step": 6448 }, { "epoch": 0.88, "grad_norm": 3.540889340100673, "learning_rate": 6.224448511046367e-06, "loss": 0.2324, "step": 6449 }, { "epoch": 0.88, "grad_norm": 3.8564285905209394, "learning_rate": 6.223380481998046e-06, "loss": 0.2467, "step": 6450 }, { "epoch": 0.88, "grad_norm": 3.4458653522931737, "learning_rate": 6.2223123935721785e-06, "loss": 0.2416, "step": 6451 }, { "epoch": 0.88, "grad_norm": 2.994357883096625, "learning_rate": 6.2212442458206065e-06, "loss": 0.178, "step": 6452 }, { "epoch": 0.88, "grad_norm": 2.96424550505415, "learning_rate": 6.220176038795174e-06, "loss": 0.1883, "step": 6453 }, { "epoch": 0.88, "grad_norm": 2.9069805614336053, "learning_rate": 6.2191077725477255e-06, "loss": 0.1935, "step": 6454 }, { "epoch": 0.88, "grad_norm": 4.242720887190229, "learning_rate": 6.218039447130109e-06, "loss": 0.2381, "step": 6455 }, { "epoch": 0.88, "grad_norm": 3.3901513003303023, "learning_rate": 6.216971062594179e-06, "loss": 0.2182, "step": 6456 }, { "epoch": 0.88, "grad_norm": 3.462039108678252, "learning_rate": 6.215902618991789e-06, "loss": 0.1739, "step": 6457 }, { "epoch": 0.88, "grad_norm": 3.5758895590770092, "learning_rate": 6.214834116374795e-06, "loss": 0.2307, "step": 6458 }, { "epoch": 0.88, "grad_norm": 3.3571151139069455, "learning_rate": 6.213765554795061e-06, "loss": 0.2715, "step": 6459 }, { "epoch": 0.88, "grad_norm": 3.761314973679387, "learning_rate": 6.2126969343044464e-06, "loss": 0.2428, "step": 6460 }, { "epoch": 0.88, "grad_norm": 4.378555438511868, "learning_rate": 6.21162825495482e-06, "loss": 0.2311, "step": 6461 }, { "epoch": 0.88, "grad_norm": 4.631084361861994, "learning_rate": 6.21055951679805e-06, "loss": 0.246, "step": 6462 }, { "epoch": 0.88, "grad_norm": 3.9789909678305277, "learning_rate": 6.209490719886009e-06, "loss": 0.2679, "step": 6463 }, { "epoch": 0.88, "grad_norm": 3.4931649977422357, "learning_rate": 6.20842186427057e-06, "loss": 0.2149, "step": 6464 }, { "epoch": 0.88, "grad_norm": 3.8575846154672964, "learning_rate": 6.207352950003611e-06, "loss": 0.2471, "step": 6465 }, { "epoch": 0.88, "grad_norm": 4.074446541800772, "learning_rate": 6.2062839771370135e-06, "loss": 0.2409, "step": 6466 }, { "epoch": 0.88, "grad_norm": 4.154211669266572, "learning_rate": 6.20521494572266e-06, "loss": 0.2556, "step": 6467 }, { "epoch": 0.88, "grad_norm": 3.861266856671814, "learning_rate": 6.204145855812439e-06, "loss": 0.2289, "step": 6468 }, { "epoch": 0.88, "grad_norm": 3.7333681651149138, "learning_rate": 6.203076707458236e-06, "loss": 0.2522, "step": 6469 }, { "epoch": 0.88, "grad_norm": 3.567074711139495, "learning_rate": 6.202007500711944e-06, "loss": 0.255, "step": 6470 }, { "epoch": 0.88, "grad_norm": 3.66620226534803, "learning_rate": 6.2009382356254585e-06, "loss": 0.1882, "step": 6471 }, { "epoch": 0.88, "grad_norm": 3.995637510812192, "learning_rate": 6.1998689122506765e-06, "loss": 0.2442, "step": 6472 }, { "epoch": 0.88, "grad_norm": 4.087232642272171, "learning_rate": 6.198799530639497e-06, "loss": 0.2648, "step": 6473 }, { "epoch": 0.88, "grad_norm": 3.9275027295657816, "learning_rate": 6.197730090843824e-06, "loss": 0.2031, "step": 6474 }, { "epoch": 0.88, "grad_norm": 2.717176595437863, "learning_rate": 6.196660592915564e-06, "loss": 0.1861, "step": 6475 }, { "epoch": 0.88, "grad_norm": 3.344338081338973, "learning_rate": 6.1955910369066265e-06, "loss": 0.2275, "step": 6476 }, { "epoch": 0.88, "grad_norm": 3.0500630949301812, "learning_rate": 6.19452142286892e-06, "loss": 0.1848, "step": 6477 }, { "epoch": 0.88, "grad_norm": 2.981930476967746, "learning_rate": 6.193451750854362e-06, "loss": 0.1894, "step": 6478 }, { "epoch": 0.88, "grad_norm": 4.003034099007735, "learning_rate": 6.192382020914869e-06, "loss": 0.2222, "step": 6479 }, { "epoch": 0.88, "grad_norm": 3.4943418476124326, "learning_rate": 6.19131223310236e-06, "loss": 0.2301, "step": 6480 }, { "epoch": 0.88, "grad_norm": 3.8589000500712407, "learning_rate": 6.1902423874687575e-06, "loss": 0.2108, "step": 6481 }, { "epoch": 0.88, "grad_norm": 3.290503526099538, "learning_rate": 6.1891724840659895e-06, "loss": 0.2081, "step": 6482 }, { "epoch": 0.88, "grad_norm": 4.452943416785804, "learning_rate": 6.188102522945982e-06, "loss": 0.1843, "step": 6483 }, { "epoch": 0.88, "grad_norm": 4.117552604608875, "learning_rate": 6.187032504160667e-06, "loss": 0.2416, "step": 6484 }, { "epoch": 0.88, "grad_norm": 4.058813344068317, "learning_rate": 6.185962427761979e-06, "loss": 0.1846, "step": 6485 }, { "epoch": 0.88, "grad_norm": 5.185076200238153, "learning_rate": 6.184892293801855e-06, "loss": 0.2645, "step": 6486 }, { "epoch": 0.88, "grad_norm": 4.1086621357656465, "learning_rate": 6.183822102332234e-06, "loss": 0.2423, "step": 6487 }, { "epoch": 0.88, "grad_norm": 4.686194781556498, "learning_rate": 6.182751853405059e-06, "loss": 0.2211, "step": 6488 }, { "epoch": 0.88, "grad_norm": 4.107044066564524, "learning_rate": 6.1816815470722745e-06, "loss": 0.2427, "step": 6489 }, { "epoch": 0.88, "grad_norm": 4.028719959709479, "learning_rate": 6.18061118338583e-06, "loss": 0.224, "step": 6490 }, { "epoch": 0.88, "grad_norm": 3.473302925908054, "learning_rate": 6.179540762397672e-06, "loss": 0.2444, "step": 6491 }, { "epoch": 0.88, "grad_norm": 4.195428912149643, "learning_rate": 6.17847028415976e-06, "loss": 0.2755, "step": 6492 }, { "epoch": 0.88, "grad_norm": 3.517947098310261, "learning_rate": 6.177399748724047e-06, "loss": 0.1931, "step": 6493 }, { "epoch": 0.88, "grad_norm": 3.6155276477136784, "learning_rate": 6.176329156142492e-06, "loss": 0.2468, "step": 6494 }, { "epoch": 0.88, "grad_norm": 3.8097348571880065, "learning_rate": 6.175258506467058e-06, "loss": 0.2489, "step": 6495 }, { "epoch": 0.88, "grad_norm": 4.197488116957262, "learning_rate": 6.174187799749709e-06, "loss": 0.2236, "step": 6496 }, { "epoch": 0.88, "grad_norm": 3.1832097762030367, "learning_rate": 6.1731170360424116e-06, "loss": 0.2278, "step": 6497 }, { "epoch": 0.88, "grad_norm": 4.371929536345536, "learning_rate": 6.172046215397137e-06, "loss": 0.2774, "step": 6498 }, { "epoch": 0.88, "grad_norm": 3.3767118796680533, "learning_rate": 6.17097533786586e-06, "loss": 0.2447, "step": 6499 }, { "epoch": 0.88, "grad_norm": 3.691761971261659, "learning_rate": 6.169904403500553e-06, "loss": 0.2136, "step": 6500 }, { "epoch": 0.88, "grad_norm": 3.6688107742094687, "learning_rate": 6.1688334123531964e-06, "loss": 0.2327, "step": 6501 }, { "epoch": 0.88, "grad_norm": 3.1540243614243653, "learning_rate": 6.1677623644757715e-06, "loss": 0.2537, "step": 6502 }, { "epoch": 0.88, "grad_norm": 2.938454583635569, "learning_rate": 6.166691259920261e-06, "loss": 0.2088, "step": 6503 }, { "epoch": 0.88, "grad_norm": 3.6398745300114137, "learning_rate": 6.165620098738653e-06, "loss": 0.2159, "step": 6504 }, { "epoch": 0.88, "grad_norm": 4.168887765049372, "learning_rate": 6.164548880982935e-06, "loss": 0.2013, "step": 6505 }, { "epoch": 0.88, "grad_norm": 4.229781478124319, "learning_rate": 6.163477606705101e-06, "loss": 0.2428, "step": 6506 }, { "epoch": 0.89, "grad_norm": 3.138633631880758, "learning_rate": 6.162406275957147e-06, "loss": 0.2464, "step": 6507 }, { "epoch": 0.89, "grad_norm": 4.288788862020617, "learning_rate": 6.161334888791069e-06, "loss": 0.21, "step": 6508 }, { "epoch": 0.89, "grad_norm": 4.107240412418441, "learning_rate": 6.160263445258867e-06, "loss": 0.2424, "step": 6509 }, { "epoch": 0.89, "grad_norm": 4.093825809060403, "learning_rate": 6.159191945412546e-06, "loss": 0.2176, "step": 6510 }, { "epoch": 0.89, "grad_norm": 3.700978903972459, "learning_rate": 6.15812038930411e-06, "loss": 0.2249, "step": 6511 }, { "epoch": 0.89, "grad_norm": 4.267680959078023, "learning_rate": 6.157048776985568e-06, "loss": 0.2158, "step": 6512 }, { "epoch": 0.89, "grad_norm": 3.4054213961182063, "learning_rate": 6.1559771085089325e-06, "loss": 0.2154, "step": 6513 }, { "epoch": 0.89, "grad_norm": 4.1337745798057925, "learning_rate": 6.154905383926218e-06, "loss": 0.2511, "step": 6514 }, { "epoch": 0.89, "grad_norm": 3.2186374472696326, "learning_rate": 6.15383360328944e-06, "loss": 0.1825, "step": 6515 }, { "epoch": 0.89, "grad_norm": 3.976362982472996, "learning_rate": 6.152761766650616e-06, "loss": 0.2848, "step": 6516 }, { "epoch": 0.89, "grad_norm": 3.92191165543486, "learning_rate": 6.151689874061773e-06, "loss": 0.2182, "step": 6517 }, { "epoch": 0.89, "grad_norm": 3.5441828645627114, "learning_rate": 6.1506179255749335e-06, "loss": 0.1727, "step": 6518 }, { "epoch": 0.89, "grad_norm": 3.673993944489214, "learning_rate": 6.149545921242124e-06, "loss": 0.237, "step": 6519 }, { "epoch": 0.89, "grad_norm": 4.0192288761617005, "learning_rate": 6.148473861115377e-06, "loss": 0.26, "step": 6520 }, { "epoch": 0.89, "grad_norm": 3.6352068456626934, "learning_rate": 6.147401745246726e-06, "loss": 0.211, "step": 6521 }, { "epoch": 0.89, "grad_norm": 3.3431773538741094, "learning_rate": 6.1463295736882045e-06, "loss": 0.2495, "step": 6522 }, { "epoch": 0.89, "grad_norm": 3.4054736602989926, "learning_rate": 6.145257346491853e-06, "loss": 0.2314, "step": 6523 }, { "epoch": 0.89, "grad_norm": 4.1783726095480445, "learning_rate": 6.14418506370971e-06, "loss": 0.1943, "step": 6524 }, { "epoch": 0.89, "grad_norm": 8.234041966905686, "learning_rate": 6.143112725393823e-06, "loss": 0.2813, "step": 6525 }, { "epoch": 0.89, "grad_norm": 4.489901409566596, "learning_rate": 6.142040331596236e-06, "loss": 0.2745, "step": 6526 }, { "epoch": 0.89, "grad_norm": 3.770213770918874, "learning_rate": 6.140967882369001e-06, "loss": 0.2296, "step": 6527 }, { "epoch": 0.89, "grad_norm": 3.6582911473174313, "learning_rate": 6.139895377764167e-06, "loss": 0.3016, "step": 6528 }, { "epoch": 0.89, "grad_norm": 3.9216800770075575, "learning_rate": 6.138822817833791e-06, "loss": 0.2289, "step": 6529 }, { "epoch": 0.89, "grad_norm": 3.7200533337810606, "learning_rate": 6.1377502026299286e-06, "loss": 0.2327, "step": 6530 }, { "epoch": 0.89, "grad_norm": 3.533632703613175, "learning_rate": 6.136677532204639e-06, "loss": 0.2445, "step": 6531 }, { "epoch": 0.89, "grad_norm": 3.069518294167137, "learning_rate": 6.135604806609988e-06, "loss": 0.2192, "step": 6532 }, { "epoch": 0.89, "grad_norm": 3.357489546426807, "learning_rate": 6.13453202589804e-06, "loss": 0.198, "step": 6533 }, { "epoch": 0.89, "grad_norm": 3.283942168129679, "learning_rate": 6.1334591901208625e-06, "loss": 0.2082, "step": 6534 }, { "epoch": 0.89, "grad_norm": 3.0212607315400724, "learning_rate": 6.132386299330527e-06, "loss": 0.219, "step": 6535 }, { "epoch": 0.89, "grad_norm": 3.098935143711196, "learning_rate": 6.131313353579106e-06, "loss": 0.2288, "step": 6536 }, { "epoch": 0.89, "grad_norm": 3.0178475191121246, "learning_rate": 6.130240352918675e-06, "loss": 0.2074, "step": 6537 }, { "epoch": 0.89, "grad_norm": 3.7991362739361216, "learning_rate": 6.129167297401314e-06, "loss": 0.2672, "step": 6538 }, { "epoch": 0.89, "grad_norm": 3.911446957274481, "learning_rate": 6.1280941870791044e-06, "loss": 0.2652, "step": 6539 }, { "epoch": 0.89, "grad_norm": 3.167716000445224, "learning_rate": 6.12702102200413e-06, "loss": 0.2216, "step": 6540 }, { "epoch": 0.89, "grad_norm": 2.831240206000621, "learning_rate": 6.125947802228478e-06, "loss": 0.1998, "step": 6541 }, { "epoch": 0.89, "grad_norm": 3.405196665617814, "learning_rate": 6.1248745278042375e-06, "loss": 0.2028, "step": 6542 }, { "epoch": 0.89, "grad_norm": 3.210109506532337, "learning_rate": 6.1238011987835e-06, "loss": 0.2072, "step": 6543 }, { "epoch": 0.89, "grad_norm": 4.967674178626271, "learning_rate": 6.122727815218359e-06, "loss": 0.2654, "step": 6544 }, { "epoch": 0.89, "grad_norm": 3.4857180991272623, "learning_rate": 6.121654377160916e-06, "loss": 0.1661, "step": 6545 }, { "epoch": 0.89, "grad_norm": 3.821381147520317, "learning_rate": 6.1205808846632675e-06, "loss": 0.2283, "step": 6546 }, { "epoch": 0.89, "grad_norm": 3.6948398092798524, "learning_rate": 6.119507337777517e-06, "loss": 0.1992, "step": 6547 }, { "epoch": 0.89, "grad_norm": 3.967091206488266, "learning_rate": 6.1184337365557715e-06, "loss": 0.2308, "step": 6548 }, { "epoch": 0.89, "grad_norm": 3.6795161634919906, "learning_rate": 6.1173600810501355e-06, "loss": 0.2803, "step": 6549 }, { "epoch": 0.89, "grad_norm": 3.5512680869796034, "learning_rate": 6.116286371312722e-06, "loss": 0.2463, "step": 6550 }, { "epoch": 0.89, "grad_norm": 5.739228981505699, "learning_rate": 6.115212607395643e-06, "loss": 0.2361, "step": 6551 }, { "epoch": 0.89, "grad_norm": 3.3987617379801667, "learning_rate": 6.114138789351015e-06, "loss": 0.2283, "step": 6552 }, { "epoch": 0.89, "grad_norm": 4.2638811758585415, "learning_rate": 6.113064917230957e-06, "loss": 0.1705, "step": 6553 }, { "epoch": 0.89, "grad_norm": 2.966668969961645, "learning_rate": 6.111990991087588e-06, "loss": 0.2342, "step": 6554 }, { "epoch": 0.89, "grad_norm": 3.6204920638023026, "learning_rate": 6.110917010973034e-06, "loss": 0.2471, "step": 6555 }, { "epoch": 0.89, "grad_norm": 3.4016228541289193, "learning_rate": 6.109842976939421e-06, "loss": 0.1875, "step": 6556 }, { "epoch": 0.89, "grad_norm": 2.8698963723122763, "learning_rate": 6.108768889038875e-06, "loss": 0.1852, "step": 6557 }, { "epoch": 0.89, "grad_norm": 3.8398205498413094, "learning_rate": 6.1076947473235315e-06, "loss": 0.2286, "step": 6558 }, { "epoch": 0.89, "grad_norm": 3.3244241565116117, "learning_rate": 6.106620551845523e-06, "loss": 0.2846, "step": 6559 }, { "epoch": 0.89, "grad_norm": 3.3691640510482364, "learning_rate": 6.105546302656986e-06, "loss": 0.2094, "step": 6560 }, { "epoch": 0.89, "grad_norm": 2.998565766029202, "learning_rate": 6.10447199981006e-06, "loss": 0.215, "step": 6561 }, { "epoch": 0.89, "grad_norm": 3.8003316774852, "learning_rate": 6.103397643356888e-06, "loss": 0.2449, "step": 6562 }, { "epoch": 0.89, "grad_norm": 3.6460215568073457, "learning_rate": 6.102323233349612e-06, "loss": 0.2579, "step": 6563 }, { "epoch": 0.89, "grad_norm": 3.481456756099336, "learning_rate": 6.10124876984038e-06, "loss": 0.233, "step": 6564 }, { "epoch": 0.89, "grad_norm": 4.164168818872592, "learning_rate": 6.100174252881344e-06, "loss": 0.2223, "step": 6565 }, { "epoch": 0.89, "grad_norm": 4.825759486490017, "learning_rate": 6.099099682524655e-06, "loss": 0.2615, "step": 6566 }, { "epoch": 0.89, "grad_norm": 3.543110249905847, "learning_rate": 6.098025058822467e-06, "loss": 0.2133, "step": 6567 }, { "epoch": 0.89, "grad_norm": 4.052716123486945, "learning_rate": 6.09695038182694e-06, "loss": 0.2784, "step": 6568 }, { "epoch": 0.89, "grad_norm": 2.848799711114001, "learning_rate": 6.0958756515902314e-06, "loss": 0.2248, "step": 6569 }, { "epoch": 0.89, "grad_norm": 3.0421270843256996, "learning_rate": 6.094800868164505e-06, "loss": 0.214, "step": 6570 }, { "epoch": 0.89, "grad_norm": 4.135842581324452, "learning_rate": 6.093726031601926e-06, "loss": 0.261, "step": 6571 }, { "epoch": 0.89, "grad_norm": 2.71115066534489, "learning_rate": 6.092651141954663e-06, "loss": 0.165, "step": 6572 }, { "epoch": 0.89, "grad_norm": 2.9681057864641143, "learning_rate": 6.091576199274886e-06, "loss": 0.1707, "step": 6573 }, { "epoch": 0.89, "grad_norm": 6.279033595635916, "learning_rate": 6.090501203614767e-06, "loss": 0.2265, "step": 6574 }, { "epoch": 0.89, "grad_norm": 3.6129063951761817, "learning_rate": 6.089426155026483e-06, "loss": 0.2102, "step": 6575 }, { "epoch": 0.89, "grad_norm": 3.1835835738264175, "learning_rate": 6.0883510535622116e-06, "loss": 0.2396, "step": 6576 }, { "epoch": 0.89, "grad_norm": 3.1926490399425074, "learning_rate": 6.087275899274132e-06, "loss": 0.2593, "step": 6577 }, { "epoch": 0.89, "grad_norm": 4.454858095207467, "learning_rate": 6.086200692214431e-06, "loss": 0.3, "step": 6578 }, { "epoch": 0.89, "grad_norm": 3.2802307545552885, "learning_rate": 6.085125432435291e-06, "loss": 0.168, "step": 6579 }, { "epoch": 0.9, "grad_norm": 3.9272863340201356, "learning_rate": 6.084050119988905e-06, "loss": 0.2153, "step": 6580 }, { "epoch": 0.9, "grad_norm": 4.6198173290202424, "learning_rate": 6.082974754927458e-06, "loss": 0.2525, "step": 6581 }, { "epoch": 0.9, "grad_norm": 3.876869722633778, "learning_rate": 6.081899337303148e-06, "loss": 0.2298, "step": 6582 }, { "epoch": 0.9, "grad_norm": 2.884097851600057, "learning_rate": 6.080823867168169e-06, "loss": 0.2435, "step": 6583 }, { "epoch": 0.9, "grad_norm": 3.661868686980185, "learning_rate": 6.079748344574717e-06, "loss": 0.208, "step": 6584 }, { "epoch": 0.9, "grad_norm": 2.7927727463192706, "learning_rate": 6.078672769575e-06, "loss": 0.1832, "step": 6585 }, { "epoch": 0.9, "grad_norm": 3.620381830148462, "learning_rate": 6.077597142221216e-06, "loss": 0.25, "step": 6586 }, { "epoch": 0.9, "grad_norm": 3.1833265078199076, "learning_rate": 6.076521462565575e-06, "loss": 0.2333, "step": 6587 }, { "epoch": 0.9, "grad_norm": 3.9233434422556948, "learning_rate": 6.075445730660283e-06, "loss": 0.2156, "step": 6588 }, { "epoch": 0.9, "grad_norm": 2.9604789744276077, "learning_rate": 6.074369946557553e-06, "loss": 0.2012, "step": 6589 }, { "epoch": 0.9, "grad_norm": 3.4253711448234068, "learning_rate": 6.073294110309599e-06, "loss": 0.252, "step": 6590 }, { "epoch": 0.9, "grad_norm": 3.0909621839984656, "learning_rate": 6.0722182219686355e-06, "loss": 0.215, "step": 6591 }, { "epoch": 0.9, "grad_norm": 3.1901618833027054, "learning_rate": 6.071142281586883e-06, "loss": 0.2317, "step": 6592 }, { "epoch": 0.9, "grad_norm": 3.474890021268909, "learning_rate": 6.070066289216561e-06, "loss": 0.2446, "step": 6593 }, { "epoch": 0.9, "grad_norm": 4.164021119024509, "learning_rate": 6.068990244909896e-06, "loss": 0.2528, "step": 6594 }, { "epoch": 0.9, "grad_norm": 3.336082969974824, "learning_rate": 6.067914148719114e-06, "loss": 0.2133, "step": 6595 }, { "epoch": 0.9, "grad_norm": 3.7480993345217364, "learning_rate": 6.066838000696442e-06, "loss": 0.2345, "step": 6596 }, { "epoch": 0.9, "grad_norm": 3.174742624600892, "learning_rate": 6.0657618008941135e-06, "loss": 0.2331, "step": 6597 }, { "epoch": 0.9, "grad_norm": 3.279393676402045, "learning_rate": 6.064685549364363e-06, "loss": 0.1876, "step": 6598 }, { "epoch": 0.9, "grad_norm": 3.9921540002287337, "learning_rate": 6.063609246159426e-06, "loss": 0.2682, "step": 6599 }, { "epoch": 0.9, "grad_norm": 3.841123623660411, "learning_rate": 6.0625328913315405e-06, "loss": 0.2118, "step": 6600 }, { "epoch": 0.9, "grad_norm": 3.478582719875925, "learning_rate": 6.06145648493295e-06, "loss": 0.2645, "step": 6601 }, { "epoch": 0.9, "grad_norm": 3.644981538721234, "learning_rate": 6.060380027015897e-06, "loss": 0.2221, "step": 6602 }, { "epoch": 0.9, "grad_norm": 2.9837749676496745, "learning_rate": 6.059303517632629e-06, "loss": 0.2214, "step": 6603 }, { "epoch": 0.9, "grad_norm": 6.063059825274013, "learning_rate": 6.0582269568353945e-06, "loss": 0.2355, "step": 6604 }, { "epoch": 0.9, "grad_norm": 2.9323623927579705, "learning_rate": 6.057150344676447e-06, "loss": 0.1973, "step": 6605 }, { "epoch": 0.9, "grad_norm": 3.444729566111621, "learning_rate": 6.056073681208038e-06, "loss": 0.2019, "step": 6606 }, { "epoch": 0.9, "grad_norm": 3.24253225276716, "learning_rate": 6.054996966482425e-06, "loss": 0.2542, "step": 6607 }, { "epoch": 0.9, "grad_norm": 3.649813375733312, "learning_rate": 6.053920200551867e-06, "loss": 0.2168, "step": 6608 }, { "epoch": 0.9, "grad_norm": 3.741675996607745, "learning_rate": 6.052843383468626e-06, "loss": 0.2607, "step": 6609 }, { "epoch": 0.9, "grad_norm": 3.0434608933005367, "learning_rate": 6.051766515284965e-06, "loss": 0.1624, "step": 6610 }, { "epoch": 0.9, "grad_norm": 2.8941963187612076, "learning_rate": 6.050689596053151e-06, "loss": 0.229, "step": 6611 }, { "epoch": 0.9, "grad_norm": 3.742535298548074, "learning_rate": 6.049612625825454e-06, "loss": 0.2442, "step": 6612 }, { "epoch": 0.9, "grad_norm": 3.6620350862774376, "learning_rate": 6.048535604654143e-06, "loss": 0.2154, "step": 6613 }, { "epoch": 0.9, "grad_norm": 3.2225754335446464, "learning_rate": 6.047458532591495e-06, "loss": 0.202, "step": 6614 }, { "epoch": 0.9, "grad_norm": 3.735234939988591, "learning_rate": 6.046381409689783e-06, "loss": 0.234, "step": 6615 }, { "epoch": 0.9, "grad_norm": 3.927108354705711, "learning_rate": 6.045304236001289e-06, "loss": 0.2631, "step": 6616 }, { "epoch": 0.9, "grad_norm": 3.836165793614645, "learning_rate": 6.044227011578292e-06, "loss": 0.2541, "step": 6617 }, { "epoch": 0.9, "grad_norm": 4.184133199968599, "learning_rate": 6.043149736473076e-06, "loss": 0.2619, "step": 6618 }, { "epoch": 0.9, "grad_norm": 3.5630964185038567, "learning_rate": 6.042072410737929e-06, "loss": 0.2844, "step": 6619 }, { "epoch": 0.9, "grad_norm": 3.3339299368987767, "learning_rate": 6.040995034425139e-06, "loss": 0.2599, "step": 6620 }, { "epoch": 0.9, "grad_norm": 5.425714087602323, "learning_rate": 6.039917607586996e-06, "loss": 0.2744, "step": 6621 }, { "epoch": 0.9, "grad_norm": 3.271552342171359, "learning_rate": 6.038840130275795e-06, "loss": 0.2055, "step": 6622 }, { "epoch": 0.9, "grad_norm": 4.287378342585542, "learning_rate": 6.03776260254383e-06, "loss": 0.2752, "step": 6623 }, { "epoch": 0.9, "grad_norm": 3.2000938790912774, "learning_rate": 6.036685024443401e-06, "loss": 0.2236, "step": 6624 }, { "epoch": 0.9, "grad_norm": 3.7632006920871235, "learning_rate": 6.035607396026809e-06, "loss": 0.2199, "step": 6625 }, { "epoch": 0.9, "grad_norm": 3.2489690545345904, "learning_rate": 6.034529717346358e-06, "loss": 0.2234, "step": 6626 }, { "epoch": 0.9, "grad_norm": 3.3644804686581904, "learning_rate": 6.033451988454352e-06, "loss": 0.1829, "step": 6627 }, { "epoch": 0.9, "grad_norm": 4.208968017495093, "learning_rate": 6.0323742094031e-06, "loss": 0.2375, "step": 6628 }, { "epoch": 0.9, "grad_norm": 3.668828609868732, "learning_rate": 6.031296380244915e-06, "loss": 0.1813, "step": 6629 }, { "epoch": 0.9, "grad_norm": 3.6173092290603024, "learning_rate": 6.0302185010321055e-06, "loss": 0.1788, "step": 6630 }, { "epoch": 0.9, "grad_norm": 3.081630385297069, "learning_rate": 6.029140571816991e-06, "loss": 0.2377, "step": 6631 }, { "epoch": 0.9, "grad_norm": 4.0285711097432255, "learning_rate": 6.0280625926518865e-06, "loss": 0.2174, "step": 6632 }, { "epoch": 0.9, "grad_norm": 4.0432681519964, "learning_rate": 6.026984563589116e-06, "loss": 0.247, "step": 6633 }, { "epoch": 0.9, "grad_norm": 3.8714193794323593, "learning_rate": 6.025906484680999e-06, "loss": 0.1934, "step": 6634 }, { "epoch": 0.9, "grad_norm": 3.8873509369123664, "learning_rate": 6.024828355979861e-06, "loss": 0.2038, "step": 6635 }, { "epoch": 0.9, "grad_norm": 2.7809675246090935, "learning_rate": 6.023750177538032e-06, "loss": 0.1987, "step": 6636 }, { "epoch": 0.9, "grad_norm": 2.8620953916247007, "learning_rate": 6.02267194940784e-06, "loss": 0.1914, "step": 6637 }, { "epoch": 0.9, "grad_norm": 3.700825033025504, "learning_rate": 6.021593671641619e-06, "loss": 0.1839, "step": 6638 }, { "epoch": 0.9, "grad_norm": 3.6828109848921846, "learning_rate": 6.020515344291703e-06, "loss": 0.2153, "step": 6639 }, { "epoch": 0.9, "grad_norm": 4.659431269092203, "learning_rate": 6.019436967410429e-06, "loss": 0.2344, "step": 6640 }, { "epoch": 0.9, "grad_norm": 3.517113221815538, "learning_rate": 6.018358541050138e-06, "loss": 0.2596, "step": 6641 }, { "epoch": 0.9, "grad_norm": 4.191196892485941, "learning_rate": 6.0172800652631706e-06, "loss": 0.238, "step": 6642 }, { "epoch": 0.9, "grad_norm": 3.214689733502385, "learning_rate": 6.01620154010187e-06, "loss": 0.1554, "step": 6643 }, { "epoch": 0.9, "grad_norm": 3.5810358175202386, "learning_rate": 6.015122965618587e-06, "loss": 0.2203, "step": 6644 }, { "epoch": 0.9, "grad_norm": 3.4769990858972077, "learning_rate": 6.014044341865669e-06, "loss": 0.2035, "step": 6645 }, { "epoch": 0.9, "grad_norm": 3.916999972494209, "learning_rate": 6.012965668895467e-06, "loss": 0.1999, "step": 6646 }, { "epoch": 0.9, "grad_norm": 4.073215394281154, "learning_rate": 6.011886946760337e-06, "loss": 0.2755, "step": 6647 }, { "epoch": 0.9, "grad_norm": 3.8422127139623723, "learning_rate": 6.0108081755126335e-06, "loss": 0.2131, "step": 6648 }, { "epoch": 0.9, "grad_norm": 2.694086457222537, "learning_rate": 6.009729355204715e-06, "loss": 0.1817, "step": 6649 }, { "epoch": 0.9, "grad_norm": 3.3174919602456514, "learning_rate": 6.008650485888944e-06, "loss": 0.2536, "step": 6650 }, { "epoch": 0.9, "grad_norm": 4.870970058288082, "learning_rate": 6.007571567617684e-06, "loss": 0.2364, "step": 6651 }, { "epoch": 0.9, "grad_norm": 6.4838431620215715, "learning_rate": 6.006492600443301e-06, "loss": 0.2194, "step": 6652 }, { "epoch": 0.9, "grad_norm": 3.08783813679172, "learning_rate": 6.005413584418163e-06, "loss": 0.205, "step": 6653 }, { "epoch": 0.91, "grad_norm": 3.4238963301854235, "learning_rate": 6.0043345195946415e-06, "loss": 0.2472, "step": 6654 }, { "epoch": 0.91, "grad_norm": 4.163904669530158, "learning_rate": 6.003255406025108e-06, "loss": 0.2245, "step": 6655 }, { "epoch": 0.91, "grad_norm": 3.6540382796765654, "learning_rate": 6.002176243761939e-06, "loss": 0.2464, "step": 6656 }, { "epoch": 0.91, "grad_norm": 4.035902376882191, "learning_rate": 6.001097032857513e-06, "loss": 0.277, "step": 6657 }, { "epoch": 0.91, "grad_norm": 3.531206811244184, "learning_rate": 6.000017773364208e-06, "loss": 0.1968, "step": 6658 }, { "epoch": 0.91, "grad_norm": 4.250731662722503, "learning_rate": 5.99893846533441e-06, "loss": 0.2169, "step": 6659 }, { "epoch": 0.91, "grad_norm": 3.8052793320687983, "learning_rate": 5.997859108820501e-06, "loss": 0.2483, "step": 6660 }, { "epoch": 0.91, "grad_norm": 3.3789290848568476, "learning_rate": 5.996779703874869e-06, "loss": 0.2059, "step": 6661 }, { "epoch": 0.91, "grad_norm": 3.0742055215866846, "learning_rate": 5.995700250549903e-06, "loss": 0.2011, "step": 6662 }, { "epoch": 0.91, "grad_norm": 4.090392062972495, "learning_rate": 5.994620748897994e-06, "loss": 0.2244, "step": 6663 }, { "epoch": 0.91, "grad_norm": 4.760128913109692, "learning_rate": 5.993541198971542e-06, "loss": 0.206, "step": 6664 }, { "epoch": 0.91, "grad_norm": 3.718354809529005, "learning_rate": 5.992461600822938e-06, "loss": 0.2476, "step": 6665 }, { "epoch": 0.91, "grad_norm": 3.4452717931499093, "learning_rate": 5.991381954504583e-06, "loss": 0.2131, "step": 6666 }, { "epoch": 0.91, "grad_norm": 3.591647408640617, "learning_rate": 5.990302260068877e-06, "loss": 0.2086, "step": 6667 }, { "epoch": 0.91, "grad_norm": 4.282009392268352, "learning_rate": 5.989222517568225e-06, "loss": 0.2308, "step": 6668 }, { "epoch": 0.91, "grad_norm": 7.4756769371006655, "learning_rate": 5.988142727055032e-06, "loss": 0.1993, "step": 6669 }, { "epoch": 0.91, "grad_norm": 3.3525239695279723, "learning_rate": 5.9870628885817065e-06, "loss": 0.2241, "step": 6670 }, { "epoch": 0.91, "grad_norm": 4.71118680046654, "learning_rate": 5.985983002200661e-06, "loss": 0.1979, "step": 6671 }, { "epoch": 0.91, "grad_norm": 3.601817212936563, "learning_rate": 5.9849030679643075e-06, "loss": 0.2284, "step": 6672 }, { "epoch": 0.91, "grad_norm": 5.126605958903285, "learning_rate": 5.983823085925059e-06, "loss": 0.1916, "step": 6673 }, { "epoch": 0.91, "grad_norm": 3.4158879166160268, "learning_rate": 5.982743056135335e-06, "loss": 0.2172, "step": 6674 }, { "epoch": 0.91, "grad_norm": 3.6686617248630577, "learning_rate": 5.981662978647556e-06, "loss": 0.1654, "step": 6675 }, { "epoch": 0.91, "grad_norm": 3.9916164677904042, "learning_rate": 5.980582853514142e-06, "loss": 0.2417, "step": 6676 }, { "epoch": 0.91, "grad_norm": 6.286272917080506, "learning_rate": 5.97950268078752e-06, "loss": 0.1638, "step": 6677 }, { "epoch": 0.91, "grad_norm": 3.382013849745684, "learning_rate": 5.978422460520117e-06, "loss": 0.1833, "step": 6678 }, { "epoch": 0.91, "grad_norm": 3.3105625598707547, "learning_rate": 5.97734219276436e-06, "loss": 0.2672, "step": 6679 }, { "epoch": 0.91, "grad_norm": 3.2547526280961105, "learning_rate": 5.976261877572682e-06, "loss": 0.2515, "step": 6680 }, { "epoch": 0.91, "grad_norm": 3.4782256849093045, "learning_rate": 5.9751815149975164e-06, "loss": 0.2516, "step": 6681 }, { "epoch": 0.91, "grad_norm": 3.2508092035734037, "learning_rate": 5.9741011050913e-06, "loss": 0.216, "step": 6682 }, { "epoch": 0.91, "grad_norm": 2.984135188139394, "learning_rate": 5.973020647906469e-06, "loss": 0.1358, "step": 6683 }, { "epoch": 0.91, "grad_norm": 3.4644517447116496, "learning_rate": 5.971940143495466e-06, "loss": 0.1861, "step": 6684 }, { "epoch": 0.91, "grad_norm": 3.8659009723384656, "learning_rate": 5.9708595919107336e-06, "loss": 0.2665, "step": 6685 }, { "epoch": 0.91, "grad_norm": 3.3927620848000237, "learning_rate": 5.969778993204716e-06, "loss": 0.2294, "step": 6686 }, { "epoch": 0.91, "grad_norm": 3.58166865839124, "learning_rate": 5.968698347429864e-06, "loss": 0.231, "step": 6687 }, { "epoch": 0.91, "grad_norm": 4.4016014619843, "learning_rate": 5.967617654638624e-06, "loss": 0.2202, "step": 6688 }, { "epoch": 0.91, "grad_norm": 3.568563497925939, "learning_rate": 5.966536914883449e-06, "loss": 0.2509, "step": 6689 }, { "epoch": 0.91, "grad_norm": 3.4359465959298543, "learning_rate": 5.965456128216793e-06, "loss": 0.2759, "step": 6690 }, { "epoch": 0.91, "grad_norm": 3.2192263797569773, "learning_rate": 5.964375294691114e-06, "loss": 0.2214, "step": 6691 }, { "epoch": 0.91, "grad_norm": 2.846862930507162, "learning_rate": 5.96329441435887e-06, "loss": 0.1566, "step": 6692 }, { "epoch": 0.91, "grad_norm": 4.469377556531327, "learning_rate": 5.9622134872725224e-06, "loss": 0.2146, "step": 6693 }, { "epoch": 0.91, "grad_norm": 3.156873876238389, "learning_rate": 5.961132513484535e-06, "loss": 0.1833, "step": 6694 }, { "epoch": 0.91, "grad_norm": 3.2121693610803987, "learning_rate": 5.960051493047374e-06, "loss": 0.1884, "step": 6695 }, { "epoch": 0.91, "grad_norm": 3.691763507717969, "learning_rate": 5.958970426013504e-06, "loss": 0.2072, "step": 6696 }, { "epoch": 0.91, "grad_norm": 3.4244319296220658, "learning_rate": 5.9578893124354e-06, "loss": 0.1909, "step": 6697 }, { "epoch": 0.91, "grad_norm": 3.519276430418742, "learning_rate": 5.956808152365532e-06, "loss": 0.2461, "step": 6698 }, { "epoch": 0.91, "grad_norm": 3.055983855187936, "learning_rate": 5.955726945856376e-06, "loss": 0.1908, "step": 6699 }, { "epoch": 0.91, "grad_norm": 3.8643642951189277, "learning_rate": 5.954645692960408e-06, "loss": 0.2049, "step": 6700 }, { "epoch": 0.91, "grad_norm": 5.046519821776372, "learning_rate": 5.953564393730107e-06, "loss": 0.2895, "step": 6701 }, { "epoch": 0.91, "grad_norm": 3.359647972138292, "learning_rate": 5.9524830482179565e-06, "loss": 0.2193, "step": 6702 }, { "epoch": 0.91, "grad_norm": 2.585012995020418, "learning_rate": 5.951401656476437e-06, "loss": 0.1587, "step": 6703 }, { "epoch": 0.91, "grad_norm": 3.4585629137308547, "learning_rate": 5.950320218558037e-06, "loss": 0.2241, "step": 6704 }, { "epoch": 0.91, "grad_norm": 4.423368449382002, "learning_rate": 5.949238734515243e-06, "loss": 0.2482, "step": 6705 }, { "epoch": 0.91, "grad_norm": 3.840367176445236, "learning_rate": 5.9481572044005494e-06, "loss": 0.2406, "step": 6706 }, { "epoch": 0.91, "grad_norm": 5.532792200104694, "learning_rate": 5.9470756282664455e-06, "loss": 0.1755, "step": 6707 }, { "epoch": 0.91, "grad_norm": 2.711660060832868, "learning_rate": 5.945994006165426e-06, "loss": 0.1832, "step": 6708 }, { "epoch": 0.91, "grad_norm": 4.041401901187734, "learning_rate": 5.944912338149991e-06, "loss": 0.1873, "step": 6709 }, { "epoch": 0.91, "grad_norm": 8.119909426295566, "learning_rate": 5.943830624272635e-06, "loss": 0.2445, "step": 6710 }, { "epoch": 0.91, "grad_norm": 3.7472429961318334, "learning_rate": 5.9427488645858665e-06, "loss": 0.1622, "step": 6711 }, { "epoch": 0.91, "grad_norm": 3.1024241246606077, "learning_rate": 5.941667059142184e-06, "loss": 0.2338, "step": 6712 }, { "epoch": 0.91, "grad_norm": 4.899665305871125, "learning_rate": 5.940585207994095e-06, "loss": 0.2813, "step": 6713 }, { "epoch": 0.91, "grad_norm": 4.05011583996387, "learning_rate": 5.939503311194108e-06, "loss": 0.2122, "step": 6714 }, { "epoch": 0.91, "grad_norm": 3.4141656409881187, "learning_rate": 5.938421368794733e-06, "loss": 0.2411, "step": 6715 }, { "epoch": 0.91, "grad_norm": 3.3767559831974827, "learning_rate": 5.937339380848485e-06, "loss": 0.1998, "step": 6716 }, { "epoch": 0.91, "grad_norm": 4.187233448134121, "learning_rate": 5.936257347407877e-06, "loss": 0.19, "step": 6717 }, { "epoch": 0.91, "grad_norm": 3.0683019914406247, "learning_rate": 5.935175268525426e-06, "loss": 0.1921, "step": 6718 }, { "epoch": 0.91, "grad_norm": 4.242763896858309, "learning_rate": 5.934093144253652e-06, "loss": 0.1998, "step": 6719 }, { "epoch": 0.91, "grad_norm": 4.17950294072646, "learning_rate": 5.933010974645077e-06, "loss": 0.2276, "step": 6720 }, { "epoch": 0.91, "grad_norm": 3.4250806054309297, "learning_rate": 5.931928759752224e-06, "loss": 0.2433, "step": 6721 }, { "epoch": 0.91, "grad_norm": 3.260483451715874, "learning_rate": 5.9308464996276195e-06, "loss": 0.2497, "step": 6722 }, { "epoch": 0.91, "grad_norm": 3.9252968909719677, "learning_rate": 5.929764194323792e-06, "loss": 0.2892, "step": 6723 }, { "epoch": 0.91, "grad_norm": 3.754195079498557, "learning_rate": 5.92868184389327e-06, "loss": 0.2104, "step": 6724 }, { "epoch": 0.91, "grad_norm": 3.1594431482626684, "learning_rate": 5.927599448388588e-06, "loss": 0.2008, "step": 6725 }, { "epoch": 0.91, "grad_norm": 3.1245199051629906, "learning_rate": 5.92651700786228e-06, "loss": 0.1886, "step": 6726 }, { "epoch": 0.92, "grad_norm": 3.7924562598903626, "learning_rate": 5.925434522366884e-06, "loss": 0.292, "step": 6727 }, { "epoch": 0.92, "grad_norm": 3.3291126469053216, "learning_rate": 5.924351991954938e-06, "loss": 0.2142, "step": 6728 }, { "epoch": 0.92, "grad_norm": 4.105339080513771, "learning_rate": 5.923269416678981e-06, "loss": 0.1946, "step": 6729 }, { "epoch": 0.92, "grad_norm": 3.3456389489504743, "learning_rate": 5.9221867965915625e-06, "loss": 0.2045, "step": 6730 }, { "epoch": 0.92, "grad_norm": 3.22889379411696, "learning_rate": 5.921104131745224e-06, "loss": 0.2553, "step": 6731 }, { "epoch": 0.92, "grad_norm": 3.4931674331942073, "learning_rate": 5.920021422192512e-06, "loss": 0.268, "step": 6732 }, { "epoch": 0.92, "grad_norm": 3.6712675213374815, "learning_rate": 5.91893866798598e-06, "loss": 0.2478, "step": 6733 }, { "epoch": 0.92, "grad_norm": 3.9330877078363753, "learning_rate": 5.91785586917818e-06, "loss": 0.1983, "step": 6734 }, { "epoch": 0.92, "grad_norm": 3.242869909187944, "learning_rate": 5.916773025821662e-06, "loss": 0.221, "step": 6735 }, { "epoch": 0.92, "grad_norm": 4.344286039626464, "learning_rate": 5.9156901379689854e-06, "loss": 0.2492, "step": 6736 }, { "epoch": 0.92, "grad_norm": 2.720592451340553, "learning_rate": 5.914607205672711e-06, "loss": 0.1446, "step": 6737 }, { "epoch": 0.92, "grad_norm": 4.131154360520335, "learning_rate": 5.913524228985397e-06, "loss": 0.1754, "step": 6738 }, { "epoch": 0.92, "grad_norm": 6.893537636002985, "learning_rate": 5.9124412079596074e-06, "loss": 0.2198, "step": 6739 }, { "epoch": 0.92, "grad_norm": 3.9475906161736862, "learning_rate": 5.911358142647905e-06, "loss": 0.2116, "step": 6740 }, { "epoch": 0.92, "grad_norm": 4.28602276769936, "learning_rate": 5.91027503310286e-06, "loss": 0.2623, "step": 6741 }, { "epoch": 0.92, "grad_norm": 3.687777806247047, "learning_rate": 5.909191879377041e-06, "loss": 0.2555, "step": 6742 }, { "epoch": 0.92, "grad_norm": 3.7310549373271624, "learning_rate": 5.908108681523017e-06, "loss": 0.2383, "step": 6743 }, { "epoch": 0.92, "grad_norm": 3.8202266447440096, "learning_rate": 5.907025439593366e-06, "loss": 0.2438, "step": 6744 }, { "epoch": 0.92, "grad_norm": 3.847882468355467, "learning_rate": 5.9059421536406606e-06, "loss": 0.2304, "step": 6745 }, { "epoch": 0.92, "grad_norm": 2.86855588676171, "learning_rate": 5.904858823717481e-06, "loss": 0.1823, "step": 6746 }, { "epoch": 0.92, "grad_norm": 2.893544911354656, "learning_rate": 5.903775449876406e-06, "loss": 0.1949, "step": 6747 }, { "epoch": 0.92, "grad_norm": 2.99114916277195, "learning_rate": 5.902692032170017e-06, "loss": 0.2506, "step": 6748 }, { "epoch": 0.92, "grad_norm": 3.667799654089412, "learning_rate": 5.9016085706509e-06, "loss": 0.2571, "step": 6749 }, { "epoch": 0.92, "grad_norm": 3.389086079234657, "learning_rate": 5.900525065371639e-06, "loss": 0.1605, "step": 6750 }, { "epoch": 0.92, "grad_norm": 4.1775387081139685, "learning_rate": 5.8994415163848274e-06, "loss": 0.2629, "step": 6751 }, { "epoch": 0.92, "grad_norm": 4.07627134126584, "learning_rate": 5.898357923743052e-06, "loss": 0.2187, "step": 6752 }, { "epoch": 0.92, "grad_norm": 2.971162688053361, "learning_rate": 5.8972742874989065e-06, "loss": 0.2056, "step": 6753 }, { "epoch": 0.92, "grad_norm": 5.755943658871684, "learning_rate": 5.896190607704985e-06, "loss": 0.2338, "step": 6754 }, { "epoch": 0.92, "grad_norm": 3.679559306317724, "learning_rate": 5.895106884413887e-06, "loss": 0.2663, "step": 6755 }, { "epoch": 0.92, "grad_norm": 2.634893158296822, "learning_rate": 5.8940231176782105e-06, "loss": 0.1743, "step": 6756 }, { "epoch": 0.92, "grad_norm": 4.009412932402383, "learning_rate": 5.892939307550556e-06, "loss": 0.1958, "step": 6757 }, { "epoch": 0.92, "grad_norm": 4.002940296541158, "learning_rate": 5.8918554540835275e-06, "loss": 0.2237, "step": 6758 }, { "epoch": 0.92, "grad_norm": 3.998698490413843, "learning_rate": 5.890771557329731e-06, "loss": 0.2385, "step": 6759 }, { "epoch": 0.92, "grad_norm": 4.088492434694145, "learning_rate": 5.889687617341773e-06, "loss": 0.2506, "step": 6760 }, { "epoch": 0.92, "grad_norm": 3.0326644847407755, "learning_rate": 5.888603634172264e-06, "loss": 0.1859, "step": 6761 }, { "epoch": 0.92, "grad_norm": 3.1798386097625713, "learning_rate": 5.887519607873815e-06, "loss": 0.1849, "step": 6762 }, { "epoch": 0.92, "grad_norm": 4.104966174383422, "learning_rate": 5.886435538499041e-06, "loss": 0.2292, "step": 6763 }, { "epoch": 0.92, "grad_norm": 3.79030569486011, "learning_rate": 5.885351426100558e-06, "loss": 0.2437, "step": 6764 }, { "epoch": 0.92, "grad_norm": 3.414528346384765, "learning_rate": 5.884267270730983e-06, "loss": 0.1516, "step": 6765 }, { "epoch": 0.92, "grad_norm": 3.3554784374559774, "learning_rate": 5.883183072442938e-06, "loss": 0.1655, "step": 6766 }, { "epoch": 0.92, "grad_norm": 3.6192922070253237, "learning_rate": 5.882098831289044e-06, "loss": 0.23, "step": 6767 }, { "epoch": 0.92, "grad_norm": 4.062503234395206, "learning_rate": 5.881014547321924e-06, "loss": 0.25, "step": 6768 }, { "epoch": 0.92, "grad_norm": 4.032205366103847, "learning_rate": 5.879930220594206e-06, "loss": 0.21, "step": 6769 }, { "epoch": 0.92, "grad_norm": 3.2318788124441276, "learning_rate": 5.878845851158519e-06, "loss": 0.245, "step": 6770 }, { "epoch": 0.92, "grad_norm": 3.4684089217683036, "learning_rate": 5.877761439067493e-06, "loss": 0.2179, "step": 6771 }, { "epoch": 0.92, "grad_norm": 3.4383986068592485, "learning_rate": 5.8766769843737604e-06, "loss": 0.2549, "step": 6772 }, { "epoch": 0.92, "grad_norm": 3.4919095688640875, "learning_rate": 5.875592487129955e-06, "loss": 0.2356, "step": 6773 }, { "epoch": 0.92, "grad_norm": 3.88228398508812, "learning_rate": 5.874507947388716e-06, "loss": 0.1849, "step": 6774 }, { "epoch": 0.92, "grad_norm": 3.571541015594942, "learning_rate": 5.873423365202679e-06, "loss": 0.2018, "step": 6775 }, { "epoch": 0.92, "grad_norm": 3.5905603878976406, "learning_rate": 5.872338740624488e-06, "loss": 0.2146, "step": 6776 }, { "epoch": 0.92, "grad_norm": 3.6229264128553256, "learning_rate": 5.8712540737067835e-06, "loss": 0.2305, "step": 6777 }, { "epoch": 0.92, "grad_norm": 3.409634859083451, "learning_rate": 5.870169364502213e-06, "loss": 0.2414, "step": 6778 }, { "epoch": 0.92, "grad_norm": 3.50608131153957, "learning_rate": 5.869084613063422e-06, "loss": 0.191, "step": 6779 }, { "epoch": 0.92, "grad_norm": 3.352756781842669, "learning_rate": 5.867999819443059e-06, "loss": 0.1752, "step": 6780 }, { "epoch": 0.92, "grad_norm": 3.9081837992947364, "learning_rate": 5.866914983693775e-06, "loss": 0.3038, "step": 6781 }, { "epoch": 0.92, "grad_norm": 3.677212022445934, "learning_rate": 5.865830105868226e-06, "loss": 0.1944, "step": 6782 }, { "epoch": 0.92, "grad_norm": 3.657524794622864, "learning_rate": 5.864745186019063e-06, "loss": 0.2284, "step": 6783 }, { "epoch": 0.92, "grad_norm": 2.966060741838257, "learning_rate": 5.863660224198946e-06, "loss": 0.2026, "step": 6784 }, { "epoch": 0.92, "grad_norm": 3.3942122740725145, "learning_rate": 5.862575220460536e-06, "loss": 0.1988, "step": 6785 }, { "epoch": 0.92, "grad_norm": 4.861305699571759, "learning_rate": 5.86149017485649e-06, "loss": 0.2322, "step": 6786 }, { "epoch": 0.92, "grad_norm": 3.0589885579358724, "learning_rate": 5.860405087439475e-06, "loss": 0.237, "step": 6787 }, { "epoch": 0.92, "grad_norm": 3.5308386766133117, "learning_rate": 5.859319958262154e-06, "loss": 0.1946, "step": 6788 }, { "epoch": 0.92, "grad_norm": 4.267850739908004, "learning_rate": 5.858234787377195e-06, "loss": 0.2625, "step": 6789 }, { "epoch": 0.92, "grad_norm": 3.3556184243795535, "learning_rate": 5.857149574837269e-06, "loss": 0.2163, "step": 6790 }, { "epoch": 0.92, "grad_norm": 3.721572252278671, "learning_rate": 5.856064320695046e-06, "loss": 0.1778, "step": 6791 }, { "epoch": 0.92, "grad_norm": 3.581148713492773, "learning_rate": 5.8549790250032e-06, "loss": 0.2016, "step": 6792 }, { "epoch": 0.92, "grad_norm": 3.2401225605129813, "learning_rate": 5.853893687814407e-06, "loss": 0.2297, "step": 6793 }, { "epoch": 0.92, "grad_norm": 4.321573935788026, "learning_rate": 5.852808309181343e-06, "loss": 0.1973, "step": 6794 }, { "epoch": 0.92, "grad_norm": 3.188380841067934, "learning_rate": 5.851722889156687e-06, "loss": 0.1976, "step": 6795 }, { "epoch": 0.92, "grad_norm": 4.2723885440112594, "learning_rate": 5.850637427793123e-06, "loss": 0.271, "step": 6796 }, { "epoch": 0.92, "grad_norm": 3.9428780342210676, "learning_rate": 5.849551925143334e-06, "loss": 0.2439, "step": 6797 }, { "epoch": 0.92, "grad_norm": 3.2850411096978385, "learning_rate": 5.848466381260005e-06, "loss": 0.2014, "step": 6798 }, { "epoch": 0.92, "grad_norm": 3.964144967889648, "learning_rate": 5.847380796195824e-06, "loss": 0.2024, "step": 6799 }, { "epoch": 0.92, "grad_norm": 3.8869009956795924, "learning_rate": 5.84629517000348e-06, "loss": 0.1952, "step": 6800 }, { "epoch": 0.93, "grad_norm": 4.019335326952987, "learning_rate": 5.845209502735664e-06, "loss": 0.1998, "step": 6801 }, { "epoch": 0.93, "grad_norm": 4.622900778375476, "learning_rate": 5.84412379444507e-06, "loss": 0.2119, "step": 6802 }, { "epoch": 0.93, "grad_norm": 3.8129553549016784, "learning_rate": 5.8430380451843945e-06, "loss": 0.2133, "step": 6803 }, { "epoch": 0.93, "grad_norm": 4.073486553660503, "learning_rate": 5.841952255006334e-06, "loss": 0.2469, "step": 6804 }, { "epoch": 0.93, "grad_norm": 4.188435353241041, "learning_rate": 5.840866423963589e-06, "loss": 0.2148, "step": 6805 }, { "epoch": 0.93, "grad_norm": 4.070854714335739, "learning_rate": 5.83978055210886e-06, "loss": 0.2415, "step": 6806 }, { "epoch": 0.93, "grad_norm": 4.020326355999494, "learning_rate": 5.838694639494852e-06, "loss": 0.218, "step": 6807 }, { "epoch": 0.93, "grad_norm": 4.0200131614483725, "learning_rate": 5.837608686174266e-06, "loss": 0.203, "step": 6808 }, { "epoch": 0.93, "grad_norm": 4.612176773284736, "learning_rate": 5.836522692199816e-06, "loss": 0.2337, "step": 6809 }, { "epoch": 0.93, "grad_norm": 3.9769843571574066, "learning_rate": 5.835436657624207e-06, "loss": 0.2205, "step": 6810 }, { "epoch": 0.93, "grad_norm": 4.134021802587455, "learning_rate": 5.834350582500151e-06, "loss": 0.263, "step": 6811 }, { "epoch": 0.93, "grad_norm": 3.536114616382717, "learning_rate": 5.833264466880363e-06, "loss": 0.1938, "step": 6812 }, { "epoch": 0.93, "grad_norm": 3.335341434877661, "learning_rate": 5.832178310817556e-06, "loss": 0.2077, "step": 6813 }, { "epoch": 0.93, "grad_norm": 3.514741173953528, "learning_rate": 5.831092114364449e-06, "loss": 0.215, "step": 6814 }, { "epoch": 0.93, "grad_norm": 3.869392456210117, "learning_rate": 5.83000587757376e-06, "loss": 0.2159, "step": 6815 }, { "epoch": 0.93, "grad_norm": 3.7645177000418033, "learning_rate": 5.828919600498212e-06, "loss": 0.2449, "step": 6816 }, { "epoch": 0.93, "grad_norm": 3.410161852823566, "learning_rate": 5.827833283190527e-06, "loss": 0.2363, "step": 6817 }, { "epoch": 0.93, "grad_norm": 3.150371257910697, "learning_rate": 5.826746925703429e-06, "loss": 0.2263, "step": 6818 }, { "epoch": 0.93, "grad_norm": 4.100222242409089, "learning_rate": 5.825660528089647e-06, "loss": 0.1862, "step": 6819 }, { "epoch": 0.93, "grad_norm": 3.2751197208628287, "learning_rate": 5.824574090401909e-06, "loss": 0.2038, "step": 6820 }, { "epoch": 0.93, "grad_norm": 3.0639905380176184, "learning_rate": 5.823487612692946e-06, "loss": 0.1969, "step": 6821 }, { "epoch": 0.93, "grad_norm": 2.885441538746138, "learning_rate": 5.8224010950154895e-06, "loss": 0.1841, "step": 6822 }, { "epoch": 0.93, "grad_norm": 3.519045753157774, "learning_rate": 5.821314537422278e-06, "loss": 0.1884, "step": 6823 }, { "epoch": 0.93, "grad_norm": 3.449419400782974, "learning_rate": 5.820227939966044e-06, "loss": 0.2494, "step": 6824 }, { "epoch": 0.93, "grad_norm": 3.5660260229466294, "learning_rate": 5.819141302699531e-06, "loss": 0.2454, "step": 6825 }, { "epoch": 0.93, "grad_norm": 3.460098126665138, "learning_rate": 5.818054625675474e-06, "loss": 0.1871, "step": 6826 }, { "epoch": 0.93, "grad_norm": 3.2542558349761097, "learning_rate": 5.81696790894662e-06, "loss": 0.2115, "step": 6827 }, { "epoch": 0.93, "grad_norm": 4.192675403292818, "learning_rate": 5.815881152565712e-06, "loss": 0.2428, "step": 6828 }, { "epoch": 0.93, "grad_norm": 3.315587563089646, "learning_rate": 5.814794356585495e-06, "loss": 0.1884, "step": 6829 }, { "epoch": 0.93, "grad_norm": 3.629114714700988, "learning_rate": 5.813707521058719e-06, "loss": 0.213, "step": 6830 }, { "epoch": 0.93, "grad_norm": 3.708223466480359, "learning_rate": 5.812620646038137e-06, "loss": 0.2246, "step": 6831 }, { "epoch": 0.93, "grad_norm": 3.3495526700450613, "learning_rate": 5.811533731576494e-06, "loss": 0.1968, "step": 6832 }, { "epoch": 0.93, "grad_norm": 3.3115426174354856, "learning_rate": 5.810446777726549e-06, "loss": 0.1751, "step": 6833 }, { "epoch": 0.93, "grad_norm": 3.189743765662738, "learning_rate": 5.809359784541058e-06, "loss": 0.2918, "step": 6834 }, { "epoch": 0.93, "grad_norm": 3.094604023658766, "learning_rate": 5.808272752072776e-06, "loss": 0.1677, "step": 6835 }, { "epoch": 0.93, "grad_norm": 4.0223848824419575, "learning_rate": 5.807185680374467e-06, "loss": 0.2284, "step": 6836 }, { "epoch": 0.93, "grad_norm": 3.8596889807616925, "learning_rate": 5.806098569498892e-06, "loss": 0.1981, "step": 6837 }, { "epoch": 0.93, "grad_norm": 4.065124330571884, "learning_rate": 5.8050114194988104e-06, "loss": 0.2337, "step": 6838 }, { "epoch": 0.93, "grad_norm": 3.422125797905312, "learning_rate": 5.8039242304269915e-06, "loss": 0.2047, "step": 6839 }, { "epoch": 0.93, "grad_norm": 3.8374955654607255, "learning_rate": 5.802837002336202e-06, "loss": 0.2455, "step": 6840 }, { "epoch": 0.93, "grad_norm": 4.333052862967823, "learning_rate": 5.801749735279209e-06, "loss": 0.2264, "step": 6841 }, { "epoch": 0.93, "grad_norm": 3.79683986482354, "learning_rate": 5.800662429308787e-06, "loss": 0.2204, "step": 6842 }, { "epoch": 0.93, "grad_norm": 3.538842018248204, "learning_rate": 5.799575084477707e-06, "loss": 0.2089, "step": 6843 }, { "epoch": 0.93, "grad_norm": 4.168697571637044, "learning_rate": 5.798487700838747e-06, "loss": 0.1932, "step": 6844 }, { "epoch": 0.93, "grad_norm": 2.8825462584366996, "learning_rate": 5.797400278444678e-06, "loss": 0.1586, "step": 6845 }, { "epoch": 0.93, "grad_norm": 3.4856710171152803, "learning_rate": 5.796312817348284e-06, "loss": 0.2006, "step": 6846 }, { "epoch": 0.93, "grad_norm": 3.6367636978857982, "learning_rate": 5.795225317602344e-06, "loss": 0.1883, "step": 6847 }, { "epoch": 0.93, "grad_norm": 3.4218328107296947, "learning_rate": 5.79413777925964e-06, "loss": 0.2186, "step": 6848 }, { "epoch": 0.93, "grad_norm": 3.488541706564089, "learning_rate": 5.793050202372956e-06, "loss": 0.2252, "step": 6849 }, { "epoch": 0.93, "grad_norm": 2.92975903292416, "learning_rate": 5.79196258699508e-06, "loss": 0.1829, "step": 6850 }, { "epoch": 0.93, "grad_norm": 3.6006369046354676, "learning_rate": 5.7908749331788e-06, "loss": 0.2273, "step": 6851 }, { "epoch": 0.93, "grad_norm": 3.321789847100784, "learning_rate": 5.789787240976903e-06, "loss": 0.2177, "step": 6852 }, { "epoch": 0.93, "grad_norm": 4.916853698733316, "learning_rate": 5.788699510442184e-06, "loss": 0.2194, "step": 6853 }, { "epoch": 0.93, "grad_norm": 3.187309158471487, "learning_rate": 5.787611741627435e-06, "loss": 0.2137, "step": 6854 }, { "epoch": 0.93, "grad_norm": 4.093335816254938, "learning_rate": 5.786523934585451e-06, "loss": 0.1917, "step": 6855 }, { "epoch": 0.93, "grad_norm": 3.4166457815492963, "learning_rate": 5.785436089369031e-06, "loss": 0.2145, "step": 6856 }, { "epoch": 0.93, "grad_norm": 3.3745639992646908, "learning_rate": 5.784348206030974e-06, "loss": 0.1957, "step": 6857 }, { "epoch": 0.93, "grad_norm": 3.831240272079237, "learning_rate": 5.783260284624082e-06, "loss": 0.2104, "step": 6858 }, { "epoch": 0.93, "grad_norm": 4.064298814490961, "learning_rate": 5.782172325201155e-06, "loss": 0.2073, "step": 6859 }, { "epoch": 0.93, "grad_norm": 3.450661809499048, "learning_rate": 5.781084327815001e-06, "loss": 0.2278, "step": 6860 }, { "epoch": 0.93, "grad_norm": 3.353943137430165, "learning_rate": 5.779996292518424e-06, "loss": 0.2161, "step": 6861 }, { "epoch": 0.93, "grad_norm": 3.674353529613038, "learning_rate": 5.778908219364234e-06, "loss": 0.2128, "step": 6862 }, { "epoch": 0.93, "grad_norm": 4.530166950219238, "learning_rate": 5.777820108405241e-06, "loss": 0.1898, "step": 6863 }, { "epoch": 0.93, "grad_norm": 3.706205926842248, "learning_rate": 5.7767319596942575e-06, "loss": 0.2308, "step": 6864 }, { "epoch": 0.93, "grad_norm": 3.530591743956461, "learning_rate": 5.775643773284096e-06, "loss": 0.1694, "step": 6865 }, { "epoch": 0.93, "grad_norm": 3.221624073569427, "learning_rate": 5.774555549227575e-06, "loss": 0.1982, "step": 6866 }, { "epoch": 0.93, "grad_norm": 3.5457776085852855, "learning_rate": 5.77346728757751e-06, "loss": 0.227, "step": 6867 }, { "epoch": 0.93, "grad_norm": 3.5031872740993135, "learning_rate": 5.772378988386721e-06, "loss": 0.2211, "step": 6868 }, { "epoch": 0.93, "grad_norm": 3.2358520524680388, "learning_rate": 5.771290651708028e-06, "loss": 0.2102, "step": 6869 }, { "epoch": 0.93, "grad_norm": 3.279652808714236, "learning_rate": 5.770202277594258e-06, "loss": 0.168, "step": 6870 }, { "epoch": 0.93, "grad_norm": 3.774121833918722, "learning_rate": 5.769113866098232e-06, "loss": 0.1721, "step": 6871 }, { "epoch": 0.93, "grad_norm": 4.317081353900552, "learning_rate": 5.768025417272779e-06, "loss": 0.2588, "step": 6872 }, { "epoch": 0.93, "grad_norm": 4.533871266561943, "learning_rate": 5.766936931170727e-06, "loss": 0.2342, "step": 6873 }, { "epoch": 0.94, "grad_norm": 3.8999166865660015, "learning_rate": 5.765848407844904e-06, "loss": 0.2358, "step": 6874 }, { "epoch": 0.94, "grad_norm": 4.974279674763923, "learning_rate": 5.764759847348145e-06, "loss": 0.1809, "step": 6875 }, { "epoch": 0.94, "grad_norm": 2.920802030621108, "learning_rate": 5.763671249733284e-06, "loss": 0.2402, "step": 6876 }, { "epoch": 0.94, "grad_norm": 3.438525836021489, "learning_rate": 5.762582615053155e-06, "loss": 0.2209, "step": 6877 }, { "epoch": 0.94, "grad_norm": 4.595337128055558, "learning_rate": 5.7614939433605976e-06, "loss": 0.2408, "step": 6878 }, { "epoch": 0.94, "grad_norm": 3.4958440763801732, "learning_rate": 5.760405234708449e-06, "loss": 0.2722, "step": 6879 }, { "epoch": 0.94, "grad_norm": 3.5277961305322876, "learning_rate": 5.759316489149552e-06, "loss": 0.28, "step": 6880 }, { "epoch": 0.94, "grad_norm": 3.6772049229883, "learning_rate": 5.758227706736747e-06, "loss": 0.1941, "step": 6881 }, { "epoch": 0.94, "grad_norm": 3.128856106201631, "learning_rate": 5.757138887522884e-06, "loss": 0.142, "step": 6882 }, { "epoch": 0.94, "grad_norm": 3.604467724157245, "learning_rate": 5.756050031560805e-06, "loss": 0.2866, "step": 6883 }, { "epoch": 0.94, "grad_norm": 4.087700069728421, "learning_rate": 5.75496113890336e-06, "loss": 0.2189, "step": 6884 }, { "epoch": 0.94, "grad_norm": 5.467044791116477, "learning_rate": 5.753872209603399e-06, "loss": 0.2282, "step": 6885 }, { "epoch": 0.94, "grad_norm": 3.679100868309521, "learning_rate": 5.752783243713772e-06, "loss": 0.2587, "step": 6886 }, { "epoch": 0.94, "grad_norm": 4.82406102876043, "learning_rate": 5.751694241287336e-06, "loss": 0.2349, "step": 6887 }, { "epoch": 0.94, "grad_norm": 3.189274249085616, "learning_rate": 5.750605202376942e-06, "loss": 0.1396, "step": 6888 }, { "epoch": 0.94, "grad_norm": 3.3929179792977497, "learning_rate": 5.749516127035451e-06, "loss": 0.2376, "step": 6889 }, { "epoch": 0.94, "grad_norm": 3.1478379642479135, "learning_rate": 5.7484270153157215e-06, "loss": 0.2057, "step": 6890 }, { "epoch": 0.94, "grad_norm": 3.7646954830814168, "learning_rate": 5.747337867270614e-06, "loss": 0.2445, "step": 6891 }, { "epoch": 0.94, "grad_norm": 3.38130946227022, "learning_rate": 5.7462486829529895e-06, "loss": 0.2029, "step": 6892 }, { "epoch": 0.94, "grad_norm": 3.7868840861791657, "learning_rate": 5.745159462415714e-06, "loss": 0.2178, "step": 6893 }, { "epoch": 0.94, "grad_norm": 3.1712007404632763, "learning_rate": 5.7440702057116525e-06, "loss": 0.1982, "step": 6894 }, { "epoch": 0.94, "grad_norm": 4.299645757340226, "learning_rate": 5.742980912893672e-06, "loss": 0.2121, "step": 6895 }, { "epoch": 0.94, "grad_norm": 3.2978718925292276, "learning_rate": 5.7418915840146435e-06, "loss": 0.2193, "step": 6896 }, { "epoch": 0.94, "grad_norm": 3.6574207879269136, "learning_rate": 5.7408022191274385e-06, "loss": 0.2388, "step": 6897 }, { "epoch": 0.94, "grad_norm": 4.125938158836213, "learning_rate": 5.739712818284929e-06, "loss": 0.2151, "step": 6898 }, { "epoch": 0.94, "grad_norm": 4.086235266601118, "learning_rate": 5.738623381539991e-06, "loss": 0.1683, "step": 6899 }, { "epoch": 0.94, "grad_norm": 3.9106190395350438, "learning_rate": 5.737533908945498e-06, "loss": 0.2185, "step": 6900 }, { "epoch": 0.94, "grad_norm": 3.8005309951722035, "learning_rate": 5.736444400554332e-06, "loss": 0.2116, "step": 6901 }, { "epoch": 0.94, "grad_norm": 4.339263809831289, "learning_rate": 5.735354856419371e-06, "loss": 0.2246, "step": 6902 }, { "epoch": 0.94, "grad_norm": 3.702899528195779, "learning_rate": 5.734265276593496e-06, "loss": 0.2046, "step": 6903 }, { "epoch": 0.94, "grad_norm": 3.200924017688428, "learning_rate": 5.733175661129592e-06, "loss": 0.2238, "step": 6904 }, { "epoch": 0.94, "grad_norm": 3.5976186879823744, "learning_rate": 5.732086010080544e-06, "loss": 0.1919, "step": 6905 }, { "epoch": 0.94, "grad_norm": 3.570724627908074, "learning_rate": 5.730996323499238e-06, "loss": 0.2411, "step": 6906 }, { "epoch": 0.94, "grad_norm": 4.061483011154527, "learning_rate": 5.729906601438564e-06, "loss": 0.2659, "step": 6907 }, { "epoch": 0.94, "grad_norm": 3.923165388241353, "learning_rate": 5.728816843951411e-06, "loss": 0.2366, "step": 6908 }, { "epoch": 0.94, "grad_norm": 4.489565523745927, "learning_rate": 5.727727051090672e-06, "loss": 0.2175, "step": 6909 }, { "epoch": 0.94, "grad_norm": 4.0094228731501875, "learning_rate": 5.726637222909239e-06, "loss": 0.1832, "step": 6910 }, { "epoch": 0.94, "grad_norm": 4.1575572341047105, "learning_rate": 5.725547359460011e-06, "loss": 0.2082, "step": 6911 }, { "epoch": 0.94, "grad_norm": 3.9496232569364556, "learning_rate": 5.724457460795883e-06, "loss": 0.2293, "step": 6912 }, { "epoch": 0.94, "grad_norm": 4.496561619384507, "learning_rate": 5.723367526969753e-06, "loss": 0.2327, "step": 6913 }, { "epoch": 0.94, "grad_norm": 3.975961405673502, "learning_rate": 5.722277558034521e-06, "loss": 0.1898, "step": 6914 }, { "epoch": 0.94, "grad_norm": 3.367134356061725, "learning_rate": 5.721187554043093e-06, "loss": 0.2139, "step": 6915 }, { "epoch": 0.94, "grad_norm": 3.796770085483675, "learning_rate": 5.720097515048371e-06, "loss": 0.2489, "step": 6916 }, { "epoch": 0.94, "grad_norm": 3.5495166245581244, "learning_rate": 5.71900744110326e-06, "loss": 0.2026, "step": 6917 }, { "epoch": 0.94, "grad_norm": 4.014327716545353, "learning_rate": 5.717917332260668e-06, "loss": 0.2083, "step": 6918 }, { "epoch": 0.94, "grad_norm": 3.3647987639131394, "learning_rate": 5.716827188573506e-06, "loss": 0.173, "step": 6919 }, { "epoch": 0.94, "grad_norm": 3.5021183696640925, "learning_rate": 5.715737010094682e-06, "loss": 0.2015, "step": 6920 }, { "epoch": 0.94, "grad_norm": 3.507399007825457, "learning_rate": 5.714646796877108e-06, "loss": 0.2528, "step": 6921 }, { "epoch": 0.94, "grad_norm": 4.5074573960065285, "learning_rate": 5.713556548973701e-06, "loss": 0.2258, "step": 6922 }, { "epoch": 0.94, "grad_norm": 3.785149646442372, "learning_rate": 5.712466266437377e-06, "loss": 0.2117, "step": 6923 }, { "epoch": 0.94, "grad_norm": 3.876640733095144, "learning_rate": 5.711375949321051e-06, "loss": 0.2206, "step": 6924 }, { "epoch": 0.94, "grad_norm": 3.627210239635762, "learning_rate": 5.710285597677643e-06, "loss": 0.1869, "step": 6925 }, { "epoch": 0.94, "grad_norm": 4.833246490983718, "learning_rate": 5.709195211560074e-06, "loss": 0.235, "step": 6926 }, { "epoch": 0.94, "grad_norm": 3.1353172282798467, "learning_rate": 5.708104791021267e-06, "loss": 0.234, "step": 6927 }, { "epoch": 0.94, "grad_norm": 3.953275156381884, "learning_rate": 5.707014336114147e-06, "loss": 0.1871, "step": 6928 }, { "epoch": 0.94, "grad_norm": 3.8703086907030984, "learning_rate": 5.7059238468916404e-06, "loss": 0.2487, "step": 6929 }, { "epoch": 0.94, "grad_norm": 3.785534843790175, "learning_rate": 5.704833323406671e-06, "loss": 0.1895, "step": 6930 }, { "epoch": 0.94, "grad_norm": 3.686714201367293, "learning_rate": 5.703742765712173e-06, "loss": 0.2253, "step": 6931 }, { "epoch": 0.94, "grad_norm": 3.565469795395229, "learning_rate": 5.702652173861073e-06, "loss": 0.1997, "step": 6932 }, { "epoch": 0.94, "grad_norm": 3.371436343585872, "learning_rate": 5.701561547906306e-06, "loss": 0.2063, "step": 6933 }, { "epoch": 0.94, "grad_norm": 3.682482433619368, "learning_rate": 5.700470887900804e-06, "loss": 0.2367, "step": 6934 }, { "epoch": 0.94, "grad_norm": 5.903854666370239, "learning_rate": 5.699380193897506e-06, "loss": 0.1993, "step": 6935 }, { "epoch": 0.94, "grad_norm": 8.483734325122953, "learning_rate": 5.698289465949348e-06, "loss": 0.2381, "step": 6936 }, { "epoch": 0.94, "grad_norm": 3.185497601936989, "learning_rate": 5.697198704109269e-06, "loss": 0.2105, "step": 6937 }, { "epoch": 0.94, "grad_norm": 4.030008795270314, "learning_rate": 5.6961079084302104e-06, "loss": 0.2282, "step": 6938 }, { "epoch": 0.94, "grad_norm": 3.9350188907713233, "learning_rate": 5.695017078965114e-06, "loss": 0.2311, "step": 6939 }, { "epoch": 0.94, "grad_norm": 3.644758903313459, "learning_rate": 5.693926215766924e-06, "loss": 0.1538, "step": 6940 }, { "epoch": 0.94, "grad_norm": 4.376089128832222, "learning_rate": 5.692835318888585e-06, "loss": 0.2225, "step": 6941 }, { "epoch": 0.94, "grad_norm": 4.023462657870995, "learning_rate": 5.691744388383047e-06, "loss": 0.243, "step": 6942 }, { "epoch": 0.94, "grad_norm": 3.6116960762863384, "learning_rate": 5.690653424303258e-06, "loss": 0.1864, "step": 6943 }, { "epoch": 0.94, "grad_norm": 4.672122898923783, "learning_rate": 5.689562426702169e-06, "loss": 0.2567, "step": 6944 }, { "epoch": 0.94, "grad_norm": 3.8133436657757827, "learning_rate": 5.68847139563273e-06, "loss": 0.2082, "step": 6945 }, { "epoch": 0.94, "grad_norm": 3.5408068152703924, "learning_rate": 5.687380331147897e-06, "loss": 0.2488, "step": 6946 }, { "epoch": 0.94, "grad_norm": 3.7113595463358457, "learning_rate": 5.686289233300625e-06, "loss": 0.1978, "step": 6947 }, { "epoch": 0.95, "grad_norm": 3.760837713226239, "learning_rate": 5.68519810214387e-06, "loss": 0.2128, "step": 6948 }, { "epoch": 0.95, "grad_norm": 3.673303984899044, "learning_rate": 5.684106937730593e-06, "loss": 0.1855, "step": 6949 }, { "epoch": 0.95, "grad_norm": 3.4215691759778255, "learning_rate": 5.6830157401137535e-06, "loss": 0.2814, "step": 6950 }, { "epoch": 0.95, "grad_norm": 4.233750391833272, "learning_rate": 5.681924509346312e-06, "loss": 0.2879, "step": 6951 }, { "epoch": 0.95, "grad_norm": 4.112093850287105, "learning_rate": 5.680833245481234e-06, "loss": 0.2235, "step": 6952 }, { "epoch": 0.95, "grad_norm": 3.1821371433203973, "learning_rate": 5.679741948571483e-06, "loss": 0.1686, "step": 6953 }, { "epoch": 0.95, "grad_norm": 4.042464991752967, "learning_rate": 5.678650618670027e-06, "loss": 0.2399, "step": 6954 }, { "epoch": 0.95, "grad_norm": 3.457957665774741, "learning_rate": 5.6775592558298345e-06, "loss": 0.1673, "step": 6955 }, { "epoch": 0.95, "grad_norm": 3.6031736428104186, "learning_rate": 5.676467860103875e-06, "loss": 0.2197, "step": 6956 }, { "epoch": 0.95, "grad_norm": 3.1317664832749994, "learning_rate": 5.6753764315451196e-06, "loss": 0.1506, "step": 6957 }, { "epoch": 0.95, "grad_norm": 5.09455811550276, "learning_rate": 5.6742849702065426e-06, "loss": 0.2472, "step": 6958 }, { "epoch": 0.95, "grad_norm": 4.173410942221551, "learning_rate": 5.673193476141119e-06, "loss": 0.2414, "step": 6959 }, { "epoch": 0.95, "grad_norm": 2.873222812185007, "learning_rate": 5.672101949401821e-06, "loss": 0.1846, "step": 6960 }, { "epoch": 0.95, "grad_norm": 3.122440761107316, "learning_rate": 5.671010390041633e-06, "loss": 0.2, "step": 6961 }, { "epoch": 0.95, "grad_norm": 4.76848284423737, "learning_rate": 5.669918798113531e-06, "loss": 0.2092, "step": 6962 }, { "epoch": 0.95, "grad_norm": 4.0690393771428, "learning_rate": 5.668827173670496e-06, "loss": 0.2171, "step": 6963 }, { "epoch": 0.95, "grad_norm": 2.927220104788602, "learning_rate": 5.66773551676551e-06, "loss": 0.1837, "step": 6964 }, { "epoch": 0.95, "grad_norm": 3.9291443082806055, "learning_rate": 5.66664382745156e-06, "loss": 0.1718, "step": 6965 }, { "epoch": 0.95, "grad_norm": 3.7890782067920976, "learning_rate": 5.66555210578163e-06, "loss": 0.2518, "step": 6966 }, { "epoch": 0.95, "grad_norm": 3.1965444316574523, "learning_rate": 5.664460351808706e-06, "loss": 0.2118, "step": 6967 }, { "epoch": 0.95, "grad_norm": 3.4677452630823193, "learning_rate": 5.663368565585781e-06, "loss": 0.2189, "step": 6968 }, { "epoch": 0.95, "grad_norm": 4.296753187234513, "learning_rate": 5.662276747165843e-06, "loss": 0.2305, "step": 6969 }, { "epoch": 0.95, "grad_norm": 3.156302207755229, "learning_rate": 5.6611848966018825e-06, "loss": 0.1555, "step": 6970 }, { "epoch": 0.95, "grad_norm": 3.4214385925926964, "learning_rate": 5.660093013946896e-06, "loss": 0.2322, "step": 6971 }, { "epoch": 0.95, "grad_norm": 3.4912139058471103, "learning_rate": 5.659001099253878e-06, "loss": 0.2663, "step": 6972 }, { "epoch": 0.95, "grad_norm": 3.1778016208588955, "learning_rate": 5.657909152575825e-06, "loss": 0.1863, "step": 6973 }, { "epoch": 0.95, "grad_norm": 4.276070647327757, "learning_rate": 5.656817173965733e-06, "loss": 0.2384, "step": 6974 }, { "epoch": 0.95, "grad_norm": 4.593478351370709, "learning_rate": 5.655725163476605e-06, "loss": 0.1945, "step": 6975 }, { "epoch": 0.95, "grad_norm": 5.673658630009673, "learning_rate": 5.654633121161442e-06, "loss": 0.2478, "step": 6976 }, { "epoch": 0.95, "grad_norm": 3.311944898564154, "learning_rate": 5.653541047073248e-06, "loss": 0.1815, "step": 6977 }, { "epoch": 0.95, "grad_norm": 3.0700567964504573, "learning_rate": 5.6524489412650254e-06, "loss": 0.1651, "step": 6978 }, { "epoch": 0.95, "grad_norm": 4.035663952692029, "learning_rate": 5.651356803789779e-06, "loss": 0.2378, "step": 6979 }, { "epoch": 0.95, "grad_norm": 3.4968538520308776, "learning_rate": 5.650264634700519e-06, "loss": 0.2222, "step": 6980 }, { "epoch": 0.95, "grad_norm": 3.72462402301374, "learning_rate": 5.649172434050254e-06, "loss": 0.1965, "step": 6981 }, { "epoch": 0.95, "grad_norm": 4.2786248913387315, "learning_rate": 5.648080201891994e-06, "loss": 0.2279, "step": 6982 }, { "epoch": 0.95, "grad_norm": 4.331498051184058, "learning_rate": 5.646987938278753e-06, "loss": 0.2755, "step": 6983 }, { "epoch": 0.95, "grad_norm": 4.0131990957437464, "learning_rate": 5.6458956432635424e-06, "loss": 0.2359, "step": 6984 }, { "epoch": 0.95, "grad_norm": 3.630077902245068, "learning_rate": 5.644803316899377e-06, "loss": 0.2601, "step": 6985 }, { "epoch": 0.95, "grad_norm": 3.7013142669916843, "learning_rate": 5.643710959239277e-06, "loss": 0.2328, "step": 6986 }, { "epoch": 0.95, "grad_norm": 4.5398439076765476, "learning_rate": 5.642618570336258e-06, "loss": 0.2042, "step": 6987 }, { "epoch": 0.95, "grad_norm": 3.209895677920497, "learning_rate": 5.6415261502433385e-06, "loss": 0.2027, "step": 6988 }, { "epoch": 0.95, "grad_norm": 3.339808420001843, "learning_rate": 5.640433699013543e-06, "loss": 0.1977, "step": 6989 }, { "epoch": 0.95, "grad_norm": 3.5524369075815363, "learning_rate": 5.639341216699893e-06, "loss": 0.2191, "step": 6990 }, { "epoch": 0.95, "grad_norm": 2.6841728946389103, "learning_rate": 5.638248703355413e-06, "loss": 0.1952, "step": 6991 }, { "epoch": 0.95, "grad_norm": 3.693586828236783, "learning_rate": 5.637156159033129e-06, "loss": 0.2604, "step": 6992 }, { "epoch": 0.95, "grad_norm": 5.6363085774359565, "learning_rate": 5.636063583786067e-06, "loss": 0.2597, "step": 6993 }, { "epoch": 0.95, "grad_norm": 3.3908788504744627, "learning_rate": 5.6349709776672565e-06, "loss": 0.2144, "step": 6994 }, { "epoch": 0.95, "grad_norm": 3.150253268272893, "learning_rate": 5.633878340729728e-06, "loss": 0.201, "step": 6995 }, { "epoch": 0.95, "grad_norm": 2.9220589451006544, "learning_rate": 5.632785673026514e-06, "loss": 0.1911, "step": 6996 }, { "epoch": 0.95, "grad_norm": 3.0206212749941574, "learning_rate": 5.631692974610647e-06, "loss": 0.1745, "step": 6997 }, { "epoch": 0.95, "grad_norm": 6.154618986271628, "learning_rate": 5.630600245535163e-06, "loss": 0.245, "step": 6998 }, { "epoch": 0.95, "grad_norm": 3.2089011728061085, "learning_rate": 5.629507485853095e-06, "loss": 0.239, "step": 6999 }, { "epoch": 0.95, "grad_norm": 3.0511462727759078, "learning_rate": 5.6284146956174834e-06, "loss": 0.2614, "step": 7000 }, { "epoch": 0.95, "grad_norm": 3.488896193889658, "learning_rate": 5.627321874881368e-06, "loss": 0.2184, "step": 7001 }, { "epoch": 0.95, "grad_norm": 3.668010439014741, "learning_rate": 5.626229023697789e-06, "loss": 0.206, "step": 7002 }, { "epoch": 0.95, "grad_norm": 3.9173780054320817, "learning_rate": 5.625136142119788e-06, "loss": 0.2299, "step": 7003 }, { "epoch": 0.95, "grad_norm": 3.756187405744926, "learning_rate": 5.62404323020041e-06, "loss": 0.1819, "step": 7004 }, { "epoch": 0.95, "grad_norm": 3.5636677695058046, "learning_rate": 5.622950287992698e-06, "loss": 0.228, "step": 7005 }, { "epoch": 0.95, "grad_norm": 3.145995753938061, "learning_rate": 5.6218573155497004e-06, "loss": 0.175, "step": 7006 }, { "epoch": 0.95, "grad_norm": 3.481693654399858, "learning_rate": 5.6207643129244625e-06, "loss": 0.2073, "step": 7007 }, { "epoch": 0.95, "grad_norm": 3.45837047014158, "learning_rate": 5.619671280170038e-06, "loss": 0.1653, "step": 7008 }, { "epoch": 0.95, "grad_norm": 2.8748229210508587, "learning_rate": 5.618578217339476e-06, "loss": 0.1613, "step": 7009 }, { "epoch": 0.95, "grad_norm": 3.612637285866414, "learning_rate": 5.61748512448583e-06, "loss": 0.1969, "step": 7010 }, { "epoch": 0.95, "grad_norm": 3.9788049711790783, "learning_rate": 5.616392001662152e-06, "loss": 0.2153, "step": 7011 }, { "epoch": 0.95, "grad_norm": 3.562806844433076, "learning_rate": 5.6152988489214985e-06, "loss": 0.2214, "step": 7012 }, { "epoch": 0.95, "grad_norm": 4.087685536621118, "learning_rate": 5.614205666316926e-06, "loss": 0.229, "step": 7013 }, { "epoch": 0.95, "grad_norm": 4.686502969303487, "learning_rate": 5.613112453901493e-06, "loss": 0.2223, "step": 7014 }, { "epoch": 0.95, "grad_norm": 3.784145095895249, "learning_rate": 5.612019211728259e-06, "loss": 0.2098, "step": 7015 }, { "epoch": 0.95, "grad_norm": 3.858316930288659, "learning_rate": 5.6109259398502865e-06, "loss": 0.2456, "step": 7016 }, { "epoch": 0.95, "grad_norm": 3.459181990315122, "learning_rate": 5.609832638320637e-06, "loss": 0.2153, "step": 7017 }, { "epoch": 0.95, "grad_norm": 3.69318225734213, "learning_rate": 5.6087393071923745e-06, "loss": 0.2351, "step": 7018 }, { "epoch": 0.95, "grad_norm": 4.049517541995418, "learning_rate": 5.607645946518565e-06, "loss": 0.2515, "step": 7019 }, { "epoch": 0.95, "grad_norm": 3.5475460687460076, "learning_rate": 5.606552556352275e-06, "loss": 0.2429, "step": 7020 }, { "epoch": 0.96, "grad_norm": 3.617928917788265, "learning_rate": 5.605459136746573e-06, "loss": 0.2216, "step": 7021 }, { "epoch": 0.96, "grad_norm": 3.0120585150024364, "learning_rate": 5.604365687754528e-06, "loss": 0.191, "step": 7022 }, { "epoch": 0.96, "grad_norm": 4.0698582703535005, "learning_rate": 5.603272209429213e-06, "loss": 0.1754, "step": 7023 }, { "epoch": 0.96, "grad_norm": 3.037266360120911, "learning_rate": 5.602178701823701e-06, "loss": 0.1662, "step": 7024 }, { "epoch": 0.96, "grad_norm": 4.224832463590437, "learning_rate": 5.601085164991064e-06, "loss": 0.1789, "step": 7025 }, { "epoch": 0.96, "grad_norm": 2.8822329697554623, "learning_rate": 5.599991598984378e-06, "loss": 0.2348, "step": 7026 }, { "epoch": 0.96, "grad_norm": 3.205486611065246, "learning_rate": 5.59889800385672e-06, "loss": 0.1974, "step": 7027 }, { "epoch": 0.96, "grad_norm": 4.748772735018688, "learning_rate": 5.597804379661169e-06, "loss": 0.2633, "step": 7028 }, { "epoch": 0.96, "grad_norm": 3.429144027487984, "learning_rate": 5.596710726450805e-06, "loss": 0.2067, "step": 7029 }, { "epoch": 0.96, "grad_norm": 4.189394753639759, "learning_rate": 5.595617044278708e-06, "loss": 0.2311, "step": 7030 }, { "epoch": 0.96, "grad_norm": 3.992627505320246, "learning_rate": 5.594523333197962e-06, "loss": 0.2361, "step": 7031 }, { "epoch": 0.96, "grad_norm": 3.4572822269425645, "learning_rate": 5.59342959326165e-06, "loss": 0.2027, "step": 7032 }, { "epoch": 0.96, "grad_norm": 4.241218005019291, "learning_rate": 5.592335824522856e-06, "loss": 0.2444, "step": 7033 }, { "epoch": 0.96, "grad_norm": 3.689805314754987, "learning_rate": 5.591242027034669e-06, "loss": 0.2428, "step": 7034 }, { "epoch": 0.96, "grad_norm": 3.494207080636756, "learning_rate": 5.590148200850176e-06, "loss": 0.2595, "step": 7035 }, { "epoch": 0.96, "grad_norm": 4.277083581281413, "learning_rate": 5.589054346022468e-06, "loss": 0.2088, "step": 7036 }, { "epoch": 0.96, "grad_norm": 3.9506152963199703, "learning_rate": 5.587960462604634e-06, "loss": 0.218, "step": 7037 }, { "epoch": 0.96, "grad_norm": 3.8466709332373874, "learning_rate": 5.586866550649766e-06, "loss": 0.205, "step": 7038 }, { "epoch": 0.96, "grad_norm": 3.0462926185262313, "learning_rate": 5.58577261021096e-06, "loss": 0.182, "step": 7039 }, { "epoch": 0.96, "grad_norm": 4.167142130142875, "learning_rate": 5.584678641341309e-06, "loss": 0.22, "step": 7040 }, { "epoch": 0.96, "grad_norm": 4.009632011515552, "learning_rate": 5.583584644093911e-06, "loss": 0.1623, "step": 7041 }, { "epoch": 0.96, "grad_norm": 3.7415979340249956, "learning_rate": 5.582490618521864e-06, "loss": 0.2336, "step": 7042 }, { "epoch": 0.96, "grad_norm": 5.47241291106459, "learning_rate": 5.581396564678266e-06, "loss": 0.2321, "step": 7043 }, { "epoch": 0.96, "grad_norm": 3.5525982736456263, "learning_rate": 5.580302482616217e-06, "loss": 0.2483, "step": 7044 }, { "epoch": 0.96, "grad_norm": 4.4350572776053285, "learning_rate": 5.579208372388822e-06, "loss": 0.261, "step": 7045 }, { "epoch": 0.96, "grad_norm": 3.349440796327708, "learning_rate": 5.57811423404918e-06, "loss": 0.1845, "step": 7046 }, { "epoch": 0.96, "grad_norm": 4.376292825470864, "learning_rate": 5.5770200676504e-06, "loss": 0.2051, "step": 7047 }, { "epoch": 0.96, "grad_norm": 3.068530403599741, "learning_rate": 5.575925873245586e-06, "loss": 0.2369, "step": 7048 }, { "epoch": 0.96, "grad_norm": 3.367106764132734, "learning_rate": 5.5748316508878445e-06, "loss": 0.171, "step": 7049 }, { "epoch": 0.96, "grad_norm": 3.4956309833702424, "learning_rate": 5.5737374006302855e-06, "loss": 0.2343, "step": 7050 }, { "epoch": 0.96, "grad_norm": 3.904005214660078, "learning_rate": 5.57264312252602e-06, "loss": 0.2171, "step": 7051 }, { "epoch": 0.96, "grad_norm": 3.8550126906945184, "learning_rate": 5.571548816628159e-06, "loss": 0.2138, "step": 7052 }, { "epoch": 0.96, "grad_norm": 3.3545741166114107, "learning_rate": 5.570454482989811e-06, "loss": 0.2248, "step": 7053 }, { "epoch": 0.96, "grad_norm": 3.6364077676650397, "learning_rate": 5.569360121664097e-06, "loss": 0.2198, "step": 7054 }, { "epoch": 0.96, "grad_norm": 3.1804249295772253, "learning_rate": 5.568265732704129e-06, "loss": 0.224, "step": 7055 }, { "epoch": 0.96, "grad_norm": 3.4381610812432144, "learning_rate": 5.567171316163024e-06, "loss": 0.1952, "step": 7056 }, { "epoch": 0.96, "grad_norm": 3.89221345362529, "learning_rate": 5.5660768720939e-06, "loss": 0.2194, "step": 7057 }, { "epoch": 0.96, "grad_norm": 4.2751278526320515, "learning_rate": 5.564982400549878e-06, "loss": 0.2301, "step": 7058 }, { "epoch": 0.96, "grad_norm": 4.50744537053683, "learning_rate": 5.563887901584078e-06, "loss": 0.2287, "step": 7059 }, { "epoch": 0.96, "grad_norm": 3.5773330635910146, "learning_rate": 5.56279337524962e-06, "loss": 0.2117, "step": 7060 }, { "epoch": 0.96, "grad_norm": 2.909469420052198, "learning_rate": 5.561698821599632e-06, "loss": 0.1988, "step": 7061 }, { "epoch": 0.96, "grad_norm": 4.159269976772127, "learning_rate": 5.560604240687235e-06, "loss": 0.1918, "step": 7062 }, { "epoch": 0.96, "grad_norm": 3.9733092259352527, "learning_rate": 5.559509632565558e-06, "loss": 0.1923, "step": 7063 }, { "epoch": 0.96, "grad_norm": 4.132260305191379, "learning_rate": 5.558414997287727e-06, "loss": 0.2169, "step": 7064 }, { "epoch": 0.96, "grad_norm": 3.5412489278102988, "learning_rate": 5.55732033490687e-06, "loss": 0.2071, "step": 7065 }, { "epoch": 0.96, "grad_norm": 3.742944327112776, "learning_rate": 5.556225645476119e-06, "loss": 0.1945, "step": 7066 }, { "epoch": 0.96, "grad_norm": 4.386780264270831, "learning_rate": 5.555130929048603e-06, "loss": 0.2121, "step": 7067 }, { "epoch": 0.96, "grad_norm": 4.8722910476755255, "learning_rate": 5.554036185677459e-06, "loss": 0.2892, "step": 7068 }, { "epoch": 0.96, "grad_norm": 4.039137815286021, "learning_rate": 5.552941415415816e-06, "loss": 0.2148, "step": 7069 }, { "epoch": 0.96, "grad_norm": 3.04572252734137, "learning_rate": 5.551846618316812e-06, "loss": 0.1808, "step": 7070 }, { "epoch": 0.96, "grad_norm": 3.160163807005509, "learning_rate": 5.550751794433584e-06, "loss": 0.1887, "step": 7071 }, { "epoch": 0.96, "grad_norm": 3.2254501473208026, "learning_rate": 5.5496569438192695e-06, "loss": 0.2335, "step": 7072 }, { "epoch": 0.96, "grad_norm": 3.5504782885194914, "learning_rate": 5.548562066527008e-06, "loss": 0.2148, "step": 7073 }, { "epoch": 0.96, "grad_norm": 3.160847380475455, "learning_rate": 5.547467162609938e-06, "loss": 0.1918, "step": 7074 }, { "epoch": 0.96, "grad_norm": 3.656298448686997, "learning_rate": 5.546372232121205e-06, "loss": 0.217, "step": 7075 }, { "epoch": 0.96, "grad_norm": 3.770561489659454, "learning_rate": 5.5452772751139496e-06, "loss": 0.267, "step": 7076 }, { "epoch": 0.96, "grad_norm": 2.9211998893314792, "learning_rate": 5.544182291641317e-06, "loss": 0.2001, "step": 7077 }, { "epoch": 0.96, "grad_norm": 3.1593059535863097, "learning_rate": 5.5430872817564515e-06, "loss": 0.1962, "step": 7078 }, { "epoch": 0.96, "grad_norm": 3.4827095713421423, "learning_rate": 5.541992245512502e-06, "loss": 0.2241, "step": 7079 }, { "epoch": 0.96, "grad_norm": 4.045036754511347, "learning_rate": 5.5408971829626164e-06, "loss": 0.2128, "step": 7080 }, { "epoch": 0.96, "grad_norm": 2.868155860506081, "learning_rate": 5.539802094159944e-06, "loss": 0.2055, "step": 7081 }, { "epoch": 0.96, "grad_norm": 4.293442829836599, "learning_rate": 5.538706979157635e-06, "loss": 0.2252, "step": 7082 }, { "epoch": 0.96, "grad_norm": 4.121257156365731, "learning_rate": 5.537611838008843e-06, "loss": 0.2513, "step": 7083 }, { "epoch": 0.96, "grad_norm": 4.2017274953810855, "learning_rate": 5.536516670766721e-06, "loss": 0.2346, "step": 7084 }, { "epoch": 0.96, "grad_norm": 3.9532219782449474, "learning_rate": 5.535421477484422e-06, "loss": 0.2269, "step": 7085 }, { "epoch": 0.96, "grad_norm": 5.2053049683718315, "learning_rate": 5.534326258215101e-06, "loss": 0.1853, "step": 7086 }, { "epoch": 0.96, "grad_norm": 4.362399447702479, "learning_rate": 5.533231013011919e-06, "loss": 0.2271, "step": 7087 }, { "epoch": 0.96, "grad_norm": 5.026425196177224, "learning_rate": 5.532135741928034e-06, "loss": 0.2269, "step": 7088 }, { "epoch": 0.96, "grad_norm": 5.2426142435379885, "learning_rate": 5.531040445016604e-06, "loss": 0.2069, "step": 7089 }, { "epoch": 0.96, "grad_norm": 3.2954295997760426, "learning_rate": 5.529945122330789e-06, "loss": 0.2324, "step": 7090 }, { "epoch": 0.96, "grad_norm": 4.634981195439009, "learning_rate": 5.5288497739237525e-06, "loss": 0.2244, "step": 7091 }, { "epoch": 0.96, "grad_norm": 4.0707760802280974, "learning_rate": 5.527754399848657e-06, "loss": 0.205, "step": 7092 }, { "epoch": 0.96, "grad_norm": 3.417141869579415, "learning_rate": 5.526659000158668e-06, "loss": 0.1386, "step": 7093 }, { "epoch": 0.96, "grad_norm": 4.132154595764808, "learning_rate": 5.525563574906952e-06, "loss": 0.2451, "step": 7094 }, { "epoch": 0.97, "grad_norm": 3.687138914578974, "learning_rate": 5.524468124146677e-06, "loss": 0.2288, "step": 7095 }, { "epoch": 0.97, "grad_norm": 4.508646517330651, "learning_rate": 5.523372647931008e-06, "loss": 0.3289, "step": 7096 }, { "epoch": 0.97, "grad_norm": 4.7431531224708525, "learning_rate": 5.522277146313117e-06, "loss": 0.2223, "step": 7097 }, { "epoch": 0.97, "grad_norm": 3.4969268864476657, "learning_rate": 5.521181619346176e-06, "loss": 0.1966, "step": 7098 }, { "epoch": 0.97, "grad_norm": 3.5712627546564835, "learning_rate": 5.520086067083353e-06, "loss": 0.2099, "step": 7099 }, { "epoch": 0.97, "grad_norm": 4.27126894801677, "learning_rate": 5.518990489577824e-06, "loss": 0.2197, "step": 7100 }, { "epoch": 0.97, "grad_norm": 3.0638732333927696, "learning_rate": 5.5178948868827655e-06, "loss": 0.2273, "step": 7101 }, { "epoch": 0.97, "grad_norm": 4.411783437137981, "learning_rate": 5.51679925905135e-06, "loss": 0.2352, "step": 7102 }, { "epoch": 0.97, "grad_norm": 4.257620004142174, "learning_rate": 5.515703606136755e-06, "loss": 0.1906, "step": 7103 }, { "epoch": 0.97, "grad_norm": 2.967073394877385, "learning_rate": 5.514607928192161e-06, "loss": 0.2099, "step": 7104 }, { "epoch": 0.97, "grad_norm": 4.665590456873979, "learning_rate": 5.513512225270744e-06, "loss": 0.2294, "step": 7105 }, { "epoch": 0.97, "grad_norm": 3.6622281571910964, "learning_rate": 5.512416497425687e-06, "loss": 0.1892, "step": 7106 }, { "epoch": 0.97, "grad_norm": 3.3037478729367957, "learning_rate": 5.511320744710171e-06, "loss": 0.1671, "step": 7107 }, { "epoch": 0.97, "grad_norm": 3.6509656605992764, "learning_rate": 5.51022496717738e-06, "loss": 0.1804, "step": 7108 }, { "epoch": 0.97, "grad_norm": 3.7118098405880477, "learning_rate": 5.509129164880498e-06, "loss": 0.1556, "step": 7109 }, { "epoch": 0.97, "grad_norm": 3.3197640990975703, "learning_rate": 5.508033337872709e-06, "loss": 0.174, "step": 7110 }, { "epoch": 0.97, "grad_norm": 3.3736915877930427, "learning_rate": 5.506937486207201e-06, "loss": 0.2128, "step": 7111 }, { "epoch": 0.97, "grad_norm": 3.066588780575234, "learning_rate": 5.505841609937162e-06, "loss": 0.1847, "step": 7112 }, { "epoch": 0.97, "grad_norm": 4.069277274536981, "learning_rate": 5.50474570911578e-06, "loss": 0.1844, "step": 7113 }, { "epoch": 0.97, "grad_norm": 5.336700807189944, "learning_rate": 5.503649783796246e-06, "loss": 0.232, "step": 7114 }, { "epoch": 0.97, "grad_norm": 3.2895633292310134, "learning_rate": 5.502553834031751e-06, "loss": 0.2088, "step": 7115 }, { "epoch": 0.97, "grad_norm": 3.689359406493356, "learning_rate": 5.501457859875488e-06, "loss": 0.1681, "step": 7116 }, { "epoch": 0.97, "grad_norm": 3.838514148006987, "learning_rate": 5.500361861380651e-06, "loss": 0.1974, "step": 7117 }, { "epoch": 0.97, "grad_norm": 2.860524233761681, "learning_rate": 5.499265838600434e-06, "loss": 0.1571, "step": 7118 }, { "epoch": 0.97, "grad_norm": 3.78591209209985, "learning_rate": 5.498169791588033e-06, "loss": 0.2255, "step": 7119 }, { "epoch": 0.97, "grad_norm": 3.7009470161954052, "learning_rate": 5.4970737203966476e-06, "loss": 0.2208, "step": 7120 }, { "epoch": 0.97, "grad_norm": 3.966225496759399, "learning_rate": 5.495977625079475e-06, "loss": 0.1651, "step": 7121 }, { "epoch": 0.97, "grad_norm": 4.10576657835346, "learning_rate": 5.494881505689714e-06, "loss": 0.2627, "step": 7122 }, { "epoch": 0.97, "grad_norm": 4.0343112972803095, "learning_rate": 5.493785362280566e-06, "loss": 0.1983, "step": 7123 }, { "epoch": 0.97, "grad_norm": 3.921045292996157, "learning_rate": 5.492689194905234e-06, "loss": 0.1455, "step": 7124 }, { "epoch": 0.97, "grad_norm": 3.655441138438337, "learning_rate": 5.49159300361692e-06, "loss": 0.2317, "step": 7125 }, { "epoch": 0.97, "grad_norm": 3.715725686294216, "learning_rate": 5.490496788468828e-06, "loss": 0.2141, "step": 7126 }, { "epoch": 0.97, "grad_norm": 3.687951961804995, "learning_rate": 5.489400549514165e-06, "loss": 0.208, "step": 7127 }, { "epoch": 0.97, "grad_norm": 2.64346753504147, "learning_rate": 5.488304286806136e-06, "loss": 0.1924, "step": 7128 }, { "epoch": 0.97, "grad_norm": 4.111557278517589, "learning_rate": 5.48720800039795e-06, "loss": 0.1849, "step": 7129 }, { "epoch": 0.97, "grad_norm": 3.4576361731862595, "learning_rate": 5.486111690342816e-06, "loss": 0.2451, "step": 7130 }, { "epoch": 0.97, "grad_norm": 3.80094184290647, "learning_rate": 5.485015356693944e-06, "loss": 0.1756, "step": 7131 }, { "epoch": 0.97, "grad_norm": 3.2811662665926242, "learning_rate": 5.483918999504544e-06, "loss": 0.2602, "step": 7132 }, { "epoch": 0.97, "grad_norm": 4.3263877431451405, "learning_rate": 5.482822618827829e-06, "loss": 0.2364, "step": 7133 }, { "epoch": 0.97, "grad_norm": 3.5131714112956636, "learning_rate": 5.481726214717013e-06, "loss": 0.1967, "step": 7134 }, { "epoch": 0.97, "grad_norm": 4.032773600826826, "learning_rate": 5.480629787225312e-06, "loss": 0.1935, "step": 7135 }, { "epoch": 0.97, "grad_norm": 4.505152962507535, "learning_rate": 5.47953333640594e-06, "loss": 0.2598, "step": 7136 }, { "epoch": 0.97, "grad_norm": 4.370284834720705, "learning_rate": 5.478436862312113e-06, "loss": 0.229, "step": 7137 }, { "epoch": 0.97, "grad_norm": 4.30642956351738, "learning_rate": 5.477340364997051e-06, "loss": 0.2393, "step": 7138 }, { "epoch": 0.97, "grad_norm": 3.473017718714929, "learning_rate": 5.476243844513973e-06, "loss": 0.2161, "step": 7139 }, { "epoch": 0.97, "grad_norm": 3.591741538891302, "learning_rate": 5.475147300916097e-06, "loss": 0.1788, "step": 7140 }, { "epoch": 0.97, "grad_norm": 3.1029284890894275, "learning_rate": 5.4740507342566485e-06, "loss": 0.2322, "step": 7141 }, { "epoch": 0.97, "grad_norm": 3.4510673758408874, "learning_rate": 5.472954144588847e-06, "loss": 0.181, "step": 7142 }, { "epoch": 0.97, "grad_norm": 3.624224059180171, "learning_rate": 5.471857531965918e-06, "loss": 0.1942, "step": 7143 }, { "epoch": 0.97, "grad_norm": 3.857092321563762, "learning_rate": 5.470760896441084e-06, "loss": 0.2258, "step": 7144 }, { "epoch": 0.97, "grad_norm": 5.988054847834331, "learning_rate": 5.4696642380675725e-06, "loss": 0.1622, "step": 7145 }, { "epoch": 0.97, "grad_norm": 3.2289142224937284, "learning_rate": 5.468567556898609e-06, "loss": 0.2197, "step": 7146 }, { "epoch": 0.97, "grad_norm": 4.153532276674517, "learning_rate": 5.467470852987424e-06, "loss": 0.2021, "step": 7147 }, { "epoch": 0.97, "grad_norm": 3.8682573507674776, "learning_rate": 5.466374126387246e-06, "loss": 0.182, "step": 7148 }, { "epoch": 0.97, "grad_norm": 3.8950594281961672, "learning_rate": 5.465277377151304e-06, "loss": 0.2065, "step": 7149 }, { "epoch": 0.97, "grad_norm": 3.9516114761483925, "learning_rate": 5.4641806053328306e-06, "loss": 0.2039, "step": 7150 }, { "epoch": 0.97, "grad_norm": 3.9110550658379433, "learning_rate": 5.4630838109850584e-06, "loss": 0.2334, "step": 7151 }, { "epoch": 0.97, "grad_norm": 3.3885481807919784, "learning_rate": 5.4619869941612204e-06, "loss": 0.2328, "step": 7152 }, { "epoch": 0.97, "grad_norm": 3.035142613202944, "learning_rate": 5.460890154914551e-06, "loss": 0.1809, "step": 7153 }, { "epoch": 0.97, "grad_norm": 5.00070279283035, "learning_rate": 5.459793293298288e-06, "loss": 0.273, "step": 7154 }, { "epoch": 0.97, "grad_norm": 3.1701431690285653, "learning_rate": 5.458696409365665e-06, "loss": 0.1942, "step": 7155 }, { "epoch": 0.97, "grad_norm": 3.8049361687840255, "learning_rate": 5.457599503169922e-06, "loss": 0.2082, "step": 7156 }, { "epoch": 0.97, "grad_norm": 4.282918322094789, "learning_rate": 5.456502574764299e-06, "loss": 0.1883, "step": 7157 }, { "epoch": 0.97, "grad_norm": 3.733358451712411, "learning_rate": 5.455405624202032e-06, "loss": 0.2014, "step": 7158 }, { "epoch": 0.97, "grad_norm": 4.004373878163023, "learning_rate": 5.454308651536366e-06, "loss": 0.1701, "step": 7159 }, { "epoch": 0.97, "grad_norm": 4.078833510060361, "learning_rate": 5.453211656820544e-06, "loss": 0.2453, "step": 7160 }, { "epoch": 0.97, "grad_norm": 4.135795859362289, "learning_rate": 5.452114640107807e-06, "loss": 0.2277, "step": 7161 }, { "epoch": 0.97, "grad_norm": 3.7318404804212473, "learning_rate": 5.4510176014514e-06, "loss": 0.2074, "step": 7162 }, { "epoch": 0.97, "grad_norm": 4.257223343075828, "learning_rate": 5.449920540904568e-06, "loss": 0.2417, "step": 7163 }, { "epoch": 0.97, "grad_norm": 3.095377551551065, "learning_rate": 5.448823458520558e-06, "loss": 0.2216, "step": 7164 }, { "epoch": 0.97, "grad_norm": 4.309615300727246, "learning_rate": 5.447726354352618e-06, "loss": 0.2504, "step": 7165 }, { "epoch": 0.97, "grad_norm": 3.361031772678561, "learning_rate": 5.4466292284539955e-06, "loss": 0.2389, "step": 7166 }, { "epoch": 0.97, "grad_norm": 3.365023833981162, "learning_rate": 5.445532080877942e-06, "loss": 0.2368, "step": 7167 }, { "epoch": 0.98, "grad_norm": 3.4411886468348625, "learning_rate": 5.444434911677706e-06, "loss": 0.206, "step": 7168 }, { "epoch": 0.98, "grad_norm": 3.270415797361557, "learning_rate": 5.443337720906542e-06, "loss": 0.2396, "step": 7169 }, { "epoch": 0.98, "grad_norm": 4.258874146228967, "learning_rate": 5.442240508617701e-06, "loss": 0.2196, "step": 7170 }, { "epoch": 0.98, "grad_norm": 2.759375180083678, "learning_rate": 5.4411432748644365e-06, "loss": 0.1946, "step": 7171 }, { "epoch": 0.98, "grad_norm": 4.0220704291555665, "learning_rate": 5.440046019700004e-06, "loss": 0.1901, "step": 7172 }, { "epoch": 0.98, "grad_norm": 3.717264992867564, "learning_rate": 5.43894874317766e-06, "loss": 0.2211, "step": 7173 }, { "epoch": 0.98, "grad_norm": 3.4618897391884045, "learning_rate": 5.4378514453506615e-06, "loss": 0.1975, "step": 7174 }, { "epoch": 0.98, "grad_norm": 2.7921773391005, "learning_rate": 5.436754126272267e-06, "loss": 0.1404, "step": 7175 }, { "epoch": 0.98, "grad_norm": 3.0658557047824253, "learning_rate": 5.4356567859957335e-06, "loss": 0.1746, "step": 7176 }, { "epoch": 0.98, "grad_norm": 3.1120437684345585, "learning_rate": 5.434559424574323e-06, "loss": 0.1981, "step": 7177 }, { "epoch": 0.98, "grad_norm": 3.076556577459402, "learning_rate": 5.433462042061296e-06, "loss": 0.2, "step": 7178 }, { "epoch": 0.98, "grad_norm": 3.016486142940108, "learning_rate": 5.432364638509913e-06, "loss": 0.2117, "step": 7179 }, { "epoch": 0.98, "grad_norm": 3.9443834478545083, "learning_rate": 5.43126721397344e-06, "loss": 0.2063, "step": 7180 }, { "epoch": 0.98, "grad_norm": 4.182516879283463, "learning_rate": 5.430169768505141e-06, "loss": 0.2523, "step": 7181 }, { "epoch": 0.98, "grad_norm": 3.849862278217405, "learning_rate": 5.429072302158279e-06, "loss": 0.1771, "step": 7182 }, { "epoch": 0.98, "grad_norm": 4.694842361084979, "learning_rate": 5.427974814986123e-06, "loss": 0.22, "step": 7183 }, { "epoch": 0.98, "grad_norm": 3.249133282024179, "learning_rate": 5.426877307041938e-06, "loss": 0.2047, "step": 7184 }, { "epoch": 0.98, "grad_norm": 3.6414949012543287, "learning_rate": 5.425779778378992e-06, "loss": 0.1866, "step": 7185 }, { "epoch": 0.98, "grad_norm": 4.391030120343949, "learning_rate": 5.424682229050555e-06, "loss": 0.2027, "step": 7186 }, { "epoch": 0.98, "grad_norm": 3.7185242802713887, "learning_rate": 5.4235846591098995e-06, "loss": 0.2108, "step": 7187 }, { "epoch": 0.98, "grad_norm": 3.3506174682586827, "learning_rate": 5.422487068610293e-06, "loss": 0.2412, "step": 7188 }, { "epoch": 0.98, "grad_norm": 3.095168048660803, "learning_rate": 5.421389457605011e-06, "loss": 0.1505, "step": 7189 }, { "epoch": 0.98, "grad_norm": 4.53902248359907, "learning_rate": 5.420291826147324e-06, "loss": 0.1987, "step": 7190 }, { "epoch": 0.98, "grad_norm": 2.4965043695531337, "learning_rate": 5.419194174290508e-06, "loss": 0.1875, "step": 7191 }, { "epoch": 0.98, "grad_norm": 3.6233539753545574, "learning_rate": 5.4180965020878365e-06, "loss": 0.2218, "step": 7192 }, { "epoch": 0.98, "grad_norm": 3.6445743744066967, "learning_rate": 5.416998809592589e-06, "loss": 0.1767, "step": 7193 }, { "epoch": 0.98, "grad_norm": 3.311676417553703, "learning_rate": 5.415901096858038e-06, "loss": 0.1822, "step": 7194 }, { "epoch": 0.98, "grad_norm": 3.6701845671057804, "learning_rate": 5.414803363937466e-06, "loss": 0.2137, "step": 7195 }, { "epoch": 0.98, "grad_norm": 3.8459581276497485, "learning_rate": 5.41370561088415e-06, "loss": 0.2506, "step": 7196 }, { "epoch": 0.98, "grad_norm": 3.0271931551854534, "learning_rate": 5.41260783775137e-06, "loss": 0.1878, "step": 7197 }, { "epoch": 0.98, "grad_norm": 3.3814201491250486, "learning_rate": 5.4115100445924074e-06, "loss": 0.209, "step": 7198 }, { "epoch": 0.98, "grad_norm": 3.6512357033779534, "learning_rate": 5.410412231460545e-06, "loss": 0.1979, "step": 7199 }, { "epoch": 0.98, "grad_norm": 4.00234632640867, "learning_rate": 5.409314398409067e-06, "loss": 0.2346, "step": 7200 }, { "epoch": 0.98, "grad_norm": 3.8624841533799392, "learning_rate": 5.408216545491254e-06, "loss": 0.1951, "step": 7201 }, { "epoch": 0.98, "grad_norm": 4.051614355033294, "learning_rate": 5.407118672760393e-06, "loss": 0.2022, "step": 7202 }, { "epoch": 0.98, "grad_norm": 3.386291926104393, "learning_rate": 5.406020780269771e-06, "loss": 0.2033, "step": 7203 }, { "epoch": 0.98, "grad_norm": 3.805277514161475, "learning_rate": 5.404922868072673e-06, "loss": 0.2253, "step": 7204 }, { "epoch": 0.98, "grad_norm": 3.682686044553083, "learning_rate": 5.403824936222386e-06, "loss": 0.2032, "step": 7205 }, { "epoch": 0.98, "grad_norm": 3.2615478343505426, "learning_rate": 5.4027269847722006e-06, "loss": 0.1905, "step": 7206 }, { "epoch": 0.98, "grad_norm": 3.5104979262988385, "learning_rate": 5.401629013775408e-06, "loss": 0.2095, "step": 7207 }, { "epoch": 0.98, "grad_norm": 3.72594757982031, "learning_rate": 5.4005310232852966e-06, "loss": 0.2417, "step": 7208 }, { "epoch": 0.98, "grad_norm": 4.203741371018409, "learning_rate": 5.399433013355159e-06, "loss": 0.2347, "step": 7209 }, { "epoch": 0.98, "grad_norm": 5.7742149936917615, "learning_rate": 5.398334984038287e-06, "loss": 0.19, "step": 7210 }, { "epoch": 0.98, "grad_norm": 4.726012475520461, "learning_rate": 5.397236935387975e-06, "loss": 0.2366, "step": 7211 }, { "epoch": 0.98, "grad_norm": 4.6534923960925445, "learning_rate": 5.396138867457517e-06, "loss": 0.2691, "step": 7212 }, { "epoch": 0.98, "grad_norm": 3.5084463020784407, "learning_rate": 5.395040780300208e-06, "loss": 0.1853, "step": 7213 }, { "epoch": 0.98, "grad_norm": 3.049021522840113, "learning_rate": 5.393942673969345e-06, "loss": 0.2097, "step": 7214 }, { "epoch": 0.98, "grad_norm": 5.649974654254955, "learning_rate": 5.3928445485182254e-06, "loss": 0.2594, "step": 7215 }, { "epoch": 0.98, "grad_norm": 3.9364335752350956, "learning_rate": 5.391746404000148e-06, "loss": 0.2218, "step": 7216 }, { "epoch": 0.98, "grad_norm": 6.702149348571896, "learning_rate": 5.39064824046841e-06, "loss": 0.2498, "step": 7217 }, { "epoch": 0.98, "grad_norm": 3.74766142718837, "learning_rate": 5.3895500579763135e-06, "loss": 0.23, "step": 7218 }, { "epoch": 0.98, "grad_norm": 4.5667554415750535, "learning_rate": 5.388451856577157e-06, "loss": 0.2471, "step": 7219 }, { "epoch": 0.98, "grad_norm": 3.177079751674192, "learning_rate": 5.387353636324246e-06, "loss": 0.199, "step": 7220 }, { "epoch": 0.98, "grad_norm": 4.156181630815838, "learning_rate": 5.386255397270881e-06, "loss": 0.1812, "step": 7221 }, { "epoch": 0.98, "grad_norm": 3.029667448258276, "learning_rate": 5.385157139470365e-06, "loss": 0.1921, "step": 7222 }, { "epoch": 0.98, "grad_norm": 3.5413886508262715, "learning_rate": 5.384058862976005e-06, "loss": 0.2179, "step": 7223 }, { "epoch": 0.98, "grad_norm": 3.9343061693416517, "learning_rate": 5.3829605678411035e-06, "loss": 0.1623, "step": 7224 }, { "epoch": 0.98, "grad_norm": 3.640971782625613, "learning_rate": 5.381862254118971e-06, "loss": 0.2147, "step": 7225 }, { "epoch": 0.98, "grad_norm": 3.991229129362742, "learning_rate": 5.380763921862909e-06, "loss": 0.1843, "step": 7226 }, { "epoch": 0.98, "grad_norm": 3.3001139153304164, "learning_rate": 5.379665571126232e-06, "loss": 0.1829, "step": 7227 }, { "epoch": 0.98, "grad_norm": 3.440484172243165, "learning_rate": 5.378567201962246e-06, "loss": 0.2736, "step": 7228 }, { "epoch": 0.98, "grad_norm": 3.4513788533664633, "learning_rate": 5.3774688144242605e-06, "loss": 0.1719, "step": 7229 }, { "epoch": 0.98, "grad_norm": 3.5427766091924653, "learning_rate": 5.376370408565589e-06, "loss": 0.2271, "step": 7230 }, { "epoch": 0.98, "grad_norm": 3.8523323109476135, "learning_rate": 5.375271984439541e-06, "loss": 0.2146, "step": 7231 }, { "epoch": 0.98, "grad_norm": 3.6236889345845276, "learning_rate": 5.374173542099429e-06, "loss": 0.2061, "step": 7232 }, { "epoch": 0.98, "grad_norm": 2.986024576574895, "learning_rate": 5.3730750815985685e-06, "loss": 0.202, "step": 7233 }, { "epoch": 0.98, "grad_norm": 3.242524403153061, "learning_rate": 5.371976602990273e-06, "loss": 0.1905, "step": 7234 }, { "epoch": 0.98, "grad_norm": 3.1678280405192245, "learning_rate": 5.370878106327857e-06, "loss": 0.1874, "step": 7235 }, { "epoch": 0.98, "grad_norm": 8.73627270010317, "learning_rate": 5.369779591664639e-06, "loss": 0.2006, "step": 7236 }, { "epoch": 0.98, "grad_norm": 3.6713424128661365, "learning_rate": 5.368681059053934e-06, "loss": 0.2048, "step": 7237 }, { "epoch": 0.98, "grad_norm": 2.9482164745805037, "learning_rate": 5.367582508549058e-06, "loss": 0.1955, "step": 7238 }, { "epoch": 0.98, "grad_norm": 4.468031184628961, "learning_rate": 5.366483940203336e-06, "loss": 0.2009, "step": 7239 }, { "epoch": 0.98, "grad_norm": 3.713548327429414, "learning_rate": 5.365385354070083e-06, "loss": 0.2011, "step": 7240 }, { "epoch": 0.98, "grad_norm": 3.931563640116795, "learning_rate": 5.3642867502026205e-06, "loss": 0.204, "step": 7241 }, { "epoch": 0.99, "grad_norm": 3.3605902451392162, "learning_rate": 5.363188128654272e-06, "loss": 0.1696, "step": 7242 }, { "epoch": 0.99, "grad_norm": 3.141492237753093, "learning_rate": 5.362089489478356e-06, "loss": 0.2263, "step": 7243 }, { "epoch": 0.99, "grad_norm": 3.367519056143933, "learning_rate": 5.360990832728199e-06, "loss": 0.1858, "step": 7244 }, { "epoch": 0.99, "grad_norm": 4.854374712356326, "learning_rate": 5.3598921584571215e-06, "loss": 0.2344, "step": 7245 }, { "epoch": 0.99, "grad_norm": 4.31142178550291, "learning_rate": 5.358793466718452e-06, "loss": 0.2529, "step": 7246 }, { "epoch": 0.99, "grad_norm": 4.1804821537099075, "learning_rate": 5.357694757565515e-06, "loss": 0.2087, "step": 7247 }, { "epoch": 0.99, "grad_norm": 3.679340622628517, "learning_rate": 5.356596031051636e-06, "loss": 0.1898, "step": 7248 }, { "epoch": 0.99, "grad_norm": 3.8655179905910813, "learning_rate": 5.355497287230144e-06, "loss": 0.1976, "step": 7249 }, { "epoch": 0.99, "grad_norm": 3.8360133329092814, "learning_rate": 5.354398526154365e-06, "loss": 0.2128, "step": 7250 }, { "epoch": 0.99, "grad_norm": 5.706819656603499, "learning_rate": 5.35329974787763e-06, "loss": 0.2748, "step": 7251 }, { "epoch": 0.99, "grad_norm": 3.6118662998970783, "learning_rate": 5.352200952453268e-06, "loss": 0.2073, "step": 7252 }, { "epoch": 0.99, "grad_norm": 4.637595068543485, "learning_rate": 5.351102139934611e-06, "loss": 0.2155, "step": 7253 }, { "epoch": 0.99, "grad_norm": 4.1320949030457905, "learning_rate": 5.350003310374989e-06, "loss": 0.198, "step": 7254 }, { "epoch": 0.99, "grad_norm": 3.5096916726548844, "learning_rate": 5.348904463827735e-06, "loss": 0.1969, "step": 7255 }, { "epoch": 0.99, "grad_norm": 3.8561149539753328, "learning_rate": 5.347805600346183e-06, "loss": 0.2723, "step": 7256 }, { "epoch": 0.99, "grad_norm": 3.1910172710473623, "learning_rate": 5.3467067199836665e-06, "loss": 0.1931, "step": 7257 }, { "epoch": 0.99, "grad_norm": 4.701948921477372, "learning_rate": 5.345607822793519e-06, "loss": 0.1837, "step": 7258 }, { "epoch": 0.99, "grad_norm": 3.681171356321577, "learning_rate": 5.344508908829076e-06, "loss": 0.2364, "step": 7259 }, { "epoch": 0.99, "grad_norm": 6.034220452332489, "learning_rate": 5.343409978143676e-06, "loss": 0.2013, "step": 7260 }, { "epoch": 0.99, "grad_norm": 3.341051264775717, "learning_rate": 5.342311030790658e-06, "loss": 0.1874, "step": 7261 }, { "epoch": 0.99, "grad_norm": 3.667157931423144, "learning_rate": 5.341212066823356e-06, "loss": 0.2031, "step": 7262 }, { "epoch": 0.99, "grad_norm": 4.229493846692746, "learning_rate": 5.34011308629511e-06, "loss": 0.181, "step": 7263 }, { "epoch": 0.99, "grad_norm": 3.4732356788250534, "learning_rate": 5.339014089259261e-06, "loss": 0.2177, "step": 7264 }, { "epoch": 0.99, "grad_norm": 3.8927282183944976, "learning_rate": 5.337915075769148e-06, "loss": 0.2065, "step": 7265 }, { "epoch": 0.99, "grad_norm": 4.1616938947737765, "learning_rate": 5.336816045878112e-06, "loss": 0.2625, "step": 7266 }, { "epoch": 0.99, "grad_norm": 3.065566509948505, "learning_rate": 5.335716999639499e-06, "loss": 0.1978, "step": 7267 }, { "epoch": 0.99, "grad_norm": 4.59732579047424, "learning_rate": 5.334617937106649e-06, "loss": 0.2438, "step": 7268 }, { "epoch": 0.99, "grad_norm": 4.197829855672333, "learning_rate": 5.3335188583329046e-06, "loss": 0.225, "step": 7269 }, { "epoch": 0.99, "grad_norm": 3.0859474690127753, "learning_rate": 5.332419763371611e-06, "loss": 0.2081, "step": 7270 }, { "epoch": 0.99, "grad_norm": 4.188654774212709, "learning_rate": 5.331320652276115e-06, "loss": 0.2281, "step": 7271 }, { "epoch": 0.99, "grad_norm": 3.0206084459585987, "learning_rate": 5.330221525099761e-06, "loss": 0.1947, "step": 7272 }, { "epoch": 0.99, "grad_norm": 4.036507703935227, "learning_rate": 5.329122381895897e-06, "loss": 0.2331, "step": 7273 }, { "epoch": 0.99, "grad_norm": 4.064617016084963, "learning_rate": 5.32802322271787e-06, "loss": 0.284, "step": 7274 }, { "epoch": 0.99, "grad_norm": 3.3110531656620386, "learning_rate": 5.326924047619029e-06, "loss": 0.159, "step": 7275 }, { "epoch": 0.99, "grad_norm": 3.452892194668126, "learning_rate": 5.325824856652721e-06, "loss": 0.1575, "step": 7276 }, { "epoch": 0.99, "grad_norm": 4.081221806190579, "learning_rate": 5.3247256498722985e-06, "loss": 0.2857, "step": 7277 }, { "epoch": 0.99, "grad_norm": 3.2491065067868594, "learning_rate": 5.32362642733111e-06, "loss": 0.2203, "step": 7278 }, { "epoch": 0.99, "grad_norm": 3.667876878678596, "learning_rate": 5.322527189082509e-06, "loss": 0.1921, "step": 7279 }, { "epoch": 0.99, "grad_norm": 2.6509505472857757, "learning_rate": 5.321427935179848e-06, "loss": 0.1925, "step": 7280 }, { "epoch": 0.99, "grad_norm": 4.12397488028433, "learning_rate": 5.3203286656764784e-06, "loss": 0.2252, "step": 7281 }, { "epoch": 0.99, "grad_norm": 3.538111135291421, "learning_rate": 5.319229380625754e-06, "loss": 0.2258, "step": 7282 }, { "epoch": 0.99, "grad_norm": 4.249082432448482, "learning_rate": 5.3181300800810295e-06, "loss": 0.194, "step": 7283 }, { "epoch": 0.99, "grad_norm": 6.7003318049281395, "learning_rate": 5.317030764095661e-06, "loss": 0.2147, "step": 7284 }, { "epoch": 0.99, "grad_norm": 3.462128791410551, "learning_rate": 5.315931432723003e-06, "loss": 0.2037, "step": 7285 }, { "epoch": 0.99, "grad_norm": 3.9219641983777733, "learning_rate": 5.314832086016414e-06, "loss": 0.1509, "step": 7286 }, { "epoch": 0.99, "grad_norm": 5.047391847747456, "learning_rate": 5.31373272402925e-06, "loss": 0.1821, "step": 7287 }, { "epoch": 0.99, "grad_norm": 2.7397523124661016, "learning_rate": 5.3126333468148705e-06, "loss": 0.1896, "step": 7288 }, { "epoch": 0.99, "grad_norm": 4.674774428127592, "learning_rate": 5.311533954426633e-06, "loss": 0.3029, "step": 7289 }, { "epoch": 0.99, "grad_norm": 4.538405845252676, "learning_rate": 5.310434546917899e-06, "loss": 0.2158, "step": 7290 }, { "epoch": 0.99, "grad_norm": 4.408226018180218, "learning_rate": 5.309335124342027e-06, "loss": 0.2648, "step": 7291 }, { "epoch": 0.99, "grad_norm": 3.275074903396142, "learning_rate": 5.308235686752379e-06, "loss": 0.1936, "step": 7292 }, { "epoch": 0.99, "grad_norm": 3.9769812742050026, "learning_rate": 5.307136234202318e-06, "loss": 0.2103, "step": 7293 }, { "epoch": 0.99, "grad_norm": 3.225560629840356, "learning_rate": 5.3060367667452055e-06, "loss": 0.1906, "step": 7294 }, { "epoch": 0.99, "grad_norm": 3.6446529907865215, "learning_rate": 5.304937284434405e-06, "loss": 0.2022, "step": 7295 }, { "epoch": 0.99, "grad_norm": 3.826229325181535, "learning_rate": 5.30383778732328e-06, "loss": 0.2271, "step": 7296 }, { "epoch": 0.99, "grad_norm": 3.6575579231706397, "learning_rate": 5.302738275465196e-06, "loss": 0.2392, "step": 7297 }, { "epoch": 0.99, "grad_norm": 3.3040956225488607, "learning_rate": 5.301638748913518e-06, "loss": 0.1812, "step": 7298 }, { "epoch": 0.99, "grad_norm": 3.2359619766172396, "learning_rate": 5.3005392077216136e-06, "loss": 0.1825, "step": 7299 }, { "epoch": 0.99, "grad_norm": 3.9215457139926566, "learning_rate": 5.299439651942847e-06, "loss": 0.2078, "step": 7300 }, { "epoch": 0.99, "grad_norm": 3.489848991742789, "learning_rate": 5.2983400816305895e-06, "loss": 0.2387, "step": 7301 }, { "epoch": 0.99, "grad_norm": 3.527196228094341, "learning_rate": 5.297240496838206e-06, "loss": 0.2639, "step": 7302 }, { "epoch": 0.99, "grad_norm": 3.786817138325032, "learning_rate": 5.296140897619069e-06, "loss": 0.2137, "step": 7303 }, { "epoch": 0.99, "grad_norm": 3.4253133138773033, "learning_rate": 5.295041284026544e-06, "loss": 0.2054, "step": 7304 }, { "epoch": 0.99, "grad_norm": 3.408165029436988, "learning_rate": 5.293941656114004e-06, "loss": 0.2512, "step": 7305 }, { "epoch": 0.99, "grad_norm": 3.538293602622635, "learning_rate": 5.2928420139348205e-06, "loss": 0.2735, "step": 7306 }, { "epoch": 0.99, "grad_norm": 3.126598874991761, "learning_rate": 5.291742357542364e-06, "loss": 0.2384, "step": 7307 }, { "epoch": 0.99, "grad_norm": 4.661240036063662, "learning_rate": 5.2906426869900086e-06, "loss": 0.1922, "step": 7308 }, { "epoch": 0.99, "grad_norm": 3.2869867294436106, "learning_rate": 5.289543002331127e-06, "loss": 0.212, "step": 7309 }, { "epoch": 0.99, "grad_norm": 4.740308210617787, "learning_rate": 5.288443303619092e-06, "loss": 0.2447, "step": 7310 }, { "epoch": 0.99, "grad_norm": 4.625551801619504, "learning_rate": 5.287343590907279e-06, "loss": 0.208, "step": 7311 }, { "epoch": 0.99, "grad_norm": 4.464141154023131, "learning_rate": 5.2862438642490634e-06, "loss": 0.2258, "step": 7312 }, { "epoch": 0.99, "grad_norm": 3.761274730625357, "learning_rate": 5.285144123697819e-06, "loss": 0.2031, "step": 7313 }, { "epoch": 0.99, "grad_norm": 3.2950564328707093, "learning_rate": 5.284044369306927e-06, "loss": 0.1905, "step": 7314 }, { "epoch": 1.0, "grad_norm": 3.5042230249510697, "learning_rate": 5.2829446011297605e-06, "loss": 0.2432, "step": 7315 }, { "epoch": 1.0, "grad_norm": 3.7047399412553896, "learning_rate": 5.281844819219698e-06, "loss": 0.2503, "step": 7316 }, { "epoch": 1.0, "grad_norm": 4.099114195573189, "learning_rate": 5.280745023630119e-06, "loss": 0.1899, "step": 7317 }, { "epoch": 1.0, "grad_norm": 3.5121968804314183, "learning_rate": 5.279645214414404e-06, "loss": 0.1799, "step": 7318 }, { "epoch": 1.0, "grad_norm": 3.713807865722439, "learning_rate": 5.278545391625931e-06, "loss": 0.2258, "step": 7319 }, { "epoch": 1.0, "grad_norm": 3.6309656475650844, "learning_rate": 5.277445555318081e-06, "loss": 0.1892, "step": 7320 }, { "epoch": 1.0, "grad_norm": 2.787518597666823, "learning_rate": 5.276345705544237e-06, "loss": 0.2172, "step": 7321 }, { "epoch": 1.0, "grad_norm": 3.3805859466449593, "learning_rate": 5.275245842357778e-06, "loss": 0.206, "step": 7322 }, { "epoch": 1.0, "grad_norm": 4.060690792972819, "learning_rate": 5.274145965812088e-06, "loss": 0.2304, "step": 7323 }, { "epoch": 1.0, "grad_norm": 2.7609276771467215, "learning_rate": 5.27304607596055e-06, "loss": 0.1884, "step": 7324 }, { "epoch": 1.0, "grad_norm": 3.489855020531112, "learning_rate": 5.271946172856549e-06, "loss": 0.2437, "step": 7325 }, { "epoch": 1.0, "grad_norm": 4.712164023967028, "learning_rate": 5.270846256553468e-06, "loss": 0.2667, "step": 7326 }, { "epoch": 1.0, "grad_norm": 3.037150822332647, "learning_rate": 5.269746327104693e-06, "loss": 0.2036, "step": 7327 }, { "epoch": 1.0, "grad_norm": 3.4361810280536296, "learning_rate": 5.268646384563609e-06, "loss": 0.2128, "step": 7328 }, { "epoch": 1.0, "grad_norm": 3.694972240364652, "learning_rate": 5.2675464289836046e-06, "loss": 0.1821, "step": 7329 }, { "epoch": 1.0, "grad_norm": 3.2580989357208576, "learning_rate": 5.266446460418063e-06, "loss": 0.1767, "step": 7330 }, { "epoch": 1.0, "grad_norm": 4.172036504465267, "learning_rate": 5.265346478920374e-06, "loss": 0.2194, "step": 7331 }, { "epoch": 1.0, "grad_norm": 3.6688696760670223, "learning_rate": 5.264246484543926e-06, "loss": 0.2126, "step": 7332 }, { "epoch": 1.0, "grad_norm": 3.291548633635999, "learning_rate": 5.2631464773421085e-06, "loss": 0.2271, "step": 7333 }, { "epoch": 1.0, "grad_norm": 2.915058908123231, "learning_rate": 5.2620464573683095e-06, "loss": 0.1847, "step": 7334 }, { "epoch": 1.0, "grad_norm": 3.6431047467140005, "learning_rate": 5.26094642467592e-06, "loss": 0.2505, "step": 7335 }, { "epoch": 1.0, "grad_norm": 3.393701918177582, "learning_rate": 5.259846379318332e-06, "loss": 0.2446, "step": 7336 }, { "epoch": 1.0, "grad_norm": 3.818083505144832, "learning_rate": 5.258746321348934e-06, "loss": 0.2194, "step": 7337 }, { "epoch": 1.0, "grad_norm": 4.771612417932317, "learning_rate": 5.25764625082112e-06, "loss": 0.2369, "step": 7338 }, { "epoch": 1.0, "grad_norm": 4.4608319507997525, "learning_rate": 5.256546167788283e-06, "loss": 0.2589, "step": 7339 }, { "epoch": 1.0, "grad_norm": 3.9144877224924386, "learning_rate": 5.255446072303815e-06, "loss": 0.2273, "step": 7340 }, { "epoch": 1.0, "grad_norm": 3.982302637059972, "learning_rate": 5.25434596442111e-06, "loss": 0.206, "step": 7341 }, { "epoch": 1.0, "grad_norm": 3.6452479073700124, "learning_rate": 5.253245844193564e-06, "loss": 0.2033, "step": 7342 }, { "epoch": 1.0, "grad_norm": 3.7945710930136896, "learning_rate": 5.25214571167457e-06, "loss": 0.2402, "step": 7343 }, { "epoch": 1.0, "grad_norm": 3.7165195048644533, "learning_rate": 5.251045566917525e-06, "loss": 0.2128, "step": 7344 }, { "epoch": 1.0, "grad_norm": 4.6838825401342925, "learning_rate": 5.249945409975823e-06, "loss": 0.2136, "step": 7345 }, { "epoch": 1.0, "grad_norm": 3.609705253290724, "learning_rate": 5.2488452409028645e-06, "loss": 0.2242, "step": 7346 }, { "epoch": 1.0, "grad_norm": 3.7992003327552397, "learning_rate": 5.247745059752044e-06, "loss": 0.2039, "step": 7347 }, { "epoch": 1.0, "grad_norm": 4.18689726495723, "learning_rate": 5.24664486657676e-06, "loss": 0.2496, "step": 7348 }, { "epoch": 1.0, "grad_norm": 3.231171337106068, "learning_rate": 5.245544661430413e-06, "loss": 0.2266, "step": 7349 }, { "epoch": 1.0, "grad_norm": 2.528784392147147, "learning_rate": 5.2444444443663985e-06, "loss": 0.1756, "step": 7350 }, { "epoch": 1.0, "grad_norm": 3.21030855573136, "learning_rate": 5.2433442154381196e-06, "loss": 0.1945, "step": 7351 }, { "epoch": 1.0, "eval_EK100_accuracy": 0.6482054455445545, "step": 7351 }, { "epoch": 1.0, "grad_norm": 3.074261966615747, "learning_rate": 5.242243974698975e-06, "loss": 0.3476, "step": 7352 }, { "epoch": 1.0, "grad_norm": 3.2896460520931554, "learning_rate": 5.241143722202367e-06, "loss": 0.2092, "step": 7353 }, { "epoch": 1.0, "grad_norm": 3.470320330675441, "learning_rate": 5.240043458001694e-06, "loss": 0.1966, "step": 7354 }, { "epoch": 1.0, "grad_norm": 3.211384326099183, "learning_rate": 5.238943182150361e-06, "loss": 0.1623, "step": 7355 }, { "epoch": 1.0, "grad_norm": 3.2080213312496793, "learning_rate": 5.237842894701769e-06, "loss": 0.2107, "step": 7356 }, { "epoch": 1.0, "grad_norm": 3.2716436996612632, "learning_rate": 5.236742595709321e-06, "loss": 0.1698, "step": 7357 }, { "epoch": 1.0, "grad_norm": 2.829862058998519, "learning_rate": 5.235642285226422e-06, "loss": 0.1189, "step": 7358 }, { "epoch": 1.0, "grad_norm": 3.181898739609208, "learning_rate": 5.234541963306476e-06, "loss": 0.169, "step": 7359 }, { "epoch": 1.0, "grad_norm": 2.7795802052947893, "learning_rate": 5.2334416300028886e-06, "loss": 0.142, "step": 7360 }, { "epoch": 1.0, "grad_norm": 2.9836527178909233, "learning_rate": 5.232341285369063e-06, "loss": 0.2055, "step": 7361 }, { "epoch": 1.0, "grad_norm": 4.001005946878919, "learning_rate": 5.231240929458406e-06, "loss": 0.1564, "step": 7362 }, { "epoch": 1.0, "grad_norm": 3.315578596280007, "learning_rate": 5.230140562324325e-06, "loss": 0.2064, "step": 7363 }, { "epoch": 1.0, "grad_norm": 3.2758380456467484, "learning_rate": 5.229040184020224e-06, "loss": 0.1499, "step": 7364 }, { "epoch": 1.0, "grad_norm": 4.124491214715152, "learning_rate": 5.227939794599514e-06, "loss": 0.1732, "step": 7365 }, { "epoch": 1.0, "grad_norm": 3.203280428541086, "learning_rate": 5.226839394115604e-06, "loss": 0.1583, "step": 7366 }, { "epoch": 1.0, "grad_norm": 2.8777121610510648, "learning_rate": 5.225738982621898e-06, "loss": 0.1456, "step": 7367 }, { "epoch": 1.0, "grad_norm": 3.0592041749607053, "learning_rate": 5.22463856017181e-06, "loss": 0.1971, "step": 7368 }, { "epoch": 1.0, "grad_norm": 3.2920393081191968, "learning_rate": 5.223538126818746e-06, "loss": 0.1462, "step": 7369 }, { "epoch": 1.0, "grad_norm": 3.8375776968569557, "learning_rate": 5.2224376826161185e-06, "loss": 0.1825, "step": 7370 }, { "epoch": 1.0, "grad_norm": 4.739091011807606, "learning_rate": 5.2213372276173345e-06, "loss": 0.1764, "step": 7371 }, { "epoch": 1.0, "grad_norm": 3.0171075917473114, "learning_rate": 5.220236761875811e-06, "loss": 0.1457, "step": 7372 }, { "epoch": 1.0, "grad_norm": 3.870828407252211, "learning_rate": 5.2191362854449565e-06, "loss": 0.1824, "step": 7373 }, { "epoch": 1.0, "grad_norm": 3.890580499425639, "learning_rate": 5.218035798378184e-06, "loss": 0.1863, "step": 7374 }, { "epoch": 1.0, "grad_norm": 3.270487124418797, "learning_rate": 5.216935300728906e-06, "loss": 0.1324, "step": 7375 }, { "epoch": 1.0, "grad_norm": 3.0485116801176733, "learning_rate": 5.215834792550536e-06, "loss": 0.1675, "step": 7376 }, { "epoch": 1.0, "grad_norm": 2.9178338099355936, "learning_rate": 5.214734273896488e-06, "loss": 0.1467, "step": 7377 }, { "epoch": 1.0, "grad_norm": 3.504644869021338, "learning_rate": 5.213633744820174e-06, "loss": 0.1843, "step": 7378 }, { "epoch": 1.0, "grad_norm": 3.3984829482861776, "learning_rate": 5.212533205375013e-06, "loss": 0.1754, "step": 7379 }, { "epoch": 1.0, "grad_norm": 3.8093165537838796, "learning_rate": 5.211432655614419e-06, "loss": 0.1852, "step": 7380 }, { "epoch": 1.0, "grad_norm": 3.3734285301142255, "learning_rate": 5.210332095591807e-06, "loss": 0.1838, "step": 7381 }, { "epoch": 1.0, "grad_norm": 3.875802715192318, "learning_rate": 5.209231525360594e-06, "loss": 0.2077, "step": 7382 }, { "epoch": 1.0, "grad_norm": 3.424322881966081, "learning_rate": 5.208130944974196e-06, "loss": 0.1872, "step": 7383 }, { "epoch": 1.0, "grad_norm": 4.003847268322232, "learning_rate": 5.20703035448603e-06, "loss": 0.1691, "step": 7384 }, { "epoch": 1.0, "grad_norm": 3.0563832598576246, "learning_rate": 5.205929753949516e-06, "loss": 0.174, "step": 7385 }, { "epoch": 1.0, "grad_norm": 3.8039120377762834, "learning_rate": 5.204829143418072e-06, "loss": 0.2004, "step": 7386 }, { "epoch": 1.0, "grad_norm": 4.652015841521336, "learning_rate": 5.203728522945115e-06, "loss": 0.1787, "step": 7387 }, { "epoch": 1.0, "grad_norm": 4.9435197886736955, "learning_rate": 5.2026278925840656e-06, "loss": 0.1824, "step": 7388 }, { "epoch": 1.01, "grad_norm": 3.223426860723198, "learning_rate": 5.201527252388343e-06, "loss": 0.1836, "step": 7389 }, { "epoch": 1.01, "grad_norm": 3.166922510595105, "learning_rate": 5.200426602411368e-06, "loss": 0.1442, "step": 7390 }, { "epoch": 1.01, "grad_norm": 3.038487917537227, "learning_rate": 5.19932594270656e-06, "loss": 0.1572, "step": 7391 }, { "epoch": 1.01, "grad_norm": 4.49351281120141, "learning_rate": 5.198225273327343e-06, "loss": 0.1886, "step": 7392 }, { "epoch": 1.01, "grad_norm": 3.2630313174452867, "learning_rate": 5.197124594327137e-06, "loss": 0.1852, "step": 7393 }, { "epoch": 1.01, "grad_norm": 3.082425904767994, "learning_rate": 5.196023905759363e-06, "loss": 0.1966, "step": 7394 }, { "epoch": 1.01, "grad_norm": 3.7830336755032272, "learning_rate": 5.194923207677446e-06, "loss": 0.2213, "step": 7395 }, { "epoch": 1.01, "grad_norm": 3.626712654050683, "learning_rate": 5.193822500134806e-06, "loss": 0.1682, "step": 7396 }, { "epoch": 1.01, "grad_norm": 4.1576999233351035, "learning_rate": 5.1927217831848685e-06, "loss": 0.1719, "step": 7397 }, { "epoch": 1.01, "grad_norm": 4.7375858747250605, "learning_rate": 5.191621056881059e-06, "loss": 0.2174, "step": 7398 }, { "epoch": 1.01, "grad_norm": 4.404620945488908, "learning_rate": 5.190520321276799e-06, "loss": 0.1581, "step": 7399 }, { "epoch": 1.01, "grad_norm": 3.415640285507794, "learning_rate": 5.189419576425516e-06, "loss": 0.166, "step": 7400 }, { "epoch": 1.01, "grad_norm": 4.133487234389966, "learning_rate": 5.1883188223806336e-06, "loss": 0.2353, "step": 7401 }, { "epoch": 1.01, "grad_norm": 3.217963371587039, "learning_rate": 5.187218059195578e-06, "loss": 0.1603, "step": 7402 }, { "epoch": 1.01, "grad_norm": 2.9932329198062346, "learning_rate": 5.186117286923775e-06, "loss": 0.1414, "step": 7403 }, { "epoch": 1.01, "grad_norm": 3.738667566505425, "learning_rate": 5.185016505618652e-06, "loss": 0.1561, "step": 7404 }, { "epoch": 1.01, "grad_norm": 4.036506663660283, "learning_rate": 5.183915715333636e-06, "loss": 0.2163, "step": 7405 }, { "epoch": 1.01, "grad_norm": 2.5832867103497605, "learning_rate": 5.182814916122154e-06, "loss": 0.1416, "step": 7406 }, { "epoch": 1.01, "grad_norm": 4.630256277570794, "learning_rate": 5.181714108037635e-06, "loss": 0.1762, "step": 7407 }, { "epoch": 1.01, "grad_norm": 3.9564786382212476, "learning_rate": 5.180613291133507e-06, "loss": 0.2077, "step": 7408 }, { "epoch": 1.01, "grad_norm": 3.4112389654106794, "learning_rate": 5.179512465463199e-06, "loss": 0.1604, "step": 7409 }, { "epoch": 1.01, "grad_norm": 4.533803252708708, "learning_rate": 5.178411631080139e-06, "loss": 0.18, "step": 7410 }, { "epoch": 1.01, "grad_norm": 6.843882310000426, "learning_rate": 5.177310788037758e-06, "loss": 0.1741, "step": 7411 }, { "epoch": 1.01, "grad_norm": 3.066338171627138, "learning_rate": 5.176209936389485e-06, "loss": 0.1425, "step": 7412 }, { "epoch": 1.01, "grad_norm": 3.693098368061454, "learning_rate": 5.175109076188754e-06, "loss": 0.1971, "step": 7413 }, { "epoch": 1.01, "grad_norm": 3.443261224840282, "learning_rate": 5.174008207488991e-06, "loss": 0.1554, "step": 7414 }, { "epoch": 1.01, "grad_norm": 2.8892517027351743, "learning_rate": 5.1729073303436304e-06, "loss": 0.1285, "step": 7415 }, { "epoch": 1.01, "grad_norm": 2.822173788126479, "learning_rate": 5.171806444806102e-06, "loss": 0.1726, "step": 7416 }, { "epoch": 1.01, "grad_norm": 3.1119571367467525, "learning_rate": 5.17070555092984e-06, "loss": 0.1384, "step": 7417 }, { "epoch": 1.01, "grad_norm": 3.049062530419466, "learning_rate": 5.169604648768276e-06, "loss": 0.1553, "step": 7418 }, { "epoch": 1.01, "grad_norm": 3.5899920812317485, "learning_rate": 5.168503738374843e-06, "loss": 0.1445, "step": 7419 }, { "epoch": 1.01, "grad_norm": 3.4126860644475827, "learning_rate": 5.1674028198029746e-06, "loss": 0.2031, "step": 7420 }, { "epoch": 1.01, "grad_norm": 3.481972559111792, "learning_rate": 5.166301893106105e-06, "loss": 0.2155, "step": 7421 }, { "epoch": 1.01, "grad_norm": 3.320257911823644, "learning_rate": 5.1652009583376676e-06, "loss": 0.1771, "step": 7422 }, { "epoch": 1.01, "grad_norm": 3.518824592374053, "learning_rate": 5.164100015551098e-06, "loss": 0.2153, "step": 7423 }, { "epoch": 1.01, "grad_norm": 3.4133824118092413, "learning_rate": 5.162999064799827e-06, "loss": 0.1603, "step": 7424 }, { "epoch": 1.01, "grad_norm": 4.170392393485278, "learning_rate": 5.161898106137297e-06, "loss": 0.1975, "step": 7425 }, { "epoch": 1.01, "grad_norm": 3.514874697486768, "learning_rate": 5.160797139616939e-06, "loss": 0.1829, "step": 7426 }, { "epoch": 1.01, "grad_norm": 3.0902104740856577, "learning_rate": 5.159696165292189e-06, "loss": 0.1234, "step": 7427 }, { "epoch": 1.01, "grad_norm": 4.158671212096517, "learning_rate": 5.158595183216486e-06, "loss": 0.2017, "step": 7428 }, { "epoch": 1.01, "grad_norm": 3.89725904696487, "learning_rate": 5.157494193443266e-06, "loss": 0.1776, "step": 7429 }, { "epoch": 1.01, "grad_norm": 3.127760676631335, "learning_rate": 5.156393196025964e-06, "loss": 0.2227, "step": 7430 }, { "epoch": 1.01, "grad_norm": 3.4546313939218423, "learning_rate": 5.1552921910180195e-06, "loss": 0.1683, "step": 7431 }, { "epoch": 1.01, "grad_norm": 3.8665688739851527, "learning_rate": 5.154191178472873e-06, "loss": 0.1884, "step": 7432 }, { "epoch": 1.01, "grad_norm": 3.5266000164344433, "learning_rate": 5.153090158443958e-06, "loss": 0.1485, "step": 7433 }, { "epoch": 1.01, "grad_norm": 4.082700157227225, "learning_rate": 5.151989130984715e-06, "loss": 0.2027, "step": 7434 }, { "epoch": 1.01, "grad_norm": 2.9425910511547024, "learning_rate": 5.150888096148584e-06, "loss": 0.1595, "step": 7435 }, { "epoch": 1.01, "grad_norm": 3.805682455509553, "learning_rate": 5.149787053989003e-06, "loss": 0.1808, "step": 7436 }, { "epoch": 1.01, "grad_norm": 3.6028811734349366, "learning_rate": 5.148686004559412e-06, "loss": 0.1625, "step": 7437 }, { "epoch": 1.01, "grad_norm": 3.38076753359894, "learning_rate": 5.147584947913253e-06, "loss": 0.1731, "step": 7438 }, { "epoch": 1.01, "grad_norm": 2.7667139246086845, "learning_rate": 5.146483884103967e-06, "loss": 0.1192, "step": 7439 }, { "epoch": 1.01, "grad_norm": 3.653384483422841, "learning_rate": 5.1453828131849904e-06, "loss": 0.1863, "step": 7440 }, { "epoch": 1.01, "grad_norm": 3.6362011893647237, "learning_rate": 5.144281735209768e-06, "loss": 0.1546, "step": 7441 }, { "epoch": 1.01, "grad_norm": 3.734453000540021, "learning_rate": 5.143180650231741e-06, "loss": 0.1517, "step": 7442 }, { "epoch": 1.01, "grad_norm": 3.8710373167055, "learning_rate": 5.142079558304348e-06, "loss": 0.1801, "step": 7443 }, { "epoch": 1.01, "grad_norm": 4.375538944439668, "learning_rate": 5.140978459481036e-06, "loss": 0.1656, "step": 7444 }, { "epoch": 1.01, "grad_norm": 3.006270767839707, "learning_rate": 5.139877353815245e-06, "loss": 0.1958, "step": 7445 }, { "epoch": 1.01, "grad_norm": 3.3485960842664437, "learning_rate": 5.138776241360418e-06, "loss": 0.1717, "step": 7446 }, { "epoch": 1.01, "grad_norm": 4.117267427918887, "learning_rate": 5.13767512217e-06, "loss": 0.1818, "step": 7447 }, { "epoch": 1.01, "grad_norm": 3.6774317019242195, "learning_rate": 5.136573996297431e-06, "loss": 0.1199, "step": 7448 }, { "epoch": 1.01, "grad_norm": 2.9431425446677606, "learning_rate": 5.135472863796157e-06, "loss": 0.1308, "step": 7449 }, { "epoch": 1.01, "grad_norm": 3.8650620649566476, "learning_rate": 5.13437172471962e-06, "loss": 0.2085, "step": 7450 }, { "epoch": 1.01, "grad_norm": 3.913178073434799, "learning_rate": 5.133270579121269e-06, "loss": 0.2114, "step": 7451 }, { "epoch": 1.01, "grad_norm": 4.142601089116117, "learning_rate": 5.1321694270545455e-06, "loss": 0.2167, "step": 7452 }, { "epoch": 1.01, "grad_norm": 4.070042288966545, "learning_rate": 5.131068268572895e-06, "loss": 0.1966, "step": 7453 }, { "epoch": 1.01, "grad_norm": 3.4171236295972034, "learning_rate": 5.1299671037297636e-06, "loss": 0.1837, "step": 7454 }, { "epoch": 1.01, "grad_norm": 4.150122375198186, "learning_rate": 5.128865932578597e-06, "loss": 0.1905, "step": 7455 }, { "epoch": 1.01, "grad_norm": 3.671360169601104, "learning_rate": 5.12776475517284e-06, "loss": 0.1858, "step": 7456 }, { "epoch": 1.01, "grad_norm": 3.4677296064722443, "learning_rate": 5.12666357156594e-06, "loss": 0.1675, "step": 7457 }, { "epoch": 1.01, "grad_norm": 3.5807842023313583, "learning_rate": 5.1255623818113435e-06, "loss": 0.1712, "step": 7458 }, { "epoch": 1.01, "grad_norm": 3.9821246070561798, "learning_rate": 5.124461185962497e-06, "loss": 0.2212, "step": 7459 }, { "epoch": 1.01, "grad_norm": 22.110258087567964, "learning_rate": 5.1233599840728495e-06, "loss": 0.1613, "step": 7460 }, { "epoch": 1.01, "grad_norm": 2.8127158743866065, "learning_rate": 5.122258776195846e-06, "loss": 0.1939, "step": 7461 }, { "epoch": 1.02, "grad_norm": 3.3072354684331877, "learning_rate": 5.121157562384936e-06, "loss": 0.1501, "step": 7462 }, { "epoch": 1.02, "grad_norm": 6.02833593156409, "learning_rate": 5.120056342693566e-06, "loss": 0.202, "step": 7463 }, { "epoch": 1.02, "grad_norm": 3.863067202608734, "learning_rate": 5.118955117175186e-06, "loss": 0.1765, "step": 7464 }, { "epoch": 1.02, "grad_norm": 3.115495935232462, "learning_rate": 5.117853885883245e-06, "loss": 0.1644, "step": 7465 }, { "epoch": 1.02, "grad_norm": 3.202347914810263, "learning_rate": 5.11675264887119e-06, "loss": 0.1493, "step": 7466 }, { "epoch": 1.02, "grad_norm": 3.3443516720437745, "learning_rate": 5.115651406192473e-06, "loss": 0.1435, "step": 7467 }, { "epoch": 1.02, "grad_norm": 3.7510916703304837, "learning_rate": 5.1145501579005405e-06, "loss": 0.1807, "step": 7468 }, { "epoch": 1.02, "grad_norm": 3.1167761849702877, "learning_rate": 5.1134489040488446e-06, "loss": 0.1879, "step": 7469 }, { "epoch": 1.02, "grad_norm": 3.0008856520456004, "learning_rate": 5.112347644690834e-06, "loss": 0.1859, "step": 7470 }, { "epoch": 1.02, "grad_norm": 2.7135528169531025, "learning_rate": 5.111246379879958e-06, "loss": 0.1312, "step": 7471 }, { "epoch": 1.02, "grad_norm": 3.1727126744710095, "learning_rate": 5.110145109669671e-06, "loss": 0.1532, "step": 7472 }, { "epoch": 1.02, "grad_norm": 3.485300573357951, "learning_rate": 5.10904383411342e-06, "loss": 0.2013, "step": 7473 }, { "epoch": 1.02, "grad_norm": 2.9486541530569745, "learning_rate": 5.107942553264658e-06, "loss": 0.1524, "step": 7474 }, { "epoch": 1.02, "grad_norm": 2.8174870645551717, "learning_rate": 5.106841267176837e-06, "loss": 0.1727, "step": 7475 }, { "epoch": 1.02, "grad_norm": 4.076145158406884, "learning_rate": 5.105739975903405e-06, "loss": 0.2135, "step": 7476 }, { "epoch": 1.02, "grad_norm": 3.7036369752005722, "learning_rate": 5.104638679497818e-06, "loss": 0.1679, "step": 7477 }, { "epoch": 1.02, "grad_norm": 3.868592607764868, "learning_rate": 5.103537378013526e-06, "loss": 0.2192, "step": 7478 }, { "epoch": 1.02, "grad_norm": 2.825934687297598, "learning_rate": 5.102436071503983e-06, "loss": 0.133, "step": 7479 }, { "epoch": 1.02, "grad_norm": 4.084981159119862, "learning_rate": 5.101334760022639e-06, "loss": 0.1803, "step": 7480 }, { "epoch": 1.02, "grad_norm": 3.791485287458514, "learning_rate": 5.100233443622948e-06, "loss": 0.1755, "step": 7481 }, { "epoch": 1.02, "grad_norm": 3.940445376594086, "learning_rate": 5.0991321223583655e-06, "loss": 0.2263, "step": 7482 }, { "epoch": 1.02, "grad_norm": 5.0048891605413806, "learning_rate": 5.09803079628234e-06, "loss": 0.1971, "step": 7483 }, { "epoch": 1.02, "grad_norm": 3.0088006885991385, "learning_rate": 5.09692946544833e-06, "loss": 0.1623, "step": 7484 }, { "epoch": 1.02, "grad_norm": 3.4283228865297675, "learning_rate": 5.0958281299097865e-06, "loss": 0.1821, "step": 7485 }, { "epoch": 1.02, "grad_norm": 2.954166234538328, "learning_rate": 5.0947267897201646e-06, "loss": 0.1753, "step": 7486 }, { "epoch": 1.02, "grad_norm": 3.88589413812273, "learning_rate": 5.093625444932917e-06, "loss": 0.2079, "step": 7487 }, { "epoch": 1.02, "grad_norm": 3.315818002799784, "learning_rate": 5.0925240956015e-06, "loss": 0.1862, "step": 7488 }, { "epoch": 1.02, "grad_norm": 3.4099433811595676, "learning_rate": 5.091422741779367e-06, "loss": 0.1747, "step": 7489 }, { "epoch": 1.02, "grad_norm": 3.0686093663957195, "learning_rate": 5.090321383519973e-06, "loss": 0.1618, "step": 7490 }, { "epoch": 1.02, "grad_norm": 3.5762493950170002, "learning_rate": 5.089220020876775e-06, "loss": 0.1723, "step": 7491 }, { "epoch": 1.02, "grad_norm": 3.550914906940341, "learning_rate": 5.088118653903225e-06, "loss": 0.1853, "step": 7492 }, { "epoch": 1.02, "grad_norm": 3.0638382741117693, "learning_rate": 5.087017282652782e-06, "loss": 0.1932, "step": 7493 }, { "epoch": 1.02, "grad_norm": 3.3455316055188393, "learning_rate": 5.0859159071789e-06, "loss": 0.1994, "step": 7494 }, { "epoch": 1.02, "grad_norm": 4.2840594899035676, "learning_rate": 5.084814527535035e-06, "loss": 0.1872, "step": 7495 }, { "epoch": 1.02, "grad_norm": 3.3344477994136987, "learning_rate": 5.083713143774643e-06, "loss": 0.1808, "step": 7496 }, { "epoch": 1.02, "grad_norm": 4.145846031773107, "learning_rate": 5.08261175595118e-06, "loss": 0.1872, "step": 7497 }, { "epoch": 1.02, "grad_norm": 3.5178902184317113, "learning_rate": 5.081510364118104e-06, "loss": 0.1901, "step": 7498 }, { "epoch": 1.02, "grad_norm": 3.6229311573701324, "learning_rate": 5.080408968328871e-06, "loss": 0.1675, "step": 7499 }, { "epoch": 1.02, "grad_norm": 3.39270148228864, "learning_rate": 5.079307568636938e-06, "loss": 0.1799, "step": 7500 }, { "epoch": 1.02, "grad_norm": 3.6634062988591176, "learning_rate": 5.0782061650957615e-06, "loss": 0.198, "step": 7501 }, { "epoch": 1.02, "grad_norm": 3.569669868173138, "learning_rate": 5.0771047577587995e-06, "loss": 0.1683, "step": 7502 }, { "epoch": 1.02, "grad_norm": 3.807402260559991, "learning_rate": 5.076003346679508e-06, "loss": 0.1881, "step": 7503 }, { "epoch": 1.02, "grad_norm": 3.2377410003396525, "learning_rate": 5.074901931911348e-06, "loss": 0.1568, "step": 7504 }, { "epoch": 1.02, "grad_norm": 3.741180689443553, "learning_rate": 5.073800513507775e-06, "loss": 0.1667, "step": 7505 }, { "epoch": 1.02, "grad_norm": 3.686661327312419, "learning_rate": 5.0726990915222466e-06, "loss": 0.1926, "step": 7506 }, { "epoch": 1.02, "grad_norm": 4.432468754798908, "learning_rate": 5.071597666008223e-06, "loss": 0.1639, "step": 7507 }, { "epoch": 1.02, "grad_norm": 3.397188434191225, "learning_rate": 5.0704962370191605e-06, "loss": 0.1937, "step": 7508 }, { "epoch": 1.02, "grad_norm": 3.2531094420066333, "learning_rate": 5.069394804608518e-06, "loss": 0.1418, "step": 7509 }, { "epoch": 1.02, "grad_norm": 3.4275261317963643, "learning_rate": 5.068293368829755e-06, "loss": 0.2094, "step": 7510 }, { "epoch": 1.02, "grad_norm": 2.887688353438669, "learning_rate": 5.067191929736331e-06, "loss": 0.1215, "step": 7511 }, { "epoch": 1.02, "grad_norm": 3.548954609414749, "learning_rate": 5.066090487381705e-06, "loss": 0.207, "step": 7512 }, { "epoch": 1.02, "grad_norm": 3.07408531630127, "learning_rate": 5.064989041819335e-06, "loss": 0.1571, "step": 7513 }, { "epoch": 1.02, "grad_norm": 3.5258024308806943, "learning_rate": 5.06388759310268e-06, "loss": 0.1665, "step": 7514 }, { "epoch": 1.02, "grad_norm": 2.932330863372142, "learning_rate": 5.0627861412852e-06, "loss": 0.1663, "step": 7515 }, { "epoch": 1.02, "grad_norm": 3.6736619297814754, "learning_rate": 5.0616846864203535e-06, "loss": 0.2407, "step": 7516 }, { "epoch": 1.02, "grad_norm": 3.887852135740344, "learning_rate": 5.060583228561604e-06, "loss": 0.2148, "step": 7517 }, { "epoch": 1.02, "grad_norm": 3.8364732513105118, "learning_rate": 5.059481767762409e-06, "loss": 0.1501, "step": 7518 }, { "epoch": 1.02, "grad_norm": 3.4656050820118374, "learning_rate": 5.058380304076229e-06, "loss": 0.2291, "step": 7519 }, { "epoch": 1.02, "grad_norm": 2.9440912043305656, "learning_rate": 5.057278837556524e-06, "loss": 0.158, "step": 7520 }, { "epoch": 1.02, "grad_norm": 3.8104590458363994, "learning_rate": 5.0561773682567535e-06, "loss": 0.2082, "step": 7521 }, { "epoch": 1.02, "grad_norm": 2.7068358489582365, "learning_rate": 5.055075896230379e-06, "loss": 0.1716, "step": 7522 }, { "epoch": 1.02, "grad_norm": 3.5862608596457966, "learning_rate": 5.053974421530861e-06, "loss": 0.1829, "step": 7523 }, { "epoch": 1.02, "grad_norm": 3.6832480445260396, "learning_rate": 5.0528729442116605e-06, "loss": 0.202, "step": 7524 }, { "epoch": 1.02, "grad_norm": 3.446000003019011, "learning_rate": 5.0517714643262384e-06, "loss": 0.1911, "step": 7525 }, { "epoch": 1.02, "grad_norm": 4.020768015478963, "learning_rate": 5.050669981928056e-06, "loss": 0.1633, "step": 7526 }, { "epoch": 1.02, "grad_norm": 3.3471493669313697, "learning_rate": 5.0495684970705725e-06, "loss": 0.2004, "step": 7527 }, { "epoch": 1.02, "grad_norm": 3.3482961833785905, "learning_rate": 5.048467009807253e-06, "loss": 0.1894, "step": 7528 }, { "epoch": 1.02, "grad_norm": 3.1493016443857167, "learning_rate": 5.0473655201915536e-06, "loss": 0.1452, "step": 7529 }, { "epoch": 1.02, "grad_norm": 8.57876586590602, "learning_rate": 5.046264028276941e-06, "loss": 0.1929, "step": 7530 }, { "epoch": 1.02, "grad_norm": 2.704726186635844, "learning_rate": 5.045162534116873e-06, "loss": 0.1397, "step": 7531 }, { "epoch": 1.02, "grad_norm": 3.268764964465726, "learning_rate": 5.044061037764814e-06, "loss": 0.1786, "step": 7532 }, { "epoch": 1.02, "grad_norm": 3.38649452548677, "learning_rate": 5.042959539274225e-06, "loss": 0.1885, "step": 7533 }, { "epoch": 1.02, "grad_norm": 3.1433953651966644, "learning_rate": 5.041858038698567e-06, "loss": 0.1566, "step": 7534 }, { "epoch": 1.02, "grad_norm": 3.0207917945658616, "learning_rate": 5.0407565360913025e-06, "loss": 0.1654, "step": 7535 }, { "epoch": 1.03, "grad_norm": 3.4832636466691516, "learning_rate": 5.039655031505892e-06, "loss": 0.2098, "step": 7536 }, { "epoch": 1.03, "grad_norm": 3.465270699537622, "learning_rate": 5.0385535249958015e-06, "loss": 0.1842, "step": 7537 }, { "epoch": 1.03, "grad_norm": 4.061160427406062, "learning_rate": 5.037452016614491e-06, "loss": 0.167, "step": 7538 }, { "epoch": 1.03, "grad_norm": 3.2751399789675757, "learning_rate": 5.0363505064154235e-06, "loss": 0.2284, "step": 7539 }, { "epoch": 1.03, "grad_norm": 3.3613110193051186, "learning_rate": 5.035248994452061e-06, "loss": 0.1755, "step": 7540 }, { "epoch": 1.03, "grad_norm": 3.656983753264975, "learning_rate": 5.034147480777867e-06, "loss": 0.178, "step": 7541 }, { "epoch": 1.03, "grad_norm": 3.539235341600329, "learning_rate": 5.033045965446303e-06, "loss": 0.1951, "step": 7542 }, { "epoch": 1.03, "grad_norm": 4.019446879116782, "learning_rate": 5.031944448510831e-06, "loss": 0.1977, "step": 7543 }, { "epoch": 1.03, "grad_norm": 3.9856974707485633, "learning_rate": 5.030842930024915e-06, "loss": 0.1792, "step": 7544 }, { "epoch": 1.03, "grad_norm": 3.498448943011798, "learning_rate": 5.029741410042019e-06, "loss": 0.1851, "step": 7545 }, { "epoch": 1.03, "grad_norm": 3.260506267249716, "learning_rate": 5.028639888615604e-06, "loss": 0.2145, "step": 7546 }, { "epoch": 1.03, "grad_norm": 3.029071673837048, "learning_rate": 5.027538365799135e-06, "loss": 0.2025, "step": 7547 }, { "epoch": 1.03, "grad_norm": 3.46195088322552, "learning_rate": 5.026436841646074e-06, "loss": 0.1848, "step": 7548 }, { "epoch": 1.03, "grad_norm": 3.061180238649803, "learning_rate": 5.025335316209883e-06, "loss": 0.1898, "step": 7549 }, { "epoch": 1.03, "grad_norm": 3.000411298277338, "learning_rate": 5.024233789544026e-06, "loss": 0.1582, "step": 7550 }, { "epoch": 1.03, "grad_norm": 3.1587190101298828, "learning_rate": 5.023132261701969e-06, "loss": 0.1681, "step": 7551 }, { "epoch": 1.03, "grad_norm": 4.516266897152448, "learning_rate": 5.022030732737172e-06, "loss": 0.2239, "step": 7552 }, { "epoch": 1.03, "grad_norm": 2.981847460273148, "learning_rate": 5.020929202703099e-06, "loss": 0.154, "step": 7553 }, { "epoch": 1.03, "grad_norm": 3.8103669467847046, "learning_rate": 5.019827671653215e-06, "loss": 0.1461, "step": 7554 }, { "epoch": 1.03, "grad_norm": 3.521691389600576, "learning_rate": 5.01872613964098e-06, "loss": 0.1955, "step": 7555 }, { "epoch": 1.03, "grad_norm": 3.151586331879022, "learning_rate": 5.017624606719862e-06, "loss": 0.1538, "step": 7556 }, { "epoch": 1.03, "grad_norm": 3.938901623558807, "learning_rate": 5.016523072943321e-06, "loss": 0.2137, "step": 7557 }, { "epoch": 1.03, "grad_norm": 3.5281689429385694, "learning_rate": 5.015421538364824e-06, "loss": 0.1878, "step": 7558 }, { "epoch": 1.03, "grad_norm": 3.5018415836995613, "learning_rate": 5.014320003037832e-06, "loss": 0.2352, "step": 7559 }, { "epoch": 1.03, "grad_norm": 4.034865982015675, "learning_rate": 5.01321846701581e-06, "loss": 0.2052, "step": 7560 }, { "epoch": 1.03, "grad_norm": 2.9389702878654806, "learning_rate": 5.012116930352221e-06, "loss": 0.1935, "step": 7561 }, { "epoch": 1.03, "grad_norm": 4.081039744523408, "learning_rate": 5.011015393100529e-06, "loss": 0.167, "step": 7562 }, { "epoch": 1.03, "grad_norm": 3.2686334472506022, "learning_rate": 5.009913855314198e-06, "loss": 0.2072, "step": 7563 }, { "epoch": 1.03, "grad_norm": 2.664596103060852, "learning_rate": 5.008812317046692e-06, "loss": 0.1734, "step": 7564 }, { "epoch": 1.03, "grad_norm": 3.340925635659004, "learning_rate": 5.007710778351474e-06, "loss": 0.1993, "step": 7565 }, { "epoch": 1.03, "grad_norm": 3.6194311058209796, "learning_rate": 5.00660923928201e-06, "loss": 0.1993, "step": 7566 }, { "epoch": 1.03, "grad_norm": 3.4219554410503656, "learning_rate": 5.00550769989176e-06, "loss": 0.146, "step": 7567 }, { "epoch": 1.03, "grad_norm": 4.788178251610908, "learning_rate": 5.004406160234193e-06, "loss": 0.1345, "step": 7568 }, { "epoch": 1.03, "grad_norm": 3.7659112821117535, "learning_rate": 5.0033046203627675e-06, "loss": 0.1981, "step": 7569 }, { "epoch": 1.03, "grad_norm": 3.1289039124383637, "learning_rate": 5.002203080330952e-06, "loss": 0.1864, "step": 7570 }, { "epoch": 1.03, "grad_norm": 3.6050763389468203, "learning_rate": 5.001101540192208e-06, "loss": 0.1552, "step": 7571 }, { "epoch": 1.03, "grad_norm": 3.334158202163251, "learning_rate": 5e-06, "loss": 0.1885, "step": 7572 }, { "epoch": 1.03, "grad_norm": 3.5796410883872123, "learning_rate": 4.998898459807794e-06, "loss": 0.2202, "step": 7573 }, { "epoch": 1.03, "grad_norm": 3.5125830330038053, "learning_rate": 4.9977969196690495e-06, "loss": 0.1994, "step": 7574 }, { "epoch": 1.03, "grad_norm": 4.103708592452068, "learning_rate": 4.996695379637233e-06, "loss": 0.2155, "step": 7575 }, { "epoch": 1.03, "grad_norm": 3.5231266988932557, "learning_rate": 4.995593839765808e-06, "loss": 0.1462, "step": 7576 }, { "epoch": 1.03, "grad_norm": 3.5659005802736576, "learning_rate": 4.994492300108241e-06, "loss": 0.1872, "step": 7577 }, { "epoch": 1.03, "grad_norm": 3.6526457913382333, "learning_rate": 4.993390760717992e-06, "loss": 0.1317, "step": 7578 }, { "epoch": 1.03, "grad_norm": 3.4262335015559393, "learning_rate": 4.992289221648528e-06, "loss": 0.2048, "step": 7579 }, { "epoch": 1.03, "grad_norm": 4.709621760871637, "learning_rate": 4.99118768295331e-06, "loss": 0.2163, "step": 7580 }, { "epoch": 1.03, "grad_norm": 3.3792813185974744, "learning_rate": 4.990086144685803e-06, "loss": 0.1742, "step": 7581 }, { "epoch": 1.03, "grad_norm": 3.9512841958005116, "learning_rate": 4.988984606899473e-06, "loss": 0.1884, "step": 7582 }, { "epoch": 1.03, "grad_norm": 4.21490915779159, "learning_rate": 4.987883069647781e-06, "loss": 0.225, "step": 7583 }, { "epoch": 1.03, "grad_norm": 3.7946215243399997, "learning_rate": 4.986781532984191e-06, "loss": 0.1345, "step": 7584 }, { "epoch": 1.03, "grad_norm": 4.247916711058012, "learning_rate": 4.98567999696217e-06, "loss": 0.2036, "step": 7585 }, { "epoch": 1.03, "grad_norm": 3.3894221308741552, "learning_rate": 4.9845784616351775e-06, "loss": 0.1844, "step": 7586 }, { "epoch": 1.03, "grad_norm": 2.791352228298574, "learning_rate": 4.9834769270566805e-06, "loss": 0.1324, "step": 7587 }, { "epoch": 1.03, "grad_norm": 3.4356192319754806, "learning_rate": 4.98237539328014e-06, "loss": 0.1942, "step": 7588 }, { "epoch": 1.03, "grad_norm": 3.0651832286076686, "learning_rate": 4.981273860359021e-06, "loss": 0.1399, "step": 7589 }, { "epoch": 1.03, "grad_norm": 3.9951165102251864, "learning_rate": 4.980172328346787e-06, "loss": 0.1671, "step": 7590 }, { "epoch": 1.03, "grad_norm": 3.514730549199286, "learning_rate": 4.979070797296903e-06, "loss": 0.1923, "step": 7591 }, { "epoch": 1.03, "grad_norm": 3.8685236282548456, "learning_rate": 4.977969267262829e-06, "loss": 0.1998, "step": 7592 }, { "epoch": 1.03, "grad_norm": 3.274383010625312, "learning_rate": 4.976867738298034e-06, "loss": 0.2352, "step": 7593 }, { "epoch": 1.03, "grad_norm": 3.284608834928669, "learning_rate": 4.975766210455974e-06, "loss": 0.1835, "step": 7594 }, { "epoch": 1.03, "grad_norm": 3.8086962677990366, "learning_rate": 4.974664683790118e-06, "loss": 0.1723, "step": 7595 }, { "epoch": 1.03, "grad_norm": 3.491817285394366, "learning_rate": 4.9735631583539265e-06, "loss": 0.1478, "step": 7596 }, { "epoch": 1.03, "grad_norm": 3.584228762499245, "learning_rate": 4.972461634200866e-06, "loss": 0.2157, "step": 7597 }, { "epoch": 1.03, "grad_norm": 3.5040898567630805, "learning_rate": 4.971360111384396e-06, "loss": 0.1742, "step": 7598 }, { "epoch": 1.03, "grad_norm": 4.442025432187443, "learning_rate": 4.9702585899579825e-06, "loss": 0.1445, "step": 7599 }, { "epoch": 1.03, "grad_norm": 3.780429485728008, "learning_rate": 4.9691570699750855e-06, "loss": 0.1734, "step": 7600 }, { "epoch": 1.03, "grad_norm": 3.130000543112811, "learning_rate": 4.968055551489171e-06, "loss": 0.1724, "step": 7601 }, { "epoch": 1.03, "grad_norm": 4.095420882531934, "learning_rate": 4.966954034553699e-06, "loss": 0.1727, "step": 7602 }, { "epoch": 1.03, "grad_norm": 4.463811279959471, "learning_rate": 4.965852519222135e-06, "loss": 0.1955, "step": 7603 }, { "epoch": 1.03, "grad_norm": 4.4744589980467, "learning_rate": 4.964751005547938e-06, "loss": 0.209, "step": 7604 }, { "epoch": 1.03, "grad_norm": 3.9194515609563587, "learning_rate": 4.963649493584577e-06, "loss": 0.1765, "step": 7605 }, { "epoch": 1.03, "grad_norm": 3.6051015524770134, "learning_rate": 4.962547983385509e-06, "loss": 0.1603, "step": 7606 }, { "epoch": 1.03, "grad_norm": 3.7187368933641913, "learning_rate": 4.961446475004199e-06, "loss": 0.185, "step": 7607 }, { "epoch": 1.03, "grad_norm": 2.99442482332671, "learning_rate": 4.960344968494108e-06, "loss": 0.1695, "step": 7608 }, { "epoch": 1.04, "grad_norm": 3.4735169028831736, "learning_rate": 4.959243463908699e-06, "loss": 0.1509, "step": 7609 }, { "epoch": 1.04, "grad_norm": 2.867329054741006, "learning_rate": 4.958141961301434e-06, "loss": 0.1333, "step": 7610 }, { "epoch": 1.04, "grad_norm": 3.9793978539700197, "learning_rate": 4.957040460725777e-06, "loss": 0.1822, "step": 7611 }, { "epoch": 1.04, "grad_norm": 3.3734081751689846, "learning_rate": 4.955938962235186e-06, "loss": 0.197, "step": 7612 }, { "epoch": 1.04, "grad_norm": 3.9231646083054184, "learning_rate": 4.954837465883129e-06, "loss": 0.1933, "step": 7613 }, { "epoch": 1.04, "grad_norm": 4.937722949851983, "learning_rate": 4.953735971723061e-06, "loss": 0.13, "step": 7614 }, { "epoch": 1.04, "grad_norm": 3.3991973323910174, "learning_rate": 4.952634479808448e-06, "loss": 0.1664, "step": 7615 }, { "epoch": 1.04, "grad_norm": 3.235820308636576, "learning_rate": 4.95153299019275e-06, "loss": 0.203, "step": 7616 }, { "epoch": 1.04, "grad_norm": 4.4497808740077955, "learning_rate": 4.950431502929428e-06, "loss": 0.2112, "step": 7617 }, { "epoch": 1.04, "grad_norm": 3.9800689352090526, "learning_rate": 4.949330018071947e-06, "loss": 0.1319, "step": 7618 }, { "epoch": 1.04, "grad_norm": 3.5515348422862094, "learning_rate": 4.948228535673763e-06, "loss": 0.1195, "step": 7619 }, { "epoch": 1.04, "grad_norm": 3.799744238636778, "learning_rate": 4.947127055788342e-06, "loss": 0.177, "step": 7620 }, { "epoch": 1.04, "grad_norm": 3.460628385969619, "learning_rate": 4.946025578469141e-06, "loss": 0.1564, "step": 7621 }, { "epoch": 1.04, "grad_norm": 3.541273430875488, "learning_rate": 4.944924103769623e-06, "loss": 0.1823, "step": 7622 }, { "epoch": 1.04, "grad_norm": 3.9848115815057286, "learning_rate": 4.943822631743248e-06, "loss": 0.1889, "step": 7623 }, { "epoch": 1.04, "grad_norm": 4.151091756910597, "learning_rate": 4.942721162443479e-06, "loss": 0.169, "step": 7624 }, { "epoch": 1.04, "grad_norm": 3.179574286422232, "learning_rate": 4.941619695923773e-06, "loss": 0.1665, "step": 7625 }, { "epoch": 1.04, "grad_norm": 2.926263332923441, "learning_rate": 4.940518232237593e-06, "loss": 0.1363, "step": 7626 }, { "epoch": 1.04, "grad_norm": 4.019822487061127, "learning_rate": 4.939416771438397e-06, "loss": 0.2122, "step": 7627 }, { "epoch": 1.04, "grad_norm": 4.805512076072556, "learning_rate": 4.938315313579647e-06, "loss": 0.1802, "step": 7628 }, { "epoch": 1.04, "grad_norm": 2.886013098404519, "learning_rate": 4.937213858714801e-06, "loss": 0.1577, "step": 7629 }, { "epoch": 1.04, "grad_norm": 2.940447782373196, "learning_rate": 4.9361124068973225e-06, "loss": 0.1802, "step": 7630 }, { "epoch": 1.04, "grad_norm": 3.8137381449811905, "learning_rate": 4.9350109581806665e-06, "loss": 0.2095, "step": 7631 }, { "epoch": 1.04, "grad_norm": 3.8989357661384654, "learning_rate": 4.933909512618298e-06, "loss": 0.1637, "step": 7632 }, { "epoch": 1.04, "grad_norm": 4.0030929632307455, "learning_rate": 4.93280807026367e-06, "loss": 0.14, "step": 7633 }, { "epoch": 1.04, "grad_norm": 3.0155529622806787, "learning_rate": 4.931706631170246e-06, "loss": 0.1626, "step": 7634 }, { "epoch": 1.04, "grad_norm": 3.153527117435354, "learning_rate": 4.930605195391483e-06, "loss": 0.1557, "step": 7635 }, { "epoch": 1.04, "grad_norm": 2.6947264816644148, "learning_rate": 4.929503762980841e-06, "loss": 0.1781, "step": 7636 }, { "epoch": 1.04, "grad_norm": 3.6007147450953263, "learning_rate": 4.928402333991777e-06, "loss": 0.1754, "step": 7637 }, { "epoch": 1.04, "grad_norm": 4.083404075279069, "learning_rate": 4.927300908477754e-06, "loss": 0.1611, "step": 7638 }, { "epoch": 1.04, "grad_norm": 3.293112595179533, "learning_rate": 4.926199486492226e-06, "loss": 0.1603, "step": 7639 }, { "epoch": 1.04, "grad_norm": 3.307691653103007, "learning_rate": 4.925098068088654e-06, "loss": 0.1615, "step": 7640 }, { "epoch": 1.04, "grad_norm": 3.6042264062380402, "learning_rate": 4.923996653320492e-06, "loss": 0.2039, "step": 7641 }, { "epoch": 1.04, "grad_norm": 3.034552277679939, "learning_rate": 4.922895242241202e-06, "loss": 0.1572, "step": 7642 }, { "epoch": 1.04, "grad_norm": 3.786942091155407, "learning_rate": 4.9217938349042385e-06, "loss": 0.2094, "step": 7643 }, { "epoch": 1.04, "grad_norm": 3.7758296515933347, "learning_rate": 4.920692431363064e-06, "loss": 0.1714, "step": 7644 }, { "epoch": 1.04, "grad_norm": 4.545301362944007, "learning_rate": 4.919591031671129e-06, "loss": 0.1902, "step": 7645 }, { "epoch": 1.04, "grad_norm": 3.6556490498852328, "learning_rate": 4.918489635881897e-06, "loss": 0.1985, "step": 7646 }, { "epoch": 1.04, "grad_norm": 3.4352503300603208, "learning_rate": 4.91738824404882e-06, "loss": 0.2128, "step": 7647 }, { "epoch": 1.04, "grad_norm": 3.5681412357361855, "learning_rate": 4.916286856225358e-06, "loss": 0.158, "step": 7648 }, { "epoch": 1.04, "grad_norm": 3.5284262515220854, "learning_rate": 4.915185472464966e-06, "loss": 0.1583, "step": 7649 }, { "epoch": 1.04, "grad_norm": 3.344116587945537, "learning_rate": 4.9140840928211014e-06, "loss": 0.165, "step": 7650 }, { "epoch": 1.04, "grad_norm": 3.5531518030639937, "learning_rate": 4.9129827173472175e-06, "loss": 0.1897, "step": 7651 }, { "epoch": 1.04, "grad_norm": 3.591618269124747, "learning_rate": 4.9118813460967754e-06, "loss": 0.1796, "step": 7652 }, { "epoch": 1.04, "grad_norm": 3.3207197900467853, "learning_rate": 4.910779979123228e-06, "loss": 0.1935, "step": 7653 }, { "epoch": 1.04, "grad_norm": 3.7268857072586616, "learning_rate": 4.909678616480028e-06, "loss": 0.2129, "step": 7654 }, { "epoch": 1.04, "grad_norm": 3.6696463608093466, "learning_rate": 4.9085772582206345e-06, "loss": 0.1752, "step": 7655 }, { "epoch": 1.04, "grad_norm": 3.2895261798409345, "learning_rate": 4.907475904398502e-06, "loss": 0.1437, "step": 7656 }, { "epoch": 1.04, "grad_norm": 3.680578966992358, "learning_rate": 4.906374555067085e-06, "loss": 0.172, "step": 7657 }, { "epoch": 1.04, "grad_norm": 3.398415243814765, "learning_rate": 4.905273210279837e-06, "loss": 0.2135, "step": 7658 }, { "epoch": 1.04, "grad_norm": 3.6068108802209364, "learning_rate": 4.904171870090216e-06, "loss": 0.2107, "step": 7659 }, { "epoch": 1.04, "grad_norm": 2.9658056583234735, "learning_rate": 4.9030705345516715e-06, "loss": 0.1941, "step": 7660 }, { "epoch": 1.04, "grad_norm": 2.951147885235272, "learning_rate": 4.9019692037176605e-06, "loss": 0.1744, "step": 7661 }, { "epoch": 1.04, "grad_norm": 4.04429731860773, "learning_rate": 4.900867877641636e-06, "loss": 0.1889, "step": 7662 }, { "epoch": 1.04, "grad_norm": 3.5082641708756954, "learning_rate": 4.899766556377054e-06, "loss": 0.16, "step": 7663 }, { "epoch": 1.04, "grad_norm": 3.1020082885854032, "learning_rate": 4.8986652399773625e-06, "loss": 0.1377, "step": 7664 }, { "epoch": 1.04, "grad_norm": 3.304458374514646, "learning_rate": 4.89756392849602e-06, "loss": 0.2014, "step": 7665 }, { "epoch": 1.04, "grad_norm": 2.8663148460300136, "learning_rate": 4.896462621986475e-06, "loss": 0.1564, "step": 7666 }, { "epoch": 1.04, "grad_norm": 3.8901435753884077, "learning_rate": 4.895361320502185e-06, "loss": 0.2026, "step": 7667 }, { "epoch": 1.04, "grad_norm": 3.6152670290589883, "learning_rate": 4.894260024096596e-06, "loss": 0.1979, "step": 7668 }, { "epoch": 1.04, "grad_norm": 3.7356590404274996, "learning_rate": 4.893158732823166e-06, "loss": 0.1771, "step": 7669 }, { "epoch": 1.04, "grad_norm": 3.063122083286444, "learning_rate": 4.8920574467353425e-06, "loss": 0.1683, "step": 7670 }, { "epoch": 1.04, "grad_norm": 3.2062229691416295, "learning_rate": 4.890956165886581e-06, "loss": 0.1601, "step": 7671 }, { "epoch": 1.04, "grad_norm": 2.696118270858077, "learning_rate": 4.88985489033033e-06, "loss": 0.1644, "step": 7672 }, { "epoch": 1.04, "grad_norm": 3.5037859774384192, "learning_rate": 4.888753620120043e-06, "loss": 0.1732, "step": 7673 }, { "epoch": 1.04, "grad_norm": 4.031505017091345, "learning_rate": 4.887652355309168e-06, "loss": 0.2059, "step": 7674 }, { "epoch": 1.04, "grad_norm": 3.140396407007228, "learning_rate": 4.886551095951156e-06, "loss": 0.1689, "step": 7675 }, { "epoch": 1.04, "grad_norm": 2.929558066296402, "learning_rate": 4.885449842099459e-06, "loss": 0.1832, "step": 7676 }, { "epoch": 1.04, "grad_norm": 2.3316198557237438, "learning_rate": 4.8843485938075286e-06, "loss": 0.1405, "step": 7677 }, { "epoch": 1.04, "grad_norm": 3.499593080768129, "learning_rate": 4.88324735112881e-06, "loss": 0.1567, "step": 7678 }, { "epoch": 1.04, "grad_norm": 3.9276312096556265, "learning_rate": 4.882146114116756e-06, "loss": 0.1713, "step": 7679 }, { "epoch": 1.04, "grad_norm": 4.672708805479629, "learning_rate": 4.881044882824814e-06, "loss": 0.217, "step": 7680 }, { "epoch": 1.04, "grad_norm": 3.0426702181531855, "learning_rate": 4.8799436573064344e-06, "loss": 0.1774, "step": 7681 }, { "epoch": 1.04, "grad_norm": 4.077134468573316, "learning_rate": 4.878842437615065e-06, "loss": 0.1625, "step": 7682 }, { "epoch": 1.05, "grad_norm": 3.851999630197205, "learning_rate": 4.877741223804155e-06, "loss": 0.181, "step": 7683 }, { "epoch": 1.05, "grad_norm": 5.385340891164031, "learning_rate": 4.876640015927151e-06, "loss": 0.2344, "step": 7684 }, { "epoch": 1.05, "grad_norm": 4.2226019020726815, "learning_rate": 4.875538814037504e-06, "loss": 0.1472, "step": 7685 }, { "epoch": 1.05, "grad_norm": 3.1390486062812344, "learning_rate": 4.8744376181886564e-06, "loss": 0.1473, "step": 7686 }, { "epoch": 1.05, "grad_norm": 2.8576817515757713, "learning_rate": 4.873336428434062e-06, "loss": 0.1648, "step": 7687 }, { "epoch": 1.05, "grad_norm": 4.201109935921006, "learning_rate": 4.872235244827161e-06, "loss": 0.2253, "step": 7688 }, { "epoch": 1.05, "grad_norm": 3.2030428265313886, "learning_rate": 4.871134067421404e-06, "loss": 0.1655, "step": 7689 }, { "epoch": 1.05, "grad_norm": 3.9154576395509673, "learning_rate": 4.870032896270236e-06, "loss": 0.2001, "step": 7690 }, { "epoch": 1.05, "grad_norm": 3.553047439388201, "learning_rate": 4.8689317314271055e-06, "loss": 0.2059, "step": 7691 }, { "epoch": 1.05, "grad_norm": 3.1492988651565126, "learning_rate": 4.8678305729454545e-06, "loss": 0.1836, "step": 7692 }, { "epoch": 1.05, "grad_norm": 3.8504414180102144, "learning_rate": 4.866729420878732e-06, "loss": 0.1749, "step": 7693 }, { "epoch": 1.05, "grad_norm": 3.7293153924851454, "learning_rate": 4.86562827528038e-06, "loss": 0.2249, "step": 7694 }, { "epoch": 1.05, "grad_norm": 4.0103013745888845, "learning_rate": 4.864527136203845e-06, "loss": 0.2133, "step": 7695 }, { "epoch": 1.05, "grad_norm": 4.54751884863037, "learning_rate": 4.863426003702572e-06, "loss": 0.1535, "step": 7696 }, { "epoch": 1.05, "grad_norm": 2.8520569202701407, "learning_rate": 4.862324877830003e-06, "loss": 0.0962, "step": 7697 }, { "epoch": 1.05, "grad_norm": 3.8162496684801503, "learning_rate": 4.861223758639584e-06, "loss": 0.1755, "step": 7698 }, { "epoch": 1.05, "grad_norm": 3.5806659604677122, "learning_rate": 4.860122646184756e-06, "loss": 0.1748, "step": 7699 }, { "epoch": 1.05, "grad_norm": 4.331350507863893, "learning_rate": 4.859021540518965e-06, "loss": 0.197, "step": 7700 }, { "epoch": 1.05, "grad_norm": 2.7924110214122555, "learning_rate": 4.857920441695653e-06, "loss": 0.1362, "step": 7701 }, { "epoch": 1.05, "grad_norm": 5.687942631299294, "learning_rate": 4.856819349768262e-06, "loss": 0.1639, "step": 7702 }, { "epoch": 1.05, "grad_norm": 3.771414923532772, "learning_rate": 4.855718264790234e-06, "loss": 0.2328, "step": 7703 }, { "epoch": 1.05, "grad_norm": 3.097936407480054, "learning_rate": 4.854617186815012e-06, "loss": 0.191, "step": 7704 }, { "epoch": 1.05, "grad_norm": 4.768751469055325, "learning_rate": 4.853516115896035e-06, "loss": 0.2044, "step": 7705 }, { "epoch": 1.05, "grad_norm": 3.79448388915816, "learning_rate": 4.852415052086748e-06, "loss": 0.1884, "step": 7706 }, { "epoch": 1.05, "grad_norm": 3.908776242044154, "learning_rate": 4.851313995440589e-06, "loss": 0.199, "step": 7707 }, { "epoch": 1.05, "grad_norm": 3.6389821743200517, "learning_rate": 4.8502129460109995e-06, "loss": 0.1649, "step": 7708 }, { "epoch": 1.05, "grad_norm": 3.7809443066833883, "learning_rate": 4.8491119038514176e-06, "loss": 0.2279, "step": 7709 }, { "epoch": 1.05, "grad_norm": 3.724128694411542, "learning_rate": 4.848010869015288e-06, "loss": 0.1787, "step": 7710 }, { "epoch": 1.05, "grad_norm": 4.0440212334432815, "learning_rate": 4.846909841556044e-06, "loss": 0.184, "step": 7711 }, { "epoch": 1.05, "grad_norm": 3.4824434796376993, "learning_rate": 4.845808821527131e-06, "loss": 0.1729, "step": 7712 }, { "epoch": 1.05, "grad_norm": 3.8575758150886603, "learning_rate": 4.844707808981981e-06, "loss": 0.22, "step": 7713 }, { "epoch": 1.05, "grad_norm": 2.967406919147095, "learning_rate": 4.843606803974037e-06, "loss": 0.1719, "step": 7714 }, { "epoch": 1.05, "grad_norm": 3.1898432461951693, "learning_rate": 4.842505806556736e-06, "loss": 0.176, "step": 7715 }, { "epoch": 1.05, "grad_norm": 4.103373648131052, "learning_rate": 4.841404816783514e-06, "loss": 0.2297, "step": 7716 }, { "epoch": 1.05, "grad_norm": 3.3342780723742913, "learning_rate": 4.840303834707811e-06, "loss": 0.163, "step": 7717 }, { "epoch": 1.05, "grad_norm": 2.698356174401702, "learning_rate": 4.839202860383063e-06, "loss": 0.1708, "step": 7718 }, { "epoch": 1.05, "grad_norm": 4.236419438187128, "learning_rate": 4.838101893862704e-06, "loss": 0.1543, "step": 7719 }, { "epoch": 1.05, "grad_norm": 3.738475754329417, "learning_rate": 4.8370009352001735e-06, "loss": 0.2013, "step": 7720 }, { "epoch": 1.05, "grad_norm": 3.0197540330053494, "learning_rate": 4.835899984448904e-06, "loss": 0.1907, "step": 7721 }, { "epoch": 1.05, "grad_norm": 3.914312616348704, "learning_rate": 4.834799041662333e-06, "loss": 0.1781, "step": 7722 }, { "epoch": 1.05, "grad_norm": 4.095230824798781, "learning_rate": 4.8336981068938956e-06, "loss": 0.1957, "step": 7723 }, { "epoch": 1.05, "grad_norm": 3.7790566317454712, "learning_rate": 4.832597180197026e-06, "loss": 0.1922, "step": 7724 }, { "epoch": 1.05, "grad_norm": 2.485861429996248, "learning_rate": 4.831496261625157e-06, "loss": 0.1226, "step": 7725 }, { "epoch": 1.05, "grad_norm": 4.258665421011555, "learning_rate": 4.8303953512317255e-06, "loss": 0.181, "step": 7726 }, { "epoch": 1.05, "grad_norm": 3.1657618177142965, "learning_rate": 4.829294449070161e-06, "loss": 0.1879, "step": 7727 }, { "epoch": 1.05, "grad_norm": 3.2678701816503213, "learning_rate": 4.828193555193899e-06, "loss": 0.1794, "step": 7728 }, { "epoch": 1.05, "grad_norm": 3.9058186338754997, "learning_rate": 4.82709266965637e-06, "loss": 0.1812, "step": 7729 }, { "epoch": 1.05, "grad_norm": 3.8223178044910733, "learning_rate": 4.825991792511011e-06, "loss": 0.1805, "step": 7730 }, { "epoch": 1.05, "grad_norm": 3.5751184813929404, "learning_rate": 4.824890923811247e-06, "loss": 0.1443, "step": 7731 }, { "epoch": 1.05, "grad_norm": 4.3602522541621935, "learning_rate": 4.8237900636105154e-06, "loss": 0.197, "step": 7732 }, { "epoch": 1.05, "grad_norm": 3.855855253334914, "learning_rate": 4.822689211962243e-06, "loss": 0.2126, "step": 7733 }, { "epoch": 1.05, "grad_norm": 3.5601124126919483, "learning_rate": 4.821588368919862e-06, "loss": 0.1681, "step": 7734 }, { "epoch": 1.05, "grad_norm": 3.7363779996539757, "learning_rate": 4.8204875345368035e-06, "loss": 0.129, "step": 7735 }, { "epoch": 1.05, "grad_norm": 2.691318256750541, "learning_rate": 4.819386708866494e-06, "loss": 0.1454, "step": 7736 }, { "epoch": 1.05, "grad_norm": 4.8664098637252025, "learning_rate": 4.818285891962367e-06, "loss": 0.1844, "step": 7737 }, { "epoch": 1.05, "grad_norm": 3.242986467694443, "learning_rate": 4.817185083877847e-06, "loss": 0.1743, "step": 7738 }, { "epoch": 1.05, "grad_norm": 3.8274587123612998, "learning_rate": 4.8160842846663664e-06, "loss": 0.1967, "step": 7739 }, { "epoch": 1.05, "grad_norm": 5.069347816281265, "learning_rate": 4.81498349438135e-06, "loss": 0.1727, "step": 7740 }, { "epoch": 1.05, "grad_norm": 3.804037290635544, "learning_rate": 4.813882713076227e-06, "loss": 0.1335, "step": 7741 }, { "epoch": 1.05, "grad_norm": 4.888080497405556, "learning_rate": 4.812781940804424e-06, "loss": 0.2182, "step": 7742 }, { "epoch": 1.05, "grad_norm": 3.8122493845645575, "learning_rate": 4.811681177619369e-06, "loss": 0.1868, "step": 7743 }, { "epoch": 1.05, "grad_norm": 3.1862873091004804, "learning_rate": 4.810580423574485e-06, "loss": 0.158, "step": 7744 }, { "epoch": 1.05, "grad_norm": 3.0448455760638753, "learning_rate": 4.809479678723203e-06, "loss": 0.1691, "step": 7745 }, { "epoch": 1.05, "grad_norm": 3.3434325697331992, "learning_rate": 4.808378943118943e-06, "loss": 0.1783, "step": 7746 }, { "epoch": 1.05, "grad_norm": 3.930862274675985, "learning_rate": 4.807278216815132e-06, "loss": 0.1876, "step": 7747 }, { "epoch": 1.05, "grad_norm": 4.598508456605708, "learning_rate": 4.806177499865195e-06, "loss": 0.2077, "step": 7748 }, { "epoch": 1.05, "grad_norm": 3.2852538009734276, "learning_rate": 4.805076792322557e-06, "loss": 0.171, "step": 7749 }, { "epoch": 1.05, "grad_norm": 3.4824550784160784, "learning_rate": 4.803976094240639e-06, "loss": 0.2326, "step": 7750 }, { "epoch": 1.05, "grad_norm": 3.8722714907580347, "learning_rate": 4.802875405672866e-06, "loss": 0.2111, "step": 7751 }, { "epoch": 1.05, "grad_norm": 3.7427418505721906, "learning_rate": 4.801774726672658e-06, "loss": 0.2072, "step": 7752 }, { "epoch": 1.05, "grad_norm": 2.920077156470755, "learning_rate": 4.800674057293442e-06, "loss": 0.1726, "step": 7753 }, { "epoch": 1.05, "grad_norm": 3.040433503023259, "learning_rate": 4.799573397588634e-06, "loss": 0.1509, "step": 7754 }, { "epoch": 1.05, "grad_norm": 3.47045835267721, "learning_rate": 4.798472747611659e-06, "loss": 0.1315, "step": 7755 }, { "epoch": 1.06, "grad_norm": 3.268958694592541, "learning_rate": 4.797372107415935e-06, "loss": 0.1558, "step": 7756 }, { "epoch": 1.06, "grad_norm": 3.9820885524815957, "learning_rate": 4.796271477054887e-06, "loss": 0.2309, "step": 7757 }, { "epoch": 1.06, "grad_norm": 3.743556024472095, "learning_rate": 4.795170856581929e-06, "loss": 0.1622, "step": 7758 }, { "epoch": 1.06, "grad_norm": 3.2042528795358614, "learning_rate": 4.794070246050485e-06, "loss": 0.1788, "step": 7759 }, { "epoch": 1.06, "grad_norm": 3.4394654781168588, "learning_rate": 4.79296964551397e-06, "loss": 0.1458, "step": 7760 }, { "epoch": 1.06, "grad_norm": 2.759173805056433, "learning_rate": 4.791869055025805e-06, "loss": 0.1496, "step": 7761 }, { "epoch": 1.06, "grad_norm": 3.8309140275623466, "learning_rate": 4.790768474639407e-06, "loss": 0.2081, "step": 7762 }, { "epoch": 1.06, "grad_norm": 3.4167782983145494, "learning_rate": 4.789667904408194e-06, "loss": 0.1552, "step": 7763 }, { "epoch": 1.06, "grad_norm": 2.4074054262290714, "learning_rate": 4.788567344385582e-06, "loss": 0.1588, "step": 7764 }, { "epoch": 1.06, "grad_norm": 4.308100572212933, "learning_rate": 4.7874667946249874e-06, "loss": 0.2001, "step": 7765 }, { "epoch": 1.06, "grad_norm": 3.1656396879531172, "learning_rate": 4.786366255179826e-06, "loss": 0.1605, "step": 7766 }, { "epoch": 1.06, "grad_norm": 3.489982945545099, "learning_rate": 4.785265726103514e-06, "loss": 0.209, "step": 7767 }, { "epoch": 1.06, "grad_norm": 3.9813548830699803, "learning_rate": 4.784165207449466e-06, "loss": 0.1791, "step": 7768 }, { "epoch": 1.06, "grad_norm": 3.570067140359181, "learning_rate": 4.783064699271095e-06, "loss": 0.2084, "step": 7769 }, { "epoch": 1.06, "grad_norm": 3.434980524574219, "learning_rate": 4.781964201621816e-06, "loss": 0.204, "step": 7770 }, { "epoch": 1.06, "grad_norm": 4.505936517861693, "learning_rate": 4.780863714555044e-06, "loss": 0.1536, "step": 7771 }, { "epoch": 1.06, "grad_norm": 2.7231794046714604, "learning_rate": 4.77976323812419e-06, "loss": 0.1705, "step": 7772 }, { "epoch": 1.06, "grad_norm": 3.409087722557883, "learning_rate": 4.778662772382666e-06, "loss": 0.1858, "step": 7773 }, { "epoch": 1.06, "grad_norm": 3.558297557029293, "learning_rate": 4.777562317383885e-06, "loss": 0.1721, "step": 7774 }, { "epoch": 1.06, "grad_norm": 4.5375009203036605, "learning_rate": 4.776461873181256e-06, "loss": 0.1701, "step": 7775 }, { "epoch": 1.06, "grad_norm": 3.6121859724087657, "learning_rate": 4.775361439828193e-06, "loss": 0.1833, "step": 7776 }, { "epoch": 1.06, "grad_norm": 3.7322155761000015, "learning_rate": 4.7742610173781025e-06, "loss": 0.2023, "step": 7777 }, { "epoch": 1.06, "grad_norm": 4.346112139943989, "learning_rate": 4.7731606058843986e-06, "loss": 0.2231, "step": 7778 }, { "epoch": 1.06, "grad_norm": 3.381134317017337, "learning_rate": 4.7720602054004865e-06, "loss": 0.1603, "step": 7779 }, { "epoch": 1.06, "grad_norm": 3.210717658395106, "learning_rate": 4.770959815979777e-06, "loss": 0.1825, "step": 7780 }, { "epoch": 1.06, "grad_norm": 3.4239721626660744, "learning_rate": 4.7698594376756765e-06, "loss": 0.1544, "step": 7781 }, { "epoch": 1.06, "grad_norm": 3.27734941938506, "learning_rate": 4.768759070541596e-06, "loss": 0.1523, "step": 7782 }, { "epoch": 1.06, "grad_norm": 4.046125168986535, "learning_rate": 4.767658714630938e-06, "loss": 0.138, "step": 7783 }, { "epoch": 1.06, "grad_norm": 4.077324796315115, "learning_rate": 4.766558369997114e-06, "loss": 0.2125, "step": 7784 }, { "epoch": 1.06, "grad_norm": 4.291680451702757, "learning_rate": 4.7654580366935245e-06, "loss": 0.1304, "step": 7785 }, { "epoch": 1.06, "grad_norm": 3.4675634405688873, "learning_rate": 4.76435771477358e-06, "loss": 0.2101, "step": 7786 }, { "epoch": 1.06, "grad_norm": 3.86811508001649, "learning_rate": 4.76325740429068e-06, "loss": 0.1864, "step": 7787 }, { "epoch": 1.06, "grad_norm": 3.7136330090113887, "learning_rate": 4.762157105298233e-06, "loss": 0.1914, "step": 7788 }, { "epoch": 1.06, "grad_norm": 3.121771226823521, "learning_rate": 4.7610568178496405e-06, "loss": 0.1702, "step": 7789 }, { "epoch": 1.06, "grad_norm": 3.819633849099075, "learning_rate": 4.759956541998309e-06, "loss": 0.2027, "step": 7790 }, { "epoch": 1.06, "grad_norm": 4.376911559188062, "learning_rate": 4.758856277797636e-06, "loss": 0.2253, "step": 7791 }, { "epoch": 1.06, "grad_norm": 4.763449606415976, "learning_rate": 4.7577560253010275e-06, "loss": 0.2144, "step": 7792 }, { "epoch": 1.06, "grad_norm": 3.716380545650803, "learning_rate": 4.756655784561881e-06, "loss": 0.1894, "step": 7793 }, { "epoch": 1.06, "grad_norm": 3.8778218849665165, "learning_rate": 4.755555555633602e-06, "loss": 0.1957, "step": 7794 }, { "epoch": 1.06, "grad_norm": 4.180545821542154, "learning_rate": 4.754455338569588e-06, "loss": 0.2133, "step": 7795 }, { "epoch": 1.06, "grad_norm": 2.898398753023193, "learning_rate": 4.753355133423241e-06, "loss": 0.1574, "step": 7796 }, { "epoch": 1.06, "grad_norm": 2.759185840332088, "learning_rate": 4.752254940247956e-06, "loss": 0.1352, "step": 7797 }, { "epoch": 1.06, "grad_norm": 4.109978471996031, "learning_rate": 4.751154759097137e-06, "loss": 0.2076, "step": 7798 }, { "epoch": 1.06, "grad_norm": 3.327136617996667, "learning_rate": 4.750054590024177e-06, "loss": 0.1847, "step": 7799 }, { "epoch": 1.06, "grad_norm": 3.537638145813789, "learning_rate": 4.748954433082476e-06, "loss": 0.1577, "step": 7800 }, { "epoch": 1.06, "grad_norm": 3.301455458051591, "learning_rate": 4.747854288325431e-06, "loss": 0.1615, "step": 7801 }, { "epoch": 1.06, "grad_norm": 3.4816195185990564, "learning_rate": 4.746754155806437e-06, "loss": 0.172, "step": 7802 }, { "epoch": 1.06, "grad_norm": 3.058234223183594, "learning_rate": 4.745654035578889e-06, "loss": 0.1384, "step": 7803 }, { "epoch": 1.06, "grad_norm": 3.170410592891377, "learning_rate": 4.744553927696186e-06, "loss": 0.172, "step": 7804 }, { "epoch": 1.06, "grad_norm": 3.1534392536432563, "learning_rate": 4.743453832211717e-06, "loss": 0.1559, "step": 7805 }, { "epoch": 1.06, "grad_norm": 3.328515515635191, "learning_rate": 4.742353749178881e-06, "loss": 0.1579, "step": 7806 }, { "epoch": 1.06, "grad_norm": 3.858583584445215, "learning_rate": 4.741253678651067e-06, "loss": 0.1699, "step": 7807 }, { "epoch": 1.06, "grad_norm": 2.7168960269074613, "learning_rate": 4.74015362068167e-06, "loss": 0.1226, "step": 7808 }, { "epoch": 1.06, "grad_norm": 3.396223537660466, "learning_rate": 4.73905357532408e-06, "loss": 0.1918, "step": 7809 }, { "epoch": 1.06, "grad_norm": 3.729117533414145, "learning_rate": 4.737953542631692e-06, "loss": 0.1803, "step": 7810 }, { "epoch": 1.06, "grad_norm": 3.4280392377596893, "learning_rate": 4.736853522657892e-06, "loss": 0.1681, "step": 7811 }, { "epoch": 1.06, "grad_norm": 3.157886026407484, "learning_rate": 4.735753515456076e-06, "loss": 0.1519, "step": 7812 }, { "epoch": 1.06, "grad_norm": 3.5582196759113547, "learning_rate": 4.734653521079628e-06, "loss": 0.1844, "step": 7813 }, { "epoch": 1.06, "grad_norm": 2.8957833846643086, "learning_rate": 4.733553539581939e-06, "loss": 0.1585, "step": 7814 }, { "epoch": 1.06, "grad_norm": 2.903280244205809, "learning_rate": 4.732453571016399e-06, "loss": 0.2034, "step": 7815 }, { "epoch": 1.06, "grad_norm": 3.220766906394056, "learning_rate": 4.7313536154363915e-06, "loss": 0.1638, "step": 7816 }, { "epoch": 1.06, "grad_norm": 3.281922392096789, "learning_rate": 4.7302536728953095e-06, "loss": 0.1447, "step": 7817 }, { "epoch": 1.06, "grad_norm": 4.168208259995277, "learning_rate": 4.729153743446533e-06, "loss": 0.1997, "step": 7818 }, { "epoch": 1.06, "grad_norm": 2.9060706108138463, "learning_rate": 4.728053827143453e-06, "loss": 0.1912, "step": 7819 }, { "epoch": 1.06, "grad_norm": 3.1896580844086504, "learning_rate": 4.7269539240394505e-06, "loss": 0.1784, "step": 7820 }, { "epoch": 1.06, "grad_norm": 3.805212166840949, "learning_rate": 4.725854034187913e-06, "loss": 0.2306, "step": 7821 }, { "epoch": 1.06, "grad_norm": 3.588361370101843, "learning_rate": 4.724754157642223e-06, "loss": 0.1305, "step": 7822 }, { "epoch": 1.06, "grad_norm": 3.9420175934562263, "learning_rate": 4.723654294455765e-06, "loss": 0.2067, "step": 7823 }, { "epoch": 1.06, "grad_norm": 3.2924419429116734, "learning_rate": 4.72255444468192e-06, "loss": 0.1562, "step": 7824 }, { "epoch": 1.06, "grad_norm": 3.105110601397572, "learning_rate": 4.721454608374071e-06, "loss": 0.1427, "step": 7825 }, { "epoch": 1.06, "grad_norm": 3.3897117980007776, "learning_rate": 4.7203547855855974e-06, "loss": 0.1514, "step": 7826 }, { "epoch": 1.06, "grad_norm": 4.633859142703539, "learning_rate": 4.719254976369882e-06, "loss": 0.2081, "step": 7827 }, { "epoch": 1.06, "grad_norm": 3.8945214786919418, "learning_rate": 4.7181551807803025e-06, "loss": 0.1909, "step": 7828 }, { "epoch": 1.06, "grad_norm": 3.9116781158686655, "learning_rate": 4.717055398870243e-06, "loss": 0.1379, "step": 7829 }, { "epoch": 1.07, "grad_norm": 3.5426913400331066, "learning_rate": 4.715955630693075e-06, "loss": 0.1946, "step": 7830 }, { "epoch": 1.07, "grad_norm": 3.7347126527461545, "learning_rate": 4.714855876302183e-06, "loss": 0.1851, "step": 7831 }, { "epoch": 1.07, "grad_norm": 3.3433154518927743, "learning_rate": 4.713756135750939e-06, "loss": 0.2019, "step": 7832 }, { "epoch": 1.07, "grad_norm": 4.491094260308356, "learning_rate": 4.7126564090927225e-06, "loss": 0.224, "step": 7833 }, { "epoch": 1.07, "grad_norm": 3.158226427314637, "learning_rate": 4.711556696380909e-06, "loss": 0.137, "step": 7834 }, { "epoch": 1.07, "grad_norm": 3.732771102070206, "learning_rate": 4.710456997668875e-06, "loss": 0.1317, "step": 7835 }, { "epoch": 1.07, "grad_norm": 4.336885956964059, "learning_rate": 4.709357313009991e-06, "loss": 0.2075, "step": 7836 }, { "epoch": 1.07, "grad_norm": 2.8877884821631525, "learning_rate": 4.708257642457637e-06, "loss": 0.1536, "step": 7837 }, { "epoch": 1.07, "grad_norm": 3.175102403823832, "learning_rate": 4.70715798606518e-06, "loss": 0.1523, "step": 7838 }, { "epoch": 1.07, "grad_norm": 3.2537587838947393, "learning_rate": 4.706058343885997e-06, "loss": 0.1189, "step": 7839 }, { "epoch": 1.07, "grad_norm": 3.58570241992191, "learning_rate": 4.704958715973457e-06, "loss": 0.1899, "step": 7840 }, { "epoch": 1.07, "grad_norm": 4.31704137156166, "learning_rate": 4.703859102380934e-06, "loss": 0.1801, "step": 7841 }, { "epoch": 1.07, "grad_norm": 3.900996570805838, "learning_rate": 4.702759503161794e-06, "loss": 0.1682, "step": 7842 }, { "epoch": 1.07, "grad_norm": 3.5695890457683124, "learning_rate": 4.701659918369412e-06, "loss": 0.1695, "step": 7843 }, { "epoch": 1.07, "grad_norm": 3.2465203584379867, "learning_rate": 4.700560348057153e-06, "loss": 0.1575, "step": 7844 }, { "epoch": 1.07, "grad_norm": 4.074931578807678, "learning_rate": 4.699460792278389e-06, "loss": 0.1987, "step": 7845 }, { "epoch": 1.07, "grad_norm": 3.6197437466439615, "learning_rate": 4.698361251086483e-06, "loss": 0.1463, "step": 7846 }, { "epoch": 1.07, "grad_norm": 4.263802937768237, "learning_rate": 4.697261724534805e-06, "loss": 0.2126, "step": 7847 }, { "epoch": 1.07, "grad_norm": 3.4928593664366274, "learning_rate": 4.696162212676721e-06, "loss": 0.1711, "step": 7848 }, { "epoch": 1.07, "grad_norm": 4.211853824037259, "learning_rate": 4.695062715565597e-06, "loss": 0.1931, "step": 7849 }, { "epoch": 1.07, "grad_norm": 5.849544698039218, "learning_rate": 4.693963233254795e-06, "loss": 0.192, "step": 7850 }, { "epoch": 1.07, "grad_norm": 4.3623234248786575, "learning_rate": 4.692863765797683e-06, "loss": 0.2364, "step": 7851 }, { "epoch": 1.07, "grad_norm": 3.902754884886874, "learning_rate": 4.691764313247621e-06, "loss": 0.1818, "step": 7852 }, { "epoch": 1.07, "grad_norm": 2.8802032967891624, "learning_rate": 4.690664875657975e-06, "loss": 0.1495, "step": 7853 }, { "epoch": 1.07, "grad_norm": 3.5022997886995197, "learning_rate": 4.689565453082104e-06, "loss": 0.1733, "step": 7854 }, { "epoch": 1.07, "grad_norm": 3.8188261836091226, "learning_rate": 4.688466045573368e-06, "loss": 0.1712, "step": 7855 }, { "epoch": 1.07, "grad_norm": 4.066923490847948, "learning_rate": 4.687366653185132e-06, "loss": 0.1744, "step": 7856 }, { "epoch": 1.07, "grad_norm": 5.338400852303163, "learning_rate": 4.686267275970751e-06, "loss": 0.2104, "step": 7857 }, { "epoch": 1.07, "grad_norm": 4.51389132044729, "learning_rate": 4.685167913983589e-06, "loss": 0.177, "step": 7858 }, { "epoch": 1.07, "grad_norm": 4.91237747515091, "learning_rate": 4.684068567276999e-06, "loss": 0.1977, "step": 7859 }, { "epoch": 1.07, "grad_norm": 3.9453460560101212, "learning_rate": 4.682969235904341e-06, "loss": 0.184, "step": 7860 }, { "epoch": 1.07, "grad_norm": 4.003854038824389, "learning_rate": 4.681869919918971e-06, "loss": 0.2239, "step": 7861 }, { "epoch": 1.07, "grad_norm": 2.8657396842977727, "learning_rate": 4.680770619374248e-06, "loss": 0.1108, "step": 7862 }, { "epoch": 1.07, "grad_norm": 3.02448352183645, "learning_rate": 4.679671334323523e-06, "loss": 0.1937, "step": 7863 }, { "epoch": 1.07, "grad_norm": 4.126574887148577, "learning_rate": 4.678572064820155e-06, "loss": 0.1784, "step": 7864 }, { "epoch": 1.07, "grad_norm": 4.179737328118891, "learning_rate": 4.677472810917492e-06, "loss": 0.206, "step": 7865 }, { "epoch": 1.07, "grad_norm": 2.8769631135072617, "learning_rate": 4.676373572668891e-06, "loss": 0.1822, "step": 7866 }, { "epoch": 1.07, "grad_norm": 3.295675654714549, "learning_rate": 4.675274350127702e-06, "loss": 0.2029, "step": 7867 }, { "epoch": 1.07, "grad_norm": 4.173554239975046, "learning_rate": 4.674175143347281e-06, "loss": 0.1723, "step": 7868 }, { "epoch": 1.07, "grad_norm": 3.3834510252609085, "learning_rate": 4.673075952380974e-06, "loss": 0.166, "step": 7869 }, { "epoch": 1.07, "grad_norm": 4.056695354931802, "learning_rate": 4.6719767772821325e-06, "loss": 0.1817, "step": 7870 }, { "epoch": 1.07, "grad_norm": 3.5574033685628756, "learning_rate": 4.670877618104104e-06, "loss": 0.1991, "step": 7871 }, { "epoch": 1.07, "grad_norm": 4.224533240386411, "learning_rate": 4.669778474900241e-06, "loss": 0.1712, "step": 7872 }, { "epoch": 1.07, "grad_norm": 4.121973997232754, "learning_rate": 4.668679347723887e-06, "loss": 0.1898, "step": 7873 }, { "epoch": 1.07, "grad_norm": 3.586884695358127, "learning_rate": 4.66758023662839e-06, "loss": 0.1621, "step": 7874 }, { "epoch": 1.07, "grad_norm": 3.435969081834268, "learning_rate": 4.666481141667096e-06, "loss": 0.1609, "step": 7875 }, { "epoch": 1.07, "grad_norm": 5.010854294254534, "learning_rate": 4.6653820628933536e-06, "loss": 0.1789, "step": 7876 }, { "epoch": 1.07, "grad_norm": 4.162553059419179, "learning_rate": 4.664283000360501e-06, "loss": 0.2012, "step": 7877 }, { "epoch": 1.07, "grad_norm": 3.369305544856499, "learning_rate": 4.663183954121889e-06, "loss": 0.2014, "step": 7878 }, { "epoch": 1.07, "grad_norm": 4.043249718539066, "learning_rate": 4.662084924230853e-06, "loss": 0.1922, "step": 7879 }, { "epoch": 1.07, "grad_norm": 3.4461602408655025, "learning_rate": 4.66098591074074e-06, "loss": 0.1916, "step": 7880 }, { "epoch": 1.07, "grad_norm": 3.319241643844991, "learning_rate": 4.65988691370489e-06, "loss": 0.1572, "step": 7881 }, { "epoch": 1.07, "grad_norm": 4.215241589186328, "learning_rate": 4.6587879331766465e-06, "loss": 0.1989, "step": 7882 }, { "epoch": 1.07, "grad_norm": 3.3487609716351803, "learning_rate": 4.657688969209343e-06, "loss": 0.1333, "step": 7883 }, { "epoch": 1.07, "grad_norm": 3.6981447063918584, "learning_rate": 4.656590021856325e-06, "loss": 0.1776, "step": 7884 }, { "epoch": 1.07, "grad_norm": 3.4361510590279742, "learning_rate": 4.655491091170924e-06, "loss": 0.1581, "step": 7885 }, { "epoch": 1.07, "grad_norm": 2.467378471100607, "learning_rate": 4.654392177206483e-06, "loss": 0.1283, "step": 7886 }, { "epoch": 1.07, "grad_norm": 4.600319991657141, "learning_rate": 4.653293280016335e-06, "loss": 0.1886, "step": 7887 }, { "epoch": 1.07, "grad_norm": 3.8530758933383047, "learning_rate": 4.652194399653819e-06, "loss": 0.2401, "step": 7888 }, { "epoch": 1.07, "grad_norm": 3.670983429924608, "learning_rate": 4.651095536172265e-06, "loss": 0.215, "step": 7889 }, { "epoch": 1.07, "grad_norm": 3.8166941509465504, "learning_rate": 4.649996689625012e-06, "loss": 0.1918, "step": 7890 }, { "epoch": 1.07, "grad_norm": 4.116021153613228, "learning_rate": 4.648897860065389e-06, "loss": 0.2019, "step": 7891 }, { "epoch": 1.07, "grad_norm": 3.9082186682826734, "learning_rate": 4.647799047546733e-06, "loss": 0.1695, "step": 7892 }, { "epoch": 1.07, "grad_norm": 3.5521149356639405, "learning_rate": 4.646700252122371e-06, "loss": 0.138, "step": 7893 }, { "epoch": 1.07, "grad_norm": 3.4953366454971344, "learning_rate": 4.645601473845636e-06, "loss": 0.2195, "step": 7894 }, { "epoch": 1.07, "grad_norm": 3.26056656057969, "learning_rate": 4.644502712769859e-06, "loss": 0.1473, "step": 7895 }, { "epoch": 1.07, "grad_norm": 4.646181946550193, "learning_rate": 4.6434039689483655e-06, "loss": 0.2018, "step": 7896 }, { "epoch": 1.07, "grad_norm": 2.7723123164787213, "learning_rate": 4.642305242434488e-06, "loss": 0.1901, "step": 7897 }, { "epoch": 1.07, "grad_norm": 3.5531397822731257, "learning_rate": 4.641206533281549e-06, "loss": 0.1523, "step": 7898 }, { "epoch": 1.07, "grad_norm": 3.4443754832316675, "learning_rate": 4.64010784154288e-06, "loss": 0.1585, "step": 7899 }, { "epoch": 1.07, "grad_norm": 3.85632156921963, "learning_rate": 4.639009167271803e-06, "loss": 0.1825, "step": 7900 }, { "epoch": 1.07, "grad_norm": 3.99944280669802, "learning_rate": 4.637910510521646e-06, "loss": 0.2199, "step": 7901 }, { "epoch": 1.07, "grad_norm": 4.271865166845619, "learning_rate": 4.63681187134573e-06, "loss": 0.1899, "step": 7902 }, { "epoch": 1.08, "grad_norm": 3.3155495033839424, "learning_rate": 4.635713249797381e-06, "loss": 0.151, "step": 7903 }, { "epoch": 1.08, "grad_norm": 4.645709699319956, "learning_rate": 4.634614645929918e-06, "loss": 0.2019, "step": 7904 }, { "epoch": 1.08, "grad_norm": 3.2759013605960954, "learning_rate": 4.633516059796666e-06, "loss": 0.1871, "step": 7905 }, { "epoch": 1.08, "grad_norm": 4.815976056591968, "learning_rate": 4.6324174914509425e-06, "loss": 0.1801, "step": 7906 }, { "epoch": 1.08, "grad_norm": 4.9699773726588585, "learning_rate": 4.6313189409460694e-06, "loss": 0.1687, "step": 7907 }, { "epoch": 1.08, "grad_norm": 3.2905262720866806, "learning_rate": 4.630220408335363e-06, "loss": 0.1637, "step": 7908 }, { "epoch": 1.08, "grad_norm": 4.0853518495128425, "learning_rate": 4.629121893672145e-06, "loss": 0.1812, "step": 7909 }, { "epoch": 1.08, "grad_norm": 4.030409656611133, "learning_rate": 4.628023397009729e-06, "loss": 0.1834, "step": 7910 }, { "epoch": 1.08, "grad_norm": 4.048452771388556, "learning_rate": 4.626924918401434e-06, "loss": 0.209, "step": 7911 }, { "epoch": 1.08, "grad_norm": 3.0667448284575554, "learning_rate": 4.625826457900573e-06, "loss": 0.1477, "step": 7912 }, { "epoch": 1.08, "grad_norm": 2.79979955535786, "learning_rate": 4.624728015560461e-06, "loss": 0.1593, "step": 7913 }, { "epoch": 1.08, "grad_norm": 3.8633152816354586, "learning_rate": 4.623629591434412e-06, "loss": 0.1925, "step": 7914 }, { "epoch": 1.08, "grad_norm": 3.8810840922688294, "learning_rate": 4.62253118557574e-06, "loss": 0.1543, "step": 7915 }, { "epoch": 1.08, "grad_norm": 4.459593795932877, "learning_rate": 4.621432798037755e-06, "loss": 0.1894, "step": 7916 }, { "epoch": 1.08, "grad_norm": 2.634260402203116, "learning_rate": 4.62033442887377e-06, "loss": 0.1496, "step": 7917 }, { "epoch": 1.08, "grad_norm": 3.914836805105069, "learning_rate": 4.619236078137091e-06, "loss": 0.2429, "step": 7918 }, { "epoch": 1.08, "grad_norm": 4.63749208694284, "learning_rate": 4.618137745881032e-06, "loss": 0.1769, "step": 7919 }, { "epoch": 1.08, "grad_norm": 12.67196092427758, "learning_rate": 4.617039432158897e-06, "loss": 0.2027, "step": 7920 }, { "epoch": 1.08, "grad_norm": 3.854465107635249, "learning_rate": 4.615941137023997e-06, "loss": 0.12, "step": 7921 }, { "epoch": 1.08, "grad_norm": 3.257940495164519, "learning_rate": 4.614842860529636e-06, "loss": 0.1794, "step": 7922 }, { "epoch": 1.08, "grad_norm": 3.7378702522555645, "learning_rate": 4.613744602729121e-06, "loss": 0.191, "step": 7923 }, { "epoch": 1.08, "grad_norm": 3.7280703184028585, "learning_rate": 4.612646363675755e-06, "loss": 0.1598, "step": 7924 }, { "epoch": 1.08, "grad_norm": 3.2590866273662065, "learning_rate": 4.6115481434228436e-06, "loss": 0.1522, "step": 7925 }, { "epoch": 1.08, "grad_norm": 4.223710075073797, "learning_rate": 4.610449942023687e-06, "loss": 0.1681, "step": 7926 }, { "epoch": 1.08, "grad_norm": 3.5220684940456275, "learning_rate": 4.6093517595315906e-06, "loss": 0.1973, "step": 7927 }, { "epoch": 1.08, "grad_norm": 3.5212885948800863, "learning_rate": 4.608253595999853e-06, "loss": 0.1802, "step": 7928 }, { "epoch": 1.08, "grad_norm": 3.045762252282528, "learning_rate": 4.607155451481775e-06, "loss": 0.1577, "step": 7929 }, { "epoch": 1.08, "grad_norm": 4.06603875127779, "learning_rate": 4.606057326030655e-06, "loss": 0.1788, "step": 7930 }, { "epoch": 1.08, "grad_norm": 3.291154989482382, "learning_rate": 4.6049592196997936e-06, "loss": 0.1936, "step": 7931 }, { "epoch": 1.08, "grad_norm": 2.8409104931393374, "learning_rate": 4.603861132542484e-06, "loss": 0.133, "step": 7932 }, { "epoch": 1.08, "grad_norm": 4.580102919400802, "learning_rate": 4.602763064612026e-06, "loss": 0.1843, "step": 7933 }, { "epoch": 1.08, "grad_norm": 4.153123061227708, "learning_rate": 4.6016650159617155e-06, "loss": 0.2235, "step": 7934 }, { "epoch": 1.08, "grad_norm": 4.234402007902416, "learning_rate": 4.600566986644843e-06, "loss": 0.2008, "step": 7935 }, { "epoch": 1.08, "grad_norm": 5.5267118573625895, "learning_rate": 4.599468976714706e-06, "loss": 0.2106, "step": 7936 }, { "epoch": 1.08, "grad_norm": 3.351098142324472, "learning_rate": 4.598370986224594e-06, "loss": 0.1813, "step": 7937 }, { "epoch": 1.08, "grad_norm": 3.3858928924216816, "learning_rate": 4.5972730152278e-06, "loss": 0.1437, "step": 7938 }, { "epoch": 1.08, "grad_norm": 4.539797566985416, "learning_rate": 4.596175063777616e-06, "loss": 0.2357, "step": 7939 }, { "epoch": 1.08, "grad_norm": 3.181026133904506, "learning_rate": 4.59507713192733e-06, "loss": 0.2081, "step": 7940 }, { "epoch": 1.08, "grad_norm": 3.9388454196125107, "learning_rate": 4.593979219730231e-06, "loss": 0.1726, "step": 7941 }, { "epoch": 1.08, "grad_norm": 5.409262778350243, "learning_rate": 4.59288132723961e-06, "loss": 0.1784, "step": 7942 }, { "epoch": 1.08, "grad_norm": 3.766719121778672, "learning_rate": 4.591783454508748e-06, "loss": 0.2117, "step": 7943 }, { "epoch": 1.08, "grad_norm": 3.0669870907307692, "learning_rate": 4.5906856015909365e-06, "loss": 0.1451, "step": 7944 }, { "epoch": 1.08, "grad_norm": 3.2850994888725857, "learning_rate": 4.5895877685394555e-06, "loss": 0.2017, "step": 7945 }, { "epoch": 1.08, "grad_norm": 3.444083623259341, "learning_rate": 4.588489955407594e-06, "loss": 0.138, "step": 7946 }, { "epoch": 1.08, "grad_norm": 2.9991513769273217, "learning_rate": 4.587392162248631e-06, "loss": 0.1589, "step": 7947 }, { "epoch": 1.08, "grad_norm": 3.887821167356731, "learning_rate": 4.586294389115853e-06, "loss": 0.2016, "step": 7948 }, { "epoch": 1.08, "grad_norm": 3.1827604965477256, "learning_rate": 4.585196636062535e-06, "loss": 0.1296, "step": 7949 }, { "epoch": 1.08, "grad_norm": 3.0466510854631, "learning_rate": 4.584098903141964e-06, "loss": 0.1337, "step": 7950 }, { "epoch": 1.08, "grad_norm": 3.62654300031117, "learning_rate": 4.583001190407414e-06, "loss": 0.1673, "step": 7951 }, { "epoch": 1.08, "grad_norm": 3.453256202524257, "learning_rate": 4.581903497912164e-06, "loss": 0.1835, "step": 7952 }, { "epoch": 1.08, "grad_norm": 3.393849091884604, "learning_rate": 4.580805825709494e-06, "loss": 0.1762, "step": 7953 }, { "epoch": 1.08, "grad_norm": 4.076132780604057, "learning_rate": 4.579708173852678e-06, "loss": 0.175, "step": 7954 }, { "epoch": 1.08, "grad_norm": 3.657543015557307, "learning_rate": 4.57861054239499e-06, "loss": 0.1429, "step": 7955 }, { "epoch": 1.08, "grad_norm": 4.7241802646802835, "learning_rate": 4.577512931389708e-06, "loss": 0.1891, "step": 7956 }, { "epoch": 1.08, "grad_norm": 3.9636470714435568, "learning_rate": 4.576415340890101e-06, "loss": 0.184, "step": 7957 }, { "epoch": 1.08, "grad_norm": 4.015282212773684, "learning_rate": 4.575317770949446e-06, "loss": 0.1876, "step": 7958 }, { "epoch": 1.08, "grad_norm": 4.5040371041451905, "learning_rate": 4.574220221621009e-06, "loss": 0.1982, "step": 7959 }, { "epoch": 1.08, "grad_norm": 3.742020004208524, "learning_rate": 4.573122692958064e-06, "loss": 0.1696, "step": 7960 }, { "epoch": 1.08, "grad_norm": 3.3111836467207825, "learning_rate": 4.572025185013878e-06, "loss": 0.1514, "step": 7961 }, { "epoch": 1.08, "grad_norm": 4.862170137650048, "learning_rate": 4.570927697841722e-06, "loss": 0.2022, "step": 7962 }, { "epoch": 1.08, "grad_norm": 3.4003908013104707, "learning_rate": 4.5698302314948594e-06, "loss": 0.1673, "step": 7963 }, { "epoch": 1.08, "grad_norm": 3.4024743551953502, "learning_rate": 4.5687327860265605e-06, "loss": 0.1614, "step": 7964 }, { "epoch": 1.08, "grad_norm": 3.4093847321388893, "learning_rate": 4.567635361490087e-06, "loss": 0.1613, "step": 7965 }, { "epoch": 1.08, "grad_norm": 2.78793989776305, "learning_rate": 4.566537957938706e-06, "loss": 0.1489, "step": 7966 }, { "epoch": 1.08, "grad_norm": 4.074838361027885, "learning_rate": 4.565440575425678e-06, "loss": 0.1587, "step": 7967 }, { "epoch": 1.08, "grad_norm": 3.2837956834600712, "learning_rate": 4.564343214004268e-06, "loss": 0.2003, "step": 7968 }, { "epoch": 1.08, "grad_norm": 2.701737339396605, "learning_rate": 4.563245873727734e-06, "loss": 0.1469, "step": 7969 }, { "epoch": 1.08, "grad_norm": 3.7719769860768544, "learning_rate": 4.56214855464934e-06, "loss": 0.1944, "step": 7970 }, { "epoch": 1.08, "grad_norm": 3.6019036717412027, "learning_rate": 4.561051256822341e-06, "loss": 0.1514, "step": 7971 }, { "epoch": 1.08, "grad_norm": 3.3432670762431465, "learning_rate": 4.559953980299998e-06, "loss": 0.1549, "step": 7972 }, { "epoch": 1.08, "grad_norm": 4.315490164221221, "learning_rate": 4.558856725135566e-06, "loss": 0.1779, "step": 7973 }, { "epoch": 1.08, "grad_norm": 3.246019426263603, "learning_rate": 4.557759491382301e-06, "loss": 0.1726, "step": 7974 }, { "epoch": 1.08, "grad_norm": 2.7272214336319354, "learning_rate": 4.556662279093461e-06, "loss": 0.1599, "step": 7975 }, { "epoch": 1.08, "grad_norm": 4.7897383018337525, "learning_rate": 4.555565088322295e-06, "loss": 0.2074, "step": 7976 }, { "epoch": 1.09, "grad_norm": 3.3449929273209893, "learning_rate": 4.554467919122061e-06, "loss": 0.1584, "step": 7977 }, { "epoch": 1.09, "grad_norm": 3.8045472642882583, "learning_rate": 4.553370771546005e-06, "loss": 0.1599, "step": 7978 }, { "epoch": 1.09, "grad_norm": 4.055630468107597, "learning_rate": 4.552273645647384e-06, "loss": 0.1823, "step": 7979 }, { "epoch": 1.09, "grad_norm": 4.028504024545859, "learning_rate": 4.551176541479443e-06, "loss": 0.1796, "step": 7980 }, { "epoch": 1.09, "grad_norm": 2.855590395080271, "learning_rate": 4.550079459095434e-06, "loss": 0.1367, "step": 7981 }, { "epoch": 1.09, "grad_norm": 2.8949362192455954, "learning_rate": 4.548982398548601e-06, "loss": 0.1751, "step": 7982 }, { "epoch": 1.09, "grad_norm": 3.098789814491897, "learning_rate": 4.547885359892196e-06, "loss": 0.1414, "step": 7983 }, { "epoch": 1.09, "grad_norm": 4.419992202515063, "learning_rate": 4.5467883431794575e-06, "loss": 0.217, "step": 7984 }, { "epoch": 1.09, "grad_norm": 4.0031816421483635, "learning_rate": 4.545691348463634e-06, "loss": 0.1489, "step": 7985 }, { "epoch": 1.09, "grad_norm": 7.311390835598183, "learning_rate": 4.544594375797969e-06, "loss": 0.185, "step": 7986 }, { "epoch": 1.09, "grad_norm": 3.855670288334423, "learning_rate": 4.543497425235705e-06, "loss": 0.171, "step": 7987 }, { "epoch": 1.09, "grad_norm": 4.410822190584559, "learning_rate": 4.54240049683008e-06, "loss": 0.1885, "step": 7988 }, { "epoch": 1.09, "grad_norm": 5.282491402061368, "learning_rate": 4.541303590634338e-06, "loss": 0.2495, "step": 7989 }, { "epoch": 1.09, "grad_norm": 3.5918143735670682, "learning_rate": 4.540206706701714e-06, "loss": 0.1887, "step": 7990 }, { "epoch": 1.09, "grad_norm": 2.6212329092292754, "learning_rate": 4.539109845085451e-06, "loss": 0.1419, "step": 7991 }, { "epoch": 1.09, "grad_norm": 5.234179031028997, "learning_rate": 4.538013005838781e-06, "loss": 0.2114, "step": 7992 }, { "epoch": 1.09, "grad_norm": 3.6422365293632617, "learning_rate": 4.536916189014942e-06, "loss": 0.1698, "step": 7993 }, { "epoch": 1.09, "grad_norm": 5.431529584248141, "learning_rate": 4.535819394667169e-06, "loss": 0.1836, "step": 7994 }, { "epoch": 1.09, "grad_norm": 3.114631787612889, "learning_rate": 4.5347226228486974e-06, "loss": 0.2069, "step": 7995 }, { "epoch": 1.09, "grad_norm": 3.4777765174604083, "learning_rate": 4.533625873612755e-06, "loss": 0.2021, "step": 7996 }, { "epoch": 1.09, "grad_norm": 3.432095954545672, "learning_rate": 4.532529147012578e-06, "loss": 0.185, "step": 7997 }, { "epoch": 1.09, "grad_norm": 3.5551004105222015, "learning_rate": 4.531432443101391e-06, "loss": 0.1735, "step": 7998 }, { "epoch": 1.09, "grad_norm": 3.330762215269421, "learning_rate": 4.530335761932429e-06, "loss": 0.1772, "step": 7999 }, { "epoch": 1.09, "grad_norm": 4.022351451421466, "learning_rate": 4.529239103558917e-06, "loss": 0.2026, "step": 8000 }, { "epoch": 1.09, "grad_norm": 3.698012875471708, "learning_rate": 4.528142468034084e-06, "loss": 0.1594, "step": 8001 }, { "epoch": 1.09, "grad_norm": 3.8489484860514636, "learning_rate": 4.527045855411153e-06, "loss": 0.1684, "step": 8002 }, { "epoch": 1.09, "grad_norm": 3.565125787002791, "learning_rate": 4.525949265743353e-06, "loss": 0.1812, "step": 8003 }, { "epoch": 1.09, "grad_norm": 4.441904004283613, "learning_rate": 4.524852699083903e-06, "loss": 0.1755, "step": 8004 }, { "epoch": 1.09, "grad_norm": 3.3133599512212606, "learning_rate": 4.5237561554860284e-06, "loss": 0.1706, "step": 8005 }, { "epoch": 1.09, "grad_norm": 3.484872428848179, "learning_rate": 4.52265963500295e-06, "loss": 0.1756, "step": 8006 }, { "epoch": 1.09, "grad_norm": 2.6784815596491165, "learning_rate": 4.521563137687889e-06, "loss": 0.1708, "step": 8007 }, { "epoch": 1.09, "grad_norm": 3.7454980917162373, "learning_rate": 4.520466663594061e-06, "loss": 0.2052, "step": 8008 }, { "epoch": 1.09, "grad_norm": 3.304154000991766, "learning_rate": 4.51937021277469e-06, "loss": 0.1841, "step": 8009 }, { "epoch": 1.09, "grad_norm": 2.9560768773226975, "learning_rate": 4.518273785282987e-06, "loss": 0.1394, "step": 8010 }, { "epoch": 1.09, "grad_norm": 3.5034048819594394, "learning_rate": 4.517177381172173e-06, "loss": 0.1756, "step": 8011 }, { "epoch": 1.09, "grad_norm": 3.7325458712715287, "learning_rate": 4.516081000495458e-06, "loss": 0.16, "step": 8012 }, { "epoch": 1.09, "grad_norm": 3.6165728315361094, "learning_rate": 4.514984643306058e-06, "loss": 0.1681, "step": 8013 }, { "epoch": 1.09, "grad_norm": 3.4054661843112073, "learning_rate": 4.5138883096571865e-06, "loss": 0.2088, "step": 8014 }, { "epoch": 1.09, "grad_norm": 5.2244126812424225, "learning_rate": 4.512791999602051e-06, "loss": 0.1776, "step": 8015 }, { "epoch": 1.09, "grad_norm": 3.1575676528191625, "learning_rate": 4.511695713193867e-06, "loss": 0.1682, "step": 8016 }, { "epoch": 1.09, "grad_norm": 3.391151434877064, "learning_rate": 4.510599450485838e-06, "loss": 0.1731, "step": 8017 }, { "epoch": 1.09, "grad_norm": 3.687297053489387, "learning_rate": 4.509503211531174e-06, "loss": 0.1477, "step": 8018 }, { "epoch": 1.09, "grad_norm": 3.1489592498387298, "learning_rate": 4.508406996383082e-06, "loss": 0.1666, "step": 8019 }, { "epoch": 1.09, "grad_norm": 2.631113209702183, "learning_rate": 4.507310805094769e-06, "loss": 0.172, "step": 8020 }, { "epoch": 1.09, "grad_norm": 3.3857447777675924, "learning_rate": 4.506214637719436e-06, "loss": 0.1956, "step": 8021 }, { "epoch": 1.09, "grad_norm": 3.5500581313863493, "learning_rate": 4.505118494310289e-06, "loss": 0.197, "step": 8022 }, { "epoch": 1.09, "grad_norm": 4.268259548773264, "learning_rate": 4.504022374920527e-06, "loss": 0.1836, "step": 8023 }, { "epoch": 1.09, "grad_norm": 4.939029419916039, "learning_rate": 4.502926279603355e-06, "loss": 0.1689, "step": 8024 }, { "epoch": 1.09, "grad_norm": 4.904526112194837, "learning_rate": 4.501830208411968e-06, "loss": 0.1891, "step": 8025 }, { "epoch": 1.09, "grad_norm": 3.501427706877383, "learning_rate": 4.500734161399568e-06, "loss": 0.1264, "step": 8026 }, { "epoch": 1.09, "grad_norm": 3.7464305318338487, "learning_rate": 4.499638138619351e-06, "loss": 0.1242, "step": 8027 }, { "epoch": 1.09, "grad_norm": 4.851647073546655, "learning_rate": 4.498542140124514e-06, "loss": 0.1895, "step": 8028 }, { "epoch": 1.09, "grad_norm": 3.2256633601946016, "learning_rate": 4.497446165968251e-06, "loss": 0.195, "step": 8029 }, { "epoch": 1.09, "grad_norm": 3.658787424016432, "learning_rate": 4.496350216203757e-06, "loss": 0.1644, "step": 8030 }, { "epoch": 1.09, "grad_norm": 3.1404403794358258, "learning_rate": 4.495254290884222e-06, "loss": 0.1254, "step": 8031 }, { "epoch": 1.09, "grad_norm": 4.682534343938446, "learning_rate": 4.49415839006284e-06, "loss": 0.1732, "step": 8032 }, { "epoch": 1.09, "grad_norm": 3.1267850131159696, "learning_rate": 4.4930625137928e-06, "loss": 0.1292, "step": 8033 }, { "epoch": 1.09, "grad_norm": 4.596813418363068, "learning_rate": 4.491966662127293e-06, "loss": 0.1677, "step": 8034 }, { "epoch": 1.09, "grad_norm": 3.9233665149273285, "learning_rate": 4.490870835119503e-06, "loss": 0.218, "step": 8035 }, { "epoch": 1.09, "grad_norm": 3.557167273628464, "learning_rate": 4.489775032822621e-06, "loss": 0.1416, "step": 8036 }, { "epoch": 1.09, "grad_norm": 3.567035980399054, "learning_rate": 4.488679255289829e-06, "loss": 0.2119, "step": 8037 }, { "epoch": 1.09, "grad_norm": 3.50706853786039, "learning_rate": 4.4875835025743134e-06, "loss": 0.1332, "step": 8038 }, { "epoch": 1.09, "grad_norm": 4.233034335724912, "learning_rate": 4.4864877747292565e-06, "loss": 0.1924, "step": 8039 }, { "epoch": 1.09, "grad_norm": 2.863451854584845, "learning_rate": 4.485392071807841e-06, "loss": 0.1316, "step": 8040 }, { "epoch": 1.09, "grad_norm": 3.404932835316124, "learning_rate": 4.484296393863245e-06, "loss": 0.1707, "step": 8041 }, { "epoch": 1.09, "grad_norm": 3.1331520280041456, "learning_rate": 4.483200740948652e-06, "loss": 0.1713, "step": 8042 }, { "epoch": 1.09, "grad_norm": 3.36561474534259, "learning_rate": 4.482105113117235e-06, "loss": 0.2173, "step": 8043 }, { "epoch": 1.09, "grad_norm": 2.678060405433428, "learning_rate": 4.481009510422176e-06, "loss": 0.1722, "step": 8044 }, { "epoch": 1.09, "grad_norm": 4.847243406980245, "learning_rate": 4.479913932916648e-06, "loss": 0.1568, "step": 8045 }, { "epoch": 1.09, "grad_norm": 3.841922032486207, "learning_rate": 4.478818380653826e-06, "loss": 0.1978, "step": 8046 }, { "epoch": 1.09, "grad_norm": 3.1528473396310197, "learning_rate": 4.477722853686883e-06, "loss": 0.1862, "step": 8047 }, { "epoch": 1.09, "grad_norm": 3.539851713142328, "learning_rate": 4.476627352068993e-06, "loss": 0.1629, "step": 8048 }, { "epoch": 1.09, "grad_norm": 3.1249407471448305, "learning_rate": 4.475531875853324e-06, "loss": 0.1379, "step": 8049 }, { "epoch": 1.1, "grad_norm": 3.4461281424331167, "learning_rate": 4.4744364250930485e-06, "loss": 0.1788, "step": 8050 }, { "epoch": 1.1, "grad_norm": 4.6746831062901535, "learning_rate": 4.473340999841332e-06, "loss": 0.1734, "step": 8051 }, { "epoch": 1.1, "grad_norm": 2.750024283644142, "learning_rate": 4.472245600151344e-06, "loss": 0.1515, "step": 8052 }, { "epoch": 1.1, "grad_norm": 3.6661434275404865, "learning_rate": 4.471150226076251e-06, "loss": 0.1848, "step": 8053 }, { "epoch": 1.1, "grad_norm": 3.267743187434913, "learning_rate": 4.470054877669213e-06, "loss": 0.1477, "step": 8054 }, { "epoch": 1.1, "grad_norm": 2.630069530645783, "learning_rate": 4.4689595549834e-06, "loss": 0.1824, "step": 8055 }, { "epoch": 1.1, "grad_norm": 3.753067812873147, "learning_rate": 4.467864258071968e-06, "loss": 0.1636, "step": 8056 }, { "epoch": 1.1, "grad_norm": 3.577596448639893, "learning_rate": 4.466768986988082e-06, "loss": 0.1946, "step": 8057 }, { "epoch": 1.1, "grad_norm": 3.552089131420463, "learning_rate": 4.4656737417849e-06, "loss": 0.1925, "step": 8058 }, { "epoch": 1.1, "grad_norm": 3.6092993424019317, "learning_rate": 4.4645785225155816e-06, "loss": 0.1864, "step": 8059 }, { "epoch": 1.1, "grad_norm": 3.783673724688508, "learning_rate": 4.463483329233281e-06, "loss": 0.1731, "step": 8060 }, { "epoch": 1.1, "grad_norm": 4.363723448782299, "learning_rate": 4.462388161991159e-06, "loss": 0.1936, "step": 8061 }, { "epoch": 1.1, "grad_norm": 3.5827368309007688, "learning_rate": 4.461293020842366e-06, "loss": 0.1871, "step": 8062 }, { "epoch": 1.1, "grad_norm": 3.269313380950139, "learning_rate": 4.460197905840059e-06, "loss": 0.1453, "step": 8063 }, { "epoch": 1.1, "grad_norm": 3.466221516074698, "learning_rate": 4.459102817037385e-06, "loss": 0.1475, "step": 8064 }, { "epoch": 1.1, "grad_norm": 4.3105040993698, "learning_rate": 4.458007754487499e-06, "loss": 0.1797, "step": 8065 }, { "epoch": 1.1, "grad_norm": 4.696948718057624, "learning_rate": 4.456912718243549e-06, "loss": 0.1338, "step": 8066 }, { "epoch": 1.1, "grad_norm": 3.1584477980402488, "learning_rate": 4.4558177083586855e-06, "loss": 0.1536, "step": 8067 }, { "epoch": 1.1, "grad_norm": 3.6419887022060142, "learning_rate": 4.454722724886051e-06, "loss": 0.1666, "step": 8068 }, { "epoch": 1.1, "grad_norm": 3.481878791104291, "learning_rate": 4.453627767878798e-06, "loss": 0.1578, "step": 8069 }, { "epoch": 1.1, "grad_norm": 3.186806249808649, "learning_rate": 4.452532837390063e-06, "loss": 0.177, "step": 8070 }, { "epoch": 1.1, "grad_norm": 3.790111107513291, "learning_rate": 4.451437933472993e-06, "loss": 0.1924, "step": 8071 }, { "epoch": 1.1, "grad_norm": 3.3635127630424058, "learning_rate": 4.450343056180731e-06, "loss": 0.1702, "step": 8072 }, { "epoch": 1.1, "grad_norm": 3.234601326742822, "learning_rate": 4.449248205566417e-06, "loss": 0.2273, "step": 8073 }, { "epoch": 1.1, "grad_norm": 3.831615968300057, "learning_rate": 4.448153381683188e-06, "loss": 0.1508, "step": 8074 }, { "epoch": 1.1, "grad_norm": 3.5032772864330757, "learning_rate": 4.447058584584186e-06, "loss": 0.1291, "step": 8075 }, { "epoch": 1.1, "grad_norm": 4.103142309422496, "learning_rate": 4.445963814322542e-06, "loss": 0.1984, "step": 8076 }, { "epoch": 1.1, "grad_norm": 3.8429067089060838, "learning_rate": 4.444869070951398e-06, "loss": 0.154, "step": 8077 }, { "epoch": 1.1, "grad_norm": 3.715302972949315, "learning_rate": 4.443774354523883e-06, "loss": 0.1864, "step": 8078 }, { "epoch": 1.1, "grad_norm": 3.763801535596724, "learning_rate": 4.442679665093131e-06, "loss": 0.2087, "step": 8079 }, { "epoch": 1.1, "grad_norm": 3.6608636877545533, "learning_rate": 4.441585002712274e-06, "loss": 0.1835, "step": 8080 }, { "epoch": 1.1, "grad_norm": 4.061190788988807, "learning_rate": 4.440490367434444e-06, "loss": 0.1739, "step": 8081 }, { "epoch": 1.1, "grad_norm": 3.428489695428193, "learning_rate": 4.439395759312765e-06, "loss": 0.1481, "step": 8082 }, { "epoch": 1.1, "grad_norm": 3.6715511670684546, "learning_rate": 4.43830117840037e-06, "loss": 0.1659, "step": 8083 }, { "epoch": 1.1, "grad_norm": 3.9036749859826054, "learning_rate": 4.43720662475038e-06, "loss": 0.1976, "step": 8084 }, { "epoch": 1.1, "grad_norm": 3.3203413169412843, "learning_rate": 4.436112098415923e-06, "loss": 0.1531, "step": 8085 }, { "epoch": 1.1, "grad_norm": 3.5317595286794714, "learning_rate": 4.4350175994501225e-06, "loss": 0.1999, "step": 8086 }, { "epoch": 1.1, "grad_norm": 3.432021951541701, "learning_rate": 4.433923127906101e-06, "loss": 0.1567, "step": 8087 }, { "epoch": 1.1, "grad_norm": 2.776088681912753, "learning_rate": 4.432828683836976e-06, "loss": 0.1349, "step": 8088 }, { "epoch": 1.1, "grad_norm": 3.271348874275257, "learning_rate": 4.431734267295873e-06, "loss": 0.1516, "step": 8089 }, { "epoch": 1.1, "grad_norm": 3.2018938581652825, "learning_rate": 4.430639878335904e-06, "loss": 0.1399, "step": 8090 }, { "epoch": 1.1, "grad_norm": 3.292705954684827, "learning_rate": 4.42954551701019e-06, "loss": 0.1772, "step": 8091 }, { "epoch": 1.1, "grad_norm": 3.6381116133397637, "learning_rate": 4.428451183371844e-06, "loss": 0.1856, "step": 8092 }, { "epoch": 1.1, "grad_norm": 3.1193119398183695, "learning_rate": 4.427356877473981e-06, "loss": 0.1701, "step": 8093 }, { "epoch": 1.1, "grad_norm": 3.296737449725885, "learning_rate": 4.426262599369716e-06, "loss": 0.1827, "step": 8094 }, { "epoch": 1.1, "grad_norm": 3.712010410183806, "learning_rate": 4.425168349112156e-06, "loss": 0.1627, "step": 8095 }, { "epoch": 1.1, "grad_norm": 3.297645928182125, "learning_rate": 4.424074126754417e-06, "loss": 0.1746, "step": 8096 }, { "epoch": 1.1, "grad_norm": 4.119487040244843, "learning_rate": 4.422979932349601e-06, "loss": 0.1381, "step": 8097 }, { "epoch": 1.1, "grad_norm": 2.7911117749771823, "learning_rate": 4.421885765950821e-06, "loss": 0.1363, "step": 8098 }, { "epoch": 1.1, "grad_norm": 3.3876933783992307, "learning_rate": 4.4207916276111795e-06, "loss": 0.1409, "step": 8099 }, { "epoch": 1.1, "grad_norm": 3.5566864730584165, "learning_rate": 4.419697517383785e-06, "loss": 0.1722, "step": 8100 }, { "epoch": 1.1, "grad_norm": 3.2489927806450827, "learning_rate": 4.4186034353217355e-06, "loss": 0.1744, "step": 8101 }, { "epoch": 1.1, "grad_norm": 3.5111769251584435, "learning_rate": 4.417509381478139e-06, "loss": 0.1403, "step": 8102 }, { "epoch": 1.1, "grad_norm": 3.7108305946666094, "learning_rate": 4.41641535590609e-06, "loss": 0.1934, "step": 8103 }, { "epoch": 1.1, "grad_norm": 4.059485374918036, "learning_rate": 4.415321358658692e-06, "loss": 0.1777, "step": 8104 }, { "epoch": 1.1, "grad_norm": 3.5508350783872604, "learning_rate": 4.414227389789041e-06, "loss": 0.1745, "step": 8105 }, { "epoch": 1.1, "grad_norm": 4.366640270015502, "learning_rate": 4.413133449350236e-06, "loss": 0.2075, "step": 8106 }, { "epoch": 1.1, "grad_norm": 4.185132289133139, "learning_rate": 4.412039537395369e-06, "loss": 0.1879, "step": 8107 }, { "epoch": 1.1, "grad_norm": 3.50075695104165, "learning_rate": 4.410945653977535e-06, "loss": 0.1762, "step": 8108 }, { "epoch": 1.1, "grad_norm": 3.812861494980191, "learning_rate": 4.409851799149825e-06, "loss": 0.1657, "step": 8109 }, { "epoch": 1.1, "grad_norm": 4.82170894770665, "learning_rate": 4.4087579729653334e-06, "loss": 0.1755, "step": 8110 }, { "epoch": 1.1, "grad_norm": 3.2375706435558933, "learning_rate": 4.407664175477145e-06, "loss": 0.1673, "step": 8111 }, { "epoch": 1.1, "grad_norm": 7.77930043564186, "learning_rate": 4.4065704067383526e-06, "loss": 0.1363, "step": 8112 }, { "epoch": 1.1, "grad_norm": 4.826646596205071, "learning_rate": 4.405476666802039e-06, "loss": 0.1666, "step": 8113 }, { "epoch": 1.1, "grad_norm": 3.5505728148013644, "learning_rate": 4.404382955721293e-06, "loss": 0.1408, "step": 8114 }, { "epoch": 1.1, "grad_norm": 2.721437225492254, "learning_rate": 4.403289273549195e-06, "loss": 0.1635, "step": 8115 }, { "epoch": 1.1, "grad_norm": 3.899489682360403, "learning_rate": 4.402195620338832e-06, "loss": 0.1811, "step": 8116 }, { "epoch": 1.1, "grad_norm": 3.3967081737615827, "learning_rate": 4.401101996143281e-06, "loss": 0.1755, "step": 8117 }, { "epoch": 1.1, "grad_norm": 3.519307981633004, "learning_rate": 4.400008401015623e-06, "loss": 0.1407, "step": 8118 }, { "epoch": 1.1, "grad_norm": 3.922371959147162, "learning_rate": 4.398914835008936e-06, "loss": 0.2013, "step": 8119 }, { "epoch": 1.1, "grad_norm": 4.165563198907099, "learning_rate": 4.3978212981763e-06, "loss": 0.138, "step": 8120 }, { "epoch": 1.1, "grad_norm": 4.443302696136943, "learning_rate": 4.396727790570787e-06, "loss": 0.2137, "step": 8121 }, { "epoch": 1.1, "grad_norm": 3.397962377465378, "learning_rate": 4.395634312245473e-06, "loss": 0.2013, "step": 8122 }, { "epoch": 1.1, "grad_norm": 3.4252877693727886, "learning_rate": 4.394540863253428e-06, "loss": 0.247, "step": 8123 }, { "epoch": 1.11, "grad_norm": 3.2554002360133927, "learning_rate": 4.393447443647726e-06, "loss": 0.215, "step": 8124 }, { "epoch": 1.11, "grad_norm": 2.802140767219484, "learning_rate": 4.392354053481436e-06, "loss": 0.1272, "step": 8125 }, { "epoch": 1.11, "grad_norm": 4.095368746950995, "learning_rate": 4.391260692807627e-06, "loss": 0.2018, "step": 8126 }, { "epoch": 1.11, "grad_norm": 3.834518191394008, "learning_rate": 4.390167361679363e-06, "loss": 0.173, "step": 8127 }, { "epoch": 1.11, "grad_norm": 4.0720293678493125, "learning_rate": 4.389074060149714e-06, "loss": 0.164, "step": 8128 }, { "epoch": 1.11, "grad_norm": 4.275639996818284, "learning_rate": 4.38798078827174e-06, "loss": 0.1868, "step": 8129 }, { "epoch": 1.11, "grad_norm": 3.7579290483686387, "learning_rate": 4.386887546098509e-06, "loss": 0.1498, "step": 8130 }, { "epoch": 1.11, "grad_norm": 6.422466990527595, "learning_rate": 4.3857943336830746e-06, "loss": 0.184, "step": 8131 }, { "epoch": 1.11, "grad_norm": 4.487361409445666, "learning_rate": 4.384701151078502e-06, "loss": 0.1986, "step": 8132 }, { "epoch": 1.11, "grad_norm": 3.3047812243273076, "learning_rate": 4.38360799833785e-06, "loss": 0.1718, "step": 8133 }, { "epoch": 1.11, "grad_norm": 3.4616089847715705, "learning_rate": 4.382514875514172e-06, "loss": 0.1745, "step": 8134 }, { "epoch": 1.11, "grad_norm": 3.138603205141628, "learning_rate": 4.381421782660526e-06, "loss": 0.1895, "step": 8135 }, { "epoch": 1.11, "grad_norm": 3.466710462583503, "learning_rate": 4.380328719829963e-06, "loss": 0.1725, "step": 8136 }, { "epoch": 1.11, "grad_norm": 2.604055469056516, "learning_rate": 4.379235687075538e-06, "loss": 0.1847, "step": 8137 }, { "epoch": 1.11, "grad_norm": 3.3250104414975215, "learning_rate": 4.378142684450301e-06, "loss": 0.1227, "step": 8138 }, { "epoch": 1.11, "grad_norm": 3.2385791858558908, "learning_rate": 4.3770497120073045e-06, "loss": 0.1599, "step": 8139 }, { "epoch": 1.11, "grad_norm": 5.065919193357837, "learning_rate": 4.375956769799592e-06, "loss": 0.1936, "step": 8140 }, { "epoch": 1.11, "grad_norm": 3.109551993791051, "learning_rate": 4.374863857880214e-06, "loss": 0.1745, "step": 8141 }, { "epoch": 1.11, "grad_norm": 3.616701267686644, "learning_rate": 4.373770976302212e-06, "loss": 0.1318, "step": 8142 }, { "epoch": 1.11, "grad_norm": 3.177400604171561, "learning_rate": 4.372678125118634e-06, "loss": 0.1605, "step": 8143 }, { "epoch": 1.11, "grad_norm": 3.052060087410699, "learning_rate": 4.371585304382517e-06, "loss": 0.1466, "step": 8144 }, { "epoch": 1.11, "grad_norm": 3.442045418404134, "learning_rate": 4.3704925141469065e-06, "loss": 0.1282, "step": 8145 }, { "epoch": 1.11, "grad_norm": 3.1759663991646536, "learning_rate": 4.36939975446484e-06, "loss": 0.1902, "step": 8146 }, { "epoch": 1.11, "grad_norm": 3.9440830085237484, "learning_rate": 4.368307025389355e-06, "loss": 0.2285, "step": 8147 }, { "epoch": 1.11, "grad_norm": 3.161331822111215, "learning_rate": 4.367214326973487e-06, "loss": 0.1444, "step": 8148 }, { "epoch": 1.11, "grad_norm": 4.235773241570629, "learning_rate": 4.366121659270274e-06, "loss": 0.1987, "step": 8149 }, { "epoch": 1.11, "grad_norm": 2.9903687177576206, "learning_rate": 4.365029022332745e-06, "loss": 0.1871, "step": 8150 }, { "epoch": 1.11, "grad_norm": 3.6267160120267166, "learning_rate": 4.363936416213935e-06, "loss": 0.1705, "step": 8151 }, { "epoch": 1.11, "grad_norm": 3.8737999001542445, "learning_rate": 4.362843840966872e-06, "loss": 0.1443, "step": 8152 }, { "epoch": 1.11, "grad_norm": 3.5946023580202024, "learning_rate": 4.3617512966445885e-06, "loss": 0.1305, "step": 8153 }, { "epoch": 1.11, "grad_norm": 2.8590189193375473, "learning_rate": 4.360658783300107e-06, "loss": 0.1275, "step": 8154 }, { "epoch": 1.11, "grad_norm": 3.4655989858853715, "learning_rate": 4.359566300986458e-06, "loss": 0.1775, "step": 8155 }, { "epoch": 1.11, "grad_norm": 3.7600164545114585, "learning_rate": 4.3584738497566615e-06, "loss": 0.1774, "step": 8156 }, { "epoch": 1.11, "grad_norm": 3.9208587680579075, "learning_rate": 4.357381429663744e-06, "loss": 0.1495, "step": 8157 }, { "epoch": 1.11, "grad_norm": 4.816554310889799, "learning_rate": 4.356289040760724e-06, "loss": 0.1756, "step": 8158 }, { "epoch": 1.11, "grad_norm": 3.978940737384637, "learning_rate": 4.3551966831006235e-06, "loss": 0.1376, "step": 8159 }, { "epoch": 1.11, "grad_norm": 4.074143560233754, "learning_rate": 4.354104356736458e-06, "loss": 0.1834, "step": 8160 }, { "epoch": 1.11, "grad_norm": 3.5602339258039, "learning_rate": 4.353012061721249e-06, "loss": 0.1674, "step": 8161 }, { "epoch": 1.11, "grad_norm": 3.636072665714735, "learning_rate": 4.351919798108006e-06, "loss": 0.1482, "step": 8162 }, { "epoch": 1.11, "grad_norm": 3.631791481975064, "learning_rate": 4.350827565949747e-06, "loss": 0.1735, "step": 8163 }, { "epoch": 1.11, "grad_norm": 3.661166070057888, "learning_rate": 4.3497353652994815e-06, "loss": 0.1825, "step": 8164 }, { "epoch": 1.11, "grad_norm": 3.692670083163722, "learning_rate": 4.348643196210222e-06, "loss": 0.1918, "step": 8165 }, { "epoch": 1.11, "grad_norm": 3.459477854085365, "learning_rate": 4.347551058734975e-06, "loss": 0.1995, "step": 8166 }, { "epoch": 1.11, "grad_norm": 4.181997692143849, "learning_rate": 4.346458952926754e-06, "loss": 0.1593, "step": 8167 }, { "epoch": 1.11, "grad_norm": 3.5185039853538265, "learning_rate": 4.345366878838557e-06, "loss": 0.1526, "step": 8168 }, { "epoch": 1.11, "grad_norm": 3.102905372542252, "learning_rate": 4.344274836523396e-06, "loss": 0.1361, "step": 8169 }, { "epoch": 1.11, "grad_norm": 3.9619309053358975, "learning_rate": 4.343182826034268e-06, "loss": 0.185, "step": 8170 }, { "epoch": 1.11, "grad_norm": 3.3016389497507417, "learning_rate": 4.342090847424177e-06, "loss": 0.171, "step": 8171 }, { "epoch": 1.11, "grad_norm": 5.058986603324468, "learning_rate": 4.340998900746123e-06, "loss": 0.1768, "step": 8172 }, { "epoch": 1.11, "grad_norm": 4.690115327024011, "learning_rate": 4.339906986053105e-06, "loss": 0.1637, "step": 8173 }, { "epoch": 1.11, "grad_norm": 3.6713461986494162, "learning_rate": 4.33881510339812e-06, "loss": 0.1568, "step": 8174 }, { "epoch": 1.11, "grad_norm": 4.002972827254981, "learning_rate": 4.33772325283416e-06, "loss": 0.1498, "step": 8175 }, { "epoch": 1.11, "grad_norm": 4.1468363720854065, "learning_rate": 4.336631434414221e-06, "loss": 0.1789, "step": 8176 }, { "epoch": 1.11, "grad_norm": 4.221596396035925, "learning_rate": 4.335539648191295e-06, "loss": 0.1818, "step": 8177 }, { "epoch": 1.11, "grad_norm": 3.924298450350003, "learning_rate": 4.3344478942183726e-06, "loss": 0.1724, "step": 8178 }, { "epoch": 1.11, "grad_norm": 9.062638435368395, "learning_rate": 4.333356172548442e-06, "loss": 0.1914, "step": 8179 }, { "epoch": 1.11, "grad_norm": 3.7665871798759087, "learning_rate": 4.332264483234492e-06, "loss": 0.1824, "step": 8180 }, { "epoch": 1.11, "grad_norm": 3.6629034066875046, "learning_rate": 4.3311728263295064e-06, "loss": 0.1916, "step": 8181 }, { "epoch": 1.11, "grad_norm": 3.7130597790609516, "learning_rate": 4.330081201886473e-06, "loss": 0.1589, "step": 8182 }, { "epoch": 1.11, "grad_norm": 2.97432193149106, "learning_rate": 4.328989609958368e-06, "loss": 0.1683, "step": 8183 }, { "epoch": 1.11, "grad_norm": 3.9063181982966064, "learning_rate": 4.32789805059818e-06, "loss": 0.1634, "step": 8184 }, { "epoch": 1.11, "grad_norm": 3.935742844709096, "learning_rate": 4.326806523858883e-06, "loss": 0.1685, "step": 8185 }, { "epoch": 1.11, "grad_norm": 5.586396059077869, "learning_rate": 4.325715029793459e-06, "loss": 0.1847, "step": 8186 }, { "epoch": 1.11, "grad_norm": 3.5699078741238517, "learning_rate": 4.324623568454881e-06, "loss": 0.1845, "step": 8187 }, { "epoch": 1.11, "grad_norm": 3.545197665100302, "learning_rate": 4.323532139896127e-06, "loss": 0.1645, "step": 8188 }, { "epoch": 1.11, "grad_norm": 3.908020074153767, "learning_rate": 4.322440744170166e-06, "loss": 0.1907, "step": 8189 }, { "epoch": 1.11, "grad_norm": 3.8488786174317036, "learning_rate": 4.321349381329974e-06, "loss": 0.1982, "step": 8190 }, { "epoch": 1.11, "grad_norm": 4.672912638093843, "learning_rate": 4.3202580514285186e-06, "loss": 0.2137, "step": 8191 }, { "epoch": 1.11, "grad_norm": 4.230231841790014, "learning_rate": 4.319166754518768e-06, "loss": 0.1659, "step": 8192 }, { "epoch": 1.11, "grad_norm": 3.665395140721946, "learning_rate": 4.318075490653688e-06, "loss": 0.1831, "step": 8193 }, { "epoch": 1.11, "grad_norm": 3.336533500169663, "learning_rate": 4.316984259886249e-06, "loss": 0.1421, "step": 8194 }, { "epoch": 1.11, "grad_norm": 3.583297794596581, "learning_rate": 4.315893062269408e-06, "loss": 0.163, "step": 8195 }, { "epoch": 1.11, "grad_norm": 3.25698230267045, "learning_rate": 4.314801897856132e-06, "loss": 0.1307, "step": 8196 }, { "epoch": 1.12, "grad_norm": 3.94744170209703, "learning_rate": 4.313710766699377e-06, "loss": 0.1816, "step": 8197 }, { "epoch": 1.12, "grad_norm": 2.9277122115503986, "learning_rate": 4.312619668852105e-06, "loss": 0.1191, "step": 8198 }, { "epoch": 1.12, "grad_norm": 3.597796349156059, "learning_rate": 4.31152860436727e-06, "loss": 0.1951, "step": 8199 }, { "epoch": 1.12, "grad_norm": 2.874793669626325, "learning_rate": 4.310437573297833e-06, "loss": 0.1523, "step": 8200 }, { "epoch": 1.12, "grad_norm": 3.6011366693027718, "learning_rate": 4.309346575696743e-06, "loss": 0.1714, "step": 8201 }, { "epoch": 1.12, "grad_norm": 3.8448577094479095, "learning_rate": 4.308255611616954e-06, "loss": 0.1822, "step": 8202 }, { "epoch": 1.12, "grad_norm": 2.954659102670488, "learning_rate": 4.307164681111415e-06, "loss": 0.1554, "step": 8203 }, { "epoch": 1.12, "grad_norm": 3.2270384654917974, "learning_rate": 4.306073784233077e-06, "loss": 0.1842, "step": 8204 }, { "epoch": 1.12, "grad_norm": 2.8580200831401066, "learning_rate": 4.304982921034886e-06, "loss": 0.1128, "step": 8205 }, { "epoch": 1.12, "grad_norm": 4.563319172970982, "learning_rate": 4.303892091569791e-06, "loss": 0.1254, "step": 8206 }, { "epoch": 1.12, "grad_norm": 3.8334337266892033, "learning_rate": 4.302801295890731e-06, "loss": 0.1924, "step": 8207 }, { "epoch": 1.12, "grad_norm": 3.8392040540104975, "learning_rate": 4.3017105340506536e-06, "loss": 0.1469, "step": 8208 }, { "epoch": 1.12, "grad_norm": 3.61355580385572, "learning_rate": 4.300619806102494e-06, "loss": 0.1976, "step": 8209 }, { "epoch": 1.12, "grad_norm": 3.665152810302657, "learning_rate": 4.2995291120991975e-06, "loss": 0.1856, "step": 8210 }, { "epoch": 1.12, "grad_norm": 3.9776195145670403, "learning_rate": 4.298438452093696e-06, "loss": 0.1645, "step": 8211 }, { "epoch": 1.12, "grad_norm": 3.714231132343667, "learning_rate": 4.297347826138929e-06, "loss": 0.1579, "step": 8212 }, { "epoch": 1.12, "grad_norm": 3.4138268353493024, "learning_rate": 4.296257234287831e-06, "loss": 0.1749, "step": 8213 }, { "epoch": 1.12, "grad_norm": 4.109089775593131, "learning_rate": 4.2951666765933294e-06, "loss": 0.1749, "step": 8214 }, { "epoch": 1.12, "grad_norm": 3.505086649557002, "learning_rate": 4.294076153108363e-06, "loss": 0.1518, "step": 8215 }, { "epoch": 1.12, "grad_norm": 3.294182542055754, "learning_rate": 4.292985663885854e-06, "loss": 0.1233, "step": 8216 }, { "epoch": 1.12, "grad_norm": 4.177319132184675, "learning_rate": 4.291895208978734e-06, "loss": 0.168, "step": 8217 }, { "epoch": 1.12, "grad_norm": 2.802427833921026, "learning_rate": 4.2908047884399265e-06, "loss": 0.1401, "step": 8218 }, { "epoch": 1.12, "grad_norm": 3.8575012932893054, "learning_rate": 4.28971440232236e-06, "loss": 0.211, "step": 8219 }, { "epoch": 1.12, "grad_norm": 3.9646034353412984, "learning_rate": 4.288624050678951e-06, "loss": 0.184, "step": 8220 }, { "epoch": 1.12, "grad_norm": 3.976625569601717, "learning_rate": 4.287533733562627e-06, "loss": 0.1859, "step": 8221 }, { "epoch": 1.12, "grad_norm": 3.6210348185564434, "learning_rate": 4.2864434510263e-06, "loss": 0.2143, "step": 8222 }, { "epoch": 1.12, "grad_norm": 4.160690156844245, "learning_rate": 4.285353203122894e-06, "loss": 0.1472, "step": 8223 }, { "epoch": 1.12, "grad_norm": 3.8143266791058683, "learning_rate": 4.28426298990532e-06, "loss": 0.1983, "step": 8224 }, { "epoch": 1.12, "grad_norm": 3.6155096272048497, "learning_rate": 4.283172811426497e-06, "loss": 0.1691, "step": 8225 }, { "epoch": 1.12, "grad_norm": 3.2807728199705517, "learning_rate": 4.282082667739333e-06, "loss": 0.1693, "step": 8226 }, { "epoch": 1.12, "grad_norm": 3.719538830035363, "learning_rate": 4.280992558896742e-06, "loss": 0.1932, "step": 8227 }, { "epoch": 1.12, "grad_norm": 4.527903423488567, "learning_rate": 4.27990248495163e-06, "loss": 0.1496, "step": 8228 }, { "epoch": 1.12, "grad_norm": 3.081900734375216, "learning_rate": 4.278812445956909e-06, "loss": 0.1538, "step": 8229 }, { "epoch": 1.12, "grad_norm": 2.793430876085046, "learning_rate": 4.27772244196548e-06, "loss": 0.1334, "step": 8230 }, { "epoch": 1.12, "grad_norm": 3.4227101418369736, "learning_rate": 4.27663247303025e-06, "loss": 0.1811, "step": 8231 }, { "epoch": 1.12, "grad_norm": 8.254826624291566, "learning_rate": 4.275542539204118e-06, "loss": 0.1749, "step": 8232 }, { "epoch": 1.12, "grad_norm": 3.6128058305551454, "learning_rate": 4.274452640539991e-06, "loss": 0.2007, "step": 8233 }, { "epoch": 1.12, "grad_norm": 3.724265228993697, "learning_rate": 4.27336277709076e-06, "loss": 0.1939, "step": 8234 }, { "epoch": 1.12, "grad_norm": 2.9768762791373797, "learning_rate": 4.27227294890933e-06, "loss": 0.1512, "step": 8235 }, { "epoch": 1.12, "grad_norm": 3.4517373762604167, "learning_rate": 4.271183156048589e-06, "loss": 0.1518, "step": 8236 }, { "epoch": 1.12, "grad_norm": 2.845885812650351, "learning_rate": 4.270093398561437e-06, "loss": 0.1515, "step": 8237 }, { "epoch": 1.12, "grad_norm": 4.118554433794265, "learning_rate": 4.269003676500761e-06, "loss": 0.2205, "step": 8238 }, { "epoch": 1.12, "grad_norm": 2.7589118878450662, "learning_rate": 4.267913989919457e-06, "loss": 0.1793, "step": 8239 }, { "epoch": 1.12, "grad_norm": 3.725192201300809, "learning_rate": 4.266824338870408e-06, "loss": 0.1971, "step": 8240 }, { "epoch": 1.12, "grad_norm": 4.77384822045291, "learning_rate": 4.265734723406506e-06, "loss": 0.1767, "step": 8241 }, { "epoch": 1.12, "grad_norm": 3.7409865610088477, "learning_rate": 4.26464514358063e-06, "loss": 0.1772, "step": 8242 }, { "epoch": 1.12, "grad_norm": 4.1935339476090885, "learning_rate": 4.263555599445669e-06, "loss": 0.1795, "step": 8243 }, { "epoch": 1.12, "grad_norm": 4.654377874673795, "learning_rate": 4.2624660910545025e-06, "loss": 0.1943, "step": 8244 }, { "epoch": 1.12, "grad_norm": 3.8211174405745743, "learning_rate": 4.261376618460011e-06, "loss": 0.1685, "step": 8245 }, { "epoch": 1.12, "grad_norm": 3.290531380151182, "learning_rate": 4.260287181715072e-06, "loss": 0.1318, "step": 8246 }, { "epoch": 1.12, "grad_norm": 3.7692829091908355, "learning_rate": 4.259197780872562e-06, "loss": 0.1573, "step": 8247 }, { "epoch": 1.12, "grad_norm": 2.993136533318111, "learning_rate": 4.2581084159853565e-06, "loss": 0.1688, "step": 8248 }, { "epoch": 1.12, "grad_norm": 3.634657564575061, "learning_rate": 4.25701908710633e-06, "loss": 0.1963, "step": 8249 }, { "epoch": 1.12, "grad_norm": 3.4794475536269243, "learning_rate": 4.255929794288349e-06, "loss": 0.1806, "step": 8250 }, { "epoch": 1.12, "grad_norm": 7.90172718423494, "learning_rate": 4.254840537584287e-06, "loss": 0.1861, "step": 8251 }, { "epoch": 1.12, "grad_norm": 3.8439240706985176, "learning_rate": 4.2537513170470105e-06, "loss": 0.1735, "step": 8252 }, { "epoch": 1.12, "grad_norm": 2.958908275615361, "learning_rate": 4.252662132729388e-06, "loss": 0.1482, "step": 8253 }, { "epoch": 1.12, "grad_norm": 3.7093820640868964, "learning_rate": 4.251572984684281e-06, "loss": 0.1913, "step": 8254 }, { "epoch": 1.12, "grad_norm": 4.203711588017176, "learning_rate": 4.25048387296455e-06, "loss": 0.1545, "step": 8255 }, { "epoch": 1.12, "grad_norm": 3.3711221423278586, "learning_rate": 4.24939479762306e-06, "loss": 0.1544, "step": 8256 }, { "epoch": 1.12, "grad_norm": 3.6054319646838997, "learning_rate": 4.248305758712666e-06, "loss": 0.1472, "step": 8257 }, { "epoch": 1.12, "grad_norm": 4.815515368665709, "learning_rate": 4.24721675628623e-06, "loss": 0.1794, "step": 8258 }, { "epoch": 1.12, "grad_norm": 3.4160907992854153, "learning_rate": 4.246127790396604e-06, "loss": 0.1718, "step": 8259 }, { "epoch": 1.12, "grad_norm": 2.88532804157681, "learning_rate": 4.245038861096643e-06, "loss": 0.158, "step": 8260 }, { "epoch": 1.12, "grad_norm": 3.6175664961932164, "learning_rate": 4.243949968439196e-06, "loss": 0.1936, "step": 8261 }, { "epoch": 1.12, "grad_norm": 3.3354155450905614, "learning_rate": 4.2428611124771184e-06, "loss": 0.1619, "step": 8262 }, { "epoch": 1.12, "grad_norm": 4.2730035446738395, "learning_rate": 4.2417722932632536e-06, "loss": 0.1646, "step": 8263 }, { "epoch": 1.12, "grad_norm": 3.5683245096779626, "learning_rate": 4.24068351085045e-06, "loss": 0.1643, "step": 8264 }, { "epoch": 1.12, "grad_norm": 3.9150204619805447, "learning_rate": 4.2395947652915525e-06, "loss": 0.1851, "step": 8265 }, { "epoch": 1.12, "grad_norm": 2.680220950947211, "learning_rate": 4.238506056639406e-06, "loss": 0.1449, "step": 8266 }, { "epoch": 1.12, "grad_norm": 4.762867058087735, "learning_rate": 4.237417384946846e-06, "loss": 0.1538, "step": 8267 }, { "epoch": 1.12, "grad_norm": 3.488085598192113, "learning_rate": 4.236328750266719e-06, "loss": 0.1728, "step": 8268 }, { "epoch": 1.12, "grad_norm": 4.0290882666014625, "learning_rate": 4.235240152651857e-06, "loss": 0.1579, "step": 8269 }, { "epoch": 1.12, "grad_norm": 4.393889186540398, "learning_rate": 4.234151592155098e-06, "loss": 0.1877, "step": 8270 }, { "epoch": 1.13, "grad_norm": 3.9517514630719845, "learning_rate": 4.233063068829275e-06, "loss": 0.1626, "step": 8271 }, { "epoch": 1.13, "grad_norm": 3.94457277695254, "learning_rate": 4.231974582727223e-06, "loss": 0.193, "step": 8272 }, { "epoch": 1.13, "grad_norm": 4.013389810508216, "learning_rate": 4.230886133901769e-06, "loss": 0.1967, "step": 8273 }, { "epoch": 1.13, "grad_norm": 3.5560197042017054, "learning_rate": 4.229797722405744e-06, "loss": 0.1324, "step": 8274 }, { "epoch": 1.13, "grad_norm": 2.7458511299309922, "learning_rate": 4.228709348291972e-06, "loss": 0.1128, "step": 8275 }, { "epoch": 1.13, "grad_norm": 3.9316709512488774, "learning_rate": 4.227621011613281e-06, "loss": 0.1962, "step": 8276 }, { "epoch": 1.13, "grad_norm": 3.7924605993821277, "learning_rate": 4.226532712422492e-06, "loss": 0.1581, "step": 8277 }, { "epoch": 1.13, "grad_norm": 3.1310489100210055, "learning_rate": 4.225444450772427e-06, "loss": 0.1914, "step": 8278 }, { "epoch": 1.13, "grad_norm": 2.9649179876944074, "learning_rate": 4.224356226715904e-06, "loss": 0.1661, "step": 8279 }, { "epoch": 1.13, "grad_norm": 3.6491554557653303, "learning_rate": 4.223268040305745e-06, "loss": 0.1751, "step": 8280 }, { "epoch": 1.13, "grad_norm": 7.354250156888074, "learning_rate": 4.2221798915947594e-06, "loss": 0.16, "step": 8281 }, { "epoch": 1.13, "grad_norm": 4.637543470152349, "learning_rate": 4.221091780635768e-06, "loss": 0.1035, "step": 8282 }, { "epoch": 1.13, "grad_norm": 3.8313617169359575, "learning_rate": 4.2200037074815774e-06, "loss": 0.1789, "step": 8283 }, { "epoch": 1.13, "grad_norm": 2.931574551543204, "learning_rate": 4.218915672185001e-06, "loss": 0.1392, "step": 8284 }, { "epoch": 1.13, "grad_norm": 4.1314554402741495, "learning_rate": 4.217827674798845e-06, "loss": 0.202, "step": 8285 }, { "epoch": 1.13, "grad_norm": 3.0975084543323463, "learning_rate": 4.216739715375921e-06, "loss": 0.1371, "step": 8286 }, { "epoch": 1.13, "grad_norm": 3.9155883131762588, "learning_rate": 4.215651793969026e-06, "loss": 0.1352, "step": 8287 }, { "epoch": 1.13, "grad_norm": 4.35456747904097, "learning_rate": 4.21456391063097e-06, "loss": 0.1787, "step": 8288 }, { "epoch": 1.13, "grad_norm": 3.4047282578635443, "learning_rate": 4.21347606541455e-06, "loss": 0.2004, "step": 8289 }, { "epoch": 1.13, "grad_norm": 3.0293530365420667, "learning_rate": 4.2123882583725675e-06, "loss": 0.172, "step": 8290 }, { "epoch": 1.13, "grad_norm": 3.702660117537221, "learning_rate": 4.211300489557817e-06, "loss": 0.1496, "step": 8291 }, { "epoch": 1.13, "grad_norm": 3.4479743007586223, "learning_rate": 4.210212759023099e-06, "loss": 0.1828, "step": 8292 }, { "epoch": 1.13, "grad_norm": 2.722604377608666, "learning_rate": 4.209125066821201e-06, "loss": 0.1638, "step": 8293 }, { "epoch": 1.13, "grad_norm": 2.8642275178172625, "learning_rate": 4.2080374130049215e-06, "loss": 0.1734, "step": 8294 }, { "epoch": 1.13, "grad_norm": 6.3917842504995575, "learning_rate": 4.2069497976270445e-06, "loss": 0.1253, "step": 8295 }, { "epoch": 1.13, "grad_norm": 3.1085140942413663, "learning_rate": 4.2058622207403614e-06, "loss": 0.1739, "step": 8296 }, { "epoch": 1.13, "grad_norm": 2.9336042871637455, "learning_rate": 4.204774682397658e-06, "loss": 0.166, "step": 8297 }, { "epoch": 1.13, "grad_norm": 3.2675208358220837, "learning_rate": 4.203687182651717e-06, "loss": 0.1712, "step": 8298 }, { "epoch": 1.13, "grad_norm": 4.185633613564233, "learning_rate": 4.202599721555324e-06, "loss": 0.237, "step": 8299 }, { "epoch": 1.13, "grad_norm": 3.433481862064495, "learning_rate": 4.201512299161256e-06, "loss": 0.1427, "step": 8300 }, { "epoch": 1.13, "grad_norm": 3.7445991665876486, "learning_rate": 4.200424915522294e-06, "loss": 0.1478, "step": 8301 }, { "epoch": 1.13, "grad_norm": 3.888191505596899, "learning_rate": 4.199337570691214e-06, "loss": 0.1488, "step": 8302 }, { "epoch": 1.13, "grad_norm": 3.8324117711817673, "learning_rate": 4.198250264720792e-06, "loss": 0.1771, "step": 8303 }, { "epoch": 1.13, "grad_norm": 3.984867720124317, "learning_rate": 4.1971629976638e-06, "loss": 0.1503, "step": 8304 }, { "epoch": 1.13, "grad_norm": 3.7775477814384684, "learning_rate": 4.196075769573011e-06, "loss": 0.1863, "step": 8305 }, { "epoch": 1.13, "grad_norm": 4.944966970909835, "learning_rate": 4.194988580501191e-06, "loss": 0.1972, "step": 8306 }, { "epoch": 1.13, "grad_norm": 3.9240088965416984, "learning_rate": 4.1939014305011116e-06, "loss": 0.1921, "step": 8307 }, { "epoch": 1.13, "grad_norm": 3.0258574968805365, "learning_rate": 4.192814319625534e-06, "loss": 0.1947, "step": 8308 }, { "epoch": 1.13, "grad_norm": 5.1357394387959, "learning_rate": 4.191727247927224e-06, "loss": 0.1833, "step": 8309 }, { "epoch": 1.13, "grad_norm": 3.977746128501793, "learning_rate": 4.1906402154589445e-06, "loss": 0.1518, "step": 8310 }, { "epoch": 1.13, "grad_norm": 3.1080272816337544, "learning_rate": 4.189553222273452e-06, "loss": 0.1513, "step": 8311 }, { "epoch": 1.13, "grad_norm": 2.9547844288262657, "learning_rate": 4.188466268423507e-06, "loss": 0.1793, "step": 8312 }, { "epoch": 1.13, "grad_norm": 3.138567593772116, "learning_rate": 4.187379353961867e-06, "loss": 0.1682, "step": 8313 }, { "epoch": 1.13, "grad_norm": 4.117665343245059, "learning_rate": 4.1862924789412815e-06, "loss": 0.2118, "step": 8314 }, { "epoch": 1.13, "grad_norm": 3.874993600977922, "learning_rate": 4.185205643414507e-06, "loss": 0.1792, "step": 8315 }, { "epoch": 1.13, "grad_norm": 3.686407894371995, "learning_rate": 4.18411884743429e-06, "loss": 0.1925, "step": 8316 }, { "epoch": 1.13, "grad_norm": 3.0503703368423327, "learning_rate": 4.183032091053381e-06, "loss": 0.1361, "step": 8317 }, { "epoch": 1.13, "grad_norm": 3.594778106563644, "learning_rate": 4.181945374324526e-06, "loss": 0.1679, "step": 8318 }, { "epoch": 1.13, "grad_norm": 5.866306044037231, "learning_rate": 4.180858697300472e-06, "loss": 0.193, "step": 8319 }, { "epoch": 1.13, "grad_norm": 3.096087570525928, "learning_rate": 4.179772060033956e-06, "loss": 0.1823, "step": 8320 }, { "epoch": 1.13, "grad_norm": 4.036340518635649, "learning_rate": 4.178685462577724e-06, "loss": 0.1725, "step": 8321 }, { "epoch": 1.13, "grad_norm": 5.184206907563696, "learning_rate": 4.1775989049845105e-06, "loss": 0.1684, "step": 8322 }, { "epoch": 1.13, "grad_norm": 5.26788169347876, "learning_rate": 4.176512387307055e-06, "loss": 0.2351, "step": 8323 }, { "epoch": 1.13, "grad_norm": 3.0006255093307206, "learning_rate": 4.175425909598091e-06, "loss": 0.1742, "step": 8324 }, { "epoch": 1.13, "grad_norm": 3.7582729145579132, "learning_rate": 4.174339471910354e-06, "loss": 0.1929, "step": 8325 }, { "epoch": 1.13, "grad_norm": 3.300589385165626, "learning_rate": 4.173253074296571e-06, "loss": 0.1716, "step": 8326 }, { "epoch": 1.13, "grad_norm": 2.700126962806928, "learning_rate": 4.172166716809475e-06, "loss": 0.1373, "step": 8327 }, { "epoch": 1.13, "grad_norm": 4.3040203814836895, "learning_rate": 4.171080399501789e-06, "loss": 0.1928, "step": 8328 }, { "epoch": 1.13, "grad_norm": 3.4328078331824146, "learning_rate": 4.169994122426241e-06, "loss": 0.1517, "step": 8329 }, { "epoch": 1.13, "grad_norm": 3.1168627265637636, "learning_rate": 4.168907885635551e-06, "loss": 0.1783, "step": 8330 }, { "epoch": 1.13, "grad_norm": 3.0768020722971716, "learning_rate": 4.1678216891824445e-06, "loss": 0.1412, "step": 8331 }, { "epoch": 1.13, "grad_norm": 3.1457302511016976, "learning_rate": 4.166735533119638e-06, "loss": 0.1936, "step": 8332 }, { "epoch": 1.13, "grad_norm": 3.670753422235971, "learning_rate": 4.16564941749985e-06, "loss": 0.1855, "step": 8333 }, { "epoch": 1.13, "grad_norm": 3.4685941471251653, "learning_rate": 4.164563342375796e-06, "loss": 0.1836, "step": 8334 }, { "epoch": 1.13, "grad_norm": 3.4464713446761444, "learning_rate": 4.163477307800185e-06, "loss": 0.1929, "step": 8335 }, { "epoch": 1.13, "grad_norm": 3.2749826288211255, "learning_rate": 4.162391313825735e-06, "loss": 0.1438, "step": 8336 }, { "epoch": 1.13, "grad_norm": 3.5592647617744886, "learning_rate": 4.16130536050515e-06, "loss": 0.1447, "step": 8337 }, { "epoch": 1.13, "grad_norm": 2.7561860748556732, "learning_rate": 4.160219447891142e-06, "loss": 0.173, "step": 8338 }, { "epoch": 1.13, "grad_norm": 4.436755231771436, "learning_rate": 4.159133576036412e-06, "loss": 0.1778, "step": 8339 }, { "epoch": 1.13, "grad_norm": 4.36058603689654, "learning_rate": 4.158047744993668e-06, "loss": 0.1559, "step": 8340 }, { "epoch": 1.13, "grad_norm": 3.530349086338644, "learning_rate": 4.156961954815607e-06, "loss": 0.1601, "step": 8341 }, { "epoch": 1.13, "grad_norm": 4.0434212530566915, "learning_rate": 4.155876205554931e-06, "loss": 0.1548, "step": 8342 }, { "epoch": 1.13, "grad_norm": 2.909378551236876, "learning_rate": 4.154790497264337e-06, "loss": 0.1577, "step": 8343 }, { "epoch": 1.14, "grad_norm": 3.584664504232806, "learning_rate": 4.153704829996523e-06, "loss": 0.1893, "step": 8344 }, { "epoch": 1.14, "grad_norm": 3.5092791149321823, "learning_rate": 4.152619203804178e-06, "loss": 0.1942, "step": 8345 }, { "epoch": 1.14, "grad_norm": 3.473302788283992, "learning_rate": 4.151533618739997e-06, "loss": 0.1829, "step": 8346 }, { "epoch": 1.14, "grad_norm": 4.408592391942468, "learning_rate": 4.150448074856667e-06, "loss": 0.1746, "step": 8347 }, { "epoch": 1.14, "grad_norm": 3.617366388292701, "learning_rate": 4.149362572206879e-06, "loss": 0.1856, "step": 8348 }, { "epoch": 1.14, "grad_norm": 3.7140890338340595, "learning_rate": 4.148277110843314e-06, "loss": 0.1892, "step": 8349 }, { "epoch": 1.14, "grad_norm": 4.0209878405402915, "learning_rate": 4.14719169081866e-06, "loss": 0.2181, "step": 8350 }, { "epoch": 1.14, "grad_norm": 3.3552809740306015, "learning_rate": 4.146106312185595e-06, "loss": 0.1908, "step": 8351 }, { "epoch": 1.14, "grad_norm": 5.818408479979114, "learning_rate": 4.145020974996802e-06, "loss": 0.2221, "step": 8352 }, { "epoch": 1.14, "grad_norm": 3.862752236082263, "learning_rate": 4.143935679304955e-06, "loss": 0.1791, "step": 8353 }, { "epoch": 1.14, "grad_norm": 3.80435411253464, "learning_rate": 4.1428504251627335e-06, "loss": 0.1671, "step": 8354 }, { "epoch": 1.14, "grad_norm": 3.6455571400135702, "learning_rate": 4.141765212622806e-06, "loss": 0.1925, "step": 8355 }, { "epoch": 1.14, "grad_norm": 2.9615201095246415, "learning_rate": 4.140680041737847e-06, "loss": 0.1605, "step": 8356 }, { "epoch": 1.14, "grad_norm": 4.494268727278509, "learning_rate": 4.139594912560526e-06, "loss": 0.1695, "step": 8357 }, { "epoch": 1.14, "grad_norm": 5.035468125243935, "learning_rate": 4.138509825143511e-06, "loss": 0.1834, "step": 8358 }, { "epoch": 1.14, "grad_norm": 3.7298016967687473, "learning_rate": 4.137424779539465e-06, "loss": 0.1835, "step": 8359 }, { "epoch": 1.14, "grad_norm": 2.921274523687139, "learning_rate": 4.136339775801055e-06, "loss": 0.1678, "step": 8360 }, { "epoch": 1.14, "grad_norm": 2.7664663288431526, "learning_rate": 4.135254813980938e-06, "loss": 0.1368, "step": 8361 }, { "epoch": 1.14, "grad_norm": 4.691589547464333, "learning_rate": 4.134169894131776e-06, "loss": 0.1487, "step": 8362 }, { "epoch": 1.14, "grad_norm": 3.219248807720665, "learning_rate": 4.133085016306226e-06, "loss": 0.1753, "step": 8363 }, { "epoch": 1.14, "grad_norm": 5.005022447332038, "learning_rate": 4.1320001805569434e-06, "loss": 0.1837, "step": 8364 }, { "epoch": 1.14, "grad_norm": 3.3672861534558125, "learning_rate": 4.130915386936579e-06, "loss": 0.17, "step": 8365 }, { "epoch": 1.14, "grad_norm": 3.628854603628694, "learning_rate": 4.129830635497789e-06, "loss": 0.1464, "step": 8366 }, { "epoch": 1.14, "grad_norm": 3.3893088275963232, "learning_rate": 4.1287459262932164e-06, "loss": 0.1504, "step": 8367 }, { "epoch": 1.14, "grad_norm": 3.5873440000463783, "learning_rate": 4.127661259375514e-06, "loss": 0.1536, "step": 8368 }, { "epoch": 1.14, "grad_norm": 4.607751866034252, "learning_rate": 4.126576634797322e-06, "loss": 0.1851, "step": 8369 }, { "epoch": 1.14, "grad_norm": 3.5628878005893267, "learning_rate": 4.125492052611286e-06, "loss": 0.1691, "step": 8370 }, { "epoch": 1.14, "grad_norm": 3.7968835151700535, "learning_rate": 4.124407512870045e-06, "loss": 0.1685, "step": 8371 }, { "epoch": 1.14, "grad_norm": 3.211350461047402, "learning_rate": 4.123323015626241e-06, "loss": 0.1856, "step": 8372 }, { "epoch": 1.14, "grad_norm": 4.064347585661098, "learning_rate": 4.122238560932508e-06, "loss": 0.1868, "step": 8373 }, { "epoch": 1.14, "grad_norm": 3.327818931718727, "learning_rate": 4.121154148841482e-06, "loss": 0.1704, "step": 8374 }, { "epoch": 1.14, "grad_norm": 3.357198036210817, "learning_rate": 4.120069779405795e-06, "loss": 0.1509, "step": 8375 }, { "epoch": 1.14, "grad_norm": 3.9452297739661173, "learning_rate": 4.118985452678077e-06, "loss": 0.2255, "step": 8376 }, { "epoch": 1.14, "grad_norm": 4.363562937059293, "learning_rate": 4.11790116871096e-06, "loss": 0.211, "step": 8377 }, { "epoch": 1.14, "grad_norm": 3.376233125212524, "learning_rate": 4.116816927557063e-06, "loss": 0.1748, "step": 8378 }, { "epoch": 1.14, "grad_norm": 2.972086266387172, "learning_rate": 4.115732729269019e-06, "loss": 0.1524, "step": 8379 }, { "epoch": 1.14, "grad_norm": 2.90985936692282, "learning_rate": 4.114648573899443e-06, "loss": 0.1509, "step": 8380 }, { "epoch": 1.14, "grad_norm": 3.1842259035375466, "learning_rate": 4.113564461500962e-06, "loss": 0.1736, "step": 8381 }, { "epoch": 1.14, "grad_norm": 2.924985898834414, "learning_rate": 4.112480392126187e-06, "loss": 0.1948, "step": 8382 }, { "epoch": 1.14, "grad_norm": 3.6203043415485188, "learning_rate": 4.111396365827739e-06, "loss": 0.194, "step": 8383 }, { "epoch": 1.14, "grad_norm": 3.970536428288937, "learning_rate": 4.110312382658229e-06, "loss": 0.2127, "step": 8384 }, { "epoch": 1.14, "grad_norm": 3.7164131379889938, "learning_rate": 4.1092284426702724e-06, "loss": 0.1748, "step": 8385 }, { "epoch": 1.14, "grad_norm": 3.9242351710769823, "learning_rate": 4.108144545916474e-06, "loss": 0.2166, "step": 8386 }, { "epoch": 1.14, "grad_norm": 3.9291324512591186, "learning_rate": 4.107060692449447e-06, "loss": 0.206, "step": 8387 }, { "epoch": 1.14, "grad_norm": 3.7916962074139358, "learning_rate": 4.105976882321792e-06, "loss": 0.1558, "step": 8388 }, { "epoch": 1.14, "grad_norm": 2.9902506678423357, "learning_rate": 4.104893115586115e-06, "loss": 0.1399, "step": 8389 }, { "epoch": 1.14, "grad_norm": 2.9357253080183217, "learning_rate": 4.103809392295015e-06, "loss": 0.1673, "step": 8390 }, { "epoch": 1.14, "grad_norm": 4.4511539615012, "learning_rate": 4.102725712501095e-06, "loss": 0.1817, "step": 8391 }, { "epoch": 1.14, "grad_norm": 3.6754608337659733, "learning_rate": 4.1016420762569496e-06, "loss": 0.1799, "step": 8392 }, { "epoch": 1.14, "grad_norm": 3.960472238644347, "learning_rate": 4.100558483615174e-06, "loss": 0.1519, "step": 8393 }, { "epoch": 1.14, "grad_norm": 3.269831025005755, "learning_rate": 4.0994749346283615e-06, "loss": 0.1638, "step": 8394 }, { "epoch": 1.14, "grad_norm": 3.750755802885957, "learning_rate": 4.098391429349102e-06, "loss": 0.1795, "step": 8395 }, { "epoch": 1.14, "grad_norm": 3.4537444172870684, "learning_rate": 4.097307967829985e-06, "loss": 0.1874, "step": 8396 }, { "epoch": 1.14, "grad_norm": 3.9716047742433203, "learning_rate": 4.096224550123597e-06, "loss": 0.1848, "step": 8397 }, { "epoch": 1.14, "grad_norm": 3.6518307884054684, "learning_rate": 4.09514117628252e-06, "loss": 0.1596, "step": 8398 }, { "epoch": 1.14, "grad_norm": 3.823247860111384, "learning_rate": 4.09405784635934e-06, "loss": 0.2025, "step": 8399 }, { "epoch": 1.14, "grad_norm": 3.539130481358636, "learning_rate": 4.092974560406635e-06, "loss": 0.1687, "step": 8400 }, { "epoch": 1.14, "grad_norm": 3.2308178106481824, "learning_rate": 4.091891318476984e-06, "loss": 0.1671, "step": 8401 }, { "epoch": 1.14, "grad_norm": 4.572302611838796, "learning_rate": 4.090808120622961e-06, "loss": 0.1727, "step": 8402 }, { "epoch": 1.14, "grad_norm": 3.187415113274465, "learning_rate": 4.089724966897141e-06, "loss": 0.1775, "step": 8403 }, { "epoch": 1.14, "grad_norm": 3.7402004782641654, "learning_rate": 4.0886418573520945e-06, "loss": 0.1732, "step": 8404 }, { "epoch": 1.14, "grad_norm": 3.7916468610455887, "learning_rate": 4.087558792040395e-06, "loss": 0.1578, "step": 8405 }, { "epoch": 1.14, "grad_norm": 4.210088264987006, "learning_rate": 4.086475771014603e-06, "loss": 0.1604, "step": 8406 }, { "epoch": 1.14, "grad_norm": 3.8954421788616354, "learning_rate": 4.08539279432729e-06, "loss": 0.1574, "step": 8407 }, { "epoch": 1.14, "grad_norm": 3.154132770628636, "learning_rate": 4.084309862031014e-06, "loss": 0.1732, "step": 8408 }, { "epoch": 1.14, "grad_norm": 4.061657289919419, "learning_rate": 4.083226974178339e-06, "loss": 0.2025, "step": 8409 }, { "epoch": 1.14, "grad_norm": 3.2830484740914847, "learning_rate": 4.082144130821821e-06, "loss": 0.2022, "step": 8410 }, { "epoch": 1.14, "grad_norm": 3.144392600483447, "learning_rate": 4.081061332014021e-06, "loss": 0.1811, "step": 8411 }, { "epoch": 1.14, "grad_norm": 3.5433042840660973, "learning_rate": 4.079978577807487e-06, "loss": 0.181, "step": 8412 }, { "epoch": 1.14, "grad_norm": 3.4839495587679847, "learning_rate": 4.078895868254778e-06, "loss": 0.1575, "step": 8413 }, { "epoch": 1.14, "grad_norm": 2.4647559578458176, "learning_rate": 4.077813203408439e-06, "loss": 0.1382, "step": 8414 }, { "epoch": 1.14, "grad_norm": 2.938452969127293, "learning_rate": 4.0767305833210195e-06, "loss": 0.1463, "step": 8415 }, { "epoch": 1.14, "grad_norm": 3.7188686833607276, "learning_rate": 4.075648008045065e-06, "loss": 0.175, "step": 8416 }, { "epoch": 1.14, "grad_norm": 2.936216228704969, "learning_rate": 4.074565477633117e-06, "loss": 0.1497, "step": 8417 }, { "epoch": 1.15, "grad_norm": 3.2870572322325486, "learning_rate": 4.073482992137722e-06, "loss": 0.1537, "step": 8418 }, { "epoch": 1.15, "grad_norm": 3.454422377969144, "learning_rate": 4.072400551611413e-06, "loss": 0.1703, "step": 8419 }, { "epoch": 1.15, "grad_norm": 3.315756479056682, "learning_rate": 4.071318156106733e-06, "loss": 0.1857, "step": 8420 }, { "epoch": 1.15, "grad_norm": 3.851245163284471, "learning_rate": 4.070235805676211e-06, "loss": 0.1509, "step": 8421 }, { "epoch": 1.15, "grad_norm": 4.54349151711733, "learning_rate": 4.069153500372382e-06, "loss": 0.1421, "step": 8422 }, { "epoch": 1.15, "grad_norm": 4.057862131626687, "learning_rate": 4.0680712402477765e-06, "loss": 0.1941, "step": 8423 }, { "epoch": 1.15, "grad_norm": 3.795970574623767, "learning_rate": 4.066989025354925e-06, "loss": 0.1903, "step": 8424 }, { "epoch": 1.15, "grad_norm": 3.4591494442016497, "learning_rate": 4.065906855746349e-06, "loss": 0.1849, "step": 8425 }, { "epoch": 1.15, "grad_norm": 4.241934366318842, "learning_rate": 4.064824731474576e-06, "loss": 0.1767, "step": 8426 }, { "epoch": 1.15, "grad_norm": 4.231017737290494, "learning_rate": 4.063742652592125e-06, "loss": 0.168, "step": 8427 }, { "epoch": 1.15, "grad_norm": 3.9887755248741747, "learning_rate": 4.062660619151517e-06, "loss": 0.1394, "step": 8428 }, { "epoch": 1.15, "grad_norm": 4.49237973725227, "learning_rate": 4.061578631205268e-06, "loss": 0.2172, "step": 8429 }, { "epoch": 1.15, "grad_norm": 4.845672197372578, "learning_rate": 4.060496688805894e-06, "loss": 0.1466, "step": 8430 }, { "epoch": 1.15, "grad_norm": 4.246984536367952, "learning_rate": 4.059414792005906e-06, "loss": 0.1702, "step": 8431 }, { "epoch": 1.15, "grad_norm": 4.780709829810863, "learning_rate": 4.0583329408578185e-06, "loss": 0.2213, "step": 8432 }, { "epoch": 1.15, "grad_norm": 2.886464168469267, "learning_rate": 4.057251135414135e-06, "loss": 0.1664, "step": 8433 }, { "epoch": 1.15, "grad_norm": 3.516393329834943, "learning_rate": 4.0561693757273665e-06, "loss": 0.1657, "step": 8434 }, { "epoch": 1.15, "grad_norm": 4.88280596090049, "learning_rate": 4.0550876618500115e-06, "loss": 0.1984, "step": 8435 }, { "epoch": 1.15, "grad_norm": 2.749900670749303, "learning_rate": 4.054005993834575e-06, "loss": 0.1417, "step": 8436 }, { "epoch": 1.15, "grad_norm": 3.660925383072199, "learning_rate": 4.052924371733555e-06, "loss": 0.1715, "step": 8437 }, { "epoch": 1.15, "grad_norm": 3.297182983069375, "learning_rate": 4.051842795599452e-06, "loss": 0.1583, "step": 8438 }, { "epoch": 1.15, "grad_norm": 3.079854157174554, "learning_rate": 4.050761265484756e-06, "loss": 0.1863, "step": 8439 }, { "epoch": 1.15, "grad_norm": 3.170988019301167, "learning_rate": 4.0496797814419655e-06, "loss": 0.1683, "step": 8440 }, { "epoch": 1.15, "grad_norm": 3.197304611659776, "learning_rate": 4.048598343523564e-06, "loss": 0.1589, "step": 8441 }, { "epoch": 1.15, "grad_norm": 3.9575981180164166, "learning_rate": 4.047516951782046e-06, "loss": 0.1465, "step": 8442 }, { "epoch": 1.15, "grad_norm": 3.1953435641125356, "learning_rate": 4.046435606269893e-06, "loss": 0.1446, "step": 8443 }, { "epoch": 1.15, "grad_norm": 3.4195336691997906, "learning_rate": 4.0453543070395935e-06, "loss": 0.1313, "step": 8444 }, { "epoch": 1.15, "grad_norm": 2.765021146004408, "learning_rate": 4.044273054143624e-06, "loss": 0.179, "step": 8445 }, { "epoch": 1.15, "grad_norm": 3.5032183518641022, "learning_rate": 4.043191847634469e-06, "loss": 0.1659, "step": 8446 }, { "epoch": 1.15, "grad_norm": 2.8060266610536764, "learning_rate": 4.0421106875646e-06, "loss": 0.1631, "step": 8447 }, { "epoch": 1.15, "grad_norm": 3.5817322662553646, "learning_rate": 4.0410295739864965e-06, "loss": 0.1423, "step": 8448 }, { "epoch": 1.15, "grad_norm": 3.391456660168398, "learning_rate": 4.039948506952628e-06, "loss": 0.1867, "step": 8449 }, { "epoch": 1.15, "grad_norm": 2.8782034243467325, "learning_rate": 4.038867486515466e-06, "loss": 0.1433, "step": 8450 }, { "epoch": 1.15, "grad_norm": 4.935196526215368, "learning_rate": 4.0377865127274775e-06, "loss": 0.1816, "step": 8451 }, { "epoch": 1.15, "grad_norm": 3.7390325022905007, "learning_rate": 4.036705585641131e-06, "loss": 0.1657, "step": 8452 }, { "epoch": 1.15, "grad_norm": 2.8609522616091145, "learning_rate": 4.035624705308887e-06, "loss": 0.1303, "step": 8453 }, { "epoch": 1.15, "grad_norm": 2.7791484894086502, "learning_rate": 4.034543871783209e-06, "loss": 0.1869, "step": 8454 }, { "epoch": 1.15, "grad_norm": 4.501292164181746, "learning_rate": 4.033463085116554e-06, "loss": 0.1987, "step": 8455 }, { "epoch": 1.15, "grad_norm": 3.49512370550919, "learning_rate": 4.032382345361378e-06, "loss": 0.1753, "step": 8456 }, { "epoch": 1.15, "grad_norm": 3.969447026430148, "learning_rate": 4.031301652570139e-06, "loss": 0.1725, "step": 8457 }, { "epoch": 1.15, "grad_norm": 3.144995904938718, "learning_rate": 4.0302210067952845e-06, "loss": 0.1238, "step": 8458 }, { "epoch": 1.15, "grad_norm": 2.82987335107156, "learning_rate": 4.029140408089269e-06, "loss": 0.1899, "step": 8459 }, { "epoch": 1.15, "grad_norm": 2.7794385873408207, "learning_rate": 4.028059856504536e-06, "loss": 0.1215, "step": 8460 }, { "epoch": 1.15, "grad_norm": 4.537247821860626, "learning_rate": 4.026979352093533e-06, "loss": 0.2074, "step": 8461 }, { "epoch": 1.15, "grad_norm": 3.514833069968531, "learning_rate": 4.0258988949087015e-06, "loss": 0.1963, "step": 8462 }, { "epoch": 1.15, "grad_norm": 5.138567286948426, "learning_rate": 4.024818485002486e-06, "loss": 0.1883, "step": 8463 }, { "epoch": 1.15, "grad_norm": 3.8578753956415914, "learning_rate": 4.023738122427319e-06, "loss": 0.1565, "step": 8464 }, { "epoch": 1.15, "grad_norm": 3.2496086928008836, "learning_rate": 4.0226578072356424e-06, "loss": 0.1929, "step": 8465 }, { "epoch": 1.15, "grad_norm": 4.029797538243893, "learning_rate": 4.0215775394798845e-06, "loss": 0.2055, "step": 8466 }, { "epoch": 1.15, "grad_norm": 2.717500872071351, "learning_rate": 4.020497319212482e-06, "loss": 0.1242, "step": 8467 }, { "epoch": 1.15, "grad_norm": 3.8713026034096503, "learning_rate": 4.019417146485859e-06, "loss": 0.1752, "step": 8468 }, { "epoch": 1.15, "grad_norm": 4.141379810673315, "learning_rate": 4.018337021352447e-06, "loss": 0.16, "step": 8469 }, { "epoch": 1.15, "grad_norm": 4.117802164968161, "learning_rate": 4.017256943864666e-06, "loss": 0.1748, "step": 8470 }, { "epoch": 1.15, "grad_norm": 3.1043092304661726, "learning_rate": 4.016176914074944e-06, "loss": 0.1768, "step": 8471 }, { "epoch": 1.15, "grad_norm": 3.9232791377244816, "learning_rate": 4.015096932035695e-06, "loss": 0.2105, "step": 8472 }, { "epoch": 1.15, "grad_norm": 4.150684898261125, "learning_rate": 4.014016997799341e-06, "loss": 0.2246, "step": 8473 }, { "epoch": 1.15, "grad_norm": 2.9836076125746613, "learning_rate": 4.012937111418294e-06, "loss": 0.1504, "step": 8474 }, { "epoch": 1.15, "grad_norm": 3.5755695746449208, "learning_rate": 4.011857272944969e-06, "loss": 0.1562, "step": 8475 }, { "epoch": 1.15, "grad_norm": 3.4874800145113367, "learning_rate": 4.010777482431776e-06, "loss": 0.1629, "step": 8476 }, { "epoch": 1.15, "grad_norm": 3.58238358499681, "learning_rate": 4.009697739931125e-06, "loss": 0.1501, "step": 8477 }, { "epoch": 1.15, "grad_norm": 3.0087080390742145, "learning_rate": 4.008618045495418e-06, "loss": 0.135, "step": 8478 }, { "epoch": 1.15, "grad_norm": 2.4624508292852054, "learning_rate": 4.007538399177064e-06, "loss": 0.1344, "step": 8479 }, { "epoch": 1.15, "grad_norm": 3.7520357828432243, "learning_rate": 4.006458801028459e-06, "loss": 0.1795, "step": 8480 }, { "epoch": 1.15, "grad_norm": 3.976298204895422, "learning_rate": 4.0053792511020065e-06, "loss": 0.162, "step": 8481 }, { "epoch": 1.15, "grad_norm": 3.5509438474810247, "learning_rate": 4.004299749450099e-06, "loss": 0.1177, "step": 8482 }, { "epoch": 1.15, "grad_norm": 3.752399863532835, "learning_rate": 4.003220296125133e-06, "loss": 0.1777, "step": 8483 }, { "epoch": 1.15, "grad_norm": 3.611813273574041, "learning_rate": 4.0021408911795e-06, "loss": 0.1808, "step": 8484 }, { "epoch": 1.15, "grad_norm": 3.4435201060220533, "learning_rate": 4.001061534665593e-06, "loss": 0.1594, "step": 8485 }, { "epoch": 1.15, "grad_norm": 3.4375063601131335, "learning_rate": 3.999982226635793e-06, "loss": 0.1935, "step": 8486 }, { "epoch": 1.15, "grad_norm": 3.4080865012955157, "learning_rate": 3.99890296714249e-06, "loss": 0.1547, "step": 8487 }, { "epoch": 1.15, "grad_norm": 3.8208118962706954, "learning_rate": 3.9978237562380614e-06, "loss": 0.1919, "step": 8488 }, { "epoch": 1.15, "grad_norm": 4.402860996303749, "learning_rate": 3.996744593974893e-06, "loss": 0.1828, "step": 8489 }, { "epoch": 1.15, "grad_norm": 3.8268458895630477, "learning_rate": 3.995665480405359e-06, "loss": 0.1565, "step": 8490 }, { "epoch": 1.16, "grad_norm": 3.5617728382334493, "learning_rate": 3.994586415581838e-06, "loss": 0.1879, "step": 8491 }, { "epoch": 1.16, "grad_norm": 3.4811293935595775, "learning_rate": 3.993507399556699e-06, "loss": 0.1459, "step": 8492 }, { "epoch": 1.16, "grad_norm": 3.670315978482273, "learning_rate": 3.9924284323823175e-06, "loss": 0.2052, "step": 8493 }, { "epoch": 1.16, "grad_norm": 4.159964712917059, "learning_rate": 3.991349514111058e-06, "loss": 0.1736, "step": 8494 }, { "epoch": 1.16, "grad_norm": 3.4203438243881004, "learning_rate": 3.990270644795286e-06, "loss": 0.1655, "step": 8495 }, { "epoch": 1.16, "grad_norm": 3.587045560749234, "learning_rate": 3.98919182448737e-06, "loss": 0.1821, "step": 8496 }, { "epoch": 1.16, "grad_norm": 3.658832733525305, "learning_rate": 3.988113053239664e-06, "loss": 0.1775, "step": 8497 }, { "epoch": 1.16, "grad_norm": 4.128317133047623, "learning_rate": 3.987034331104534e-06, "loss": 0.2081, "step": 8498 }, { "epoch": 1.16, "grad_norm": 3.56891100131563, "learning_rate": 3.985955658134332e-06, "loss": 0.1313, "step": 8499 }, { "epoch": 1.16, "grad_norm": 4.275737864810722, "learning_rate": 3.984877034381415e-06, "loss": 0.1587, "step": 8500 }, { "epoch": 1.16, "grad_norm": 2.8656939522459046, "learning_rate": 3.9837984598981305e-06, "loss": 0.1864, "step": 8501 }, { "epoch": 1.16, "grad_norm": 2.979218554374374, "learning_rate": 3.982719934736832e-06, "loss": 0.1467, "step": 8502 }, { "epoch": 1.16, "grad_norm": 3.5354472033736424, "learning_rate": 3.981641458949864e-06, "loss": 0.1535, "step": 8503 }, { "epoch": 1.16, "grad_norm": 3.5046024639866102, "learning_rate": 3.980563032589573e-06, "loss": 0.1861, "step": 8504 }, { "epoch": 1.16, "grad_norm": 3.536372840646263, "learning_rate": 3.979484655708298e-06, "loss": 0.2036, "step": 8505 }, { "epoch": 1.16, "grad_norm": 4.196022296218258, "learning_rate": 3.978406328358384e-06, "loss": 0.1596, "step": 8506 }, { "epoch": 1.16, "grad_norm": 3.855772008260579, "learning_rate": 3.977328050592161e-06, "loss": 0.1685, "step": 8507 }, { "epoch": 1.16, "grad_norm": 4.95587771302192, "learning_rate": 3.976249822461969e-06, "loss": 0.169, "step": 8508 }, { "epoch": 1.16, "grad_norm": 3.554551246512132, "learning_rate": 3.97517164402014e-06, "loss": 0.1893, "step": 8509 }, { "epoch": 1.16, "grad_norm": 3.3624342358629913, "learning_rate": 3.974093515319004e-06, "loss": 0.1483, "step": 8510 }, { "epoch": 1.16, "grad_norm": 3.8937707874971808, "learning_rate": 3.973015436410886e-06, "loss": 0.1931, "step": 8511 }, { "epoch": 1.16, "grad_norm": 3.8378455332706793, "learning_rate": 3.971937407348115e-06, "loss": 0.177, "step": 8512 }, { "epoch": 1.16, "grad_norm": 3.915909923150807, "learning_rate": 3.970859428183011e-06, "loss": 0.175, "step": 8513 }, { "epoch": 1.16, "grad_norm": 4.998723070105655, "learning_rate": 3.969781498967896e-06, "loss": 0.1787, "step": 8514 }, { "epoch": 1.16, "grad_norm": 6.060183250634604, "learning_rate": 3.968703619755088e-06, "loss": 0.1759, "step": 8515 }, { "epoch": 1.16, "grad_norm": 2.5225750936376006, "learning_rate": 3.9676257905969e-06, "loss": 0.1326, "step": 8516 }, { "epoch": 1.16, "grad_norm": 5.664476478870728, "learning_rate": 3.966548011545648e-06, "loss": 0.1406, "step": 8517 }, { "epoch": 1.16, "grad_norm": 3.8431014058415127, "learning_rate": 3.965470282653644e-06, "loss": 0.1652, "step": 8518 }, { "epoch": 1.16, "grad_norm": 3.4883239923059866, "learning_rate": 3.964392603973191e-06, "loss": 0.186, "step": 8519 }, { "epoch": 1.16, "grad_norm": 3.4780364509679025, "learning_rate": 3.9633149755566e-06, "loss": 0.1619, "step": 8520 }, { "epoch": 1.16, "grad_norm": 3.593128324298106, "learning_rate": 3.962237397456171e-06, "loss": 0.1412, "step": 8521 }, { "epoch": 1.16, "grad_norm": 4.220089698854114, "learning_rate": 3.961159869724207e-06, "loss": 0.1715, "step": 8522 }, { "epoch": 1.16, "grad_norm": 3.8644535034443503, "learning_rate": 3.960082392413004e-06, "loss": 0.1627, "step": 8523 }, { "epoch": 1.16, "grad_norm": 3.9481745527644247, "learning_rate": 3.9590049655748624e-06, "loss": 0.1737, "step": 8524 }, { "epoch": 1.16, "grad_norm": 3.7722767680341223, "learning_rate": 3.95792758926207e-06, "loss": 0.1985, "step": 8525 }, { "epoch": 1.16, "grad_norm": 4.083663394526607, "learning_rate": 3.956850263526925e-06, "loss": 0.1748, "step": 8526 }, { "epoch": 1.16, "grad_norm": 3.94337095984446, "learning_rate": 3.955772988421709e-06, "loss": 0.2047, "step": 8527 }, { "epoch": 1.16, "grad_norm": 4.626840390511326, "learning_rate": 3.954695763998712e-06, "loss": 0.2318, "step": 8528 }, { "epoch": 1.16, "grad_norm": 2.8128610577208413, "learning_rate": 3.953618590310218e-06, "loss": 0.1985, "step": 8529 }, { "epoch": 1.16, "grad_norm": 2.7151878090109927, "learning_rate": 3.952541467408507e-06, "loss": 0.1488, "step": 8530 }, { "epoch": 1.16, "grad_norm": 4.144547356023027, "learning_rate": 3.951464395345857e-06, "loss": 0.2225, "step": 8531 }, { "epoch": 1.16, "grad_norm": 4.119635854624774, "learning_rate": 3.950387374174548e-06, "loss": 0.2069, "step": 8532 }, { "epoch": 1.16, "grad_norm": 3.6626285596551, "learning_rate": 3.949310403946849e-06, "loss": 0.1811, "step": 8533 }, { "epoch": 1.16, "grad_norm": 3.100790595879201, "learning_rate": 3.948233484715037e-06, "loss": 0.1637, "step": 8534 }, { "epoch": 1.16, "grad_norm": 4.303967802501797, "learning_rate": 3.947156616531376e-06, "loss": 0.2049, "step": 8535 }, { "epoch": 1.16, "grad_norm": 4.638829512367346, "learning_rate": 3.946079799448134e-06, "loss": 0.202, "step": 8536 }, { "epoch": 1.16, "grad_norm": 3.15720228387972, "learning_rate": 3.945003033517578e-06, "loss": 0.1409, "step": 8537 }, { "epoch": 1.16, "grad_norm": 3.5473135728718397, "learning_rate": 3.9439263187919635e-06, "loss": 0.1632, "step": 8538 }, { "epoch": 1.16, "grad_norm": 3.9729703195958037, "learning_rate": 3.942849655323556e-06, "loss": 0.2034, "step": 8539 }, { "epoch": 1.16, "grad_norm": 2.858300848372416, "learning_rate": 3.941773043164607e-06, "loss": 0.1908, "step": 8540 }, { "epoch": 1.16, "grad_norm": 3.812478043387878, "learning_rate": 3.9406964823673726e-06, "loss": 0.1796, "step": 8541 }, { "epoch": 1.16, "grad_norm": 3.3450603904388676, "learning_rate": 3.9396199729841044e-06, "loss": 0.1694, "step": 8542 }, { "epoch": 1.16, "grad_norm": 3.712589273508183, "learning_rate": 3.938543515067053e-06, "loss": 0.165, "step": 8543 }, { "epoch": 1.16, "grad_norm": 3.492539294210272, "learning_rate": 3.937467108668461e-06, "loss": 0.1695, "step": 8544 }, { "epoch": 1.16, "grad_norm": 3.6520947115366704, "learning_rate": 3.936390753840577e-06, "loss": 0.1564, "step": 8545 }, { "epoch": 1.16, "grad_norm": 4.044646375670856, "learning_rate": 3.935314450635639e-06, "loss": 0.1653, "step": 8546 }, { "epoch": 1.16, "grad_norm": 4.419029082179459, "learning_rate": 3.934238199105887e-06, "loss": 0.1495, "step": 8547 }, { "epoch": 1.16, "grad_norm": 3.593179659470556, "learning_rate": 3.9331619993035595e-06, "loss": 0.1859, "step": 8548 }, { "epoch": 1.16, "grad_norm": 2.890580020318886, "learning_rate": 3.9320858512808884e-06, "loss": 0.117, "step": 8549 }, { "epoch": 1.16, "grad_norm": 3.657367195591405, "learning_rate": 3.931009755090105e-06, "loss": 0.1825, "step": 8550 }, { "epoch": 1.16, "grad_norm": 3.277693721121743, "learning_rate": 3.929933710783441e-06, "loss": 0.169, "step": 8551 }, { "epoch": 1.16, "grad_norm": 3.1665206463711453, "learning_rate": 3.928857718413119e-06, "loss": 0.177, "step": 8552 }, { "epoch": 1.16, "grad_norm": 3.2609081841371554, "learning_rate": 3.927781778031368e-06, "loss": 0.1685, "step": 8553 }, { "epoch": 1.16, "grad_norm": 4.061677568287353, "learning_rate": 3.9267058896904035e-06, "loss": 0.1864, "step": 8554 }, { "epoch": 1.16, "grad_norm": 3.3746613984674614, "learning_rate": 3.925630053442448e-06, "loss": 0.1674, "step": 8555 }, { "epoch": 1.16, "grad_norm": 3.2919966838389025, "learning_rate": 3.924554269339717e-06, "loss": 0.1816, "step": 8556 }, { "epoch": 1.16, "grad_norm": 3.427677094024962, "learning_rate": 3.9234785374344264e-06, "loss": 0.1357, "step": 8557 }, { "epoch": 1.16, "grad_norm": 2.9199582490441047, "learning_rate": 3.922402857778783e-06, "loss": 0.191, "step": 8558 }, { "epoch": 1.16, "grad_norm": 3.336580543498499, "learning_rate": 3.921327230425001e-06, "loss": 0.1773, "step": 8559 }, { "epoch": 1.16, "grad_norm": 3.170549549101219, "learning_rate": 3.920251655425283e-06, "loss": 0.1779, "step": 8560 }, { "epoch": 1.16, "grad_norm": 4.2804118678074845, "learning_rate": 3.919176132831834e-06, "loss": 0.2123, "step": 8561 }, { "epoch": 1.16, "grad_norm": 3.2494923018588016, "learning_rate": 3.918100662696853e-06, "loss": 0.1828, "step": 8562 }, { "epoch": 1.16, "grad_norm": 4.625118575078036, "learning_rate": 3.917025245072543e-06, "loss": 0.2037, "step": 8563 }, { "epoch": 1.16, "grad_norm": 3.388259691459486, "learning_rate": 3.915949880011096e-06, "loss": 0.1819, "step": 8564 }, { "epoch": 1.17, "grad_norm": 3.2770136648718395, "learning_rate": 3.914874567564709e-06, "loss": 0.1564, "step": 8565 }, { "epoch": 1.17, "grad_norm": 3.6042001567522157, "learning_rate": 3.9137993077855695e-06, "loss": 0.1755, "step": 8566 }, { "epoch": 1.17, "grad_norm": 3.0481675001601625, "learning_rate": 3.9127241007258695e-06, "loss": 0.1482, "step": 8567 }, { "epoch": 1.17, "grad_norm": 3.6644239347433145, "learning_rate": 3.91164894643779e-06, "loss": 0.1926, "step": 8568 }, { "epoch": 1.17, "grad_norm": 3.000937580044883, "learning_rate": 3.9105738449735185e-06, "loss": 0.1898, "step": 8569 }, { "epoch": 1.17, "grad_norm": 3.267895657176216, "learning_rate": 3.909498796385233e-06, "loss": 0.1445, "step": 8570 }, { "epoch": 1.17, "grad_norm": 4.6401100233312125, "learning_rate": 3.908423800725117e-06, "loss": 0.1889, "step": 8571 }, { "epoch": 1.17, "grad_norm": 3.4757125709254177, "learning_rate": 3.907348858045338e-06, "loss": 0.1927, "step": 8572 }, { "epoch": 1.17, "grad_norm": 4.131367657667131, "learning_rate": 3.906273968398076e-06, "loss": 0.1509, "step": 8573 }, { "epoch": 1.17, "grad_norm": 3.86962240041514, "learning_rate": 3.905199131835497e-06, "loss": 0.1524, "step": 8574 }, { "epoch": 1.17, "grad_norm": 6.027181847161556, "learning_rate": 3.904124348409769e-06, "loss": 0.1684, "step": 8575 }, { "epoch": 1.17, "grad_norm": 4.279016509413841, "learning_rate": 3.903049618173062e-06, "loss": 0.1633, "step": 8576 }, { "epoch": 1.17, "grad_norm": 3.8004753119907106, "learning_rate": 3.9019749411775336e-06, "loss": 0.1794, "step": 8577 }, { "epoch": 1.17, "grad_norm": 3.0085067734992976, "learning_rate": 3.900900317475348e-06, "loss": 0.1325, "step": 8578 }, { "epoch": 1.17, "grad_norm": 3.669288898134942, "learning_rate": 3.899825747118657e-06, "loss": 0.1796, "step": 8579 }, { "epoch": 1.17, "grad_norm": 4.5312902294515505, "learning_rate": 3.898751230159621e-06, "loss": 0.1673, "step": 8580 }, { "epoch": 1.17, "grad_norm": 3.5310338471608524, "learning_rate": 3.89767676665039e-06, "loss": 0.2021, "step": 8581 }, { "epoch": 1.17, "grad_norm": 3.73509372258263, "learning_rate": 3.8966023566431154e-06, "loss": 0.1588, "step": 8582 }, { "epoch": 1.17, "grad_norm": 3.4408387370600737, "learning_rate": 3.8955280001899405e-06, "loss": 0.151, "step": 8583 }, { "epoch": 1.17, "grad_norm": 3.765898329931465, "learning_rate": 3.894453697343016e-06, "loss": 0.1621, "step": 8584 }, { "epoch": 1.17, "grad_norm": 2.780164678495443, "learning_rate": 3.893379448154478e-06, "loss": 0.0975, "step": 8585 }, { "epoch": 1.17, "grad_norm": 3.7622773011961694, "learning_rate": 3.89230525267647e-06, "loss": 0.2136, "step": 8586 }, { "epoch": 1.17, "grad_norm": 3.202320446999131, "learning_rate": 3.891231110961126e-06, "loss": 0.136, "step": 8587 }, { "epoch": 1.17, "grad_norm": 3.869580609551456, "learning_rate": 3.890157023060582e-06, "loss": 0.1984, "step": 8588 }, { "epoch": 1.17, "grad_norm": 4.050526051285279, "learning_rate": 3.889082989026967e-06, "loss": 0.1947, "step": 8589 }, { "epoch": 1.17, "grad_norm": 4.069495118644176, "learning_rate": 3.888009008912414e-06, "loss": 0.1554, "step": 8590 }, { "epoch": 1.17, "grad_norm": 3.1544982388660268, "learning_rate": 3.886935082769044e-06, "loss": 0.1431, "step": 8591 }, { "epoch": 1.17, "grad_norm": 4.438028671554423, "learning_rate": 3.885861210648987e-06, "loss": 0.14, "step": 8592 }, { "epoch": 1.17, "grad_norm": 3.3068903307678936, "learning_rate": 3.8847873926043585e-06, "loss": 0.1298, "step": 8593 }, { "epoch": 1.17, "grad_norm": 4.285729274842549, "learning_rate": 3.8837136286872805e-06, "loss": 0.1843, "step": 8594 }, { "epoch": 1.17, "grad_norm": 3.9148704320058965, "learning_rate": 3.882639918949865e-06, "loss": 0.1679, "step": 8595 }, { "epoch": 1.17, "grad_norm": 3.3274165669395, "learning_rate": 3.881566263444231e-06, "loss": 0.1474, "step": 8596 }, { "epoch": 1.17, "grad_norm": 3.2814526484880093, "learning_rate": 3.880492662222483e-06, "loss": 0.1427, "step": 8597 }, { "epoch": 1.17, "grad_norm": 3.5085354182089317, "learning_rate": 3.879419115336734e-06, "loss": 0.1797, "step": 8598 }, { "epoch": 1.17, "grad_norm": 4.479409697986732, "learning_rate": 3.878345622839085e-06, "loss": 0.2019, "step": 8599 }, { "epoch": 1.17, "grad_norm": 3.330579646876062, "learning_rate": 3.877272184781642e-06, "loss": 0.1815, "step": 8600 }, { "epoch": 1.17, "grad_norm": 3.834449387991809, "learning_rate": 3.876198801216502e-06, "loss": 0.1874, "step": 8601 }, { "epoch": 1.17, "grad_norm": 3.515103988689696, "learning_rate": 3.875125472195764e-06, "loss": 0.1698, "step": 8602 }, { "epoch": 1.17, "grad_norm": 4.090316395619034, "learning_rate": 3.8740521977715225e-06, "loss": 0.1726, "step": 8603 }, { "epoch": 1.17, "grad_norm": 3.6767623494196, "learning_rate": 3.872978977995871e-06, "loss": 0.1416, "step": 8604 }, { "epoch": 1.17, "grad_norm": 4.3677635975501765, "learning_rate": 3.8719058129208955e-06, "loss": 0.2192, "step": 8605 }, { "epoch": 1.17, "grad_norm": 3.381221340421027, "learning_rate": 3.870832702598688e-06, "loss": 0.1793, "step": 8606 }, { "epoch": 1.17, "grad_norm": 3.5698260793224232, "learning_rate": 3.869759647081326e-06, "loss": 0.2265, "step": 8607 }, { "epoch": 1.17, "grad_norm": 3.8133035014717063, "learning_rate": 3.868686646420896e-06, "loss": 0.1967, "step": 8608 }, { "epoch": 1.17, "grad_norm": 4.130663172327329, "learning_rate": 3.867613700669474e-06, "loss": 0.2067, "step": 8609 }, { "epoch": 1.17, "grad_norm": 2.7456517254025803, "learning_rate": 3.866540809879138e-06, "loss": 0.1407, "step": 8610 }, { "epoch": 1.17, "grad_norm": 4.054741611974739, "learning_rate": 3.86546797410196e-06, "loss": 0.1677, "step": 8611 }, { "epoch": 1.17, "grad_norm": 3.595207801830714, "learning_rate": 3.8643951933900125e-06, "loss": 0.1574, "step": 8612 }, { "epoch": 1.17, "grad_norm": 3.758309023316704, "learning_rate": 3.863322467795362e-06, "loss": 0.1877, "step": 8613 }, { "epoch": 1.17, "grad_norm": 4.164156804921388, "learning_rate": 3.862249797370073e-06, "loss": 0.1681, "step": 8614 }, { "epoch": 1.17, "grad_norm": 3.02937006857984, "learning_rate": 3.861177182166212e-06, "loss": 0.1238, "step": 8615 }, { "epoch": 1.17, "grad_norm": 3.8078248433571353, "learning_rate": 3.860104622235834e-06, "loss": 0.1929, "step": 8616 }, { "epoch": 1.17, "grad_norm": 3.2523201279155742, "learning_rate": 3.859032117631002e-06, "loss": 0.1596, "step": 8617 }, { "epoch": 1.17, "grad_norm": 3.9080719033120883, "learning_rate": 3.8579596684037645e-06, "loss": 0.2028, "step": 8618 }, { "epoch": 1.17, "grad_norm": 4.606004978088248, "learning_rate": 3.85688727460618e-06, "loss": 0.1781, "step": 8619 }, { "epoch": 1.17, "grad_norm": 2.9769614537227964, "learning_rate": 3.855814936290291e-06, "loss": 0.1469, "step": 8620 }, { "epoch": 1.17, "grad_norm": 3.798886570504344, "learning_rate": 3.85474265350815e-06, "loss": 0.1841, "step": 8621 }, { "epoch": 1.17, "grad_norm": 3.3982221377001296, "learning_rate": 3.853670426311797e-06, "loss": 0.1813, "step": 8622 }, { "epoch": 1.17, "grad_norm": 3.343623778865472, "learning_rate": 3.852598254753277e-06, "loss": 0.1633, "step": 8623 }, { "epoch": 1.17, "grad_norm": 3.471656667682342, "learning_rate": 3.8515261388846235e-06, "loss": 0.166, "step": 8624 }, { "epoch": 1.17, "grad_norm": 3.680769701462786, "learning_rate": 3.850454078757878e-06, "loss": 0.2023, "step": 8625 }, { "epoch": 1.17, "grad_norm": 5.007936903559692, "learning_rate": 3.849382074425069e-06, "loss": 0.1721, "step": 8626 }, { "epoch": 1.17, "grad_norm": 4.022973686619184, "learning_rate": 3.848310125938229e-06, "loss": 0.1817, "step": 8627 }, { "epoch": 1.17, "grad_norm": 4.1007647213146585, "learning_rate": 3.8472382333493844e-06, "loss": 0.1218, "step": 8628 }, { "epoch": 1.17, "grad_norm": 3.1333547420647627, "learning_rate": 3.8461663967105634e-06, "loss": 0.1877, "step": 8629 }, { "epoch": 1.17, "grad_norm": 3.0264762012715547, "learning_rate": 3.845094616073783e-06, "loss": 0.1717, "step": 8630 }, { "epoch": 1.17, "grad_norm": 2.797569541522881, "learning_rate": 3.844022891491069e-06, "loss": 0.1198, "step": 8631 }, { "epoch": 1.17, "grad_norm": 3.28532598273248, "learning_rate": 3.842951223014433e-06, "loss": 0.1535, "step": 8632 }, { "epoch": 1.17, "grad_norm": 2.8306326632092897, "learning_rate": 3.841879610695892e-06, "loss": 0.1301, "step": 8633 }, { "epoch": 1.17, "grad_norm": 4.07394175171676, "learning_rate": 3.840808054587456e-06, "loss": 0.1877, "step": 8634 }, { "epoch": 1.17, "grad_norm": 3.403491355928715, "learning_rate": 3.8397365547411344e-06, "loss": 0.1632, "step": 8635 }, { "epoch": 1.17, "grad_norm": 3.45019765036892, "learning_rate": 3.838665111208931e-06, "loss": 0.231, "step": 8636 }, { "epoch": 1.17, "grad_norm": 5.20247973201989, "learning_rate": 3.837593724042854e-06, "loss": 0.2377, "step": 8637 }, { "epoch": 1.17, "grad_norm": 4.4856558200217345, "learning_rate": 3.836522393294899e-06, "loss": 0.1875, "step": 8638 }, { "epoch": 1.18, "grad_norm": 3.6997672640742514, "learning_rate": 3.835451119017066e-06, "loss": 0.1195, "step": 8639 }, { "epoch": 1.18, "grad_norm": 4.344042445366426, "learning_rate": 3.834379901261348e-06, "loss": 0.2106, "step": 8640 }, { "epoch": 1.18, "grad_norm": 3.8858341213765075, "learning_rate": 3.833308740079741e-06, "loss": 0.146, "step": 8641 }, { "epoch": 1.18, "grad_norm": 3.304423859303574, "learning_rate": 3.832237635524229e-06, "loss": 0.1716, "step": 8642 }, { "epoch": 1.18, "grad_norm": 2.8040118820318845, "learning_rate": 3.831166587646804e-06, "loss": 0.1357, "step": 8643 }, { "epoch": 1.18, "grad_norm": 2.8688619346996673, "learning_rate": 3.8300955964994476e-06, "loss": 0.1645, "step": 8644 }, { "epoch": 1.18, "grad_norm": 3.5291565024657716, "learning_rate": 3.829024662134142e-06, "loss": 0.163, "step": 8645 }, { "epoch": 1.18, "grad_norm": 4.166350302849065, "learning_rate": 3.827953784602863e-06, "loss": 0.1347, "step": 8646 }, { "epoch": 1.18, "grad_norm": 3.3768917562141, "learning_rate": 3.826882963957589e-06, "loss": 0.1581, "step": 8647 }, { "epoch": 1.18, "grad_norm": 4.092359994700936, "learning_rate": 3.825812200250293e-06, "loss": 0.2149, "step": 8648 }, { "epoch": 1.18, "grad_norm": 3.914921321321153, "learning_rate": 3.824741493532944e-06, "loss": 0.1988, "step": 8649 }, { "epoch": 1.18, "grad_norm": 3.015376925340001, "learning_rate": 3.823670843857509e-06, "loss": 0.1416, "step": 8650 }, { "epoch": 1.18, "grad_norm": 3.5188206226140832, "learning_rate": 3.8226002512759555e-06, "loss": 0.174, "step": 8651 }, { "epoch": 1.18, "grad_norm": 5.071558258584201, "learning_rate": 3.821529715840241e-06, "loss": 0.1911, "step": 8652 }, { "epoch": 1.18, "grad_norm": 3.961704636600797, "learning_rate": 3.820459237602329e-06, "loss": 0.166, "step": 8653 }, { "epoch": 1.18, "grad_norm": 4.104555085842458, "learning_rate": 3.819388816614174e-06, "loss": 0.1746, "step": 8654 }, { "epoch": 1.18, "grad_norm": 4.022306527333734, "learning_rate": 3.818318452927726e-06, "loss": 0.1917, "step": 8655 }, { "epoch": 1.18, "grad_norm": 3.144557752983494, "learning_rate": 3.817248146594943e-06, "loss": 0.1869, "step": 8656 }, { "epoch": 1.18, "grad_norm": 3.010678251033131, "learning_rate": 3.816177897667767e-06, "loss": 0.1441, "step": 8657 }, { "epoch": 1.18, "grad_norm": 2.859248105124228, "learning_rate": 3.815107706198147e-06, "loss": 0.1528, "step": 8658 }, { "epoch": 1.18, "grad_norm": 2.8606815002769954, "learning_rate": 3.814037572238022e-06, "loss": 0.1637, "step": 8659 }, { "epoch": 1.18, "grad_norm": 3.0118399088808405, "learning_rate": 3.8129674958393344e-06, "loss": 0.1814, "step": 8660 }, { "epoch": 1.18, "grad_norm": 4.226272512400761, "learning_rate": 3.811897477054019e-06, "loss": 0.164, "step": 8661 }, { "epoch": 1.18, "grad_norm": 4.312123945938637, "learning_rate": 3.810827515934013e-06, "loss": 0.1806, "step": 8662 }, { "epoch": 1.18, "grad_norm": 3.581261893757698, "learning_rate": 3.8097576125312437e-06, "loss": 0.1877, "step": 8663 }, { "epoch": 1.18, "grad_norm": 3.06988618954489, "learning_rate": 3.808687766897643e-06, "loss": 0.1779, "step": 8664 }, { "epoch": 1.18, "grad_norm": 3.737750220276074, "learning_rate": 3.807617979085133e-06, "loss": 0.1508, "step": 8665 }, { "epoch": 1.18, "grad_norm": 4.296766087068115, "learning_rate": 3.8065482491456396e-06, "loss": 0.1439, "step": 8666 }, { "epoch": 1.18, "grad_norm": 4.23934512535747, "learning_rate": 3.8054785771310817e-06, "loss": 0.2172, "step": 8667 }, { "epoch": 1.18, "grad_norm": 3.9170596999819463, "learning_rate": 3.8044089630933756e-06, "loss": 0.2011, "step": 8668 }, { "epoch": 1.18, "grad_norm": 3.5799220077045137, "learning_rate": 3.8033394070844365e-06, "loss": 0.1598, "step": 8669 }, { "epoch": 1.18, "grad_norm": 3.8504742599358415, "learning_rate": 3.8022699091561776e-06, "loss": 0.1721, "step": 8670 }, { "epoch": 1.18, "grad_norm": 3.5012923383542898, "learning_rate": 3.801200469360504e-06, "loss": 0.1509, "step": 8671 }, { "epoch": 1.18, "grad_norm": 3.5099494864056577, "learning_rate": 3.8001310877493265e-06, "loss": 0.1404, "step": 8672 }, { "epoch": 1.18, "grad_norm": 4.45238454837205, "learning_rate": 3.799061764374543e-06, "loss": 0.191, "step": 8673 }, { "epoch": 1.18, "grad_norm": 3.218981103124831, "learning_rate": 3.7979924992880567e-06, "loss": 0.1844, "step": 8674 }, { "epoch": 1.18, "grad_norm": 4.444233226616933, "learning_rate": 3.796923292541764e-06, "loss": 0.126, "step": 8675 }, { "epoch": 1.18, "grad_norm": 3.475646016080487, "learning_rate": 3.7958541441875628e-06, "loss": 0.1794, "step": 8676 }, { "epoch": 1.18, "grad_norm": 4.068535782599349, "learning_rate": 3.7947850542773396e-06, "loss": 0.172, "step": 8677 }, { "epoch": 1.18, "grad_norm": 3.850670125976606, "learning_rate": 3.7937160228629877e-06, "loss": 0.1844, "step": 8678 }, { "epoch": 1.18, "grad_norm": 3.977652853055698, "learning_rate": 3.7926470499963896e-06, "loss": 0.1538, "step": 8679 }, { "epoch": 1.18, "grad_norm": 3.5259278331389203, "learning_rate": 3.791578135729432e-06, "loss": 0.1514, "step": 8680 }, { "epoch": 1.18, "grad_norm": 5.071265390205854, "learning_rate": 3.790509280113992e-06, "loss": 0.1607, "step": 8681 }, { "epoch": 1.18, "grad_norm": 3.6813560329557977, "learning_rate": 3.7894404832019514e-06, "loss": 0.1875, "step": 8682 }, { "epoch": 1.18, "grad_norm": 3.9342358003527806, "learning_rate": 3.7883717450451806e-06, "loss": 0.1721, "step": 8683 }, { "epoch": 1.18, "grad_norm": 4.645271456866317, "learning_rate": 3.7873030656955552e-06, "loss": 0.1919, "step": 8684 }, { "epoch": 1.18, "grad_norm": 3.3687969775230644, "learning_rate": 3.7862344452049402e-06, "loss": 0.2037, "step": 8685 }, { "epoch": 1.18, "grad_norm": 3.372441841679031, "learning_rate": 3.785165883625206e-06, "loss": 0.2187, "step": 8686 }, { "epoch": 1.18, "grad_norm": 3.2515550142463834, "learning_rate": 3.784097381008212e-06, "loss": 0.147, "step": 8687 }, { "epoch": 1.18, "grad_norm": 3.030207526794745, "learning_rate": 3.7830289374058214e-06, "loss": 0.1702, "step": 8688 }, { "epoch": 1.18, "grad_norm": 3.27153196832957, "learning_rate": 3.7819605528698906e-06, "loss": 0.2054, "step": 8689 }, { "epoch": 1.18, "grad_norm": 3.7062509338459324, "learning_rate": 3.7808922274522757e-06, "loss": 0.2003, "step": 8690 }, { "epoch": 1.18, "grad_norm": 2.431719884848439, "learning_rate": 3.779823961204826e-06, "loss": 0.1222, "step": 8691 }, { "epoch": 1.18, "grad_norm": 3.74591484525914, "learning_rate": 3.778755754179394e-06, "loss": 0.1454, "step": 8692 }, { "epoch": 1.18, "grad_norm": 2.9197013103833407, "learning_rate": 3.777687606427822e-06, "loss": 0.1395, "step": 8693 }, { "epoch": 1.18, "grad_norm": 3.5964340895435902, "learning_rate": 3.7766195180019556e-06, "loss": 0.1513, "step": 8694 }, { "epoch": 1.18, "grad_norm": 2.981466526171217, "learning_rate": 3.7755514889536355e-06, "loss": 0.18, "step": 8695 }, { "epoch": 1.18, "grad_norm": 3.4990638973108137, "learning_rate": 3.7744835193346962e-06, "loss": 0.1703, "step": 8696 }, { "epoch": 1.18, "grad_norm": 3.4480662059416454, "learning_rate": 3.7734156091969766e-06, "loss": 0.1672, "step": 8697 }, { "epoch": 1.18, "grad_norm": 3.9713941449596155, "learning_rate": 3.772347758592304e-06, "loss": 0.1987, "step": 8698 }, { "epoch": 1.18, "grad_norm": 3.326777640249892, "learning_rate": 3.7712799675725105e-06, "loss": 0.1731, "step": 8699 }, { "epoch": 1.18, "grad_norm": 3.5757352200007944, "learning_rate": 3.770212236189419e-06, "loss": 0.2031, "step": 8700 }, { "epoch": 1.18, "grad_norm": 3.0700890853406166, "learning_rate": 3.769144564494857e-06, "loss": 0.1402, "step": 8701 }, { "epoch": 1.18, "grad_norm": 2.8257224430869194, "learning_rate": 3.7680769525406398e-06, "loss": 0.1314, "step": 8702 }, { "epoch": 1.18, "grad_norm": 4.73950302615429, "learning_rate": 3.767009400378589e-06, "loss": 0.1517, "step": 8703 }, { "epoch": 1.18, "grad_norm": 2.6391989535048124, "learning_rate": 3.7659419080605144e-06, "loss": 0.141, "step": 8704 }, { "epoch": 1.18, "grad_norm": 3.4425564740627133, "learning_rate": 3.7648744756382317e-06, "loss": 0.1842, "step": 8705 }, { "epoch": 1.18, "grad_norm": 4.3497880117382035, "learning_rate": 3.7638071031635454e-06, "loss": 0.1873, "step": 8706 }, { "epoch": 1.18, "grad_norm": 3.3217773129280164, "learning_rate": 3.762739790688264e-06, "loss": 0.1925, "step": 8707 }, { "epoch": 1.18, "grad_norm": 4.3278278537601205, "learning_rate": 3.7616725382641883e-06, "loss": 0.1591, "step": 8708 }, { "epoch": 1.18, "grad_norm": 3.07319866871363, "learning_rate": 3.7606053459431215e-06, "loss": 0.1339, "step": 8709 }, { "epoch": 1.18, "grad_norm": 3.238634731092212, "learning_rate": 3.7595382137768556e-06, "loss": 0.1814, "step": 8710 }, { "epoch": 1.18, "grad_norm": 3.1569963156908027, "learning_rate": 3.7584711418171894e-06, "loss": 0.1694, "step": 8711 }, { "epoch": 1.19, "grad_norm": 3.457091890130337, "learning_rate": 3.757404130115909e-06, "loss": 0.2143, "step": 8712 }, { "epoch": 1.19, "grad_norm": 3.3468251775835896, "learning_rate": 3.756337178724806e-06, "loss": 0.1603, "step": 8713 }, { "epoch": 1.19, "grad_norm": 4.4478140447600465, "learning_rate": 3.7552702876956637e-06, "loss": 0.1545, "step": 8714 }, { "epoch": 1.19, "grad_norm": 3.51805757531998, "learning_rate": 3.754203457080267e-06, "loss": 0.1853, "step": 8715 }, { "epoch": 1.19, "grad_norm": 2.9258112618240992, "learning_rate": 3.7531366869303916e-06, "loss": 0.1081, "step": 8716 }, { "epoch": 1.19, "grad_norm": 3.4096025680572923, "learning_rate": 3.752069977297817e-06, "loss": 0.1639, "step": 8717 }, { "epoch": 1.19, "grad_norm": 3.1329413595275897, "learning_rate": 3.751003328234314e-06, "loss": 0.115, "step": 8718 }, { "epoch": 1.19, "grad_norm": 4.0270678901886985, "learning_rate": 3.749936739791656e-06, "loss": 0.144, "step": 8719 }, { "epoch": 1.19, "grad_norm": 2.6046214717546814, "learning_rate": 3.7488702120216063e-06, "loss": 0.1328, "step": 8720 }, { "epoch": 1.19, "grad_norm": 3.2440388721296163, "learning_rate": 3.747803744975934e-06, "loss": 0.1806, "step": 8721 }, { "epoch": 1.19, "grad_norm": 4.068722155435195, "learning_rate": 3.7467373387063973e-06, "loss": 0.2081, "step": 8722 }, { "epoch": 1.19, "grad_norm": 3.1760753501038064, "learning_rate": 3.7456709932647585e-06, "loss": 0.1447, "step": 8723 }, { "epoch": 1.19, "grad_norm": 3.366111490698261, "learning_rate": 3.744604708702769e-06, "loss": 0.1856, "step": 8724 }, { "epoch": 1.19, "grad_norm": 3.6182205763914905, "learning_rate": 3.7435384850721862e-06, "loss": 0.1363, "step": 8725 }, { "epoch": 1.19, "grad_norm": 3.383778852280856, "learning_rate": 3.7424723224247556e-06, "loss": 0.1838, "step": 8726 }, { "epoch": 1.19, "grad_norm": 3.237217915539142, "learning_rate": 3.741406220812227e-06, "loss": 0.1404, "step": 8727 }, { "epoch": 1.19, "grad_norm": 4.592988686059981, "learning_rate": 3.7403401802863425e-06, "loss": 0.1684, "step": 8728 }, { "epoch": 1.19, "grad_norm": 3.7809692018796137, "learning_rate": 3.739274200898846e-06, "loss": 0.1307, "step": 8729 }, { "epoch": 1.19, "grad_norm": 6.053505191936638, "learning_rate": 3.738208282701471e-06, "loss": 0.1788, "step": 8730 }, { "epoch": 1.19, "grad_norm": 4.614581364301069, "learning_rate": 3.7371424257459576e-06, "loss": 0.1807, "step": 8731 }, { "epoch": 1.19, "grad_norm": 2.879889433651989, "learning_rate": 3.7360766300840323e-06, "loss": 0.1469, "step": 8732 }, { "epoch": 1.19, "grad_norm": 3.8656944490891303, "learning_rate": 3.7350108957674284e-06, "loss": 0.1976, "step": 8733 }, { "epoch": 1.19, "grad_norm": 2.9538981244264773, "learning_rate": 3.7339452228478722e-06, "loss": 0.185, "step": 8734 }, { "epoch": 1.19, "grad_norm": 3.2669610781956617, "learning_rate": 3.7328796113770827e-06, "loss": 0.161, "step": 8735 }, { "epoch": 1.19, "grad_norm": 4.116990219241257, "learning_rate": 3.731814061406785e-06, "loss": 0.192, "step": 8736 }, { "epoch": 1.19, "grad_norm": 3.220671772105736, "learning_rate": 3.7307485729886917e-06, "loss": 0.1215, "step": 8737 }, { "epoch": 1.19, "grad_norm": 3.5034022494516797, "learning_rate": 3.7296831461745207e-06, "loss": 0.1631, "step": 8738 }, { "epoch": 1.19, "grad_norm": 3.2665431616112586, "learning_rate": 3.7286177810159795e-06, "loss": 0.1255, "step": 8739 }, { "epoch": 1.19, "grad_norm": 4.715789917655684, "learning_rate": 3.72755247756478e-06, "loss": 0.1811, "step": 8740 }, { "epoch": 1.19, "grad_norm": 4.131426538786198, "learning_rate": 3.726487235872624e-06, "loss": 0.1831, "step": 8741 }, { "epoch": 1.19, "grad_norm": 3.379599803762793, "learning_rate": 3.725422055991218e-06, "loss": 0.1568, "step": 8742 }, { "epoch": 1.19, "grad_norm": 4.050978360938146, "learning_rate": 3.7243569379722556e-06, "loss": 0.1932, "step": 8743 }, { "epoch": 1.19, "grad_norm": 3.7700007977578123, "learning_rate": 3.723291881867439e-06, "loss": 0.1884, "step": 8744 }, { "epoch": 1.19, "grad_norm": 5.096911270160594, "learning_rate": 3.7222268877284555e-06, "loss": 0.1902, "step": 8745 }, { "epoch": 1.19, "grad_norm": 4.591171159092901, "learning_rate": 3.7211619556069993e-06, "loss": 0.1653, "step": 8746 }, { "epoch": 1.19, "grad_norm": 7.030960343497067, "learning_rate": 3.720097085554756e-06, "loss": 0.1608, "step": 8747 }, { "epoch": 1.19, "grad_norm": 4.219840797874878, "learning_rate": 3.719032277623411e-06, "loss": 0.1764, "step": 8748 }, { "epoch": 1.19, "grad_norm": 3.177125861471255, "learning_rate": 3.7179675318646426e-06, "loss": 0.1971, "step": 8749 }, { "epoch": 1.19, "grad_norm": 2.8819938843981987, "learning_rate": 3.7169028483301333e-06, "loss": 0.1456, "step": 8750 }, { "epoch": 1.19, "grad_norm": 3.706863474428121, "learning_rate": 3.715838227071553e-06, "loss": 0.1972, "step": 8751 }, { "epoch": 1.19, "grad_norm": 4.26847420141781, "learning_rate": 3.7147736681405784e-06, "loss": 0.1688, "step": 8752 }, { "epoch": 1.19, "grad_norm": 3.8077715742766376, "learning_rate": 3.7137091715888764e-06, "loss": 0.2106, "step": 8753 }, { "epoch": 1.19, "grad_norm": 3.364898426060232, "learning_rate": 3.7126447374681126e-06, "loss": 0.1388, "step": 8754 }, { "epoch": 1.19, "grad_norm": 3.424465130890989, "learning_rate": 3.71158036582995e-06, "loss": 0.2051, "step": 8755 }, { "epoch": 1.19, "grad_norm": 4.878630790530536, "learning_rate": 3.7105160567260505e-06, "loss": 0.1435, "step": 8756 }, { "epoch": 1.19, "grad_norm": 4.169150306264246, "learning_rate": 3.709451810208068e-06, "loss": 0.1871, "step": 8757 }, { "epoch": 1.19, "grad_norm": 4.21301244946476, "learning_rate": 3.7083876263276604e-06, "loss": 0.1911, "step": 8758 }, { "epoch": 1.19, "grad_norm": 4.0436286489628905, "learning_rate": 3.707323505136474e-06, "loss": 0.2317, "step": 8759 }, { "epoch": 1.19, "grad_norm": 3.9995987767943633, "learning_rate": 3.7062594466861594e-06, "loss": 0.2205, "step": 8760 }, { "epoch": 1.19, "grad_norm": 3.044830528308862, "learning_rate": 3.70519545102836e-06, "loss": 0.1616, "step": 8761 }, { "epoch": 1.19, "grad_norm": 3.197372267028391, "learning_rate": 3.7041315182147203e-06, "loss": 0.1501, "step": 8762 }, { "epoch": 1.19, "grad_norm": 3.2013798409175, "learning_rate": 3.703067648296874e-06, "loss": 0.159, "step": 8763 }, { "epoch": 1.19, "grad_norm": 3.259979528080619, "learning_rate": 3.702003841326463e-06, "loss": 0.1683, "step": 8764 }, { "epoch": 1.19, "grad_norm": 2.747473408602392, "learning_rate": 3.7009400973551136e-06, "loss": 0.1566, "step": 8765 }, { "epoch": 1.19, "grad_norm": 2.8471921160958775, "learning_rate": 3.6998764164344592e-06, "loss": 0.1573, "step": 8766 }, { "epoch": 1.19, "grad_norm": 2.9771704902092933, "learning_rate": 3.6988127986161247e-06, "loss": 0.1493, "step": 8767 }, { "epoch": 1.19, "grad_norm": 4.012580776648179, "learning_rate": 3.697749243951735e-06, "loss": 0.1972, "step": 8768 }, { "epoch": 1.19, "grad_norm": 3.334120362865063, "learning_rate": 3.6966857524929068e-06, "loss": 0.1952, "step": 8769 }, { "epoch": 1.19, "grad_norm": 3.5938509493058532, "learning_rate": 3.6956223242912624e-06, "loss": 0.1834, "step": 8770 }, { "epoch": 1.19, "grad_norm": 3.2269899171837944, "learning_rate": 3.694558959398412e-06, "loss": 0.1238, "step": 8771 }, { "epoch": 1.19, "grad_norm": 4.416839925423448, "learning_rate": 3.6934956578659697e-06, "loss": 0.1716, "step": 8772 }, { "epoch": 1.19, "grad_norm": 3.8997982871658614, "learning_rate": 3.6924324197455397e-06, "loss": 0.1737, "step": 8773 }, { "epoch": 1.19, "grad_norm": 2.955210107495038, "learning_rate": 3.69136924508873e-06, "loss": 0.1492, "step": 8774 }, { "epoch": 1.19, "grad_norm": 3.7259112839811652, "learning_rate": 3.6903061339471423e-06, "loss": 0.1671, "step": 8775 }, { "epoch": 1.19, "grad_norm": 4.159071505242456, "learning_rate": 3.689243086372374e-06, "loss": 0.1786, "step": 8776 }, { "epoch": 1.19, "grad_norm": 3.5113766647242914, "learning_rate": 3.688180102416022e-06, "loss": 0.1923, "step": 8777 }, { "epoch": 1.19, "grad_norm": 3.413666744837211, "learning_rate": 3.687117182129677e-06, "loss": 0.1679, "step": 8778 }, { "epoch": 1.19, "grad_norm": 3.5589409698907586, "learning_rate": 3.6860543255649307e-06, "loss": 0.1884, "step": 8779 }, { "epoch": 1.19, "grad_norm": 2.550393958378905, "learning_rate": 3.684991532773367e-06, "loss": 0.1472, "step": 8780 }, { "epoch": 1.19, "grad_norm": 4.238530738992079, "learning_rate": 3.6839288038065736e-06, "loss": 0.1439, "step": 8781 }, { "epoch": 1.19, "grad_norm": 4.358913212850396, "learning_rate": 3.682866138716126e-06, "loss": 0.1924, "step": 8782 }, { "epoch": 1.19, "grad_norm": 3.8610801457843, "learning_rate": 3.6818035375536043e-06, "loss": 0.1715, "step": 8783 }, { "epoch": 1.19, "grad_norm": 2.918380273917975, "learning_rate": 3.68074100037058e-06, "loss": 0.1534, "step": 8784 }, { "epoch": 1.19, "grad_norm": 4.201936178002281, "learning_rate": 3.6796785272186275e-06, "loss": 0.2148, "step": 8785 }, { "epoch": 1.2, "grad_norm": 3.805479440671562, "learning_rate": 3.678616118149312e-06, "loss": 0.1969, "step": 8786 }, { "epoch": 1.2, "grad_norm": 3.6016743143481484, "learning_rate": 3.6775537732141986e-06, "loss": 0.1709, "step": 8787 }, { "epoch": 1.2, "grad_norm": 3.491099367038115, "learning_rate": 3.6764914924648482e-06, "loss": 0.1845, "step": 8788 }, { "epoch": 1.2, "grad_norm": 3.7554296687352235, "learning_rate": 3.675429275952822e-06, "loss": 0.2054, "step": 8789 }, { "epoch": 1.2, "grad_norm": 2.6166404977723303, "learning_rate": 3.6743671237296716e-06, "loss": 0.1773, "step": 8790 }, { "epoch": 1.2, "grad_norm": 3.8034578761563815, "learning_rate": 3.673305035846954e-06, "loss": 0.196, "step": 8791 }, { "epoch": 1.2, "grad_norm": 4.021391151756777, "learning_rate": 3.6722430123562124e-06, "loss": 0.1691, "step": 8792 }, { "epoch": 1.2, "grad_norm": 3.111599480496268, "learning_rate": 3.6711810533089975e-06, "loss": 0.1623, "step": 8793 }, { "epoch": 1.2, "grad_norm": 3.2274413719074397, "learning_rate": 3.67011915875685e-06, "loss": 0.1599, "step": 8794 }, { "epoch": 1.2, "grad_norm": 3.6700269890316117, "learning_rate": 3.669057328751311e-06, "loss": 0.1631, "step": 8795 }, { "epoch": 1.2, "grad_norm": 4.340130107937274, "learning_rate": 3.6679955633439144e-06, "loss": 0.1727, "step": 8796 }, { "epoch": 1.2, "grad_norm": 3.3991317263943035, "learning_rate": 3.6669338625861983e-06, "loss": 0.1543, "step": 8797 }, { "epoch": 1.2, "grad_norm": 3.559592502908461, "learning_rate": 3.6658722265296876e-06, "loss": 0.1863, "step": 8798 }, { "epoch": 1.2, "grad_norm": 3.484909398704474, "learning_rate": 3.6648106552259133e-06, "loss": 0.1735, "step": 8799 }, { "epoch": 1.2, "grad_norm": 4.043864884292754, "learning_rate": 3.6637491487263966e-06, "loss": 0.1401, "step": 8800 }, { "epoch": 1.2, "grad_norm": 3.4373118279304054, "learning_rate": 3.662687707082663e-06, "loss": 0.175, "step": 8801 }, { "epoch": 1.2, "grad_norm": 2.85588692563445, "learning_rate": 3.661626330346224e-06, "loss": 0.1644, "step": 8802 }, { "epoch": 1.2, "grad_norm": 3.345594921339655, "learning_rate": 3.6605650185686e-06, "loss": 0.1622, "step": 8803 }, { "epoch": 1.2, "grad_norm": 4.231329254738221, "learning_rate": 3.6595037718012983e-06, "loss": 0.1578, "step": 8804 }, { "epoch": 1.2, "grad_norm": 3.966799754192343, "learning_rate": 3.658442590095831e-06, "loss": 0.1479, "step": 8805 }, { "epoch": 1.2, "grad_norm": 3.078855896293487, "learning_rate": 3.657381473503698e-06, "loss": 0.1815, "step": 8806 }, { "epoch": 1.2, "grad_norm": 3.067718310158901, "learning_rate": 3.656320422076406e-06, "loss": 0.1428, "step": 8807 }, { "epoch": 1.2, "grad_norm": 3.1596894757350213, "learning_rate": 3.6552594358654503e-06, "loss": 0.1585, "step": 8808 }, { "epoch": 1.2, "grad_norm": 3.4375776780556224, "learning_rate": 3.654198514922331e-06, "loss": 0.1842, "step": 8809 }, { "epoch": 1.2, "grad_norm": 2.6624184363038155, "learning_rate": 3.6531376592985352e-06, "loss": 0.132, "step": 8810 }, { "epoch": 1.2, "grad_norm": 3.7568683224157478, "learning_rate": 3.6520768690455572e-06, "loss": 0.1799, "step": 8811 }, { "epoch": 1.2, "grad_norm": 3.058611675941817, "learning_rate": 3.6510161442148783e-06, "loss": 0.1577, "step": 8812 }, { "epoch": 1.2, "grad_norm": 3.856997209079434, "learning_rate": 3.649955484857986e-06, "loss": 0.2311, "step": 8813 }, { "epoch": 1.2, "grad_norm": 2.546461305447678, "learning_rate": 3.648894891026358e-06, "loss": 0.1816, "step": 8814 }, { "epoch": 1.2, "grad_norm": 3.2766690402510177, "learning_rate": 3.6478343627714695e-06, "loss": 0.1455, "step": 8815 }, { "epoch": 1.2, "grad_norm": 3.7365036902560473, "learning_rate": 3.646773900144797e-06, "loss": 0.1568, "step": 8816 }, { "epoch": 1.2, "grad_norm": 3.1004516681263863, "learning_rate": 3.6457135031978077e-06, "loss": 0.1283, "step": 8817 }, { "epoch": 1.2, "grad_norm": 3.5903649515612845, "learning_rate": 3.64465317198197e-06, "loss": 0.1528, "step": 8818 }, { "epoch": 1.2, "grad_norm": 3.252295289904094, "learning_rate": 3.643592906548747e-06, "loss": 0.1781, "step": 8819 }, { "epoch": 1.2, "grad_norm": 4.16062114346818, "learning_rate": 3.642532706949602e-06, "loss": 0.1861, "step": 8820 }, { "epoch": 1.2, "grad_norm": 3.5402640988539393, "learning_rate": 3.641472573235989e-06, "loss": 0.1609, "step": 8821 }, { "epoch": 1.2, "grad_norm": 3.3921804994084552, "learning_rate": 3.6404125054593653e-06, "loss": 0.1518, "step": 8822 }, { "epoch": 1.2, "grad_norm": 2.992429064695303, "learning_rate": 3.6393525036711783e-06, "loss": 0.1614, "step": 8823 }, { "epoch": 1.2, "grad_norm": 3.5302363393290075, "learning_rate": 3.6382925679228807e-06, "loss": 0.1266, "step": 8824 }, { "epoch": 1.2, "grad_norm": 3.725687911248855, "learning_rate": 3.637232698265912e-06, "loss": 0.1427, "step": 8825 }, { "epoch": 1.2, "grad_norm": 3.069387143990032, "learning_rate": 3.6361728947517173e-06, "loss": 0.1455, "step": 8826 }, { "epoch": 1.2, "grad_norm": 4.126036550780113, "learning_rate": 3.635113157431732e-06, "loss": 0.1609, "step": 8827 }, { "epoch": 1.2, "grad_norm": 5.24324274658361, "learning_rate": 3.634053486357395e-06, "loss": 0.1795, "step": 8828 }, { "epoch": 1.2, "grad_norm": 3.4798630031888185, "learning_rate": 3.6329938815801345e-06, "loss": 0.1602, "step": 8829 }, { "epoch": 1.2, "grad_norm": 2.3243889984313335, "learning_rate": 3.6319343431513822e-06, "loss": 0.1322, "step": 8830 }, { "epoch": 1.2, "grad_norm": 3.5249328062592364, "learning_rate": 3.6308748711225596e-06, "loss": 0.1562, "step": 8831 }, { "epoch": 1.2, "grad_norm": 4.157376981813059, "learning_rate": 3.629815465545091e-06, "loss": 0.1837, "step": 8832 }, { "epoch": 1.2, "grad_norm": 3.356474551829963, "learning_rate": 3.6287561264703953e-06, "loss": 0.153, "step": 8833 }, { "epoch": 1.2, "grad_norm": 3.0524881968310327, "learning_rate": 3.62769685394989e-06, "loss": 0.1465, "step": 8834 }, { "epoch": 1.2, "grad_norm": 4.290045609761318, "learning_rate": 3.626637648034984e-06, "loss": 0.2068, "step": 8835 }, { "epoch": 1.2, "grad_norm": 2.7116729509639255, "learning_rate": 3.62557850877709e-06, "loss": 0.1344, "step": 8836 }, { "epoch": 1.2, "grad_norm": 4.056224904246832, "learning_rate": 3.62451943622761e-06, "loss": 0.1691, "step": 8837 }, { "epoch": 1.2, "grad_norm": 3.0176799161041905, "learning_rate": 3.623460430437951e-06, "loss": 0.1728, "step": 8838 }, { "epoch": 1.2, "grad_norm": 4.084325515089186, "learning_rate": 3.6224014914595084e-06, "loss": 0.1602, "step": 8839 }, { "epoch": 1.2, "grad_norm": 3.1011206973314422, "learning_rate": 3.6213426193436824e-06, "loss": 0.1531, "step": 8840 }, { "epoch": 1.2, "grad_norm": 4.1477090680997515, "learning_rate": 3.620283814141863e-06, "loss": 0.2003, "step": 8841 }, { "epoch": 1.2, "grad_norm": 3.203546951195425, "learning_rate": 3.6192250759054427e-06, "loss": 0.1316, "step": 8842 }, { "epoch": 1.2, "grad_norm": 3.387084845690884, "learning_rate": 3.6181664046858045e-06, "loss": 0.2043, "step": 8843 }, { "epoch": 1.2, "grad_norm": 3.28268530278343, "learning_rate": 3.6171078005343363e-06, "loss": 0.1878, "step": 8844 }, { "epoch": 1.2, "grad_norm": 4.490551514184768, "learning_rate": 3.616049263502413e-06, "loss": 0.1958, "step": 8845 }, { "epoch": 1.2, "grad_norm": 3.0623763581209227, "learning_rate": 3.6149907936414153e-06, "loss": 0.1435, "step": 8846 }, { "epoch": 1.2, "grad_norm": 5.287232402346467, "learning_rate": 3.6139323910027136e-06, "loss": 0.1749, "step": 8847 }, { "epoch": 1.2, "grad_norm": 3.8672566681757545, "learning_rate": 3.612874055637683e-06, "loss": 0.1931, "step": 8848 }, { "epoch": 1.2, "grad_norm": 3.6207871671091905, "learning_rate": 3.6118157875976843e-06, "loss": 0.1582, "step": 8849 }, { "epoch": 1.2, "grad_norm": 3.257429112351275, "learning_rate": 3.6107575869340867e-06, "loss": 0.1384, "step": 8850 }, { "epoch": 1.2, "grad_norm": 3.8369013201896776, "learning_rate": 3.6096994536982456e-06, "loss": 0.1846, "step": 8851 }, { "epoch": 1.2, "grad_norm": 3.1370920447579618, "learning_rate": 3.608641387941523e-06, "loss": 0.1526, "step": 8852 }, { "epoch": 1.2, "grad_norm": 3.477965131023688, "learning_rate": 3.607583389715269e-06, "loss": 0.1697, "step": 8853 }, { "epoch": 1.2, "grad_norm": 4.2523103920202185, "learning_rate": 3.6065254590708363e-06, "loss": 0.1767, "step": 8854 }, { "epoch": 1.2, "grad_norm": 3.385175672424995, "learning_rate": 3.605467596059573e-06, "loss": 0.1836, "step": 8855 }, { "epoch": 1.2, "grad_norm": 3.858448131512708, "learning_rate": 3.60440980073282e-06, "loss": 0.2009, "step": 8856 }, { "epoch": 1.2, "grad_norm": 3.204184297305076, "learning_rate": 3.6033520731419214e-06, "loss": 0.1418, "step": 8857 }, { "epoch": 1.2, "grad_norm": 2.9039393701426492, "learning_rate": 3.602294413338211e-06, "loss": 0.1542, "step": 8858 }, { "epoch": 1.21, "grad_norm": 3.4827425145752007, "learning_rate": 3.6012368213730265e-06, "loss": 0.1313, "step": 8859 }, { "epoch": 1.21, "grad_norm": 3.437026933733021, "learning_rate": 3.6001792972976957e-06, "loss": 0.1956, "step": 8860 }, { "epoch": 1.21, "grad_norm": 3.3411047893113035, "learning_rate": 3.599121841163551e-06, "loss": 0.1687, "step": 8861 }, { "epoch": 1.21, "grad_norm": 4.201334117447246, "learning_rate": 3.598064453021911e-06, "loss": 0.1762, "step": 8862 }, { "epoch": 1.21, "grad_norm": 3.5980005905824757, "learning_rate": 3.597007132924102e-06, "loss": 0.1738, "step": 8863 }, { "epoch": 1.21, "grad_norm": 3.38294142885197, "learning_rate": 3.5959498809214366e-06, "loss": 0.1631, "step": 8864 }, { "epoch": 1.21, "grad_norm": 3.9944158609215408, "learning_rate": 3.594892697065232e-06, "loss": 0.1563, "step": 8865 }, { "epoch": 1.21, "grad_norm": 3.9530612422662954, "learning_rate": 3.5938355814067986e-06, "loss": 0.1969, "step": 8866 }, { "epoch": 1.21, "grad_norm": 4.124467920099141, "learning_rate": 3.592778533997446e-06, "loss": 0.2238, "step": 8867 }, { "epoch": 1.21, "grad_norm": 3.7697332018835596, "learning_rate": 3.5917215548884765e-06, "loss": 0.1981, "step": 8868 }, { "epoch": 1.21, "grad_norm": 3.365704204594341, "learning_rate": 3.5906646441311934e-06, "loss": 0.162, "step": 8869 }, { "epoch": 1.21, "grad_norm": 3.7267012178693357, "learning_rate": 3.58960780177689e-06, "loss": 0.1463, "step": 8870 }, { "epoch": 1.21, "grad_norm": 3.708618878108552, "learning_rate": 3.588551027876865e-06, "loss": 0.1893, "step": 8871 }, { "epoch": 1.21, "grad_norm": 3.5735088126983725, "learning_rate": 3.5874943224824097e-06, "loss": 0.169, "step": 8872 }, { "epoch": 1.21, "grad_norm": 3.480514707426302, "learning_rate": 3.58643768564481e-06, "loss": 0.196, "step": 8873 }, { "epoch": 1.21, "grad_norm": 3.2589640659782635, "learning_rate": 3.58538111741535e-06, "loss": 0.1725, "step": 8874 }, { "epoch": 1.21, "grad_norm": 3.578655639458785, "learning_rate": 3.584324617845314e-06, "loss": 0.1728, "step": 8875 }, { "epoch": 1.21, "grad_norm": 3.8710018934699426, "learning_rate": 3.5832681869859765e-06, "loss": 0.2214, "step": 8876 }, { "epoch": 1.21, "grad_norm": 4.303174253497888, "learning_rate": 3.582211824888615e-06, "loss": 0.174, "step": 8877 }, { "epoch": 1.21, "grad_norm": 4.054417576178395, "learning_rate": 3.581155531604497e-06, "loss": 0.2237, "step": 8878 }, { "epoch": 1.21, "grad_norm": 3.147292287387798, "learning_rate": 3.5800993071848945e-06, "loss": 0.1546, "step": 8879 }, { "epoch": 1.21, "grad_norm": 3.4421696068296974, "learning_rate": 3.5790431516810673e-06, "loss": 0.2111, "step": 8880 }, { "epoch": 1.21, "grad_norm": 3.3452568392399917, "learning_rate": 3.5779870651442826e-06, "loss": 0.2213, "step": 8881 }, { "epoch": 1.21, "grad_norm": 3.250731933050154, "learning_rate": 3.5769310476257935e-06, "loss": 0.1391, "step": 8882 }, { "epoch": 1.21, "grad_norm": 2.9729977430283427, "learning_rate": 3.575875099176857e-06, "loss": 0.1548, "step": 8883 }, { "epoch": 1.21, "grad_norm": 3.7170230195725242, "learning_rate": 3.5748192198487215e-06, "loss": 0.169, "step": 8884 }, { "epoch": 1.21, "grad_norm": 2.5700780135134655, "learning_rate": 3.573763409692637e-06, "loss": 0.1379, "step": 8885 }, { "epoch": 1.21, "grad_norm": 4.159386362977719, "learning_rate": 3.572707668759848e-06, "loss": 0.17, "step": 8886 }, { "epoch": 1.21, "grad_norm": 3.567939585721713, "learning_rate": 3.5716519971015947e-06, "loss": 0.1575, "step": 8887 }, { "epoch": 1.21, "grad_norm": 5.113726845935689, "learning_rate": 3.5705963947691135e-06, "loss": 0.1974, "step": 8888 }, { "epoch": 1.21, "grad_norm": 3.9260208394385607, "learning_rate": 3.569540861813643e-06, "loss": 0.179, "step": 8889 }, { "epoch": 1.21, "grad_norm": 3.6018775829851863, "learning_rate": 3.568485398286409e-06, "loss": 0.2043, "step": 8890 }, { "epoch": 1.21, "grad_norm": 3.940890423735566, "learning_rate": 3.5674300042386434e-06, "loss": 0.1958, "step": 8891 }, { "epoch": 1.21, "grad_norm": 3.80218815188737, "learning_rate": 3.5663746797215658e-06, "loss": 0.1312, "step": 8892 }, { "epoch": 1.21, "grad_norm": 3.811348269370393, "learning_rate": 3.565319424786402e-06, "loss": 0.1839, "step": 8893 }, { "epoch": 1.21, "grad_norm": 3.2403183103637425, "learning_rate": 3.564264239484367e-06, "loss": 0.1611, "step": 8894 }, { "epoch": 1.21, "grad_norm": 3.06438107205218, "learning_rate": 3.5632091238666746e-06, "loss": 0.1592, "step": 8895 }, { "epoch": 1.21, "grad_norm": 3.6710154494284755, "learning_rate": 3.5621540779845366e-06, "loss": 0.1656, "step": 8896 }, { "epoch": 1.21, "grad_norm": 2.8154881092923087, "learning_rate": 3.561099101889158e-06, "loss": 0.157, "step": 8897 }, { "epoch": 1.21, "grad_norm": 2.9370141575723907, "learning_rate": 3.560044195631746e-06, "loss": 0.165, "step": 8898 }, { "epoch": 1.21, "grad_norm": 3.009731207757748, "learning_rate": 3.558989359263498e-06, "loss": 0.1607, "step": 8899 }, { "epoch": 1.21, "grad_norm": 3.6294479942032507, "learning_rate": 3.5579345928356158e-06, "loss": 0.1728, "step": 8900 }, { "epoch": 1.21, "grad_norm": 3.9683109352452144, "learning_rate": 3.556879896399287e-06, "loss": 0.1965, "step": 8901 }, { "epoch": 1.21, "grad_norm": 3.818183934479101, "learning_rate": 3.555825270005707e-06, "loss": 0.2066, "step": 8902 }, { "epoch": 1.21, "grad_norm": 3.68856083192283, "learning_rate": 3.5547707137060593e-06, "loss": 0.1638, "step": 8903 }, { "epoch": 1.21, "grad_norm": 3.895819543150844, "learning_rate": 3.5537162275515295e-06, "loss": 0.1256, "step": 8904 }, { "epoch": 1.21, "grad_norm": 4.094199693819785, "learning_rate": 3.5526618115932974e-06, "loss": 0.1608, "step": 8905 }, { "epoch": 1.21, "grad_norm": 4.138651727147414, "learning_rate": 3.55160746588254e-06, "loss": 0.1982, "step": 8906 }, { "epoch": 1.21, "grad_norm": 3.033089496361756, "learning_rate": 3.5505531904704287e-06, "loss": 0.1597, "step": 8907 }, { "epoch": 1.21, "grad_norm": 2.9042393467733145, "learning_rate": 3.5494989854081374e-06, "loss": 0.1723, "step": 8908 }, { "epoch": 1.21, "grad_norm": 3.362842193088103, "learning_rate": 3.5484448507468273e-06, "loss": 0.1822, "step": 8909 }, { "epoch": 1.21, "grad_norm": 3.8069428544425303, "learning_rate": 3.5473907865376672e-06, "loss": 0.1692, "step": 8910 }, { "epoch": 1.21, "grad_norm": 3.7115668323929847, "learning_rate": 3.5463367928318116e-06, "loss": 0.1863, "step": 8911 }, { "epoch": 1.21, "grad_norm": 3.5229129706842137, "learning_rate": 3.5452828696804196e-06, "loss": 0.1619, "step": 8912 }, { "epoch": 1.21, "grad_norm": 5.725552759408466, "learning_rate": 3.5442290171346427e-06, "loss": 0.1783, "step": 8913 }, { "epoch": 1.21, "grad_norm": 4.5250192944437195, "learning_rate": 3.5431752352456335e-06, "loss": 0.1656, "step": 8914 }, { "epoch": 1.21, "grad_norm": 3.8472723457916786, "learning_rate": 3.5421215240645334e-06, "loss": 0.1906, "step": 8915 }, { "epoch": 1.21, "grad_norm": 3.622724499379614, "learning_rate": 3.5410678836424884e-06, "loss": 0.2044, "step": 8916 }, { "epoch": 1.21, "grad_norm": 3.9268689825266936, "learning_rate": 3.5400143140306355e-06, "loss": 0.1658, "step": 8917 }, { "epoch": 1.21, "grad_norm": 3.857637601745735, "learning_rate": 3.538960815280111e-06, "loss": 0.168, "step": 8918 }, { "epoch": 1.21, "grad_norm": 3.300909637196817, "learning_rate": 3.537907387442047e-06, "loss": 0.1343, "step": 8919 }, { "epoch": 1.21, "grad_norm": 4.666602134489714, "learning_rate": 3.5368540305675738e-06, "loss": 0.1703, "step": 8920 }, { "epoch": 1.21, "grad_norm": 4.152622976069163, "learning_rate": 3.535800744707814e-06, "loss": 0.1525, "step": 8921 }, { "epoch": 1.21, "grad_norm": 5.212260313904286, "learning_rate": 3.5347475299138932e-06, "loss": 0.1713, "step": 8922 }, { "epoch": 1.21, "grad_norm": 4.246857076275951, "learning_rate": 3.5336943862369256e-06, "loss": 0.1704, "step": 8923 }, { "epoch": 1.21, "grad_norm": 3.0206974403300486, "learning_rate": 3.53264131372803e-06, "loss": 0.1689, "step": 8924 }, { "epoch": 1.21, "grad_norm": 3.2393456695349756, "learning_rate": 3.5315883124383143e-06, "loss": 0.193, "step": 8925 }, { "epoch": 1.21, "grad_norm": 3.25715240692101, "learning_rate": 3.5305353824188893e-06, "loss": 0.1625, "step": 8926 }, { "epoch": 1.21, "grad_norm": 3.4873336414935903, "learning_rate": 3.5294825237208573e-06, "loss": 0.1589, "step": 8927 }, { "epoch": 1.21, "grad_norm": 4.238410588433453, "learning_rate": 3.528429736395324e-06, "loss": 0.1891, "step": 8928 }, { "epoch": 1.21, "grad_norm": 3.68450359159976, "learning_rate": 3.5273770204933814e-06, "loss": 0.1707, "step": 8929 }, { "epoch": 1.21, "grad_norm": 2.970474314576743, "learning_rate": 3.526324376066129e-06, "loss": 0.1241, "step": 8930 }, { "epoch": 1.21, "grad_norm": 4.504538890858875, "learning_rate": 3.525271803164653e-06, "loss": 0.1835, "step": 8931 }, { "epoch": 1.21, "grad_norm": 3.496087626687576, "learning_rate": 3.524219301840043e-06, "loss": 0.1969, "step": 8932 }, { "epoch": 1.22, "grad_norm": 3.7291341745048423, "learning_rate": 3.5231668721433808e-06, "loss": 0.2058, "step": 8933 }, { "epoch": 1.22, "grad_norm": 3.3971318720494645, "learning_rate": 3.522114514125752e-06, "loss": 0.1668, "step": 8934 }, { "epoch": 1.22, "grad_norm": 4.05468678061905, "learning_rate": 3.521062227838229e-06, "loss": 0.168, "step": 8935 }, { "epoch": 1.22, "grad_norm": 3.4741455293380117, "learning_rate": 3.5200100133318836e-06, "loss": 0.1861, "step": 8936 }, { "epoch": 1.22, "grad_norm": 4.1081499504558145, "learning_rate": 3.5189578706577896e-06, "loss": 0.1761, "step": 8937 }, { "epoch": 1.22, "grad_norm": 3.219404415109366, "learning_rate": 3.5179057998670113e-06, "loss": 0.1629, "step": 8938 }, { "epoch": 1.22, "grad_norm": 3.89261613331732, "learning_rate": 3.516853801010614e-06, "loss": 0.1525, "step": 8939 }, { "epoch": 1.22, "grad_norm": 3.1539571063891296, "learning_rate": 3.5158018741396534e-06, "loss": 0.1493, "step": 8940 }, { "epoch": 1.22, "grad_norm": 4.500310469567811, "learning_rate": 3.5147500193051896e-06, "loss": 0.1948, "step": 8941 }, { "epoch": 1.22, "grad_norm": 3.509066132230449, "learning_rate": 3.5136982365582704e-06, "loss": 0.1493, "step": 8942 }, { "epoch": 1.22, "grad_norm": 4.320769529339552, "learning_rate": 3.5126465259499497e-06, "loss": 0.1829, "step": 8943 }, { "epoch": 1.22, "grad_norm": 4.590994421113589, "learning_rate": 3.5115948875312677e-06, "loss": 0.1962, "step": 8944 }, { "epoch": 1.22, "grad_norm": 2.588956893127304, "learning_rate": 3.510543321353271e-06, "loss": 0.1418, "step": 8945 }, { "epoch": 1.22, "grad_norm": 3.9794546334359184, "learning_rate": 3.509491827466995e-06, "loss": 0.166, "step": 8946 }, { "epoch": 1.22, "grad_norm": 3.262014730516515, "learning_rate": 3.5084404059234773e-06, "loss": 0.1482, "step": 8947 }, { "epoch": 1.22, "grad_norm": 3.879539512404166, "learning_rate": 3.5073890567737457e-06, "loss": 0.1689, "step": 8948 }, { "epoch": 1.22, "grad_norm": 3.3449417261934826, "learning_rate": 3.5063377800688325e-06, "loss": 0.1617, "step": 8949 }, { "epoch": 1.22, "grad_norm": 3.583572339315948, "learning_rate": 3.5052865758597577e-06, "loss": 0.13, "step": 8950 }, { "epoch": 1.22, "grad_norm": 3.5305317325491785, "learning_rate": 3.504235444197545e-06, "loss": 0.1521, "step": 8951 }, { "epoch": 1.22, "grad_norm": 4.847829895869409, "learning_rate": 3.5031843851332105e-06, "loss": 0.1843, "step": 8952 }, { "epoch": 1.22, "grad_norm": 4.113627970641154, "learning_rate": 3.5021333987177695e-06, "loss": 0.1313, "step": 8953 }, { "epoch": 1.22, "grad_norm": 4.423642652591877, "learning_rate": 3.5010824850022295e-06, "loss": 0.1567, "step": 8954 }, { "epoch": 1.22, "grad_norm": 6.193161981155962, "learning_rate": 3.5000316440376014e-06, "loss": 0.1939, "step": 8955 }, { "epoch": 1.22, "grad_norm": 3.6484084336807383, "learning_rate": 3.4989808758748834e-06, "loss": 0.1477, "step": 8956 }, { "epoch": 1.22, "grad_norm": 4.223762341761627, "learning_rate": 3.4979301805650805e-06, "loss": 0.2036, "step": 8957 }, { "epoch": 1.22, "grad_norm": 4.681239885858964, "learning_rate": 3.4968795581591843e-06, "loss": 0.1703, "step": 8958 }, { "epoch": 1.22, "grad_norm": 3.3287062156426024, "learning_rate": 3.4958290087081893e-06, "loss": 0.1574, "step": 8959 }, { "epoch": 1.22, "grad_norm": 3.4868336703573264, "learning_rate": 3.4947785322630847e-06, "loss": 0.1956, "step": 8960 }, { "epoch": 1.22, "grad_norm": 3.403489507246488, "learning_rate": 3.493728128874857e-06, "loss": 0.1669, "step": 8961 }, { "epoch": 1.22, "grad_norm": 3.8049459988069096, "learning_rate": 3.492677798594486e-06, "loss": 0.1823, "step": 8962 }, { "epoch": 1.22, "grad_norm": 4.026972663416669, "learning_rate": 3.4916275414729527e-06, "loss": 0.1709, "step": 8963 }, { "epoch": 1.22, "grad_norm": 3.1046428773210746, "learning_rate": 3.4905773575612284e-06, "loss": 0.1448, "step": 8964 }, { "epoch": 1.22, "grad_norm": 3.9968819348864653, "learning_rate": 3.4895272469102883e-06, "loss": 0.2202, "step": 8965 }, { "epoch": 1.22, "grad_norm": 3.41277034106894, "learning_rate": 3.4884772095710962e-06, "loss": 0.1645, "step": 8966 }, { "epoch": 1.22, "grad_norm": 3.383438991476435, "learning_rate": 3.4874272455946217e-06, "loss": 0.1675, "step": 8967 }, { "epoch": 1.22, "grad_norm": 3.918297655234693, "learning_rate": 3.4863773550318204e-06, "loss": 0.173, "step": 8968 }, { "epoch": 1.22, "grad_norm": 4.124857238669712, "learning_rate": 3.485327537933653e-06, "loss": 0.1746, "step": 8969 }, { "epoch": 1.22, "grad_norm": 4.500086792153677, "learning_rate": 3.4842777943510697e-06, "loss": 0.1921, "step": 8970 }, { "epoch": 1.22, "grad_norm": 3.354356277333102, "learning_rate": 3.483228124335023e-06, "loss": 0.1953, "step": 8971 }, { "epoch": 1.22, "grad_norm": 3.3667763967581763, "learning_rate": 3.4821785279364585e-06, "loss": 0.1959, "step": 8972 }, { "epoch": 1.22, "grad_norm": 3.2479835595448554, "learning_rate": 3.4811290052063187e-06, "loss": 0.1354, "step": 8973 }, { "epoch": 1.22, "grad_norm": 4.250755853643333, "learning_rate": 3.4800795561955442e-06, "loss": 0.166, "step": 8974 }, { "epoch": 1.22, "grad_norm": 4.814488862917156, "learning_rate": 3.4790301809550676e-06, "loss": 0.1884, "step": 8975 }, { "epoch": 1.22, "grad_norm": 3.789319048681657, "learning_rate": 3.4779808795358256e-06, "loss": 0.1651, "step": 8976 }, { "epoch": 1.22, "grad_norm": 3.5284961260253125, "learning_rate": 3.476931651988742e-06, "loss": 0.1687, "step": 8977 }, { "epoch": 1.22, "grad_norm": 3.833059572662276, "learning_rate": 3.4758824983647445e-06, "loss": 0.1776, "step": 8978 }, { "epoch": 1.22, "grad_norm": 4.353398744434046, "learning_rate": 3.4748334187147527e-06, "loss": 0.1619, "step": 8979 }, { "epoch": 1.22, "grad_norm": 3.1029886615800675, "learning_rate": 3.4737844130896874e-06, "loss": 0.1757, "step": 8980 }, { "epoch": 1.22, "grad_norm": 3.271337595960714, "learning_rate": 3.472735481540459e-06, "loss": 0.1378, "step": 8981 }, { "epoch": 1.22, "grad_norm": 4.585730111056677, "learning_rate": 3.471686624117982e-06, "loss": 0.1948, "step": 8982 }, { "epoch": 1.22, "grad_norm": 3.8081974324838836, "learning_rate": 3.4706378408731587e-06, "loss": 0.1941, "step": 8983 }, { "epoch": 1.22, "grad_norm": 3.555766789934669, "learning_rate": 3.4695891318568965e-06, "loss": 0.1457, "step": 8984 }, { "epoch": 1.22, "grad_norm": 3.760644201386937, "learning_rate": 3.468540497120092e-06, "loss": 0.195, "step": 8985 }, { "epoch": 1.22, "grad_norm": 3.502584343154103, "learning_rate": 3.4674919367136454e-06, "loss": 0.198, "step": 8986 }, { "epoch": 1.22, "grad_norm": 5.262461828603563, "learning_rate": 3.466443450688445e-06, "loss": 0.1814, "step": 8987 }, { "epoch": 1.22, "grad_norm": 3.710521497923693, "learning_rate": 3.4653950390953833e-06, "loss": 0.1887, "step": 8988 }, { "epoch": 1.22, "grad_norm": 2.5770651998899528, "learning_rate": 3.4643467019853415e-06, "loss": 0.164, "step": 8989 }, { "epoch": 1.22, "grad_norm": 3.8312852237497603, "learning_rate": 3.463298439409205e-06, "loss": 0.1826, "step": 8990 }, { "epoch": 1.22, "grad_norm": 3.746462187180437, "learning_rate": 3.46225025141785e-06, "loss": 0.1927, "step": 8991 }, { "epoch": 1.22, "grad_norm": 3.8127562336636798, "learning_rate": 3.461202138062153e-06, "loss": 0.1552, "step": 8992 }, { "epoch": 1.22, "grad_norm": 3.9773018412818573, "learning_rate": 3.4601540993929805e-06, "loss": 0.1937, "step": 8993 }, { "epoch": 1.22, "grad_norm": 3.7802434864229615, "learning_rate": 3.4591061354612056e-06, "loss": 0.1855, "step": 8994 }, { "epoch": 1.22, "grad_norm": 3.9656203483240855, "learning_rate": 3.4580582463176873e-06, "loss": 0.199, "step": 8995 }, { "epoch": 1.22, "grad_norm": 4.704733627401178, "learning_rate": 3.457010432013288e-06, "loss": 0.1547, "step": 8996 }, { "epoch": 1.22, "grad_norm": 4.048492045669616, "learning_rate": 3.4559626925988623e-06, "loss": 0.191, "step": 8997 }, { "epoch": 1.22, "grad_norm": 3.803333037155936, "learning_rate": 3.4549150281252635e-06, "loss": 0.1663, "step": 8998 }, { "epoch": 1.22, "grad_norm": 3.4643925725093907, "learning_rate": 3.45386743864334e-06, "loss": 0.1875, "step": 8999 }, { "epoch": 1.22, "grad_norm": 4.321484842679558, "learning_rate": 3.452819924203941e-06, "loss": 0.1517, "step": 9000 }, { "epoch": 1.22, "grad_norm": 4.621559868673954, "learning_rate": 3.451772484857903e-06, "loss": 0.1765, "step": 9001 }, { "epoch": 1.22, "grad_norm": 3.867586501235449, "learning_rate": 3.450725120656069e-06, "loss": 0.1856, "step": 9002 }, { "epoch": 1.22, "grad_norm": 4.368087456924716, "learning_rate": 3.4496778316492685e-06, "loss": 0.2057, "step": 9003 }, { "epoch": 1.22, "grad_norm": 3.189222687975933, "learning_rate": 3.4486306178883354e-06, "loss": 0.1207, "step": 9004 }, { "epoch": 1.22, "grad_norm": 4.057635476043052, "learning_rate": 3.447583479424097e-06, "loss": 0.1714, "step": 9005 }, { "epoch": 1.23, "grad_norm": 3.471866731575164, "learning_rate": 3.4465364163073763e-06, "loss": 0.1831, "step": 9006 }, { "epoch": 1.23, "grad_norm": 4.837892391414738, "learning_rate": 3.4454894285889916e-06, "loss": 0.1538, "step": 9007 }, { "epoch": 1.23, "grad_norm": 3.7947075039079095, "learning_rate": 3.4444425163197626e-06, "loss": 0.1634, "step": 9008 }, { "epoch": 1.23, "grad_norm": 4.558018416760028, "learning_rate": 3.443395679550498e-06, "loss": 0.1692, "step": 9009 }, { "epoch": 1.23, "grad_norm": 2.9903211457406003, "learning_rate": 3.44234891833201e-06, "loss": 0.1589, "step": 9010 }, { "epoch": 1.23, "grad_norm": 4.4031424732630295, "learning_rate": 3.4413022327151012e-06, "loss": 0.1718, "step": 9011 }, { "epoch": 1.23, "grad_norm": 4.499295190070728, "learning_rate": 3.4402556227505746e-06, "loss": 0.1988, "step": 9012 }, { "epoch": 1.23, "grad_norm": 4.049931890129313, "learning_rate": 3.439209088489227e-06, "loss": 0.2152, "step": 9013 }, { "epoch": 1.23, "grad_norm": 4.056812309594317, "learning_rate": 3.438162629981855e-06, "loss": 0.1776, "step": 9014 }, { "epoch": 1.23, "grad_norm": 2.645376249091225, "learning_rate": 3.4371162472792475e-06, "loss": 0.1217, "step": 9015 }, { "epoch": 1.23, "grad_norm": 3.25935319784184, "learning_rate": 3.4360699404321895e-06, "loss": 0.1628, "step": 9016 }, { "epoch": 1.23, "grad_norm": 6.037117921095698, "learning_rate": 3.435023709491467e-06, "loss": 0.1818, "step": 9017 }, { "epoch": 1.23, "grad_norm": 3.3590837457288116, "learning_rate": 3.433977554507857e-06, "loss": 0.156, "step": 9018 }, { "epoch": 1.23, "grad_norm": 3.4087812203336822, "learning_rate": 3.4329314755321384e-06, "loss": 0.1348, "step": 9019 }, { "epoch": 1.23, "grad_norm": 3.654466818780571, "learning_rate": 3.431885472615081e-06, "loss": 0.1455, "step": 9020 }, { "epoch": 1.23, "grad_norm": 3.545733498219093, "learning_rate": 3.430839545807455e-06, "loss": 0.1627, "step": 9021 }, { "epoch": 1.23, "grad_norm": 4.312565837803306, "learning_rate": 3.4297936951600217e-06, "loss": 0.1885, "step": 9022 }, { "epoch": 1.23, "grad_norm": 8.711957536490926, "learning_rate": 3.428747920723546e-06, "loss": 0.1292, "step": 9023 }, { "epoch": 1.23, "grad_norm": 3.3051197508776653, "learning_rate": 3.4277022225487834e-06, "loss": 0.1266, "step": 9024 }, { "epoch": 1.23, "grad_norm": 3.720941164384531, "learning_rate": 3.4266566006864874e-06, "loss": 0.1856, "step": 9025 }, { "epoch": 1.23, "grad_norm": 4.531696284697464, "learning_rate": 3.4256110551874076e-06, "loss": 0.1777, "step": 9026 }, { "epoch": 1.23, "grad_norm": 3.5730552196476326, "learning_rate": 3.424565586102293e-06, "loss": 0.2089, "step": 9027 }, { "epoch": 1.23, "grad_norm": 4.586869969431808, "learning_rate": 3.423520193481882e-06, "loss": 0.1876, "step": 9028 }, { "epoch": 1.23, "grad_norm": 2.155787816552854, "learning_rate": 3.4224748773769166e-06, "loss": 0.1484, "step": 9029 }, { "epoch": 1.23, "grad_norm": 3.217254516206906, "learning_rate": 3.4214296378381296e-06, "loss": 0.1585, "step": 9030 }, { "epoch": 1.23, "grad_norm": 3.9582851439355315, "learning_rate": 3.420384474916254e-06, "loss": 0.1447, "step": 9031 }, { "epoch": 1.23, "grad_norm": 6.310721635826722, "learning_rate": 3.4193393886620153e-06, "loss": 0.1632, "step": 9032 }, { "epoch": 1.23, "grad_norm": 3.624445410251156, "learning_rate": 3.4182943791261414e-06, "loss": 0.2024, "step": 9033 }, { "epoch": 1.23, "grad_norm": 3.4738078802037142, "learning_rate": 3.4172494463593487e-06, "loss": 0.1391, "step": 9034 }, { "epoch": 1.23, "grad_norm": 3.1290853576292648, "learning_rate": 3.4162045904123563e-06, "loss": 0.1259, "step": 9035 }, { "epoch": 1.23, "grad_norm": 3.8455494117542948, "learning_rate": 3.4151598113358743e-06, "loss": 0.1808, "step": 9036 }, { "epoch": 1.23, "grad_norm": 3.4856180972695916, "learning_rate": 3.4141151091806134e-06, "loss": 0.1354, "step": 9037 }, { "epoch": 1.23, "grad_norm": 5.912307895938129, "learning_rate": 3.4130704839972773e-06, "loss": 0.1797, "step": 9038 }, { "epoch": 1.23, "grad_norm": 2.4323093634546744, "learning_rate": 3.4120259358365715e-06, "loss": 0.1366, "step": 9039 }, { "epoch": 1.23, "grad_norm": 3.4194546980357363, "learning_rate": 3.410981464749189e-06, "loss": 0.1564, "step": 9040 }, { "epoch": 1.23, "grad_norm": 3.685186739743652, "learning_rate": 3.409937070785828e-06, "loss": 0.183, "step": 9041 }, { "epoch": 1.23, "grad_norm": 3.720709741832952, "learning_rate": 3.408892753997175e-06, "loss": 0.1871, "step": 9042 }, { "epoch": 1.23, "grad_norm": 3.3440712240564126, "learning_rate": 3.40784851443392e-06, "loss": 0.1369, "step": 9043 }, { "epoch": 1.23, "grad_norm": 4.3892299736445075, "learning_rate": 3.4068043521467424e-06, "loss": 0.1989, "step": 9044 }, { "epoch": 1.23, "grad_norm": 3.538360308667543, "learning_rate": 3.4057602671863244e-06, "loss": 0.1494, "step": 9045 }, { "epoch": 1.23, "grad_norm": 4.496910645169113, "learning_rate": 3.4047162596033386e-06, "loss": 0.1887, "step": 9046 }, { "epoch": 1.23, "grad_norm": 2.895987574388931, "learning_rate": 3.40367232944846e-06, "loss": 0.1746, "step": 9047 }, { "epoch": 1.23, "grad_norm": 3.6128945175838463, "learning_rate": 3.4026284767723524e-06, "loss": 0.1585, "step": 9048 }, { "epoch": 1.23, "grad_norm": 4.397248614361258, "learning_rate": 3.4015847016256842e-06, "loss": 0.1594, "step": 9049 }, { "epoch": 1.23, "grad_norm": 5.33948417800021, "learning_rate": 3.400541004059111e-06, "loss": 0.1816, "step": 9050 }, { "epoch": 1.23, "grad_norm": 3.7408750220290816, "learning_rate": 3.3994973841232925e-06, "loss": 0.1441, "step": 9051 }, { "epoch": 1.23, "grad_norm": 5.024157308691727, "learning_rate": 3.3984538418688795e-06, "loss": 0.1655, "step": 9052 }, { "epoch": 1.23, "grad_norm": 3.226482141405985, "learning_rate": 3.3974103773465243e-06, "loss": 0.1672, "step": 9053 }, { "epoch": 1.23, "grad_norm": 3.4729304756233863, "learning_rate": 3.3963669906068687e-06, "loss": 0.1834, "step": 9054 }, { "epoch": 1.23, "grad_norm": 4.346556104950098, "learning_rate": 3.3953236817005565e-06, "loss": 0.2081, "step": 9055 }, { "epoch": 1.23, "grad_norm": 3.082196331210179, "learning_rate": 3.394280450678223e-06, "loss": 0.161, "step": 9056 }, { "epoch": 1.23, "grad_norm": 3.7031495558198597, "learning_rate": 3.3932372975905027e-06, "loss": 0.1545, "step": 9057 }, { "epoch": 1.23, "grad_norm": 3.655767998950126, "learning_rate": 3.3921942224880277e-06, "loss": 0.1782, "step": 9058 }, { "epoch": 1.23, "grad_norm": 3.4936220654651624, "learning_rate": 3.391151225421422e-06, "loss": 0.2192, "step": 9059 }, { "epoch": 1.23, "grad_norm": 3.220279400106194, "learning_rate": 3.39010830644131e-06, "loss": 0.1736, "step": 9060 }, { "epoch": 1.23, "grad_norm": 3.680424997312236, "learning_rate": 3.389065465598308e-06, "loss": 0.1728, "step": 9061 }, { "epoch": 1.23, "grad_norm": 3.9976900760845173, "learning_rate": 3.3880227029430335e-06, "loss": 0.1432, "step": 9062 }, { "epoch": 1.23, "grad_norm": 3.6429146576277067, "learning_rate": 3.386980018526095e-06, "loss": 0.1538, "step": 9063 }, { "epoch": 1.23, "grad_norm": 4.836264452463982, "learning_rate": 3.3859374123981028e-06, "loss": 0.146, "step": 9064 }, { "epoch": 1.23, "grad_norm": 5.845067705690301, "learning_rate": 3.3848948846096575e-06, "loss": 0.1766, "step": 9065 }, { "epoch": 1.23, "grad_norm": 4.064945440796079, "learning_rate": 3.383852435211363e-06, "loss": 0.1729, "step": 9066 }, { "epoch": 1.23, "grad_norm": 3.1717509431918804, "learning_rate": 3.3828100642538097e-06, "loss": 0.158, "step": 9067 }, { "epoch": 1.23, "grad_norm": 3.191050098122489, "learning_rate": 3.381767771787595e-06, "loss": 0.1755, "step": 9068 }, { "epoch": 1.23, "grad_norm": 3.3623155718326996, "learning_rate": 3.380725557863303e-06, "loss": 0.1651, "step": 9069 }, { "epoch": 1.23, "grad_norm": 3.467082032603, "learning_rate": 3.379683422531521e-06, "loss": 0.1923, "step": 9070 }, { "epoch": 1.23, "grad_norm": 3.4619625065679642, "learning_rate": 3.3786413658428276e-06, "loss": 0.1704, "step": 9071 }, { "epoch": 1.23, "grad_norm": 2.424968651412354, "learning_rate": 3.377599387847803e-06, "loss": 0.1257, "step": 9072 }, { "epoch": 1.23, "grad_norm": 3.112711662998418, "learning_rate": 3.3765574885970163e-06, "loss": 0.1461, "step": 9073 }, { "epoch": 1.23, "grad_norm": 3.560714921037106, "learning_rate": 3.3755156681410405e-06, "loss": 0.1629, "step": 9074 }, { "epoch": 1.23, "grad_norm": 3.0342781212137564, "learning_rate": 3.3744739265304367e-06, "loss": 0.146, "step": 9075 }, { "epoch": 1.23, "grad_norm": 3.7063342948537, "learning_rate": 3.373432263815771e-06, "loss": 0.1799, "step": 9076 }, { "epoch": 1.23, "grad_norm": 3.3696512378845167, "learning_rate": 3.372390680047597e-06, "loss": 0.1359, "step": 9077 }, { "epoch": 1.23, "grad_norm": 3.777256069911019, "learning_rate": 3.3713491752764715e-06, "loss": 0.1429, "step": 9078 }, { "epoch": 1.23, "grad_norm": 3.5689640931453415, "learning_rate": 3.370307749552943e-06, "loss": 0.1544, "step": 9079 }, { "epoch": 1.24, "grad_norm": 3.4398818300288547, "learning_rate": 3.3692664029275603e-06, "loss": 0.1604, "step": 9080 }, { "epoch": 1.24, "grad_norm": 3.3480105432475047, "learning_rate": 3.368225135450862e-06, "loss": 0.1373, "step": 9081 }, { "epoch": 1.24, "grad_norm": 3.9031753670708977, "learning_rate": 3.3671839471733906e-06, "loss": 0.1719, "step": 9082 }, { "epoch": 1.24, "grad_norm": 3.7479081839792507, "learning_rate": 3.3661428381456774e-06, "loss": 0.1593, "step": 9083 }, { "epoch": 1.24, "grad_norm": 3.262090537511603, "learning_rate": 3.365101808418255e-06, "loss": 0.133, "step": 9084 }, { "epoch": 1.24, "grad_norm": 4.288341571544941, "learning_rate": 3.36406085804165e-06, "loss": 0.1815, "step": 9085 }, { "epoch": 1.24, "grad_norm": 4.366517791261178, "learning_rate": 3.363019987066387e-06, "loss": 0.1683, "step": 9086 }, { "epoch": 1.24, "grad_norm": 3.159159829258511, "learning_rate": 3.3619791955429826e-06, "loss": 0.1992, "step": 9087 }, { "epoch": 1.24, "grad_norm": 3.1086167736567, "learning_rate": 3.360938483521956e-06, "loss": 0.151, "step": 9088 }, { "epoch": 1.24, "grad_norm": 5.255190085104433, "learning_rate": 3.3598978510538143e-06, "loss": 0.1811, "step": 9089 }, { "epoch": 1.24, "grad_norm": 3.655809419399234, "learning_rate": 3.358857298189069e-06, "loss": 0.1877, "step": 9090 }, { "epoch": 1.24, "grad_norm": 4.078520615795661, "learning_rate": 3.357816824978222e-06, "loss": 0.1867, "step": 9091 }, { "epoch": 1.24, "grad_norm": 2.9963723337527837, "learning_rate": 3.3567764314717744e-06, "loss": 0.1561, "step": 9092 }, { "epoch": 1.24, "grad_norm": 4.7897885682726296, "learning_rate": 3.3557361177202206e-06, "loss": 0.1471, "step": 9093 }, { "epoch": 1.24, "grad_norm": 2.631055589618682, "learning_rate": 3.3546958837740562e-06, "loss": 0.1512, "step": 9094 }, { "epoch": 1.24, "grad_norm": 3.3501140463036374, "learning_rate": 3.3536557296837673e-06, "loss": 0.1625, "step": 9095 }, { "epoch": 1.24, "grad_norm": 3.871964458723713, "learning_rate": 3.3526156554998367e-06, "loss": 0.1971, "step": 9096 }, { "epoch": 1.24, "grad_norm": 4.5032823039260395, "learning_rate": 3.351575661272749e-06, "loss": 0.1833, "step": 9097 }, { "epoch": 1.24, "grad_norm": 4.411505513100558, "learning_rate": 3.350535747052977e-06, "loss": 0.1496, "step": 9098 }, { "epoch": 1.24, "grad_norm": 2.981118318115497, "learning_rate": 3.3494959128909976e-06, "loss": 0.2013, "step": 9099 }, { "epoch": 1.24, "grad_norm": 3.034797423508213, "learning_rate": 3.3484561588372766e-06, "loss": 0.1245, "step": 9100 }, { "epoch": 1.24, "grad_norm": 3.6783148865600084, "learning_rate": 3.347416484942282e-06, "loss": 0.1619, "step": 9101 }, { "epoch": 1.24, "grad_norm": 4.422914820161794, "learning_rate": 3.346376891256471e-06, "loss": 0.2087, "step": 9102 }, { "epoch": 1.24, "grad_norm": 3.475792977842513, "learning_rate": 3.3453373778303044e-06, "loss": 0.1653, "step": 9103 }, { "epoch": 1.24, "grad_norm": 4.285781229676762, "learning_rate": 3.3442979447142343e-06, "loss": 0.185, "step": 9104 }, { "epoch": 1.24, "grad_norm": 4.624265424338046, "learning_rate": 3.3432585919587118e-06, "loss": 0.1297, "step": 9105 }, { "epoch": 1.24, "grad_norm": 3.3196224760958466, "learning_rate": 3.3422193196141796e-06, "loss": 0.1631, "step": 9106 }, { "epoch": 1.24, "grad_norm": 3.3606670470518276, "learning_rate": 3.341180127731083e-06, "loss": 0.188, "step": 9107 }, { "epoch": 1.24, "grad_norm": 4.2034953442639615, "learning_rate": 3.3401410163598557e-06, "loss": 0.1767, "step": 9108 }, { "epoch": 1.24, "grad_norm": 3.07120542182862, "learning_rate": 3.339101985550935e-06, "loss": 0.1447, "step": 9109 }, { "epoch": 1.24, "grad_norm": 3.1070026486614086, "learning_rate": 3.3380630353547494e-06, "loss": 0.1301, "step": 9110 }, { "epoch": 1.24, "grad_norm": 3.7369872160105873, "learning_rate": 3.337024165821726e-06, "loss": 0.1812, "step": 9111 }, { "epoch": 1.24, "grad_norm": 9.753361713678654, "learning_rate": 3.335985377002285e-06, "loss": 0.1879, "step": 9112 }, { "epoch": 1.24, "grad_norm": 2.6621976285628906, "learning_rate": 3.334946668946848e-06, "loss": 0.1621, "step": 9113 }, { "epoch": 1.24, "grad_norm": 2.306772645593448, "learning_rate": 3.3339080417058256e-06, "loss": 0.1123, "step": 9114 }, { "epoch": 1.24, "grad_norm": 3.1579569432592165, "learning_rate": 3.332869495329632e-06, "loss": 0.1376, "step": 9115 }, { "epoch": 1.24, "grad_norm": 3.1907115321214166, "learning_rate": 3.3318310298686694e-06, "loss": 0.1921, "step": 9116 }, { "epoch": 1.24, "grad_norm": 4.839139423308444, "learning_rate": 3.330792645373344e-06, "loss": 0.1451, "step": 9117 }, { "epoch": 1.24, "grad_norm": 4.392813817032452, "learning_rate": 3.3297543418940527e-06, "loss": 0.1893, "step": 9118 }, { "epoch": 1.24, "grad_norm": 3.459002946289681, "learning_rate": 3.328716119481192e-06, "loss": 0.1863, "step": 9119 }, { "epoch": 1.24, "grad_norm": 3.221025506927357, "learning_rate": 3.3276779781851486e-06, "loss": 0.1706, "step": 9120 }, { "epoch": 1.24, "grad_norm": 3.6109360406870734, "learning_rate": 3.3266399180563156e-06, "loss": 0.1922, "step": 9121 }, { "epoch": 1.24, "grad_norm": 3.6075755600619366, "learning_rate": 3.3256019391450696e-06, "loss": 0.1638, "step": 9122 }, { "epoch": 1.24, "grad_norm": 3.611336006261091, "learning_rate": 3.324564041501793e-06, "loss": 0.1391, "step": 9123 }, { "epoch": 1.24, "grad_norm": 3.6475875965660274, "learning_rate": 3.3235262251768607e-06, "loss": 0.1828, "step": 9124 }, { "epoch": 1.24, "grad_norm": 2.925652467869155, "learning_rate": 3.3224884902206435e-06, "loss": 0.1834, "step": 9125 }, { "epoch": 1.24, "grad_norm": 4.469242510643249, "learning_rate": 3.3214508366835067e-06, "loss": 0.2062, "step": 9126 }, { "epoch": 1.24, "grad_norm": 3.948329648504906, "learning_rate": 3.320413264615817e-06, "loss": 0.1758, "step": 9127 }, { "epoch": 1.24, "grad_norm": 3.9099307770991514, "learning_rate": 3.3193757740679307e-06, "loss": 0.1222, "step": 9128 }, { "epoch": 1.24, "grad_norm": 4.029622995450373, "learning_rate": 3.318338365090205e-06, "loss": 0.1788, "step": 9129 }, { "epoch": 1.24, "grad_norm": 3.234961405283894, "learning_rate": 3.3173010377329883e-06, "loss": 0.1759, "step": 9130 }, { "epoch": 1.24, "grad_norm": 3.114951870310546, "learning_rate": 3.316263792046631e-06, "loss": 0.1465, "step": 9131 }, { "epoch": 1.24, "grad_norm": 3.3702604154875373, "learning_rate": 3.315226628081475e-06, "loss": 0.1513, "step": 9132 }, { "epoch": 1.24, "grad_norm": 4.11963474481321, "learning_rate": 3.3141895458878613e-06, "loss": 0.2218, "step": 9133 }, { "epoch": 1.24, "grad_norm": 2.9675253103205588, "learning_rate": 3.3131525455161218e-06, "loss": 0.1731, "step": 9134 }, { "epoch": 1.24, "grad_norm": 3.1611231453250497, "learning_rate": 3.3121156270165926e-06, "loss": 0.1794, "step": 9135 }, { "epoch": 1.24, "grad_norm": 2.6571533148972306, "learning_rate": 3.311078790439598e-06, "loss": 0.131, "step": 9136 }, { "epoch": 1.24, "grad_norm": 3.4152428780235846, "learning_rate": 3.3100420358354614e-06, "loss": 0.1028, "step": 9137 }, { "epoch": 1.24, "grad_norm": 3.9055928906675006, "learning_rate": 3.309005363254505e-06, "loss": 0.2009, "step": 9138 }, { "epoch": 1.24, "grad_norm": 2.547745727929373, "learning_rate": 3.3079687727470413e-06, "loss": 0.1499, "step": 9139 }, { "epoch": 1.24, "grad_norm": 5.958868487526593, "learning_rate": 3.3069322643633845e-06, "loss": 0.2294, "step": 9140 }, { "epoch": 1.24, "grad_norm": 4.146090427211832, "learning_rate": 3.3058958381538393e-06, "loss": 0.1509, "step": 9141 }, { "epoch": 1.24, "grad_norm": 5.165134643333387, "learning_rate": 3.3048594941687117e-06, "loss": 0.1939, "step": 9142 }, { "epoch": 1.24, "grad_norm": 3.2980781501161314, "learning_rate": 3.3038232324583008e-06, "loss": 0.1713, "step": 9143 }, { "epoch": 1.24, "grad_norm": 2.9576358705161567, "learning_rate": 3.302787053072902e-06, "loss": 0.1897, "step": 9144 }, { "epoch": 1.24, "grad_norm": 3.653804938673319, "learning_rate": 3.3017509560628052e-06, "loss": 0.1948, "step": 9145 }, { "epoch": 1.24, "grad_norm": 3.482651750128789, "learning_rate": 3.3007149414783025e-06, "loss": 0.1925, "step": 9146 }, { "epoch": 1.24, "grad_norm": 3.0532197538308736, "learning_rate": 3.299679009369672e-06, "loss": 0.1573, "step": 9147 }, { "epoch": 1.24, "grad_norm": 4.082582366030529, "learning_rate": 3.2986431597871986e-06, "loss": 0.1654, "step": 9148 }, { "epoch": 1.24, "grad_norm": 3.137830220057635, "learning_rate": 3.297607392781154e-06, "loss": 0.1817, "step": 9149 }, { "epoch": 1.24, "grad_norm": 3.7808531471815305, "learning_rate": 3.296571708401811e-06, "loss": 0.1872, "step": 9150 }, { "epoch": 1.24, "grad_norm": 3.13196314461302, "learning_rate": 3.295536106699436e-06, "loss": 0.161, "step": 9151 }, { "epoch": 1.24, "grad_norm": 3.6650030584735482, "learning_rate": 3.2945005877242975e-06, "loss": 0.1737, "step": 9152 }, { "epoch": 1.25, "grad_norm": 4.015115089997471, "learning_rate": 3.2934651515266485e-06, "loss": 0.1807, "step": 9153 }, { "epoch": 1.25, "grad_norm": 3.5292002559334623, "learning_rate": 3.2924297981567497e-06, "loss": 0.1506, "step": 9154 }, { "epoch": 1.25, "grad_norm": 2.814737034627829, "learning_rate": 3.2913945276648485e-06, "loss": 0.1512, "step": 9155 }, { "epoch": 1.25, "grad_norm": 2.6993297732871193, "learning_rate": 3.2903593401011956e-06, "loss": 0.1258, "step": 9156 }, { "epoch": 1.25, "grad_norm": 4.159496860671807, "learning_rate": 3.2893242355160327e-06, "loss": 0.2115, "step": 9157 }, { "epoch": 1.25, "grad_norm": 3.10514940340221, "learning_rate": 3.288289213959601e-06, "loss": 0.136, "step": 9158 }, { "epoch": 1.25, "grad_norm": 3.761095412895185, "learning_rate": 3.2872542754821334e-06, "loss": 0.1598, "step": 9159 }, { "epoch": 1.25, "grad_norm": 3.810899919862945, "learning_rate": 3.2862194201338644e-06, "loss": 0.1294, "step": 9160 }, { "epoch": 1.25, "grad_norm": 4.168936015120684, "learning_rate": 3.2851846479650172e-06, "loss": 0.2031, "step": 9161 }, { "epoch": 1.25, "grad_norm": 3.9816110193281586, "learning_rate": 3.28414995902582e-06, "loss": 0.1586, "step": 9162 }, { "epoch": 1.25, "grad_norm": 4.02631594138965, "learning_rate": 3.283115353366488e-06, "loss": 0.1785, "step": 9163 }, { "epoch": 1.25, "grad_norm": 3.287084214999321, "learning_rate": 3.282080831037239e-06, "loss": 0.1971, "step": 9164 }, { "epoch": 1.25, "grad_norm": 3.341767852299069, "learning_rate": 3.2810463920882813e-06, "loss": 0.1471, "step": 9165 }, { "epoch": 1.25, "grad_norm": 3.9394421862431668, "learning_rate": 3.280012036569827e-06, "loss": 0.1937, "step": 9166 }, { "epoch": 1.25, "grad_norm": 3.219812663913443, "learning_rate": 3.2789777645320736e-06, "loss": 0.1869, "step": 9167 }, { "epoch": 1.25, "grad_norm": 3.232548225248116, "learning_rate": 3.2779435760252256e-06, "loss": 0.1725, "step": 9168 }, { "epoch": 1.25, "grad_norm": 4.047783385136013, "learning_rate": 3.276909471099472e-06, "loss": 0.1253, "step": 9169 }, { "epoch": 1.25, "grad_norm": 3.6540695065565565, "learning_rate": 3.2758754498050083e-06, "loss": 0.162, "step": 9170 }, { "epoch": 1.25, "grad_norm": 5.951007908033774, "learning_rate": 3.274841512192019e-06, "loss": 0.159, "step": 9171 }, { "epoch": 1.25, "grad_norm": 3.765619550231485, "learning_rate": 3.2738076583106903e-06, "loss": 0.1554, "step": 9172 }, { "epoch": 1.25, "grad_norm": 4.141759063436078, "learning_rate": 3.2727738882111958e-06, "loss": 0.1435, "step": 9173 }, { "epoch": 1.25, "grad_norm": 3.8861944655453944, "learning_rate": 3.2717402019437154e-06, "loss": 0.1551, "step": 9174 }, { "epoch": 1.25, "grad_norm": 4.037940809017984, "learning_rate": 3.2707065995584164e-06, "loss": 0.1375, "step": 9175 }, { "epoch": 1.25, "grad_norm": 3.2349843538164196, "learning_rate": 3.2696730811054644e-06, "loss": 0.1726, "step": 9176 }, { "epoch": 1.25, "grad_norm": 4.798283778328958, "learning_rate": 3.268639646635027e-06, "loss": 0.1907, "step": 9177 }, { "epoch": 1.25, "grad_norm": 4.444987194221048, "learning_rate": 3.267606296197256e-06, "loss": 0.11, "step": 9178 }, { "epoch": 1.25, "grad_norm": 3.9010098419609633, "learning_rate": 3.2665730298423117e-06, "loss": 0.1726, "step": 9179 }, { "epoch": 1.25, "grad_norm": 3.9989884772326754, "learning_rate": 3.265539847620339e-06, "loss": 0.1646, "step": 9180 }, { "epoch": 1.25, "grad_norm": 4.317896930405904, "learning_rate": 3.264506749581489e-06, "loss": 0.1614, "step": 9181 }, { "epoch": 1.25, "grad_norm": 4.462234315930826, "learning_rate": 3.2634737357758994e-06, "loss": 0.2222, "step": 9182 }, { "epoch": 1.25, "grad_norm": 3.222284965701954, "learning_rate": 3.2624408062537112e-06, "loss": 0.1626, "step": 9183 }, { "epoch": 1.25, "grad_norm": 3.207902907980184, "learning_rate": 3.261407961065056e-06, "loss": 0.1256, "step": 9184 }, { "epoch": 1.25, "grad_norm": 2.8987424267203203, "learning_rate": 3.2603752002600675e-06, "loss": 0.1151, "step": 9185 }, { "epoch": 1.25, "grad_norm": 3.638087845130162, "learning_rate": 3.2593425238888667e-06, "loss": 0.1808, "step": 9186 }, { "epoch": 1.25, "grad_norm": 3.4448451571547585, "learning_rate": 3.2583099320015787e-06, "loss": 0.1644, "step": 9187 }, { "epoch": 1.25, "grad_norm": 3.3283223870871717, "learning_rate": 3.2572774246483173e-06, "loss": 0.1706, "step": 9188 }, { "epoch": 1.25, "grad_norm": 3.1809872989157277, "learning_rate": 3.2562450018792002e-06, "loss": 0.1749, "step": 9189 }, { "epoch": 1.25, "grad_norm": 3.441225463806419, "learning_rate": 3.2552126637443332e-06, "loss": 0.1498, "step": 9190 }, { "epoch": 1.25, "grad_norm": 3.0971153498375825, "learning_rate": 3.254180410293825e-06, "loss": 0.1563, "step": 9191 }, { "epoch": 1.25, "grad_norm": 4.408645712988475, "learning_rate": 3.253148241577773e-06, "loss": 0.1658, "step": 9192 }, { "epoch": 1.25, "grad_norm": 3.551757388097542, "learning_rate": 3.2521161576462766e-06, "loss": 0.1881, "step": 9193 }, { "epoch": 1.25, "grad_norm": 3.5913541034256586, "learning_rate": 3.2510841585494267e-06, "loss": 0.1962, "step": 9194 }, { "epoch": 1.25, "grad_norm": 3.13946508401925, "learning_rate": 3.250052244337315e-06, "loss": 0.1845, "step": 9195 }, { "epoch": 1.25, "grad_norm": 3.788999539232461, "learning_rate": 3.2490204150600222e-06, "loss": 0.1611, "step": 9196 }, { "epoch": 1.25, "grad_norm": 3.5333665479686975, "learning_rate": 3.2479886707676323e-06, "loss": 0.1584, "step": 9197 }, { "epoch": 1.25, "grad_norm": 3.584581848727211, "learning_rate": 3.2469570115102188e-06, "loss": 0.1693, "step": 9198 }, { "epoch": 1.25, "grad_norm": 4.23858742938887, "learning_rate": 3.245925437337858e-06, "loss": 0.17, "step": 9199 }, { "epoch": 1.25, "grad_norm": 3.531499072063147, "learning_rate": 3.2448939483006125e-06, "loss": 0.1588, "step": 9200 }, { "epoch": 1.25, "grad_norm": 3.8456251708545586, "learning_rate": 3.2438625444485514e-06, "loss": 0.1933, "step": 9201 }, { "epoch": 1.25, "grad_norm": 3.296923787449984, "learning_rate": 3.2428312258317306e-06, "loss": 0.1924, "step": 9202 }, { "epoch": 1.25, "grad_norm": 3.6392389356322608, "learning_rate": 3.2417999925002085e-06, "loss": 0.1686, "step": 9203 }, { "epoch": 1.25, "grad_norm": 4.226552534235468, "learning_rate": 3.240768844504034e-06, "loss": 0.2001, "step": 9204 }, { "epoch": 1.25, "grad_norm": 3.4551333741300176, "learning_rate": 3.2397377818932583e-06, "loss": 0.164, "step": 9205 }, { "epoch": 1.25, "grad_norm": 3.2925089865005144, "learning_rate": 3.238706804717921e-06, "loss": 0.1816, "step": 9206 }, { "epoch": 1.25, "grad_norm": 2.990501311997187, "learning_rate": 3.2376759130280644e-06, "loss": 0.1204, "step": 9207 }, { "epoch": 1.25, "grad_norm": 4.0050426557669745, "learning_rate": 3.2366451068737204e-06, "loss": 0.2043, "step": 9208 }, { "epoch": 1.25, "grad_norm": 2.41808942014153, "learning_rate": 3.2356143863049216e-06, "loss": 0.0989, "step": 9209 }, { "epoch": 1.25, "grad_norm": 3.4579450132825937, "learning_rate": 3.234583751371695e-06, "loss": 0.1562, "step": 9210 }, { "epoch": 1.25, "grad_norm": 3.760654916348444, "learning_rate": 3.2335532021240612e-06, "loss": 0.1411, "step": 9211 }, { "epoch": 1.25, "grad_norm": 3.4963922012864974, "learning_rate": 3.23252273861204e-06, "loss": 0.1637, "step": 9212 }, { "epoch": 1.25, "grad_norm": 3.7037159294717723, "learning_rate": 3.2314923608856468e-06, "loss": 0.1841, "step": 9213 }, { "epoch": 1.25, "grad_norm": 3.1481924231564773, "learning_rate": 3.2304620689948883e-06, "loss": 0.0963, "step": 9214 }, { "epoch": 1.25, "grad_norm": 3.9389529997645862, "learning_rate": 3.2294318629897746e-06, "loss": 0.1266, "step": 9215 }, { "epoch": 1.25, "grad_norm": 3.92514979925095, "learning_rate": 3.228401742920304e-06, "loss": 0.1726, "step": 9216 }, { "epoch": 1.25, "grad_norm": 4.633053469611868, "learning_rate": 3.2273717088364743e-06, "loss": 0.1826, "step": 9217 }, { "epoch": 1.25, "grad_norm": 4.015304108842756, "learning_rate": 3.2263417607882817e-06, "loss": 0.1753, "step": 9218 }, { "epoch": 1.25, "grad_norm": 3.921113898825301, "learning_rate": 3.2253118988257115e-06, "loss": 0.1689, "step": 9219 }, { "epoch": 1.25, "grad_norm": 3.0710351865973236, "learning_rate": 3.224282122998752e-06, "loss": 0.1509, "step": 9220 }, { "epoch": 1.25, "grad_norm": 4.5209982328888225, "learning_rate": 3.2232524333573813e-06, "loss": 0.2112, "step": 9221 }, { "epoch": 1.25, "grad_norm": 2.707260921911958, "learning_rate": 3.222222829951578e-06, "loss": 0.1343, "step": 9222 }, { "epoch": 1.25, "grad_norm": 3.1282038317786967, "learning_rate": 3.221193312831313e-06, "loss": 0.1541, "step": 9223 }, { "epoch": 1.25, "grad_norm": 3.0830705973954533, "learning_rate": 3.2201638820465573e-06, "loss": 0.1668, "step": 9224 }, { "epoch": 1.25, "grad_norm": 4.342878735256577, "learning_rate": 3.2191345376472717e-06, "loss": 0.2098, "step": 9225 }, { "epoch": 1.25, "grad_norm": 4.360806977635948, "learning_rate": 3.2181052796834187e-06, "loss": 0.1357, "step": 9226 }, { "epoch": 1.26, "grad_norm": 3.5119395339767294, "learning_rate": 3.2170761082049504e-06, "loss": 0.1687, "step": 9227 }, { "epoch": 1.26, "grad_norm": 3.2693123647984605, "learning_rate": 3.2160470232618228e-06, "loss": 0.1496, "step": 9228 }, { "epoch": 1.26, "grad_norm": 2.9148219745865274, "learning_rate": 3.215018024903981e-06, "loss": 0.1808, "step": 9229 }, { "epoch": 1.26, "grad_norm": 3.694702120574861, "learning_rate": 3.213989113181368e-06, "loss": 0.1644, "step": 9230 }, { "epoch": 1.26, "grad_norm": 4.540832785790777, "learning_rate": 3.2129602881439214e-06, "loss": 0.1718, "step": 9231 }, { "epoch": 1.26, "grad_norm": 3.243496227887649, "learning_rate": 3.21193154984158e-06, "loss": 0.1372, "step": 9232 }, { "epoch": 1.26, "grad_norm": 3.403247396063677, "learning_rate": 3.210902898324269e-06, "loss": 0.1519, "step": 9233 }, { "epoch": 1.26, "grad_norm": 3.8734729496425673, "learning_rate": 3.2098743336419196e-06, "loss": 0.167, "step": 9234 }, { "epoch": 1.26, "grad_norm": 5.757059630307827, "learning_rate": 3.208845855844449e-06, "loss": 0.2405, "step": 9235 }, { "epoch": 1.26, "grad_norm": 2.7397369538164416, "learning_rate": 3.2078174649817796e-06, "loss": 0.1511, "step": 9236 }, { "epoch": 1.26, "grad_norm": 3.1390101142671627, "learning_rate": 3.2067891611038203e-06, "loss": 0.1554, "step": 9237 }, { "epoch": 1.26, "grad_norm": 2.7094199361979974, "learning_rate": 3.2057609442604864e-06, "loss": 0.1311, "step": 9238 }, { "epoch": 1.26, "grad_norm": 3.8478016521515084, "learning_rate": 3.204732814501677e-06, "loss": 0.1758, "step": 9239 }, { "epoch": 1.26, "grad_norm": 4.157915924342289, "learning_rate": 3.2037047718772983e-06, "loss": 0.1866, "step": 9240 }, { "epoch": 1.26, "grad_norm": 3.658883640101838, "learning_rate": 3.2026768164372422e-06, "loss": 0.1608, "step": 9241 }, { "epoch": 1.26, "grad_norm": 4.025372228923029, "learning_rate": 3.201648948231404e-06, "loss": 0.1908, "step": 9242 }, { "epoch": 1.26, "grad_norm": 3.13071326015748, "learning_rate": 3.200621167309672e-06, "loss": 0.1794, "step": 9243 }, { "epoch": 1.26, "grad_norm": 3.689594712909675, "learning_rate": 3.1995934737219293e-06, "loss": 0.1617, "step": 9244 }, { "epoch": 1.26, "grad_norm": 3.0747504898206266, "learning_rate": 3.1985658675180552e-06, "loss": 0.2048, "step": 9245 }, { "epoch": 1.26, "grad_norm": 3.2972073758557747, "learning_rate": 3.197538348747927e-06, "loss": 0.1231, "step": 9246 }, { "epoch": 1.26, "grad_norm": 3.9405461729738653, "learning_rate": 3.196510917461414e-06, "loss": 0.1087, "step": 9247 }, { "epoch": 1.26, "grad_norm": 3.510390182518684, "learning_rate": 3.1954835737083857e-06, "loss": 0.1831, "step": 9248 }, { "epoch": 1.26, "grad_norm": 4.488062460577727, "learning_rate": 3.194456317538701e-06, "loss": 0.2113, "step": 9249 }, { "epoch": 1.26, "grad_norm": 3.7046234213388356, "learning_rate": 3.193429149002222e-06, "loss": 0.1329, "step": 9250 }, { "epoch": 1.26, "grad_norm": 3.2458694741017924, "learning_rate": 3.1924020681488013e-06, "loss": 0.1219, "step": 9251 }, { "epoch": 1.26, "grad_norm": 4.049883215218876, "learning_rate": 3.191375075028291e-06, "loss": 0.1762, "step": 9252 }, { "epoch": 1.26, "grad_norm": 3.0808453611814057, "learning_rate": 3.1903481696905325e-06, "loss": 0.1674, "step": 9253 }, { "epoch": 1.26, "grad_norm": 4.517294609779739, "learning_rate": 3.189321352185373e-06, "loss": 0.1962, "step": 9254 }, { "epoch": 1.26, "grad_norm": 3.1328643049065947, "learning_rate": 3.1882946225626444e-06, "loss": 0.1128, "step": 9255 }, { "epoch": 1.26, "grad_norm": 3.3058750388128617, "learning_rate": 3.1872679808721816e-06, "loss": 0.1466, "step": 9256 }, { "epoch": 1.26, "grad_norm": 3.490280535743949, "learning_rate": 3.1862414271638163e-06, "loss": 0.2047, "step": 9257 }, { "epoch": 1.26, "grad_norm": 3.14712240971395, "learning_rate": 3.1852149614873683e-06, "loss": 0.146, "step": 9258 }, { "epoch": 1.26, "grad_norm": 4.119073746877832, "learning_rate": 3.184188583892661e-06, "loss": 0.1878, "step": 9259 }, { "epoch": 1.26, "grad_norm": 3.9067895326377924, "learning_rate": 3.1831622944295076e-06, "loss": 0.184, "step": 9260 }, { "epoch": 1.26, "grad_norm": 4.052414583481359, "learning_rate": 3.182136093147722e-06, "loss": 0.1726, "step": 9261 }, { "epoch": 1.26, "grad_norm": 4.516491397199536, "learning_rate": 3.181109980097111e-06, "loss": 0.1868, "step": 9262 }, { "epoch": 1.26, "grad_norm": 3.508021658941889, "learning_rate": 3.1800839553274775e-06, "loss": 0.155, "step": 9263 }, { "epoch": 1.26, "grad_norm": 3.496585158163662, "learning_rate": 3.179058018888619e-06, "loss": 0.1921, "step": 9264 }, { "epoch": 1.26, "grad_norm": 3.27343226491705, "learning_rate": 3.1780321708303323e-06, "loss": 0.1363, "step": 9265 }, { "epoch": 1.26, "grad_norm": 3.348547319553247, "learning_rate": 3.177006411202405e-06, "loss": 0.1789, "step": 9266 }, { "epoch": 1.26, "grad_norm": 3.8295058622267053, "learning_rate": 3.1759807400546266e-06, "loss": 0.2115, "step": 9267 }, { "epoch": 1.26, "grad_norm": 3.9097837991428475, "learning_rate": 3.174955157436774e-06, "loss": 0.1689, "step": 9268 }, { "epoch": 1.26, "grad_norm": 3.530166814235782, "learning_rate": 3.1739296633986285e-06, "loss": 0.1403, "step": 9269 }, { "epoch": 1.26, "grad_norm": 3.1624155591133647, "learning_rate": 3.1729042579899607e-06, "loss": 0.183, "step": 9270 }, { "epoch": 1.26, "grad_norm": 4.335891272778721, "learning_rate": 3.1718789412605412e-06, "loss": 0.152, "step": 9271 }, { "epoch": 1.26, "grad_norm": 6.932156348808156, "learning_rate": 3.1708537132601324e-06, "loss": 0.1596, "step": 9272 }, { "epoch": 1.26, "grad_norm": 4.153891388115196, "learning_rate": 3.1698285740384966e-06, "loss": 0.1669, "step": 9273 }, { "epoch": 1.26, "grad_norm": 2.7799940753553942, "learning_rate": 3.168803523645387e-06, "loss": 0.1712, "step": 9274 }, { "epoch": 1.26, "grad_norm": 3.3321475244906984, "learning_rate": 3.167778562130557e-06, "loss": 0.1841, "step": 9275 }, { "epoch": 1.26, "grad_norm": 5.462626697635277, "learning_rate": 3.1667536895437523e-06, "loss": 0.1614, "step": 9276 }, { "epoch": 1.26, "grad_norm": 3.2606326901444773, "learning_rate": 3.1657289059347184e-06, "loss": 0.1186, "step": 9277 }, { "epoch": 1.26, "grad_norm": 3.618350973810399, "learning_rate": 3.16470421135319e-06, "loss": 0.1413, "step": 9278 }, { "epoch": 1.26, "grad_norm": 4.948907504208524, "learning_rate": 3.1636796058489056e-06, "loss": 0.1954, "step": 9279 }, { "epoch": 1.26, "grad_norm": 3.002729889026598, "learning_rate": 3.1626550894715905e-06, "loss": 0.1412, "step": 9280 }, { "epoch": 1.26, "grad_norm": 3.7376438484259116, "learning_rate": 3.1616306622709747e-06, "loss": 0.179, "step": 9281 }, { "epoch": 1.26, "grad_norm": 4.974095281860676, "learning_rate": 3.1606063242967753e-06, "loss": 0.242, "step": 9282 }, { "epoch": 1.26, "grad_norm": 2.803404542406551, "learning_rate": 3.1595820755987126e-06, "loss": 0.1123, "step": 9283 }, { "epoch": 1.26, "grad_norm": 3.943145711052189, "learning_rate": 3.158557916226496e-06, "loss": 0.1637, "step": 9284 }, { "epoch": 1.26, "grad_norm": 3.296465388543447, "learning_rate": 3.1575338462298377e-06, "loss": 0.1479, "step": 9285 }, { "epoch": 1.26, "grad_norm": 2.9723157703392085, "learning_rate": 3.1565098656584375e-06, "loss": 0.1836, "step": 9286 }, { "epoch": 1.26, "grad_norm": 3.748742603087691, "learning_rate": 3.1554859745619986e-06, "loss": 0.1689, "step": 9287 }, { "epoch": 1.26, "grad_norm": 3.96386555611886, "learning_rate": 3.1544621729902125e-06, "loss": 0.2033, "step": 9288 }, { "epoch": 1.26, "grad_norm": 4.865785496314325, "learning_rate": 3.1534384609927725e-06, "loss": 0.1465, "step": 9289 }, { "epoch": 1.26, "grad_norm": 3.896362084690559, "learning_rate": 3.1524148386193644e-06, "loss": 0.1483, "step": 9290 }, { "epoch": 1.26, "grad_norm": 3.1309968321013923, "learning_rate": 3.151391305919672e-06, "loss": 0.1245, "step": 9291 }, { "epoch": 1.26, "grad_norm": 3.5564630991068764, "learning_rate": 3.15036786294337e-06, "loss": 0.1914, "step": 9292 }, { "epoch": 1.26, "grad_norm": 4.3943205214975265, "learning_rate": 3.149344509740135e-06, "loss": 0.1744, "step": 9293 }, { "epoch": 1.26, "grad_norm": 2.8919611069497537, "learning_rate": 3.1483212463596326e-06, "loss": 0.1762, "step": 9294 }, { "epoch": 1.26, "grad_norm": 3.264367519175987, "learning_rate": 3.14729807285153e-06, "loss": 0.1839, "step": 9295 }, { "epoch": 1.26, "grad_norm": 3.585138535569164, "learning_rate": 3.1462749892654894e-06, "loss": 0.1616, "step": 9296 }, { "epoch": 1.26, "grad_norm": 3.172547338270243, "learning_rate": 3.145251995651162e-06, "loss": 0.122, "step": 9297 }, { "epoch": 1.26, "grad_norm": 4.711086278868668, "learning_rate": 3.144229092058203e-06, "loss": 0.1754, "step": 9298 }, { "epoch": 1.26, "grad_norm": 4.436661913481897, "learning_rate": 3.1432062785362573e-06, "loss": 0.1621, "step": 9299 }, { "epoch": 1.27, "grad_norm": 4.072339701034021, "learning_rate": 3.1421835551349712e-06, "loss": 0.1567, "step": 9300 }, { "epoch": 1.27, "grad_norm": 4.307694832911804, "learning_rate": 3.1411609219039787e-06, "loss": 0.1936, "step": 9301 }, { "epoch": 1.27, "grad_norm": 4.374475308546362, "learning_rate": 3.1401383788929175e-06, "loss": 0.1311, "step": 9302 }, { "epoch": 1.27, "grad_norm": 6.4413819418360605, "learning_rate": 3.1391159261514153e-06, "loss": 0.1799, "step": 9303 }, { "epoch": 1.27, "grad_norm": 3.861791285601437, "learning_rate": 3.138093563729101e-06, "loss": 0.1403, "step": 9304 }, { "epoch": 1.27, "grad_norm": 4.114349315732891, "learning_rate": 3.13707129167559e-06, "loss": 0.148, "step": 9305 }, { "epoch": 1.27, "grad_norm": 2.9338448812505455, "learning_rate": 3.1360491100405045e-06, "loss": 0.1523, "step": 9306 }, { "epoch": 1.27, "grad_norm": 3.361525606704714, "learning_rate": 3.1350270188734523e-06, "loss": 0.1425, "step": 9307 }, { "epoch": 1.27, "grad_norm": 4.0632970791578265, "learning_rate": 3.1340050182240438e-06, "loss": 0.18, "step": 9308 }, { "epoch": 1.27, "grad_norm": 3.5548333921437965, "learning_rate": 3.132983108141881e-06, "loss": 0.1608, "step": 9309 }, { "epoch": 1.27, "grad_norm": 3.2713909131714862, "learning_rate": 3.131961288676566e-06, "loss": 0.1413, "step": 9310 }, { "epoch": 1.27, "grad_norm": 3.236771327676531, "learning_rate": 3.1309395598776883e-06, "loss": 0.1391, "step": 9311 }, { "epoch": 1.27, "grad_norm": 4.078101827916225, "learning_rate": 3.129917921794844e-06, "loss": 0.1667, "step": 9312 }, { "epoch": 1.27, "grad_norm": 3.309940242040016, "learning_rate": 3.1288963744776135e-06, "loss": 0.165, "step": 9313 }, { "epoch": 1.27, "grad_norm": 4.966879063756335, "learning_rate": 3.127874917975583e-06, "loss": 0.1745, "step": 9314 }, { "epoch": 1.27, "grad_norm": 3.4257513500107817, "learning_rate": 3.1268535523383246e-06, "loss": 0.149, "step": 9315 }, { "epoch": 1.27, "grad_norm": 3.4561247711593657, "learning_rate": 3.1258322776154155e-06, "loss": 0.1464, "step": 9316 }, { "epoch": 1.27, "grad_norm": 3.998754654842717, "learning_rate": 3.1248110938564202e-06, "loss": 0.1382, "step": 9317 }, { "epoch": 1.27, "grad_norm": 4.474456033466975, "learning_rate": 3.1237900011109068e-06, "loss": 0.1937, "step": 9318 }, { "epoch": 1.27, "grad_norm": 3.0353301530024503, "learning_rate": 3.1227689994284304e-06, "loss": 0.1824, "step": 9319 }, { "epoch": 1.27, "grad_norm": 2.852731147593922, "learning_rate": 3.1217480888585493e-06, "loss": 0.1295, "step": 9320 }, { "epoch": 1.27, "grad_norm": 4.017406513866336, "learning_rate": 3.1207272694508106e-06, "loss": 0.1588, "step": 9321 }, { "epoch": 1.27, "grad_norm": 3.6088172818027267, "learning_rate": 3.1197065412547632e-06, "loss": 0.1746, "step": 9322 }, { "epoch": 1.27, "grad_norm": 3.00467831344273, "learning_rate": 3.1186859043199476e-06, "loss": 0.1796, "step": 9323 }, { "epoch": 1.27, "grad_norm": 2.8014658665990724, "learning_rate": 3.1176653586959025e-06, "loss": 0.1479, "step": 9324 }, { "epoch": 1.27, "grad_norm": 3.597854529819343, "learning_rate": 3.1166449044321577e-06, "loss": 0.1781, "step": 9325 }, { "epoch": 1.27, "grad_norm": 3.4871092511569315, "learning_rate": 3.115624541578246e-06, "loss": 0.2077, "step": 9326 }, { "epoch": 1.27, "grad_norm": 3.1734449067684816, "learning_rate": 3.1146042701836865e-06, "loss": 0.1235, "step": 9327 }, { "epoch": 1.27, "grad_norm": 4.357067151354012, "learning_rate": 3.113584090298002e-06, "loss": 0.1781, "step": 9328 }, { "epoch": 1.27, "grad_norm": 3.6037061574817177, "learning_rate": 3.112564001970707e-06, "loss": 0.1675, "step": 9329 }, { "epoch": 1.27, "grad_norm": 3.5517224065569644, "learning_rate": 3.1115440052513113e-06, "loss": 0.1862, "step": 9330 }, { "epoch": 1.27, "grad_norm": 4.108837130051578, "learning_rate": 3.1105241001893206e-06, "loss": 0.1816, "step": 9331 }, { "epoch": 1.27, "grad_norm": 3.5364993917412484, "learning_rate": 3.10950428683424e-06, "loss": 0.1793, "step": 9332 }, { "epoch": 1.27, "grad_norm": 4.359619920200568, "learning_rate": 3.108484565235562e-06, "loss": 0.1347, "step": 9333 }, { "epoch": 1.27, "grad_norm": 3.4786995452634426, "learning_rate": 3.1074649354427842e-06, "loss": 0.153, "step": 9334 }, { "epoch": 1.27, "grad_norm": 3.4750711116489352, "learning_rate": 3.106445397505391e-06, "loss": 0.1906, "step": 9335 }, { "epoch": 1.27, "grad_norm": 3.9664760711384646, "learning_rate": 3.105425951472867e-06, "loss": 0.2167, "step": 9336 }, { "epoch": 1.27, "grad_norm": 2.678670017540757, "learning_rate": 3.1044065973946945e-06, "loss": 0.1461, "step": 9337 }, { "epoch": 1.27, "grad_norm": 4.62214037387823, "learning_rate": 3.1033873353203454e-06, "loss": 0.1517, "step": 9338 }, { "epoch": 1.27, "grad_norm": 3.9982116010064646, "learning_rate": 3.1023681652992925e-06, "loss": 0.1911, "step": 9339 }, { "epoch": 1.27, "grad_norm": 3.7687613980298584, "learning_rate": 3.1013490873809994e-06, "loss": 0.177, "step": 9340 }, { "epoch": 1.27, "grad_norm": 3.302080161653636, "learning_rate": 3.100330101614929e-06, "loss": 0.1959, "step": 9341 }, { "epoch": 1.27, "grad_norm": 3.069838935946812, "learning_rate": 3.0993112080505383e-06, "loss": 0.1839, "step": 9342 }, { "epoch": 1.27, "grad_norm": 3.544561553045304, "learning_rate": 3.0982924067372823e-06, "loss": 0.1659, "step": 9343 }, { "epoch": 1.27, "grad_norm": 3.266109046291048, "learning_rate": 3.0972736977246042e-06, "loss": 0.1572, "step": 9344 }, { "epoch": 1.27, "grad_norm": 4.073442245394063, "learning_rate": 3.0962550810619527e-06, "loss": 0.1683, "step": 9345 }, { "epoch": 1.27, "grad_norm": 3.778186457655172, "learning_rate": 3.095236556798763e-06, "loss": 0.1633, "step": 9346 }, { "epoch": 1.27, "grad_norm": 3.93846967832282, "learning_rate": 3.0942181249844726e-06, "loss": 0.1698, "step": 9347 }, { "epoch": 1.27, "grad_norm": 4.169258968452768, "learning_rate": 3.0931997856685104e-06, "loss": 0.1736, "step": 9348 }, { "epoch": 1.27, "grad_norm": 3.04211040125588, "learning_rate": 3.092181538900303e-06, "loss": 0.1511, "step": 9349 }, { "epoch": 1.27, "grad_norm": 3.0431641914573335, "learning_rate": 3.091163384729269e-06, "loss": 0.1774, "step": 9350 }, { "epoch": 1.27, "grad_norm": 3.1797801795935086, "learning_rate": 3.09014532320483e-06, "loss": 0.1539, "step": 9351 }, { "epoch": 1.27, "grad_norm": 3.43988262798851, "learning_rate": 3.089127354376393e-06, "loss": 0.1718, "step": 9352 }, { "epoch": 1.27, "grad_norm": 4.919299561616369, "learning_rate": 3.088109478293371e-06, "loss": 0.1624, "step": 9353 }, { "epoch": 1.27, "grad_norm": 4.895028029216088, "learning_rate": 3.0870916950051623e-06, "loss": 0.1663, "step": 9354 }, { "epoch": 1.27, "grad_norm": 3.9660494563625, "learning_rate": 3.086074004561169e-06, "loss": 0.1575, "step": 9355 }, { "epoch": 1.27, "grad_norm": 3.35703523696713, "learning_rate": 3.085056407010783e-06, "loss": 0.1382, "step": 9356 }, { "epoch": 1.27, "grad_norm": 4.820664538363801, "learning_rate": 3.084038902403398e-06, "loss": 0.1953, "step": 9357 }, { "epoch": 1.27, "grad_norm": 3.8813734992346327, "learning_rate": 3.083021490788393e-06, "loss": 0.1557, "step": 9358 }, { "epoch": 1.27, "grad_norm": 3.5590958198563065, "learning_rate": 3.082004172215155e-06, "loss": 0.1257, "step": 9359 }, { "epoch": 1.27, "grad_norm": 3.5665017160592725, "learning_rate": 3.0809869467330554e-06, "loss": 0.1749, "step": 9360 }, { "epoch": 1.27, "grad_norm": 3.7192677933968987, "learning_rate": 3.079969814391469e-06, "loss": 0.1182, "step": 9361 }, { "epoch": 1.27, "grad_norm": 5.863358753366256, "learning_rate": 3.0789527752397624e-06, "loss": 0.184, "step": 9362 }, { "epoch": 1.27, "grad_norm": 3.6712828097797967, "learning_rate": 3.077935829327297e-06, "loss": 0.1848, "step": 9363 }, { "epoch": 1.27, "grad_norm": 3.752296409903482, "learning_rate": 3.076918976703431e-06, "loss": 0.1989, "step": 9364 }, { "epoch": 1.27, "grad_norm": 4.370754967458986, "learning_rate": 3.0759022174175203e-06, "loss": 0.1873, "step": 9365 }, { "epoch": 1.27, "grad_norm": 3.6793815428113206, "learning_rate": 3.0748855515189104e-06, "loss": 0.1732, "step": 9366 }, { "epoch": 1.27, "grad_norm": 3.4148344945027946, "learning_rate": 3.07386897905695e-06, "loss": 0.1498, "step": 9367 }, { "epoch": 1.27, "grad_norm": 3.3241954696317015, "learning_rate": 3.0728525000809737e-06, "loss": 0.1842, "step": 9368 }, { "epoch": 1.27, "grad_norm": 3.532524950611294, "learning_rate": 3.071836114640322e-06, "loss": 0.1739, "step": 9369 }, { "epoch": 1.27, "grad_norm": 2.7193348837438083, "learning_rate": 3.070819822784323e-06, "loss": 0.1622, "step": 9370 }, { "epoch": 1.27, "grad_norm": 3.7769656647042464, "learning_rate": 3.069803624562305e-06, "loss": 0.1863, "step": 9371 }, { "epoch": 1.27, "grad_norm": 3.7586720881389235, "learning_rate": 3.068787520023587e-06, "loss": 0.1906, "step": 9372 }, { "epoch": 1.27, "grad_norm": 3.987227785633106, "learning_rate": 3.067771509217491e-06, "loss": 0.1428, "step": 9373 }, { "epoch": 1.28, "grad_norm": 3.355446451821208, "learning_rate": 3.0667555921933233e-06, "loss": 0.1596, "step": 9374 }, { "epoch": 1.28, "grad_norm": 3.7192612298055665, "learning_rate": 3.065739769000397e-06, "loss": 0.1477, "step": 9375 }, { "epoch": 1.28, "grad_norm": 3.405103081550875, "learning_rate": 3.064724039688015e-06, "loss": 0.1495, "step": 9376 }, { "epoch": 1.28, "grad_norm": 4.125364322329986, "learning_rate": 3.0637084043054744e-06, "loss": 0.1735, "step": 9377 }, { "epoch": 1.28, "grad_norm": 3.383167211717575, "learning_rate": 3.0626928629020724e-06, "loss": 0.1584, "step": 9378 }, { "epoch": 1.28, "grad_norm": 3.821919221791101, "learning_rate": 3.0616774155270957e-06, "loss": 0.1354, "step": 9379 }, { "epoch": 1.28, "grad_norm": 3.7691858783299566, "learning_rate": 3.0606620622298322e-06, "loss": 0.2109, "step": 9380 }, { "epoch": 1.28, "grad_norm": 3.1951742083327583, "learning_rate": 3.0596468030595616e-06, "loss": 0.1523, "step": 9381 }, { "epoch": 1.28, "grad_norm": 3.358071130936131, "learning_rate": 3.058631638065561e-06, "loss": 0.1741, "step": 9382 }, { "epoch": 1.28, "grad_norm": 3.2779833647642946, "learning_rate": 3.0576165672971004e-06, "loss": 0.2171, "step": 9383 }, { "epoch": 1.28, "grad_norm": 3.037069393174305, "learning_rate": 3.05660159080345e-06, "loss": 0.1267, "step": 9384 }, { "epoch": 1.28, "grad_norm": 3.752469266917824, "learning_rate": 3.0555867086338674e-06, "loss": 0.206, "step": 9385 }, { "epoch": 1.28, "grad_norm": 3.408554946615002, "learning_rate": 3.0545719208376166e-06, "loss": 0.1041, "step": 9386 }, { "epoch": 1.28, "grad_norm": 3.5061929900678654, "learning_rate": 3.0535572274639456e-06, "loss": 0.197, "step": 9387 }, { "epoch": 1.28, "grad_norm": 7.296508498104991, "learning_rate": 3.052542628562107e-06, "loss": 0.1841, "step": 9388 }, { "epoch": 1.28, "grad_norm": 3.3670095171570336, "learning_rate": 3.051528124181342e-06, "loss": 0.1383, "step": 9389 }, { "epoch": 1.28, "grad_norm": 4.195113874201723, "learning_rate": 3.050513714370894e-06, "loss": 0.169, "step": 9390 }, { "epoch": 1.28, "grad_norm": 4.2357148557445745, "learning_rate": 3.049499399179994e-06, "loss": 0.1313, "step": 9391 }, { "epoch": 1.28, "grad_norm": 3.565215523001136, "learning_rate": 3.048485178657875e-06, "loss": 0.1527, "step": 9392 }, { "epoch": 1.28, "grad_norm": 3.429904866282845, "learning_rate": 3.047471052853761e-06, "loss": 0.1824, "step": 9393 }, { "epoch": 1.28, "grad_norm": 3.825188533934562, "learning_rate": 3.046457021816874e-06, "loss": 0.1933, "step": 9394 }, { "epoch": 1.28, "grad_norm": 4.281616074027829, "learning_rate": 3.045443085596431e-06, "loss": 0.1968, "step": 9395 }, { "epoch": 1.28, "grad_norm": 3.0691142784276293, "learning_rate": 3.0444292442416456e-06, "loss": 0.1136, "step": 9396 }, { "epoch": 1.28, "grad_norm": 3.6785807677855464, "learning_rate": 3.0434154978017215e-06, "loss": 0.1604, "step": 9397 }, { "epoch": 1.28, "grad_norm": 3.4217737360737766, "learning_rate": 3.042401846325866e-06, "loss": 0.2263, "step": 9398 }, { "epoch": 1.28, "grad_norm": 3.3921039623130333, "learning_rate": 3.041388289863272e-06, "loss": 0.1268, "step": 9399 }, { "epoch": 1.28, "grad_norm": 3.491290349678283, "learning_rate": 3.0403748284631384e-06, "loss": 0.1642, "step": 9400 }, { "epoch": 1.28, "grad_norm": 2.6252485738108198, "learning_rate": 3.03936146217465e-06, "loss": 0.1391, "step": 9401 }, { "epoch": 1.28, "grad_norm": 3.8919808456529963, "learning_rate": 3.0383481910469936e-06, "loss": 0.1309, "step": 9402 }, { "epoch": 1.28, "grad_norm": 4.741088957071582, "learning_rate": 3.0373350151293468e-06, "loss": 0.171, "step": 9403 }, { "epoch": 1.28, "grad_norm": 3.6561564221404685, "learning_rate": 3.0363219344708887e-06, "loss": 0.2023, "step": 9404 }, { "epoch": 1.28, "grad_norm": 4.076109671322208, "learning_rate": 3.0353089491207853e-06, "loss": 0.162, "step": 9405 }, { "epoch": 1.28, "grad_norm": 3.8227174084315956, "learning_rate": 3.0342960591282066e-06, "loss": 0.1577, "step": 9406 }, { "epoch": 1.28, "grad_norm": 4.149816644496868, "learning_rate": 3.03328326454231e-06, "loss": 0.1935, "step": 9407 }, { "epoch": 1.28, "grad_norm": 3.2570848487926725, "learning_rate": 3.0322705654122543e-06, "loss": 0.1748, "step": 9408 }, { "epoch": 1.28, "grad_norm": 3.0273930952783736, "learning_rate": 3.031257961787191e-06, "loss": 0.1174, "step": 9409 }, { "epoch": 1.28, "grad_norm": 3.8032689969207283, "learning_rate": 3.0302454537162683e-06, "loss": 0.1678, "step": 9410 }, { "epoch": 1.28, "grad_norm": 3.738864712493819, "learning_rate": 3.029233041248627e-06, "loss": 0.1855, "step": 9411 }, { "epoch": 1.28, "grad_norm": 3.490262411673859, "learning_rate": 3.0282207244334084e-06, "loss": 0.1658, "step": 9412 }, { "epoch": 1.28, "grad_norm": 4.171239346344252, "learning_rate": 3.0272085033197414e-06, "loss": 0.1675, "step": 9413 }, { "epoch": 1.28, "grad_norm": 3.4483172584201927, "learning_rate": 3.0261963779567595e-06, "loss": 0.1577, "step": 9414 }, { "epoch": 1.28, "grad_norm": 4.050647827481084, "learning_rate": 3.0251843483935834e-06, "loss": 0.1506, "step": 9415 }, { "epoch": 1.28, "grad_norm": 3.9445914808659186, "learning_rate": 3.0241724146793326e-06, "loss": 0.1724, "step": 9416 }, { "epoch": 1.28, "grad_norm": 2.9810420696431517, "learning_rate": 3.0231605768631256e-06, "loss": 0.1551, "step": 9417 }, { "epoch": 1.28, "grad_norm": 3.4258516561078327, "learning_rate": 3.0221488349940683e-06, "loss": 0.1505, "step": 9418 }, { "epoch": 1.28, "grad_norm": 3.464052480157145, "learning_rate": 3.021137189121269e-06, "loss": 0.1587, "step": 9419 }, { "epoch": 1.28, "grad_norm": 2.570107675684341, "learning_rate": 3.0201256392938273e-06, "loss": 0.1339, "step": 9420 }, { "epoch": 1.28, "grad_norm": 4.6348203881864825, "learning_rate": 3.0191141855608397e-06, "loss": 0.1784, "step": 9421 }, { "epoch": 1.28, "grad_norm": 4.428932143735859, "learning_rate": 3.018102827971397e-06, "loss": 0.1923, "step": 9422 }, { "epoch": 1.28, "grad_norm": 4.176846301719107, "learning_rate": 3.0170915665745894e-06, "loss": 0.1844, "step": 9423 }, { "epoch": 1.28, "grad_norm": 2.945896446448156, "learning_rate": 3.0160804014194932e-06, "loss": 0.1248, "step": 9424 }, { "epoch": 1.28, "grad_norm": 3.863522288922549, "learning_rate": 3.0150693325551927e-06, "loss": 0.1995, "step": 9425 }, { "epoch": 1.28, "grad_norm": 3.559266068516553, "learning_rate": 3.0140583600307547e-06, "loss": 0.1733, "step": 9426 }, { "epoch": 1.28, "grad_norm": 3.9899403804827305, "learning_rate": 3.0130474838952518e-06, "loss": 0.1662, "step": 9427 }, { "epoch": 1.28, "grad_norm": 2.9288635303486403, "learning_rate": 3.012036704197744e-06, "loss": 0.1467, "step": 9428 }, { "epoch": 1.28, "grad_norm": 4.050938902928182, "learning_rate": 3.0110260209872944e-06, "loss": 0.1582, "step": 9429 }, { "epoch": 1.28, "grad_norm": 4.012116734455239, "learning_rate": 3.010015434312952e-06, "loss": 0.2011, "step": 9430 }, { "epoch": 1.28, "grad_norm": 3.6014944076166917, "learning_rate": 3.0090049442237713e-06, "loss": 0.1426, "step": 9431 }, { "epoch": 1.28, "grad_norm": 4.424023911682331, "learning_rate": 3.007994550768793e-06, "loss": 0.1683, "step": 9432 }, { "epoch": 1.28, "grad_norm": 3.9936630524781083, "learning_rate": 3.006984253997061e-06, "loss": 0.1825, "step": 9433 }, { "epoch": 1.28, "grad_norm": 5.022583046183188, "learning_rate": 3.0059740539576066e-06, "loss": 0.1494, "step": 9434 }, { "epoch": 1.28, "grad_norm": 5.156294278209015, "learning_rate": 3.004963950699463e-06, "loss": 0.1778, "step": 9435 }, { "epoch": 1.28, "grad_norm": 3.459557449581218, "learning_rate": 3.0039539442716547e-06, "loss": 0.1698, "step": 9436 }, { "epoch": 1.28, "grad_norm": 3.4226569717703264, "learning_rate": 3.0029440347232064e-06, "loss": 0.164, "step": 9437 }, { "epoch": 1.28, "grad_norm": 3.3494531442187494, "learning_rate": 3.00193422210313e-06, "loss": 0.1607, "step": 9438 }, { "epoch": 1.28, "grad_norm": 3.651830518115404, "learning_rate": 3.000924506460442e-06, "loss": 0.1925, "step": 9439 }, { "epoch": 1.28, "grad_norm": 4.355040377289473, "learning_rate": 2.999914887844145e-06, "loss": 0.1801, "step": 9440 }, { "epoch": 1.28, "grad_norm": 3.2851906151965156, "learning_rate": 2.9989053663032454e-06, "loss": 0.1659, "step": 9441 }, { "epoch": 1.28, "grad_norm": 2.6026832002190132, "learning_rate": 2.997895941886737e-06, "loss": 0.1765, "step": 9442 }, { "epoch": 1.28, "grad_norm": 3.764021420830708, "learning_rate": 2.996886614643618e-06, "loss": 0.1532, "step": 9443 }, { "epoch": 1.28, "grad_norm": 3.4247103039852, "learning_rate": 2.9958773846228717e-06, "loss": 0.1919, "step": 9444 }, { "epoch": 1.28, "grad_norm": 2.9733392050747045, "learning_rate": 2.9948682518734862e-06, "loss": 0.1525, "step": 9445 }, { "epoch": 1.28, "grad_norm": 3.79929761164534, "learning_rate": 2.993859216444436e-06, "loss": 0.1948, "step": 9446 }, { "epoch": 1.29, "grad_norm": 3.9489229620980266, "learning_rate": 2.9928502783846987e-06, "loss": 0.1971, "step": 9447 }, { "epoch": 1.29, "grad_norm": 3.781073984563136, "learning_rate": 2.9918414377432424e-06, "loss": 0.2044, "step": 9448 }, { "epoch": 1.29, "grad_norm": 4.00161714125435, "learning_rate": 2.9908326945690313e-06, "loss": 0.1529, "step": 9449 }, { "epoch": 1.29, "grad_norm": 3.8324578138596044, "learning_rate": 2.989824048911025e-06, "loss": 0.1924, "step": 9450 }, { "epoch": 1.29, "grad_norm": 2.824468846027889, "learning_rate": 2.9888155008181826e-06, "loss": 0.154, "step": 9451 }, { "epoch": 1.29, "grad_norm": 3.758312378527739, "learning_rate": 2.9878070503394484e-06, "loss": 0.15, "step": 9452 }, { "epoch": 1.29, "grad_norm": 3.298400214708752, "learning_rate": 2.9867986975237744e-06, "loss": 0.1546, "step": 9453 }, { "epoch": 1.29, "grad_norm": 3.8593305548425096, "learning_rate": 2.9857904424200967e-06, "loss": 0.1385, "step": 9454 }, { "epoch": 1.29, "grad_norm": 4.250386607495514, "learning_rate": 2.9847822850773546e-06, "loss": 0.1544, "step": 9455 }, { "epoch": 1.29, "grad_norm": 2.979693821973631, "learning_rate": 2.9837742255444793e-06, "loss": 0.1813, "step": 9456 }, { "epoch": 1.29, "grad_norm": 4.255634744441085, "learning_rate": 2.982766263870395e-06, "loss": 0.2198, "step": 9457 }, { "epoch": 1.29, "grad_norm": 3.8126565631194436, "learning_rate": 2.981758400104028e-06, "loss": 0.1491, "step": 9458 }, { "epoch": 1.29, "grad_norm": 3.070097405386647, "learning_rate": 2.9807506342942916e-06, "loss": 0.1665, "step": 9459 }, { "epoch": 1.29, "grad_norm": 4.314489788428944, "learning_rate": 2.979742966490101e-06, "loss": 0.1761, "step": 9460 }, { "epoch": 1.29, "grad_norm": 2.455138892518209, "learning_rate": 2.978735396740362e-06, "loss": 0.1437, "step": 9461 }, { "epoch": 1.29, "grad_norm": 3.1720083739257694, "learning_rate": 2.977727925093981e-06, "loss": 0.1659, "step": 9462 }, { "epoch": 1.29, "grad_norm": 3.0799184246082225, "learning_rate": 2.976720551599852e-06, "loss": 0.1742, "step": 9463 }, { "epoch": 1.29, "grad_norm": 3.3405116910801573, "learning_rate": 2.975713276306872e-06, "loss": 0.206, "step": 9464 }, { "epoch": 1.29, "grad_norm": 5.170626028444467, "learning_rate": 2.9747060992639264e-06, "loss": 0.139, "step": 9465 }, { "epoch": 1.29, "grad_norm": 3.9948266456034602, "learning_rate": 2.9736990205199023e-06, "loss": 0.1548, "step": 9466 }, { "epoch": 1.29, "grad_norm": 3.062504941867941, "learning_rate": 2.972692040123678e-06, "loss": 0.1692, "step": 9467 }, { "epoch": 1.29, "grad_norm": 4.013049007025146, "learning_rate": 2.9716851581241268e-06, "loss": 0.1779, "step": 9468 }, { "epoch": 1.29, "grad_norm": 3.4157555525215963, "learning_rate": 2.970678374570119e-06, "loss": 0.1451, "step": 9469 }, { "epoch": 1.29, "grad_norm": 4.032763124782257, "learning_rate": 2.9696716895105214e-06, "loss": 0.148, "step": 9470 }, { "epoch": 1.29, "grad_norm": 3.5374874705800474, "learning_rate": 2.96866510299419e-06, "loss": 0.1596, "step": 9471 }, { "epoch": 1.29, "grad_norm": 3.9289438488331205, "learning_rate": 2.9676586150699843e-06, "loss": 0.1371, "step": 9472 }, { "epoch": 1.29, "grad_norm": 3.0812590033467995, "learning_rate": 2.9666522257867514e-06, "loss": 0.1352, "step": 9473 }, { "epoch": 1.29, "grad_norm": 3.903563864443855, "learning_rate": 2.9656459351933394e-06, "loss": 0.1554, "step": 9474 }, { "epoch": 1.29, "grad_norm": 4.295562874524374, "learning_rate": 2.9646397433385873e-06, "loss": 0.2158, "step": 9475 }, { "epoch": 1.29, "grad_norm": 4.407890496309297, "learning_rate": 2.963633650271334e-06, "loss": 0.1939, "step": 9476 }, { "epoch": 1.29, "grad_norm": 3.7931858715943023, "learning_rate": 2.962627656040408e-06, "loss": 0.1538, "step": 9477 }, { "epoch": 1.29, "grad_norm": 3.605950429296095, "learning_rate": 2.9616217606946385e-06, "loss": 0.1376, "step": 9478 }, { "epoch": 1.29, "grad_norm": 3.505241542621623, "learning_rate": 2.9606159642828446e-06, "loss": 0.2161, "step": 9479 }, { "epoch": 1.29, "grad_norm": 3.3710287877048217, "learning_rate": 2.959610266853845e-06, "loss": 0.1566, "step": 9480 }, { "epoch": 1.29, "grad_norm": 4.34620115938035, "learning_rate": 2.9586046684564516e-06, "loss": 0.2062, "step": 9481 }, { "epoch": 1.29, "grad_norm": 4.553253126476854, "learning_rate": 2.957599169139472e-06, "loss": 0.2044, "step": 9482 }, { "epoch": 1.29, "grad_norm": 4.31250565539141, "learning_rate": 2.9565937689517068e-06, "loss": 0.1669, "step": 9483 }, { "epoch": 1.29, "grad_norm": 3.5426414293173027, "learning_rate": 2.9555884679419577e-06, "loss": 0.1728, "step": 9484 }, { "epoch": 1.29, "grad_norm": 3.5222872555905615, "learning_rate": 2.9545832661590124e-06, "loss": 0.1602, "step": 9485 }, { "epoch": 1.29, "grad_norm": 3.6931134606101925, "learning_rate": 2.953578163651665e-06, "loss": 0.1496, "step": 9486 }, { "epoch": 1.29, "grad_norm": 3.601683323220219, "learning_rate": 2.9525731604686925e-06, "loss": 0.17, "step": 9487 }, { "epoch": 1.29, "grad_norm": 3.5224370155907967, "learning_rate": 2.9515682566588776e-06, "loss": 0.1724, "step": 9488 }, { "epoch": 1.29, "grad_norm": 3.6024839038399383, "learning_rate": 2.950563452270992e-06, "loss": 0.1328, "step": 9489 }, { "epoch": 1.29, "grad_norm": 3.9940202241155065, "learning_rate": 2.9495587473538074e-06, "loss": 0.2058, "step": 9490 }, { "epoch": 1.29, "grad_norm": 3.7995017059255267, "learning_rate": 2.948554141956084e-06, "loss": 0.1541, "step": 9491 }, { "epoch": 1.29, "grad_norm": 3.13155089764686, "learning_rate": 2.9475496361265834e-06, "loss": 0.1638, "step": 9492 }, { "epoch": 1.29, "grad_norm": 3.3122518896806614, "learning_rate": 2.946545229914058e-06, "loss": 0.1762, "step": 9493 }, { "epoch": 1.29, "grad_norm": 2.6777371745355585, "learning_rate": 2.9455409233672594e-06, "loss": 0.1138, "step": 9494 }, { "epoch": 1.29, "grad_norm": 4.297862708168623, "learning_rate": 2.9445367165349324e-06, "loss": 0.16, "step": 9495 }, { "epoch": 1.29, "grad_norm": 4.796458703312733, "learning_rate": 2.9435326094658134e-06, "loss": 0.1872, "step": 9496 }, { "epoch": 1.29, "grad_norm": 3.6252457996491314, "learning_rate": 2.942528602208642e-06, "loss": 0.1373, "step": 9497 }, { "epoch": 1.29, "grad_norm": 3.0872843581214138, "learning_rate": 2.9415246948121434e-06, "loss": 0.1367, "step": 9498 }, { "epoch": 1.29, "grad_norm": 3.103805054350699, "learning_rate": 2.9405208873250473e-06, "loss": 0.1502, "step": 9499 }, { "epoch": 1.29, "grad_norm": 4.253560794454568, "learning_rate": 2.9395171797960718e-06, "loss": 0.2068, "step": 9500 }, { "epoch": 1.29, "grad_norm": 3.194953320743607, "learning_rate": 2.9385135722739334e-06, "loss": 0.1506, "step": 9501 }, { "epoch": 1.29, "grad_norm": 4.823253010969122, "learning_rate": 2.9375100648073413e-06, "loss": 0.1985, "step": 9502 }, { "epoch": 1.29, "grad_norm": 3.62862948336609, "learning_rate": 2.936506657445004e-06, "loss": 0.1701, "step": 9503 }, { "epoch": 1.29, "grad_norm": 2.645221837363177, "learning_rate": 2.93550335023562e-06, "loss": 0.146, "step": 9504 }, { "epoch": 1.29, "grad_norm": 3.812899830702469, "learning_rate": 2.934500143227888e-06, "loss": 0.1541, "step": 9505 }, { "epoch": 1.29, "grad_norm": 4.0045896142836686, "learning_rate": 2.9334970364704957e-06, "loss": 0.1687, "step": 9506 }, { "epoch": 1.29, "grad_norm": 3.0589260190422585, "learning_rate": 2.9324940300121325e-06, "loss": 0.1748, "step": 9507 }, { "epoch": 1.29, "grad_norm": 6.932700044923913, "learning_rate": 2.9314911239014776e-06, "loss": 0.1282, "step": 9508 }, { "epoch": 1.29, "grad_norm": 2.8777899886723235, "learning_rate": 2.9304883181872124e-06, "loss": 0.1597, "step": 9509 }, { "epoch": 1.29, "grad_norm": 4.4949139086477015, "learning_rate": 2.929485612918002e-06, "loss": 0.1956, "step": 9510 }, { "epoch": 1.29, "grad_norm": 2.9316894045458337, "learning_rate": 2.92848300814252e-06, "loss": 0.1759, "step": 9511 }, { "epoch": 1.29, "grad_norm": 3.769539762210618, "learning_rate": 2.9274805039094225e-06, "loss": 0.1699, "step": 9512 }, { "epoch": 1.29, "grad_norm": 4.298485253339434, "learning_rate": 2.9264781002673702e-06, "loss": 0.1867, "step": 9513 }, { "epoch": 1.29, "grad_norm": 4.4564339256526075, "learning_rate": 2.9254757972650137e-06, "loss": 0.1668, "step": 9514 }, { "epoch": 1.29, "grad_norm": 2.6772683517951883, "learning_rate": 2.924473594951004e-06, "loss": 0.1573, "step": 9515 }, { "epoch": 1.29, "grad_norm": 3.652228442452421, "learning_rate": 2.9234714933739776e-06, "loss": 0.1194, "step": 9516 }, { "epoch": 1.29, "grad_norm": 3.2588746385705187, "learning_rate": 2.922469492582578e-06, "loss": 0.1968, "step": 9517 }, { "epoch": 1.29, "grad_norm": 3.0128657129140053, "learning_rate": 2.9214675926254344e-06, "loss": 0.1329, "step": 9518 }, { "epoch": 1.29, "grad_norm": 3.7482404028268, "learning_rate": 2.9204657935511763e-06, "loss": 0.1458, "step": 9519 }, { "epoch": 1.29, "grad_norm": 3.65822079991588, "learning_rate": 2.9194640954084245e-06, "loss": 0.179, "step": 9520 }, { "epoch": 1.3, "grad_norm": 4.370079211803591, "learning_rate": 2.9184624982458e-06, "loss": 0.1917, "step": 9521 }, { "epoch": 1.3, "grad_norm": 5.230319562231815, "learning_rate": 2.9174610021119136e-06, "loss": 0.1595, "step": 9522 }, { "epoch": 1.3, "grad_norm": 4.037033664309764, "learning_rate": 2.9164596070553753e-06, "loss": 0.2139, "step": 9523 }, { "epoch": 1.3, "grad_norm": 5.205195176364139, "learning_rate": 2.915458313124787e-06, "loss": 0.1704, "step": 9524 }, { "epoch": 1.3, "grad_norm": 4.422989346641574, "learning_rate": 2.914457120368748e-06, "loss": 0.1869, "step": 9525 }, { "epoch": 1.3, "grad_norm": 3.2685472749214584, "learning_rate": 2.9134560288358494e-06, "loss": 0.1979, "step": 9526 }, { "epoch": 1.3, "grad_norm": 3.019171587801906, "learning_rate": 2.912455038574686e-06, "loss": 0.1651, "step": 9527 }, { "epoch": 1.3, "grad_norm": 2.457750774187645, "learning_rate": 2.9114541496338335e-06, "loss": 0.1439, "step": 9528 }, { "epoch": 1.3, "grad_norm": 3.825586090982986, "learning_rate": 2.910453362061878e-06, "loss": 0.1715, "step": 9529 }, { "epoch": 1.3, "grad_norm": 3.5492345669764527, "learning_rate": 2.9094526759073872e-06, "loss": 0.154, "step": 9530 }, { "epoch": 1.3, "grad_norm": 3.4425596483833876, "learning_rate": 2.908452091218934e-06, "loss": 0.1618, "step": 9531 }, { "epoch": 1.3, "grad_norm": 3.6406865749182438, "learning_rate": 2.907451608045081e-06, "loss": 0.1802, "step": 9532 }, { "epoch": 1.3, "grad_norm": 3.6229806211459983, "learning_rate": 2.906451226434388e-06, "loss": 0.1774, "step": 9533 }, { "epoch": 1.3, "grad_norm": 3.3440040673855362, "learning_rate": 2.9054509464354085e-06, "loss": 0.2274, "step": 9534 }, { "epoch": 1.3, "grad_norm": 3.2651360737331485, "learning_rate": 2.9044507680966914e-06, "loss": 0.1741, "step": 9535 }, { "epoch": 1.3, "grad_norm": 3.3012353918013004, "learning_rate": 2.9034506914667827e-06, "loss": 0.1561, "step": 9536 }, { "epoch": 1.3, "grad_norm": 2.937201154687231, "learning_rate": 2.9024507165942196e-06, "loss": 0.1551, "step": 9537 }, { "epoch": 1.3, "grad_norm": 4.0206116405665355, "learning_rate": 2.901450843527538e-06, "loss": 0.1708, "step": 9538 }, { "epoch": 1.3, "grad_norm": 3.819086077534622, "learning_rate": 2.900451072315266e-06, "loss": 0.1533, "step": 9539 }, { "epoch": 1.3, "grad_norm": 4.599371559176421, "learning_rate": 2.8994514030059296e-06, "loss": 0.1567, "step": 9540 }, { "epoch": 1.3, "grad_norm": 3.5962932771564375, "learning_rate": 2.8984518356480473e-06, "loss": 0.1621, "step": 9541 }, { "epoch": 1.3, "grad_norm": 3.674475589355425, "learning_rate": 2.8974523702901346e-06, "loss": 0.1498, "step": 9542 }, { "epoch": 1.3, "grad_norm": 4.4338661667173245, "learning_rate": 2.896453006980701e-06, "loss": 0.1538, "step": 9543 }, { "epoch": 1.3, "grad_norm": 3.4890157496099685, "learning_rate": 2.8954537457682503e-06, "loss": 0.1536, "step": 9544 }, { "epoch": 1.3, "grad_norm": 3.7739269443180063, "learning_rate": 2.8944545867012812e-06, "loss": 0.1796, "step": 9545 }, { "epoch": 1.3, "grad_norm": 2.727080442268128, "learning_rate": 2.893455529828294e-06, "loss": 0.1549, "step": 9546 }, { "epoch": 1.3, "grad_norm": 2.2806747288358884, "learning_rate": 2.892456575197771e-06, "loss": 0.1072, "step": 9547 }, { "epoch": 1.3, "grad_norm": 3.635941668844607, "learning_rate": 2.8914577228582052e-06, "loss": 0.2089, "step": 9548 }, { "epoch": 1.3, "grad_norm": 3.44950558779407, "learning_rate": 2.890458972858068e-06, "loss": 0.1716, "step": 9549 }, { "epoch": 1.3, "grad_norm": 2.508766592626429, "learning_rate": 2.8894603252458407e-06, "loss": 0.1584, "step": 9550 }, { "epoch": 1.3, "grad_norm": 3.532774438342608, "learning_rate": 2.888461780069991e-06, "loss": 0.185, "step": 9551 }, { "epoch": 1.3, "grad_norm": 5.946842527970195, "learning_rate": 2.8874633373789848e-06, "loss": 0.1426, "step": 9552 }, { "epoch": 1.3, "grad_norm": 3.3216478013085124, "learning_rate": 2.8864649972212804e-06, "loss": 0.1532, "step": 9553 }, { "epoch": 1.3, "grad_norm": 3.899595883888632, "learning_rate": 2.8854667596453345e-06, "loss": 0.1965, "step": 9554 }, { "epoch": 1.3, "grad_norm": 2.569334894371295, "learning_rate": 2.884468624699597e-06, "loss": 0.1428, "step": 9555 }, { "epoch": 1.3, "grad_norm": 3.527005745319215, "learning_rate": 2.883470592432512e-06, "loss": 0.1533, "step": 9556 }, { "epoch": 1.3, "grad_norm": 3.341035041738301, "learning_rate": 2.8824726628925204e-06, "loss": 0.1534, "step": 9557 }, { "epoch": 1.3, "grad_norm": 3.676821843647895, "learning_rate": 2.8814748361280575e-06, "loss": 0.162, "step": 9558 }, { "epoch": 1.3, "grad_norm": 3.165839965477186, "learning_rate": 2.8804771121875506e-06, "loss": 0.1924, "step": 9559 }, { "epoch": 1.3, "grad_norm": 3.2730444346389294, "learning_rate": 2.8794794911194324e-06, "loss": 0.1611, "step": 9560 }, { "epoch": 1.3, "grad_norm": 2.905363128126543, "learning_rate": 2.878481972972113e-06, "loss": 0.1312, "step": 9561 }, { "epoch": 1.3, "grad_norm": 4.783234298982898, "learning_rate": 2.877484557794017e-06, "loss": 0.1718, "step": 9562 }, { "epoch": 1.3, "grad_norm": 4.135753268175564, "learning_rate": 2.876487245633547e-06, "loss": 0.2318, "step": 9563 }, { "epoch": 1.3, "grad_norm": 4.472228582659644, "learning_rate": 2.8754900365391124e-06, "loss": 0.2101, "step": 9564 }, { "epoch": 1.3, "grad_norm": 4.18095529039824, "learning_rate": 2.874492930559113e-06, "loss": 0.1821, "step": 9565 }, { "epoch": 1.3, "grad_norm": 3.177120586979527, "learning_rate": 2.873495927741943e-06, "loss": 0.1546, "step": 9566 }, { "epoch": 1.3, "grad_norm": 3.1125336897701397, "learning_rate": 2.872499028135993e-06, "loss": 0.1759, "step": 9567 }, { "epoch": 1.3, "grad_norm": 2.6416955901354253, "learning_rate": 2.8715022317896487e-06, "loss": 0.12, "step": 9568 }, { "epoch": 1.3, "grad_norm": 3.31395223075214, "learning_rate": 2.8705055387512893e-06, "loss": 0.151, "step": 9569 }, { "epoch": 1.3, "grad_norm": 4.0841491182806875, "learning_rate": 2.8695089490692894e-06, "loss": 0.1784, "step": 9570 }, { "epoch": 1.3, "grad_norm": 3.85362396377138, "learning_rate": 2.8685124627920193e-06, "loss": 0.1359, "step": 9571 }, { "epoch": 1.3, "grad_norm": 3.7614843644931026, "learning_rate": 2.8675160799678483e-06, "loss": 0.1618, "step": 9572 }, { "epoch": 1.3, "grad_norm": 3.561870154934863, "learning_rate": 2.8665198006451276e-06, "loss": 0.182, "step": 9573 }, { "epoch": 1.3, "grad_norm": 4.293588658948316, "learning_rate": 2.865523624872223e-06, "loss": 0.1672, "step": 9574 }, { "epoch": 1.3, "grad_norm": 3.2905937990557645, "learning_rate": 2.864527552697477e-06, "loss": 0.1465, "step": 9575 }, { "epoch": 1.3, "grad_norm": 3.906194702131124, "learning_rate": 2.8635315841692368e-06, "loss": 0.1878, "step": 9576 }, { "epoch": 1.3, "grad_norm": 4.114545922771763, "learning_rate": 2.8625357193358416e-06, "loss": 0.1551, "step": 9577 }, { "epoch": 1.3, "grad_norm": 3.2462970791440697, "learning_rate": 2.8615399582456255e-06, "loss": 0.1925, "step": 9578 }, { "epoch": 1.3, "grad_norm": 3.6457931163682447, "learning_rate": 2.8605443009469237e-06, "loss": 0.144, "step": 9579 }, { "epoch": 1.3, "grad_norm": 4.4723837053564255, "learning_rate": 2.859548747488053e-06, "loss": 0.222, "step": 9580 }, { "epoch": 1.3, "grad_norm": 2.869522597756747, "learning_rate": 2.8585532979173413e-06, "loss": 0.1278, "step": 9581 }, { "epoch": 1.3, "grad_norm": 4.274148667485489, "learning_rate": 2.8575579522830965e-06, "loss": 0.1513, "step": 9582 }, { "epoch": 1.3, "grad_norm": 3.636846288277312, "learning_rate": 2.8565627106336322e-06, "loss": 0.1583, "step": 9583 }, { "epoch": 1.3, "grad_norm": 3.555062403551772, "learning_rate": 2.8555675730172526e-06, "loss": 0.1407, "step": 9584 }, { "epoch": 1.3, "grad_norm": 3.5351605015056267, "learning_rate": 2.8545725394822575e-06, "loss": 0.1835, "step": 9585 }, { "epoch": 1.3, "grad_norm": 2.5700694942545144, "learning_rate": 2.85357761007694e-06, "loss": 0.112, "step": 9586 }, { "epoch": 1.3, "grad_norm": 5.70244599644543, "learning_rate": 2.8525827848495912e-06, "loss": 0.2067, "step": 9587 }, { "epoch": 1.3, "grad_norm": 3.8049498193257483, "learning_rate": 2.851588063848495e-06, "loss": 0.1978, "step": 9588 }, { "epoch": 1.3, "grad_norm": 3.543394171987285, "learning_rate": 2.850593447121931e-06, "loss": 0.1483, "step": 9589 }, { "epoch": 1.3, "grad_norm": 4.127049312413663, "learning_rate": 2.849598934718171e-06, "loss": 0.1726, "step": 9590 }, { "epoch": 1.3, "grad_norm": 2.5408004583719186, "learning_rate": 2.84860452668549e-06, "loss": 0.1061, "step": 9591 }, { "epoch": 1.3, "grad_norm": 3.929654726545116, "learning_rate": 2.847610223072145e-06, "loss": 0.214, "step": 9592 }, { "epoch": 1.3, "grad_norm": 3.5144413176040374, "learning_rate": 2.846616023926404e-06, "loss": 0.1704, "step": 9593 }, { "epoch": 1.31, "grad_norm": 3.86141625669693, "learning_rate": 2.845621929296511e-06, "loss": 0.1785, "step": 9594 }, { "epoch": 1.31, "grad_norm": 3.122407167452016, "learning_rate": 2.8446279392307242e-06, "loss": 0.1903, "step": 9595 }, { "epoch": 1.31, "grad_norm": 3.3586977308700057, "learning_rate": 2.84363405377728e-06, "loss": 0.1587, "step": 9596 }, { "epoch": 1.31, "grad_norm": 3.66049008255653, "learning_rate": 2.842640272984422e-06, "loss": 0.1217, "step": 9597 }, { "epoch": 1.31, "grad_norm": 6.943806079861596, "learning_rate": 2.841646596900383e-06, "loss": 0.1804, "step": 9598 }, { "epoch": 1.31, "grad_norm": 2.9385017020137414, "learning_rate": 2.8406530255733913e-06, "loss": 0.1359, "step": 9599 }, { "epoch": 1.31, "grad_norm": 3.409247258191851, "learning_rate": 2.8396595590516708e-06, "loss": 0.1555, "step": 9600 }, { "epoch": 1.31, "grad_norm": 3.781585822825484, "learning_rate": 2.8386661973834397e-06, "loss": 0.1436, "step": 9601 }, { "epoch": 1.31, "grad_norm": 3.4579189790850404, "learning_rate": 2.837672940616911e-06, "loss": 0.1533, "step": 9602 }, { "epoch": 1.31, "grad_norm": 4.593060901644934, "learning_rate": 2.836679788800294e-06, "loss": 0.1789, "step": 9603 }, { "epoch": 1.31, "grad_norm": 4.081170168412006, "learning_rate": 2.835686741981789e-06, "loss": 0.1928, "step": 9604 }, { "epoch": 1.31, "grad_norm": 4.737463188465827, "learning_rate": 2.8346938002096014e-06, "loss": 0.1703, "step": 9605 }, { "epoch": 1.31, "grad_norm": 3.3766295343901245, "learning_rate": 2.833700963531916e-06, "loss": 0.1547, "step": 9606 }, { "epoch": 1.31, "grad_norm": 3.8182098249587675, "learning_rate": 2.8327082319969268e-06, "loss": 0.1526, "step": 9607 }, { "epoch": 1.31, "grad_norm": 2.5317388668104135, "learning_rate": 2.8317156056528105e-06, "loss": 0.1249, "step": 9608 }, { "epoch": 1.31, "grad_norm": 4.120313234085952, "learning_rate": 2.8307230845477514e-06, "loss": 0.195, "step": 9609 }, { "epoch": 1.31, "grad_norm": 2.5663113337892676, "learning_rate": 2.829730668729919e-06, "loss": 0.1532, "step": 9610 }, { "epoch": 1.31, "grad_norm": 3.9708740442097112, "learning_rate": 2.8287383582474803e-06, "loss": 0.1633, "step": 9611 }, { "epoch": 1.31, "grad_norm": 3.146582970457994, "learning_rate": 2.8277461531485985e-06, "loss": 0.179, "step": 9612 }, { "epoch": 1.31, "grad_norm": 3.704198538591837, "learning_rate": 2.8267540534814314e-06, "loss": 0.2032, "step": 9613 }, { "epoch": 1.31, "grad_norm": 2.8906173986416475, "learning_rate": 2.8257620592941303e-06, "loss": 0.1092, "step": 9614 }, { "epoch": 1.31, "grad_norm": 2.9905878403715604, "learning_rate": 2.8247701706348428e-06, "loss": 0.1619, "step": 9615 }, { "epoch": 1.31, "grad_norm": 3.2970475833554507, "learning_rate": 2.823778387551711e-06, "loss": 0.1687, "step": 9616 }, { "epoch": 1.31, "grad_norm": 2.9486682409314877, "learning_rate": 2.8227867100928706e-06, "loss": 0.1547, "step": 9617 }, { "epoch": 1.31, "grad_norm": 3.7243084909742032, "learning_rate": 2.8217951383064546e-06, "loss": 0.1704, "step": 9618 }, { "epoch": 1.31, "grad_norm": 3.7086525210282932, "learning_rate": 2.8208036722405895e-06, "loss": 0.1869, "step": 9619 }, { "epoch": 1.31, "grad_norm": 7.962758099671857, "learning_rate": 2.819812311943395e-06, "loss": 0.1219, "step": 9620 }, { "epoch": 1.31, "grad_norm": 2.9477813849813046, "learning_rate": 2.81882105746299e-06, "loss": 0.1722, "step": 9621 }, { "epoch": 1.31, "grad_norm": 2.8627465128098466, "learning_rate": 2.8178299088474836e-06, "loss": 0.1509, "step": 9622 }, { "epoch": 1.31, "grad_norm": 3.7115469708019, "learning_rate": 2.8168388661449807e-06, "loss": 0.1691, "step": 9623 }, { "epoch": 1.31, "grad_norm": 2.5985156656238853, "learning_rate": 2.815847929403588e-06, "loss": 0.1478, "step": 9624 }, { "epoch": 1.31, "grad_norm": 3.7997839781477922, "learning_rate": 2.8148570986713934e-06, "loss": 0.1975, "step": 9625 }, { "epoch": 1.31, "grad_norm": 4.100817525561468, "learning_rate": 2.8138663739964955e-06, "loss": 0.1542, "step": 9626 }, { "epoch": 1.31, "grad_norm": 3.4214214870330593, "learning_rate": 2.8128757554269716e-06, "loss": 0.1489, "step": 9627 }, { "epoch": 1.31, "grad_norm": 4.1840012794454315, "learning_rate": 2.8118852430109094e-06, "loss": 0.1973, "step": 9628 }, { "epoch": 1.31, "grad_norm": 3.277982613070102, "learning_rate": 2.8108948367963774e-06, "loss": 0.15, "step": 9629 }, { "epoch": 1.31, "grad_norm": 3.667568861844931, "learning_rate": 2.8099045368314504e-06, "loss": 0.1183, "step": 9630 }, { "epoch": 1.31, "grad_norm": 3.272883709764085, "learning_rate": 2.8089143431641918e-06, "loss": 0.1631, "step": 9631 }, { "epoch": 1.31, "grad_norm": 2.263278104692641, "learning_rate": 2.8079242558426612e-06, "loss": 0.1158, "step": 9632 }, { "epoch": 1.31, "grad_norm": 3.6753228377027063, "learning_rate": 2.806934274914913e-06, "loss": 0.1756, "step": 9633 }, { "epoch": 1.31, "grad_norm": 3.5628253313148397, "learning_rate": 2.805944400428997e-06, "loss": 0.1494, "step": 9634 }, { "epoch": 1.31, "grad_norm": 2.7968471232348424, "learning_rate": 2.8049546324329567e-06, "loss": 0.1266, "step": 9635 }, { "epoch": 1.31, "grad_norm": 4.335912709324329, "learning_rate": 2.803964970974832e-06, "loss": 0.1977, "step": 9636 }, { "epoch": 1.31, "grad_norm": 3.0631743400703044, "learning_rate": 2.8029754161026535e-06, "loss": 0.1165, "step": 9637 }, { "epoch": 1.31, "grad_norm": 4.209189140392156, "learning_rate": 2.8019859678644573e-06, "loss": 0.1522, "step": 9638 }, { "epoch": 1.31, "grad_norm": 3.411974926013508, "learning_rate": 2.800996626308258e-06, "loss": 0.1721, "step": 9639 }, { "epoch": 1.31, "grad_norm": 3.224650549040634, "learning_rate": 2.8000073914820813e-06, "loss": 0.1753, "step": 9640 }, { "epoch": 1.31, "grad_norm": 3.1825131549532437, "learning_rate": 2.7990182634339334e-06, "loss": 0.1416, "step": 9641 }, { "epoch": 1.31, "grad_norm": 5.048942055930336, "learning_rate": 2.7980292422118282e-06, "loss": 0.1896, "step": 9642 }, { "epoch": 1.31, "grad_norm": 3.7684093702551094, "learning_rate": 2.797040327863766e-06, "loss": 0.1359, "step": 9643 }, { "epoch": 1.31, "grad_norm": 4.415783885261657, "learning_rate": 2.7960515204377446e-06, "loss": 0.1754, "step": 9644 }, { "epoch": 1.31, "grad_norm": 3.981680933076944, "learning_rate": 2.795062819981757e-06, "loss": 0.1592, "step": 9645 }, { "epoch": 1.31, "grad_norm": 4.4560505384502305, "learning_rate": 2.794074226543789e-06, "loss": 0.1757, "step": 9646 }, { "epoch": 1.31, "grad_norm": 4.811799181363049, "learning_rate": 2.7930857401718244e-06, "loss": 0.1626, "step": 9647 }, { "epoch": 1.31, "grad_norm": 2.8619564296247484, "learning_rate": 2.7920973609138385e-06, "loss": 0.1501, "step": 9648 }, { "epoch": 1.31, "grad_norm": 3.293373192080053, "learning_rate": 2.7911090888178033e-06, "loss": 0.1559, "step": 9649 }, { "epoch": 1.31, "grad_norm": 3.7620376043131176, "learning_rate": 2.790120923931685e-06, "loss": 0.1832, "step": 9650 }, { "epoch": 1.31, "grad_norm": 3.693127212183988, "learning_rate": 2.7891328663034446e-06, "loss": 0.1701, "step": 9651 }, { "epoch": 1.31, "grad_norm": 4.100398181257106, "learning_rate": 2.7881449159810416e-06, "loss": 0.1805, "step": 9652 }, { "epoch": 1.31, "grad_norm": 4.016339683825179, "learning_rate": 2.78715707301242e-06, "loss": 0.1426, "step": 9653 }, { "epoch": 1.31, "grad_norm": 3.9719780950131804, "learning_rate": 2.7861693374455327e-06, "loss": 0.1883, "step": 9654 }, { "epoch": 1.31, "grad_norm": 3.462876638577804, "learning_rate": 2.785181709328315e-06, "loss": 0.115, "step": 9655 }, { "epoch": 1.31, "grad_norm": 2.9901020600243386, "learning_rate": 2.784194188708702e-06, "loss": 0.1566, "step": 9656 }, { "epoch": 1.31, "grad_norm": 3.443556243067071, "learning_rate": 2.7832067756346293e-06, "loss": 0.1414, "step": 9657 }, { "epoch": 1.31, "grad_norm": 4.869758752320322, "learning_rate": 2.782219470154013e-06, "loss": 0.2083, "step": 9658 }, { "epoch": 1.31, "grad_norm": 3.3203212021879733, "learning_rate": 2.7812322723147823e-06, "loss": 0.2148, "step": 9659 }, { "epoch": 1.31, "grad_norm": 3.4725454987223765, "learning_rate": 2.7802451821648414e-06, "loss": 0.1761, "step": 9660 }, { "epoch": 1.31, "grad_norm": 3.2831151444423727, "learning_rate": 2.7792581997521063e-06, "loss": 0.154, "step": 9661 }, { "epoch": 1.31, "grad_norm": 4.363774778460994, "learning_rate": 2.7782713251244797e-06, "loss": 0.1766, "step": 9662 }, { "epoch": 1.31, "grad_norm": 3.916243254582476, "learning_rate": 2.7772845583298585e-06, "loss": 0.1767, "step": 9663 }, { "epoch": 1.31, "grad_norm": 10.455056301442667, "learning_rate": 2.7762978994161372e-06, "loss": 0.1366, "step": 9664 }, { "epoch": 1.31, "grad_norm": 3.3604323474961544, "learning_rate": 2.7753113484312037e-06, "loss": 0.1331, "step": 9665 }, { "epoch": 1.31, "grad_norm": 3.1766121786972157, "learning_rate": 2.774324905422941e-06, "loss": 0.1428, "step": 9666 }, { "epoch": 1.31, "grad_norm": 3.2448093985627997, "learning_rate": 2.7733385704392257e-06, "loss": 0.165, "step": 9667 }, { "epoch": 1.32, "grad_norm": 3.6592788118049833, "learning_rate": 2.772352343527932e-06, "loss": 0.1632, "step": 9668 }, { "epoch": 1.32, "grad_norm": 3.4162478860685193, "learning_rate": 2.7713662247369256e-06, "loss": 0.174, "step": 9669 }, { "epoch": 1.32, "grad_norm": 3.0227314908370544, "learning_rate": 2.7703802141140667e-06, "loss": 0.1658, "step": 9670 }, { "epoch": 1.32, "grad_norm": 3.637897178004327, "learning_rate": 2.7693943117072186e-06, "loss": 0.1823, "step": 9671 }, { "epoch": 1.32, "grad_norm": 2.9330111023235146, "learning_rate": 2.768408517564224e-06, "loss": 0.1297, "step": 9672 }, { "epoch": 1.32, "grad_norm": 3.0398908307365327, "learning_rate": 2.7674228317329366e-06, "loss": 0.1277, "step": 9673 }, { "epoch": 1.32, "grad_norm": 5.046278874853826, "learning_rate": 2.766437254261191e-06, "loss": 0.2018, "step": 9674 }, { "epoch": 1.32, "grad_norm": 3.1958664436489816, "learning_rate": 2.7654517851968267e-06, "loss": 0.1571, "step": 9675 }, { "epoch": 1.32, "grad_norm": 2.943853285141477, "learning_rate": 2.7644664245876733e-06, "loss": 0.1145, "step": 9676 }, { "epoch": 1.32, "grad_norm": 4.043113334341724, "learning_rate": 2.763481172481556e-06, "loss": 0.1857, "step": 9677 }, { "epoch": 1.32, "grad_norm": 2.9237192394529528, "learning_rate": 2.7624960289262947e-06, "loss": 0.1688, "step": 9678 }, { "epoch": 1.32, "grad_norm": 3.3541635393489733, "learning_rate": 2.7615109939697025e-06, "loss": 0.079, "step": 9679 }, { "epoch": 1.32, "grad_norm": 3.889795632891302, "learning_rate": 2.760526067659591e-06, "loss": 0.1777, "step": 9680 }, { "epoch": 1.32, "grad_norm": 4.5880995510833245, "learning_rate": 2.7595412500437623e-06, "loss": 0.1517, "step": 9681 }, { "epoch": 1.32, "grad_norm": 3.047656622638721, "learning_rate": 2.7585565411700164e-06, "loss": 0.1882, "step": 9682 }, { "epoch": 1.32, "grad_norm": 2.7948832672130957, "learning_rate": 2.7575719410861456e-06, "loss": 0.1723, "step": 9683 }, { "epoch": 1.32, "grad_norm": 4.463313657434235, "learning_rate": 2.7565874498399375e-06, "loss": 0.2016, "step": 9684 }, { "epoch": 1.32, "grad_norm": 3.525694103238369, "learning_rate": 2.7556030674791802e-06, "loss": 0.1393, "step": 9685 }, { "epoch": 1.32, "grad_norm": 5.255007641656851, "learning_rate": 2.7546187940516424e-06, "loss": 0.1663, "step": 9686 }, { "epoch": 1.32, "grad_norm": 3.3218115008000533, "learning_rate": 2.7536346296051063e-06, "loss": 0.1199, "step": 9687 }, { "epoch": 1.32, "grad_norm": 4.124329392924494, "learning_rate": 2.75265057418733e-06, "loss": 0.1568, "step": 9688 }, { "epoch": 1.32, "grad_norm": 2.9683687576780864, "learning_rate": 2.751666627846081e-06, "loss": 0.1467, "step": 9689 }, { "epoch": 1.32, "grad_norm": 3.0548900684740494, "learning_rate": 2.750682790629113e-06, "loss": 0.1729, "step": 9690 }, { "epoch": 1.32, "grad_norm": 3.416948973172432, "learning_rate": 2.7496990625841803e-06, "loss": 0.1571, "step": 9691 }, { "epoch": 1.32, "grad_norm": 5.014514715939454, "learning_rate": 2.7487154437590252e-06, "loss": 0.1415, "step": 9692 }, { "epoch": 1.32, "grad_norm": 3.5859630115420535, "learning_rate": 2.74773193420139e-06, "loss": 0.176, "step": 9693 }, { "epoch": 1.32, "grad_norm": 3.5631367390531876, "learning_rate": 2.74674853395901e-06, "loss": 0.1775, "step": 9694 }, { "epoch": 1.32, "grad_norm": 3.278017411753832, "learning_rate": 2.7457652430796146e-06, "loss": 0.1618, "step": 9695 }, { "epoch": 1.32, "grad_norm": 3.2893745977936546, "learning_rate": 2.744782061610929e-06, "loss": 0.1391, "step": 9696 }, { "epoch": 1.32, "grad_norm": 4.489555072114231, "learning_rate": 2.743798989600672e-06, "loss": 0.1597, "step": 9697 }, { "epoch": 1.32, "grad_norm": 3.3824483530300236, "learning_rate": 2.7428160270965576e-06, "loss": 0.1974, "step": 9698 }, { "epoch": 1.32, "grad_norm": 4.035442289988728, "learning_rate": 2.741833174146295e-06, "loss": 0.1337, "step": 9699 }, { "epoch": 1.32, "grad_norm": 3.397250003699, "learning_rate": 2.7408504307975877e-06, "loss": 0.1404, "step": 9700 }, { "epoch": 1.32, "grad_norm": 3.4336998381820143, "learning_rate": 2.7398677970981324e-06, "loss": 0.1447, "step": 9701 }, { "epoch": 1.32, "grad_norm": 3.2912251207846133, "learning_rate": 2.738885273095624e-06, "loss": 0.1439, "step": 9702 }, { "epoch": 1.32, "grad_norm": 2.791860446095469, "learning_rate": 2.7379028588377466e-06, "loss": 0.1178, "step": 9703 }, { "epoch": 1.32, "grad_norm": 3.3231631110103352, "learning_rate": 2.7369205543721878e-06, "loss": 0.143, "step": 9704 }, { "epoch": 1.32, "grad_norm": 3.300898324968183, "learning_rate": 2.735938359746617e-06, "loss": 0.1714, "step": 9705 }, { "epoch": 1.32, "grad_norm": 3.5951937029489702, "learning_rate": 2.7349562750087142e-06, "loss": 0.1784, "step": 9706 }, { "epoch": 1.32, "grad_norm": 3.363103631626996, "learning_rate": 2.733974300206137e-06, "loss": 0.1396, "step": 9707 }, { "epoch": 1.32, "grad_norm": 4.837992348498634, "learning_rate": 2.732992435386552e-06, "loss": 0.2006, "step": 9708 }, { "epoch": 1.32, "grad_norm": 21.264358012422132, "learning_rate": 2.7320106805976127e-06, "loss": 0.1828, "step": 9709 }, { "epoch": 1.32, "grad_norm": 3.808419578943788, "learning_rate": 2.7310290358869694e-06, "loss": 0.1572, "step": 9710 }, { "epoch": 1.32, "grad_norm": 4.642907736966732, "learning_rate": 2.7300475013022666e-06, "loss": 0.1417, "step": 9711 }, { "epoch": 1.32, "grad_norm": 3.142406148236357, "learning_rate": 2.7290660768911435e-06, "loss": 0.1589, "step": 9712 }, { "epoch": 1.32, "grad_norm": 4.516617456450494, "learning_rate": 2.728084762701235e-06, "loss": 0.1591, "step": 9713 }, { "epoch": 1.32, "grad_norm": 3.2782193176975425, "learning_rate": 2.7271035587801695e-06, "loss": 0.1481, "step": 9714 }, { "epoch": 1.32, "grad_norm": 2.8030571047105886, "learning_rate": 2.7261224651755697e-06, "loss": 0.1628, "step": 9715 }, { "epoch": 1.32, "grad_norm": 2.7106262867904043, "learning_rate": 2.725141481935054e-06, "loss": 0.1594, "step": 9716 }, { "epoch": 1.32, "grad_norm": 3.3562087748044296, "learning_rate": 2.7241606091062334e-06, "loss": 0.1336, "step": 9717 }, { "epoch": 1.32, "grad_norm": 3.337494477586148, "learning_rate": 2.723179846736721e-06, "loss": 0.1438, "step": 9718 }, { "epoch": 1.32, "grad_norm": 3.586070156138028, "learning_rate": 2.72219919487411e-06, "loss": 0.1664, "step": 9719 }, { "epoch": 1.32, "grad_norm": 3.5298657807566616, "learning_rate": 2.7212186535660057e-06, "loss": 0.1986, "step": 9720 }, { "epoch": 1.32, "grad_norm": 3.56065288393511, "learning_rate": 2.720238222859991e-06, "loss": 0.1864, "step": 9721 }, { "epoch": 1.32, "grad_norm": 3.2095350833562724, "learning_rate": 2.719257902803658e-06, "loss": 0.1368, "step": 9722 }, { "epoch": 1.32, "grad_norm": 3.393867236431527, "learning_rate": 2.7182776934445853e-06, "loss": 0.1547, "step": 9723 }, { "epoch": 1.32, "grad_norm": 2.8424583745907985, "learning_rate": 2.717297594830347e-06, "loss": 0.132, "step": 9724 }, { "epoch": 1.32, "grad_norm": 3.78009807089218, "learning_rate": 2.716317607008514e-06, "loss": 0.1746, "step": 9725 }, { "epoch": 1.32, "grad_norm": 4.361938754834164, "learning_rate": 2.7153377300266506e-06, "loss": 0.1661, "step": 9726 }, { "epoch": 1.32, "grad_norm": 3.0481467796124138, "learning_rate": 2.7143579639323146e-06, "loss": 0.1434, "step": 9727 }, { "epoch": 1.32, "grad_norm": 3.3555738685202146, "learning_rate": 2.71337830877306e-06, "loss": 0.1481, "step": 9728 }, { "epoch": 1.32, "grad_norm": 4.047346263085089, "learning_rate": 2.7123987645964355e-06, "loss": 0.1537, "step": 9729 }, { "epoch": 1.32, "grad_norm": 3.600385490251084, "learning_rate": 2.7114193314499836e-06, "loss": 0.1953, "step": 9730 }, { "epoch": 1.32, "grad_norm": 3.3724027728836328, "learning_rate": 2.71044000938124e-06, "loss": 0.1706, "step": 9731 }, { "epoch": 1.32, "grad_norm": 3.7451479268910384, "learning_rate": 2.7094607984377423e-06, "loss": 0.1462, "step": 9732 }, { "epoch": 1.32, "grad_norm": 3.6433019826387993, "learning_rate": 2.708481698667009e-06, "loss": 0.1408, "step": 9733 }, { "epoch": 1.32, "grad_norm": 3.957562728745009, "learning_rate": 2.7075027101165706e-06, "loss": 0.1865, "step": 9734 }, { "epoch": 1.32, "grad_norm": 3.158772623685531, "learning_rate": 2.7065238328339326e-06, "loss": 0.1638, "step": 9735 }, { "epoch": 1.32, "grad_norm": 2.6855300873218177, "learning_rate": 2.7055450668666138e-06, "loss": 0.1188, "step": 9736 }, { "epoch": 1.32, "grad_norm": 3.792324467981666, "learning_rate": 2.7045664122621173e-06, "loss": 0.1789, "step": 9737 }, { "epoch": 1.32, "grad_norm": 3.753094512108422, "learning_rate": 2.7035878690679386e-06, "loss": 0.1797, "step": 9738 }, { "epoch": 1.32, "grad_norm": 3.309349948710469, "learning_rate": 2.7026094373315782e-06, "loss": 0.1672, "step": 9739 }, { "epoch": 1.32, "grad_norm": 3.3775917647071623, "learning_rate": 2.701631117100517e-06, "loss": 0.1784, "step": 9740 }, { "epoch": 1.33, "grad_norm": 3.132863810323925, "learning_rate": 2.700652908422246e-06, "loss": 0.1467, "step": 9741 }, { "epoch": 1.33, "grad_norm": 3.831681264982495, "learning_rate": 2.6996748113442397e-06, "loss": 0.2001, "step": 9742 }, { "epoch": 1.33, "grad_norm": 3.209795346332828, "learning_rate": 2.6986968259139713e-06, "loss": 0.1914, "step": 9743 }, { "epoch": 1.33, "grad_norm": 3.20579396161611, "learning_rate": 2.6977189521789083e-06, "loss": 0.1697, "step": 9744 }, { "epoch": 1.33, "grad_norm": 3.093379924483488, "learning_rate": 2.696741190186511e-06, "loss": 0.1563, "step": 9745 }, { "epoch": 1.33, "grad_norm": 3.4597889514858036, "learning_rate": 2.695763539984237e-06, "loss": 0.1661, "step": 9746 }, { "epoch": 1.33, "grad_norm": 3.785070534450054, "learning_rate": 2.6947860016195372e-06, "loss": 0.1279, "step": 9747 }, { "epoch": 1.33, "grad_norm": 2.7086545231928674, "learning_rate": 2.6938085751398555e-06, "loss": 0.1151, "step": 9748 }, { "epoch": 1.33, "grad_norm": 3.6670808344375585, "learning_rate": 2.692831260592634e-06, "loss": 0.177, "step": 9749 }, { "epoch": 1.33, "grad_norm": 4.345643542169334, "learning_rate": 2.691854058025304e-06, "loss": 0.1466, "step": 9750 }, { "epoch": 1.33, "grad_norm": 3.259104110569208, "learning_rate": 2.6908769674853007e-06, "loss": 0.1424, "step": 9751 }, { "epoch": 1.33, "grad_norm": 4.611727606773025, "learning_rate": 2.6898999890200405e-06, "loss": 0.2052, "step": 9752 }, { "epoch": 1.33, "grad_norm": 3.5270658211597596, "learning_rate": 2.6889231226769487e-06, "loss": 0.1876, "step": 9753 }, { "epoch": 1.33, "grad_norm": 2.8436480067132033, "learning_rate": 2.6879463685034313e-06, "loss": 0.1683, "step": 9754 }, { "epoch": 1.33, "grad_norm": 3.582385246515161, "learning_rate": 2.6869697265469003e-06, "loss": 0.1732, "step": 9755 }, { "epoch": 1.33, "grad_norm": 3.071355303229652, "learning_rate": 2.685993196854757e-06, "loss": 0.1594, "step": 9756 }, { "epoch": 1.33, "grad_norm": 3.840078738056331, "learning_rate": 2.6850167794743966e-06, "loss": 0.1557, "step": 9757 }, { "epoch": 1.33, "grad_norm": 3.6520563737115883, "learning_rate": 2.684040474453211e-06, "loss": 0.1413, "step": 9758 }, { "epoch": 1.33, "grad_norm": 3.738372259474919, "learning_rate": 2.6830642818385856e-06, "loss": 0.1476, "step": 9759 }, { "epoch": 1.33, "grad_norm": 3.281197736416157, "learning_rate": 2.6820882016779004e-06, "loss": 0.1695, "step": 9760 }, { "epoch": 1.33, "grad_norm": 3.9665067599647728, "learning_rate": 2.6811122340185307e-06, "loss": 0.1431, "step": 9761 }, { "epoch": 1.33, "grad_norm": 3.111087083826477, "learning_rate": 2.680136378907845e-06, "loss": 0.1101, "step": 9762 }, { "epoch": 1.33, "grad_norm": 2.9162985658704246, "learning_rate": 2.6791606363932067e-06, "loss": 0.1417, "step": 9763 }, { "epoch": 1.33, "grad_norm": 3.506722502601708, "learning_rate": 2.6781850065219727e-06, "loss": 0.15, "step": 9764 }, { "epoch": 1.33, "grad_norm": 4.07813962715479, "learning_rate": 2.6772094893415025e-06, "loss": 0.1442, "step": 9765 }, { "epoch": 1.33, "grad_norm": 2.894437073153869, "learning_rate": 2.6762340848991342e-06, "loss": 0.1493, "step": 9766 }, { "epoch": 1.33, "grad_norm": 4.015855922307633, "learning_rate": 2.6752587932422175e-06, "loss": 0.1828, "step": 9767 }, { "epoch": 1.33, "grad_norm": 4.184238314012869, "learning_rate": 2.6742836144180815e-06, "loss": 0.1658, "step": 9768 }, { "epoch": 1.33, "grad_norm": 4.814444665843123, "learning_rate": 2.673308548474064e-06, "loss": 0.1521, "step": 9769 }, { "epoch": 1.33, "grad_norm": 3.917931079598829, "learning_rate": 2.672333595457487e-06, "loss": 0.1415, "step": 9770 }, { "epoch": 1.33, "grad_norm": 4.291152967547721, "learning_rate": 2.6713587554156708e-06, "loss": 0.1705, "step": 9771 }, { "epoch": 1.33, "grad_norm": 3.489782277249291, "learning_rate": 2.67038402839593e-06, "loss": 0.1863, "step": 9772 }, { "epoch": 1.33, "grad_norm": 3.4928247565430786, "learning_rate": 2.669409414445574e-06, "loss": 0.167, "step": 9773 }, { "epoch": 1.33, "grad_norm": 3.8379331249174187, "learning_rate": 2.668434913611905e-06, "loss": 0.1843, "step": 9774 }, { "epoch": 1.33, "grad_norm": 3.7489517641173737, "learning_rate": 2.6674605259422227e-06, "loss": 0.1558, "step": 9775 }, { "epoch": 1.33, "grad_norm": 3.0942681585673704, "learning_rate": 2.6664862514838184e-06, "loss": 0.1437, "step": 9776 }, { "epoch": 1.33, "grad_norm": 4.2272361233769375, "learning_rate": 2.6655120902839802e-06, "loss": 0.1682, "step": 9777 }, { "epoch": 1.33, "grad_norm": 4.039500306394681, "learning_rate": 2.6645380423899886e-06, "loss": 0.1507, "step": 9778 }, { "epoch": 1.33, "grad_norm": 4.300458574793609, "learning_rate": 2.66356410784912e-06, "loss": 0.2053, "step": 9779 }, { "epoch": 1.33, "grad_norm": 5.954558053295967, "learning_rate": 2.662590286708645e-06, "loss": 0.167, "step": 9780 }, { "epoch": 1.33, "grad_norm": 3.590533041860657, "learning_rate": 2.6616165790158287e-06, "loss": 0.1883, "step": 9781 }, { "epoch": 1.33, "grad_norm": 2.864740650372252, "learning_rate": 2.6606429848179306e-06, "loss": 0.134, "step": 9782 }, { "epoch": 1.33, "grad_norm": 4.366192006004476, "learning_rate": 2.659669504162202e-06, "loss": 0.1492, "step": 9783 }, { "epoch": 1.33, "grad_norm": 2.9980229678055568, "learning_rate": 2.6586961370958984e-06, "loss": 0.1578, "step": 9784 }, { "epoch": 1.33, "grad_norm": 3.606877339341124, "learning_rate": 2.657722883666255e-06, "loss": 0.1591, "step": 9785 }, { "epoch": 1.33, "grad_norm": 3.47811811582927, "learning_rate": 2.656749743920516e-06, "loss": 0.1559, "step": 9786 }, { "epoch": 1.33, "grad_norm": 3.797349594602242, "learning_rate": 2.655776717905906e-06, "loss": 0.1731, "step": 9787 }, { "epoch": 1.33, "grad_norm": 3.7089424634518284, "learning_rate": 2.654803805669658e-06, "loss": 0.1789, "step": 9788 }, { "epoch": 1.33, "grad_norm": 3.2569786552744198, "learning_rate": 2.6538310072589903e-06, "loss": 0.1384, "step": 9789 }, { "epoch": 1.33, "grad_norm": 3.3304646899193986, "learning_rate": 2.652858322721118e-06, "loss": 0.1648, "step": 9790 }, { "epoch": 1.33, "grad_norm": 3.266983055377033, "learning_rate": 2.6518857521032516e-06, "loss": 0.1235, "step": 9791 }, { "epoch": 1.33, "grad_norm": 3.606547572501652, "learning_rate": 2.6509132954525946e-06, "loss": 0.1328, "step": 9792 }, { "epoch": 1.33, "grad_norm": 3.369121829235077, "learning_rate": 2.6499409528163468e-06, "loss": 0.1621, "step": 9793 }, { "epoch": 1.33, "grad_norm": 3.6844255432907635, "learning_rate": 2.6489687242417004e-06, "loss": 0.1778, "step": 9794 }, { "epoch": 1.33, "grad_norm": 2.784507011118925, "learning_rate": 2.6479966097758447e-06, "loss": 0.1349, "step": 9795 }, { "epoch": 1.33, "grad_norm": 4.090359773593999, "learning_rate": 2.64702460946596e-06, "loss": 0.1327, "step": 9796 }, { "epoch": 1.33, "grad_norm": 3.763282481260912, "learning_rate": 2.6460527233592225e-06, "loss": 0.13, "step": 9797 }, { "epoch": 1.33, "grad_norm": 3.5422673809796668, "learning_rate": 2.645080951502809e-06, "loss": 0.1924, "step": 9798 }, { "epoch": 1.33, "grad_norm": 3.283919385570894, "learning_rate": 2.6441092939438763e-06, "loss": 0.1548, "step": 9799 }, { "epoch": 1.33, "grad_norm": 3.593240735826599, "learning_rate": 2.643137750729593e-06, "loss": 0.1465, "step": 9800 }, { "epoch": 1.33, "grad_norm": 3.6214643327058975, "learning_rate": 2.642166321907105e-06, "loss": 0.1394, "step": 9801 }, { "epoch": 1.33, "grad_norm": 4.088833366462343, "learning_rate": 2.641195007523568e-06, "loss": 0.194, "step": 9802 }, { "epoch": 1.33, "grad_norm": 3.6660278154142407, "learning_rate": 2.640223807626123e-06, "loss": 0.1682, "step": 9803 }, { "epoch": 1.33, "grad_norm": 11.883883820858891, "learning_rate": 2.6392527222619078e-06, "loss": 0.1548, "step": 9804 }, { "epoch": 1.33, "grad_norm": 3.6880852289438457, "learning_rate": 2.6382817514780546e-06, "loss": 0.1761, "step": 9805 }, { "epoch": 1.33, "grad_norm": 3.6838667878475766, "learning_rate": 2.6373108953216895e-06, "loss": 0.1969, "step": 9806 }, { "epoch": 1.33, "grad_norm": 2.6473166315551313, "learning_rate": 2.636340153839935e-06, "loss": 0.1455, "step": 9807 }, { "epoch": 1.33, "grad_norm": 3.3019538400004302, "learning_rate": 2.6353695270799064e-06, "loss": 0.1733, "step": 9808 }, { "epoch": 1.33, "grad_norm": 3.01507641217303, "learning_rate": 2.63439901508871e-06, "loss": 0.1606, "step": 9809 }, { "epoch": 1.33, "grad_norm": 3.5265378495247472, "learning_rate": 2.633428617913458e-06, "loss": 0.1783, "step": 9810 }, { "epoch": 1.33, "grad_norm": 3.129319622642661, "learning_rate": 2.63245833560124e-06, "loss": 0.1409, "step": 9811 }, { "epoch": 1.33, "grad_norm": 3.814444898775483, "learning_rate": 2.631488168199159e-06, "loss": 0.1405, "step": 9812 }, { "epoch": 1.33, "grad_norm": 2.83995723198981, "learning_rate": 2.6305181157542924e-06, "loss": 0.1136, "step": 9813 }, { "epoch": 1.33, "grad_norm": 2.7947469839610775, "learning_rate": 2.629548178313732e-06, "loss": 0.1245, "step": 9814 }, { "epoch": 1.34, "grad_norm": 2.9063595014365182, "learning_rate": 2.6285783559245463e-06, "loss": 0.1309, "step": 9815 }, { "epoch": 1.34, "grad_norm": 3.27729367557177, "learning_rate": 2.6276086486338115e-06, "loss": 0.153, "step": 9816 }, { "epoch": 1.34, "grad_norm": 4.336330308130508, "learning_rate": 2.626639056488593e-06, "loss": 0.1806, "step": 9817 }, { "epoch": 1.34, "grad_norm": 5.113835622222904, "learning_rate": 2.625669579535945e-06, "loss": 0.1376, "step": 9818 }, { "epoch": 1.34, "grad_norm": 3.033859747450413, "learning_rate": 2.6247002178229292e-06, "loss": 0.1461, "step": 9819 }, { "epoch": 1.34, "grad_norm": 3.277690993981617, "learning_rate": 2.6237309713965865e-06, "loss": 0.2062, "step": 9820 }, { "epoch": 1.34, "grad_norm": 5.895098632480136, "learning_rate": 2.6227618403039656e-06, "loss": 0.1795, "step": 9821 }, { "epoch": 1.34, "grad_norm": 4.250039203604746, "learning_rate": 2.621792824592103e-06, "loss": 0.1841, "step": 9822 }, { "epoch": 1.34, "grad_norm": 4.579908811307248, "learning_rate": 2.620823924308029e-06, "loss": 0.1691, "step": 9823 }, { "epoch": 1.34, "grad_norm": 2.807142028718095, "learning_rate": 2.6198551394987704e-06, "loss": 0.1483, "step": 9824 }, { "epoch": 1.34, "grad_norm": 4.477800871823432, "learning_rate": 2.6188864702113477e-06, "loss": 0.1814, "step": 9825 }, { "epoch": 1.34, "grad_norm": 3.5111940403812616, "learning_rate": 2.617917916492776e-06, "loss": 0.1384, "step": 9826 }, { "epoch": 1.34, "grad_norm": 3.191845743093558, "learning_rate": 2.616949478390065e-06, "loss": 0.134, "step": 9827 }, { "epoch": 1.34, "grad_norm": 2.8801847300105923, "learning_rate": 2.615981155950216e-06, "loss": 0.1288, "step": 9828 }, { "epoch": 1.34, "grad_norm": 3.282012804921121, "learning_rate": 2.6150129492202326e-06, "loss": 0.1399, "step": 9829 }, { "epoch": 1.34, "grad_norm": 3.5091326704702173, "learning_rate": 2.6140448582471e-06, "loss": 0.1305, "step": 9830 }, { "epoch": 1.34, "grad_norm": 3.05552251688017, "learning_rate": 2.613076883077814e-06, "loss": 0.1385, "step": 9831 }, { "epoch": 1.34, "grad_norm": 5.210771245726726, "learning_rate": 2.612109023759346e-06, "loss": 0.1334, "step": 9832 }, { "epoch": 1.34, "grad_norm": 3.9948316028318507, "learning_rate": 2.6111412803386816e-06, "loss": 0.1611, "step": 9833 }, { "epoch": 1.34, "grad_norm": 3.0644042708783426, "learning_rate": 2.6101736528627807e-06, "loss": 0.1915, "step": 9834 }, { "epoch": 1.34, "grad_norm": 3.441492390316164, "learning_rate": 2.6092061413786158e-06, "loss": 0.1556, "step": 9835 }, { "epoch": 1.34, "grad_norm": 3.788083357622129, "learning_rate": 2.608238745933142e-06, "loss": 0.1523, "step": 9836 }, { "epoch": 1.34, "grad_norm": 3.207055221984269, "learning_rate": 2.6072714665733135e-06, "loss": 0.1707, "step": 9837 }, { "epoch": 1.34, "grad_norm": 3.5105303021481467, "learning_rate": 2.6063043033460777e-06, "loss": 0.1504, "step": 9838 }, { "epoch": 1.34, "grad_norm": 3.476667885034783, "learning_rate": 2.6053372562983757e-06, "loss": 0.2019, "step": 9839 }, { "epoch": 1.34, "grad_norm": 3.4736951707730848, "learning_rate": 2.6043703254771453e-06, "loss": 0.2198, "step": 9840 }, { "epoch": 1.34, "grad_norm": 3.2197363087372914, "learning_rate": 2.6034035109293156e-06, "loss": 0.1326, "step": 9841 }, { "epoch": 1.34, "grad_norm": 2.9785215317953226, "learning_rate": 2.60243681270181e-06, "loss": 0.1262, "step": 9842 }, { "epoch": 1.34, "grad_norm": 3.4528269207117304, "learning_rate": 2.601470230841554e-06, "loss": 0.1651, "step": 9843 }, { "epoch": 1.34, "grad_norm": 4.260759312463259, "learning_rate": 2.6005037653954533e-06, "loss": 0.1875, "step": 9844 }, { "epoch": 1.34, "grad_norm": 3.901119864632295, "learning_rate": 2.5995374164104235e-06, "loss": 0.1617, "step": 9845 }, { "epoch": 1.34, "grad_norm": 3.7886200033109354, "learning_rate": 2.5985711839333593e-06, "loss": 0.208, "step": 9846 }, { "epoch": 1.34, "grad_norm": 3.458387275256299, "learning_rate": 2.597605068011163e-06, "loss": 0.142, "step": 9847 }, { "epoch": 1.34, "grad_norm": 3.074545403758327, "learning_rate": 2.596639068690725e-06, "loss": 0.1607, "step": 9848 }, { "epoch": 1.34, "grad_norm": 3.606496406064435, "learning_rate": 2.595673186018929e-06, "loss": 0.1403, "step": 9849 }, { "epoch": 1.34, "grad_norm": 3.273803437676999, "learning_rate": 2.5947074200426555e-06, "loss": 0.1612, "step": 9850 }, { "epoch": 1.34, "grad_norm": 3.974170198128514, "learning_rate": 2.593741770808778e-06, "loss": 0.16, "step": 9851 }, { "epoch": 1.34, "grad_norm": 2.6739429242245167, "learning_rate": 2.5927762383641657e-06, "loss": 0.132, "step": 9852 }, { "epoch": 1.34, "grad_norm": 3.792781813590713, "learning_rate": 2.591810822755681e-06, "loss": 0.1697, "step": 9853 }, { "epoch": 1.34, "grad_norm": 2.786833767327841, "learning_rate": 2.5908455240301813e-06, "loss": 0.1596, "step": 9854 }, { "epoch": 1.34, "grad_norm": 2.7723325483691004, "learning_rate": 2.5898803422345163e-06, "loss": 0.0837, "step": 9855 }, { "epoch": 1.34, "grad_norm": 3.5394036990034885, "learning_rate": 2.5889152774155346e-06, "loss": 0.1741, "step": 9856 }, { "epoch": 1.34, "grad_norm": 6.063641125510461, "learning_rate": 2.5879503296200736e-06, "loss": 0.1571, "step": 9857 }, { "epoch": 1.34, "grad_norm": 3.498900036596966, "learning_rate": 2.5869854988949696e-06, "loss": 0.1853, "step": 9858 }, { "epoch": 1.34, "grad_norm": 4.1179728873780315, "learning_rate": 2.58602078528705e-06, "loss": 0.2251, "step": 9859 }, { "epoch": 1.34, "grad_norm": 2.7132036662747234, "learning_rate": 2.5850561888431382e-06, "loss": 0.1172, "step": 9860 }, { "epoch": 1.34, "grad_norm": 3.9934609091653384, "learning_rate": 2.5840917096100495e-06, "loss": 0.175, "step": 9861 }, { "epoch": 1.34, "grad_norm": 4.355719508166345, "learning_rate": 2.583127347634601e-06, "loss": 0.1602, "step": 9862 }, { "epoch": 1.34, "grad_norm": 3.4210168784849375, "learning_rate": 2.5821631029635913e-06, "loss": 0.1293, "step": 9863 }, { "epoch": 1.34, "grad_norm": 4.347959102436607, "learning_rate": 2.581198975643828e-06, "loss": 0.166, "step": 9864 }, { "epoch": 1.34, "grad_norm": 4.28912578390331, "learning_rate": 2.5802349657220994e-06, "loss": 0.1976, "step": 9865 }, { "epoch": 1.34, "grad_norm": 3.818431514069011, "learning_rate": 2.5792710732452e-06, "loss": 0.1872, "step": 9866 }, { "epoch": 1.34, "grad_norm": 3.5731893400199466, "learning_rate": 2.5783072982599057e-06, "loss": 0.1682, "step": 9867 }, { "epoch": 1.34, "grad_norm": 4.022750110059192, "learning_rate": 2.5773436408130002e-06, "loss": 0.1355, "step": 9868 }, { "epoch": 1.34, "grad_norm": 4.911499842058928, "learning_rate": 2.5763801009512534e-06, "loss": 0.1755, "step": 9869 }, { "epoch": 1.34, "grad_norm": 4.131161969838432, "learning_rate": 2.575416678721431e-06, "loss": 0.1912, "step": 9870 }, { "epoch": 1.34, "grad_norm": 4.084350548625819, "learning_rate": 2.5744533741702936e-06, "loss": 0.1637, "step": 9871 }, { "epoch": 1.34, "grad_norm": 3.6020491454463146, "learning_rate": 2.573490187344596e-06, "loss": 0.1713, "step": 9872 }, { "epoch": 1.34, "grad_norm": 3.3844196614458095, "learning_rate": 2.572527118291086e-06, "loss": 0.1683, "step": 9873 }, { "epoch": 1.34, "grad_norm": 4.2120742132366065, "learning_rate": 2.571564167056507e-06, "loss": 0.1669, "step": 9874 }, { "epoch": 1.34, "grad_norm": 3.0132972758194265, "learning_rate": 2.5706013336875956e-06, "loss": 0.1454, "step": 9875 }, { "epoch": 1.34, "grad_norm": 3.1465827715850563, "learning_rate": 2.569638618231089e-06, "loss": 0.1365, "step": 9876 }, { "epoch": 1.34, "grad_norm": 4.531113065220299, "learning_rate": 2.5686760207337045e-06, "loss": 0.1841, "step": 9877 }, { "epoch": 1.34, "grad_norm": 5.991220945010895, "learning_rate": 2.5677135412421713e-06, "loss": 0.2212, "step": 9878 }, { "epoch": 1.34, "grad_norm": 3.1173086181097736, "learning_rate": 2.5667511798031954e-06, "loss": 0.1503, "step": 9879 }, { "epoch": 1.34, "grad_norm": 2.991907605700408, "learning_rate": 2.5657889364634915e-06, "loss": 0.1407, "step": 9880 }, { "epoch": 1.34, "grad_norm": 3.0071401174736185, "learning_rate": 2.5648268112697605e-06, "loss": 0.1352, "step": 9881 }, { "epoch": 1.34, "grad_norm": 4.68149406233241, "learning_rate": 2.563864804268701e-06, "loss": 0.1694, "step": 9882 }, { "epoch": 1.34, "grad_norm": 3.4456327343652275, "learning_rate": 2.5629029155070036e-06, "loss": 0.1439, "step": 9883 }, { "epoch": 1.34, "grad_norm": 4.8139421429287985, "learning_rate": 2.5619411450313542e-06, "loss": 0.1701, "step": 9884 }, { "epoch": 1.34, "grad_norm": 3.0310064205260283, "learning_rate": 2.5609794928884337e-06, "loss": 0.1726, "step": 9885 }, { "epoch": 1.34, "grad_norm": 4.330588947572431, "learning_rate": 2.5600179591249154e-06, "loss": 0.1154, "step": 9886 }, { "epoch": 1.34, "grad_norm": 2.849591813640901, "learning_rate": 2.559056543787468e-06, "loss": 0.1461, "step": 9887 }, { "epoch": 1.35, "grad_norm": 3.9044866945129186, "learning_rate": 2.558095246922756e-06, "loss": 0.1736, "step": 9888 }, { "epoch": 1.35, "grad_norm": 2.988159675840357, "learning_rate": 2.5571340685774325e-06, "loss": 0.1582, "step": 9889 }, { "epoch": 1.35, "grad_norm": 3.3570251351507645, "learning_rate": 2.5561730087981563e-06, "loss": 0.1461, "step": 9890 }, { "epoch": 1.35, "grad_norm": 3.069886881755472, "learning_rate": 2.5552120676315647e-06, "loss": 0.1638, "step": 9891 }, { "epoch": 1.35, "grad_norm": 2.865422639122483, "learning_rate": 2.554251245124305e-06, "loss": 0.1453, "step": 9892 }, { "epoch": 1.35, "grad_norm": 4.123809813462452, "learning_rate": 2.5532905413230042e-06, "loss": 0.1985, "step": 9893 }, { "epoch": 1.35, "grad_norm": 3.4745288732930097, "learning_rate": 2.552329956274296e-06, "loss": 0.1111, "step": 9894 }, { "epoch": 1.35, "grad_norm": 2.683969091078941, "learning_rate": 2.551369490024801e-06, "loss": 0.123, "step": 9895 }, { "epoch": 1.35, "grad_norm": 4.395673837478057, "learning_rate": 2.550409142621137e-06, "loss": 0.1748, "step": 9896 }, { "epoch": 1.35, "grad_norm": 3.437241446767395, "learning_rate": 2.5494489141099155e-06, "loss": 0.167, "step": 9897 }, { "epoch": 1.35, "grad_norm": 3.01446302815164, "learning_rate": 2.5484888045377375e-06, "loss": 0.1179, "step": 9898 }, { "epoch": 1.35, "grad_norm": 3.0125972237758267, "learning_rate": 2.547528813951208e-06, "loss": 0.1381, "step": 9899 }, { "epoch": 1.35, "grad_norm": 3.714752448010135, "learning_rate": 2.5465689423969176e-06, "loss": 0.145, "step": 9900 }, { "epoch": 1.35, "grad_norm": 3.313455519168896, "learning_rate": 2.5456091899214554e-06, "loss": 0.1515, "step": 9901 }, { "epoch": 1.35, "grad_norm": 2.9150642847016566, "learning_rate": 2.5446495565714024e-06, "loss": 0.1446, "step": 9902 }, { "epoch": 1.35, "grad_norm": 3.375033465578935, "learning_rate": 2.543690042393337e-06, "loss": 0.1144, "step": 9903 }, { "epoch": 1.35, "grad_norm": 3.1749054092400004, "learning_rate": 2.542730647433829e-06, "loss": 0.1647, "step": 9904 }, { "epoch": 1.35, "grad_norm": 3.1499532213962738, "learning_rate": 2.541771371739442e-06, "loss": 0.2044, "step": 9905 }, { "epoch": 1.35, "grad_norm": 3.8280995558781896, "learning_rate": 2.5408122153567362e-06, "loss": 0.1826, "step": 9906 }, { "epoch": 1.35, "grad_norm": 3.503580159277526, "learning_rate": 2.539853178332265e-06, "loss": 0.1594, "step": 9907 }, { "epoch": 1.35, "grad_norm": 3.460534358114501, "learning_rate": 2.538894260712573e-06, "loss": 0.1566, "step": 9908 }, { "epoch": 1.35, "grad_norm": 3.5868939247091207, "learning_rate": 2.5379354625442093e-06, "loss": 0.1844, "step": 9909 }, { "epoch": 1.35, "grad_norm": 3.915437089470869, "learning_rate": 2.5369767838737007e-06, "loss": 0.2221, "step": 9910 }, { "epoch": 1.35, "grad_norm": 3.3200818851267604, "learning_rate": 2.536018224747585e-06, "loss": 0.1767, "step": 9911 }, { "epoch": 1.35, "grad_norm": 4.623138063151779, "learning_rate": 2.5350597852123798e-06, "loss": 0.2059, "step": 9912 }, { "epoch": 1.35, "grad_norm": 2.6459986501138046, "learning_rate": 2.534101465314608e-06, "loss": 0.1603, "step": 9913 }, { "epoch": 1.35, "grad_norm": 3.883874177623336, "learning_rate": 2.5331432651007813e-06, "loss": 0.1715, "step": 9914 }, { "epoch": 1.35, "grad_norm": 2.908200472969693, "learning_rate": 2.5321851846174063e-06, "loss": 0.1605, "step": 9915 }, { "epoch": 1.35, "grad_norm": 3.5602531487272375, "learning_rate": 2.5312272239109837e-06, "loss": 0.1724, "step": 9916 }, { "epoch": 1.35, "grad_norm": 4.371110821533881, "learning_rate": 2.530269383028009e-06, "loss": 0.1571, "step": 9917 }, { "epoch": 1.35, "grad_norm": 3.361908219027599, "learning_rate": 2.529311662014972e-06, "loss": 0.1725, "step": 9918 }, { "epoch": 1.35, "grad_norm": 3.845663946361894, "learning_rate": 2.5283540609183553e-06, "loss": 0.2044, "step": 9919 }, { "epoch": 1.35, "grad_norm": 2.82313691175172, "learning_rate": 2.5273965797846385e-06, "loss": 0.1727, "step": 9920 }, { "epoch": 1.35, "grad_norm": 3.2106749373271612, "learning_rate": 2.5264392186602916e-06, "loss": 0.1217, "step": 9921 }, { "epoch": 1.35, "grad_norm": 3.3564913395745526, "learning_rate": 2.5254819775917795e-06, "loss": 0.1128, "step": 9922 }, { "epoch": 1.35, "grad_norm": 3.307011812628694, "learning_rate": 2.5245248566255682e-06, "loss": 0.1486, "step": 9923 }, { "epoch": 1.35, "grad_norm": 5.515464637002597, "learning_rate": 2.5235678558081044e-06, "loss": 0.1742, "step": 9924 }, { "epoch": 1.35, "grad_norm": 3.3488491858733416, "learning_rate": 2.5226109751858453e-06, "loss": 0.1827, "step": 9925 }, { "epoch": 1.35, "grad_norm": 3.3023924213760085, "learning_rate": 2.5216542148052247e-06, "loss": 0.1856, "step": 9926 }, { "epoch": 1.35, "grad_norm": 2.9468857444550514, "learning_rate": 2.5206975747126873e-06, "loss": 0.1662, "step": 9927 }, { "epoch": 1.35, "grad_norm": 4.088656424309194, "learning_rate": 2.5197410549546598e-06, "loss": 0.1411, "step": 9928 }, { "epoch": 1.35, "grad_norm": 4.145431678839631, "learning_rate": 2.5187846555775687e-06, "loss": 0.109, "step": 9929 }, { "epoch": 1.35, "grad_norm": 5.541744361967899, "learning_rate": 2.5178283766278334e-06, "loss": 0.1466, "step": 9930 }, { "epoch": 1.35, "grad_norm": 2.817689428359878, "learning_rate": 2.5168722181518677e-06, "loss": 0.1765, "step": 9931 }, { "epoch": 1.35, "grad_norm": 3.913509413847282, "learning_rate": 2.51591618019608e-06, "loss": 0.1971, "step": 9932 }, { "epoch": 1.35, "grad_norm": 3.439194499498096, "learning_rate": 2.51496026280687e-06, "loss": 0.1627, "step": 9933 }, { "epoch": 1.35, "grad_norm": 3.528512998536025, "learning_rate": 2.5140044660306362e-06, "loss": 0.1535, "step": 9934 }, { "epoch": 1.35, "grad_norm": 3.0996485802817473, "learning_rate": 2.513048789913768e-06, "loss": 0.1392, "step": 9935 }, { "epoch": 1.35, "grad_norm": 3.632894684815202, "learning_rate": 2.5120932345026483e-06, "loss": 0.188, "step": 9936 }, { "epoch": 1.35, "grad_norm": 4.08736341738178, "learning_rate": 2.511137799843658e-06, "loss": 0.1716, "step": 9937 }, { "epoch": 1.35, "grad_norm": 3.2904809449823302, "learning_rate": 2.510182485983168e-06, "loss": 0.1784, "step": 9938 }, { "epoch": 1.35, "grad_norm": 4.1811911343865535, "learning_rate": 2.5092272929675454e-06, "loss": 0.2048, "step": 9939 }, { "epoch": 1.35, "grad_norm": 3.4074765308563606, "learning_rate": 2.508272220843151e-06, "loss": 0.1637, "step": 9940 }, { "epoch": 1.35, "grad_norm": 3.263176466274988, "learning_rate": 2.5073172696563387e-06, "loss": 0.1588, "step": 9941 }, { "epoch": 1.35, "grad_norm": 3.823129348676097, "learning_rate": 2.506362439453463e-06, "loss": 0.171, "step": 9942 }, { "epoch": 1.35, "grad_norm": 3.1103570969271592, "learning_rate": 2.5054077302808592e-06, "loss": 0.1753, "step": 9943 }, { "epoch": 1.35, "grad_norm": 3.3672641330977466, "learning_rate": 2.504453142184873e-06, "loss": 0.1379, "step": 9944 }, { "epoch": 1.35, "grad_norm": 4.531632718628713, "learning_rate": 2.5034986752118275e-06, "loss": 0.1474, "step": 9945 }, { "epoch": 1.35, "grad_norm": 3.8694288044258816, "learning_rate": 2.5025443294080543e-06, "loss": 0.1591, "step": 9946 }, { "epoch": 1.35, "grad_norm": 6.659685621509158, "learning_rate": 2.5015901048198716e-06, "loss": 0.1602, "step": 9947 }, { "epoch": 1.35, "grad_norm": 4.297796342079544, "learning_rate": 2.500636001493593e-06, "loss": 0.1824, "step": 9948 }, { "epoch": 1.35, "grad_norm": 3.075865486697603, "learning_rate": 2.4996820194755273e-06, "loss": 0.1852, "step": 9949 }, { "epoch": 1.35, "grad_norm": 3.255689735130472, "learning_rate": 2.498728158811975e-06, "loss": 0.1747, "step": 9950 }, { "epoch": 1.35, "grad_norm": 3.9067606679462927, "learning_rate": 2.4977744195492336e-06, "loss": 0.2055, "step": 9951 }, { "epoch": 1.35, "grad_norm": 3.8390597450860247, "learning_rate": 2.4968208017335936e-06, "loss": 0.1783, "step": 9952 }, { "epoch": 1.35, "grad_norm": 3.7923038739547175, "learning_rate": 2.4958673054113387e-06, "loss": 0.1444, "step": 9953 }, { "epoch": 1.35, "grad_norm": 4.067610729213518, "learning_rate": 2.494913930628748e-06, "loss": 0.185, "step": 9954 }, { "epoch": 1.35, "grad_norm": 3.7769284641287166, "learning_rate": 2.493960677432091e-06, "loss": 0.1435, "step": 9955 }, { "epoch": 1.35, "grad_norm": 3.17049621366568, "learning_rate": 2.4930075458676424e-06, "loss": 0.1494, "step": 9956 }, { "epoch": 1.35, "grad_norm": 2.2343940305595305, "learning_rate": 2.4920545359816533e-06, "loss": 0.1566, "step": 9957 }, { "epoch": 1.35, "grad_norm": 3.4326818281073663, "learning_rate": 2.4911016478203877e-06, "loss": 0.1415, "step": 9958 }, { "epoch": 1.35, "grad_norm": 6.299695542032369, "learning_rate": 2.4901488814300855e-06, "loss": 0.1724, "step": 9959 }, { "epoch": 1.35, "grad_norm": 3.200651647922709, "learning_rate": 2.489196236856997e-06, "loss": 0.1802, "step": 9960 }, { "epoch": 1.35, "grad_norm": 6.2144441931066385, "learning_rate": 2.4882437141473576e-06, "loss": 0.1828, "step": 9961 }, { "epoch": 1.36, "grad_norm": 3.198277391289283, "learning_rate": 2.487291313347397e-06, "loss": 0.1603, "step": 9962 }, { "epoch": 1.36, "grad_norm": 3.5370939488031445, "learning_rate": 2.486339034503342e-06, "loss": 0.1157, "step": 9963 }, { "epoch": 1.36, "grad_norm": 2.812562420524277, "learning_rate": 2.4853868776614117e-06, "loss": 0.1664, "step": 9964 }, { "epoch": 1.36, "grad_norm": 3.0909463735612075, "learning_rate": 2.484434842867819e-06, "loss": 0.1723, "step": 9965 }, { "epoch": 1.36, "grad_norm": 3.9456644387809834, "learning_rate": 2.483482930168772e-06, "loss": 0.2292, "step": 9966 }, { "epoch": 1.36, "grad_norm": 3.433721982577075, "learning_rate": 2.4825311396104727e-06, "loss": 0.2022, "step": 9967 }, { "epoch": 1.36, "grad_norm": 3.0444721864633695, "learning_rate": 2.481579471239117e-06, "loss": 0.1435, "step": 9968 }, { "epoch": 1.36, "grad_norm": 2.9778080966427334, "learning_rate": 2.4806279251008914e-06, "loss": 0.1433, "step": 9969 }, { "epoch": 1.36, "grad_norm": 3.603230528434105, "learning_rate": 2.479676501241987e-06, "loss": 0.2179, "step": 9970 }, { "epoch": 1.36, "grad_norm": 3.244433252326145, "learning_rate": 2.4787251997085737e-06, "loss": 0.1876, "step": 9971 }, { "epoch": 1.36, "grad_norm": 6.588304518368366, "learning_rate": 2.477774020546831e-06, "loss": 0.14, "step": 9972 }, { "epoch": 1.36, "grad_norm": 3.2623607407956063, "learning_rate": 2.4768229638029175e-06, "loss": 0.126, "step": 9973 }, { "epoch": 1.36, "grad_norm": 3.073979636088751, "learning_rate": 2.4758720295229987e-06, "loss": 0.1424, "step": 9974 }, { "epoch": 1.36, "grad_norm": 2.650406197303741, "learning_rate": 2.474921217753228e-06, "loss": 0.156, "step": 9975 }, { "epoch": 1.36, "grad_norm": 3.4782533497343513, "learning_rate": 2.473970528539752e-06, "loss": 0.1564, "step": 9976 }, { "epoch": 1.36, "grad_norm": 4.061479921960192, "learning_rate": 2.473019961928716e-06, "loss": 0.1509, "step": 9977 }, { "epoch": 1.36, "grad_norm": 3.7294970655722897, "learning_rate": 2.47206951796625e-06, "loss": 0.137, "step": 9978 }, { "epoch": 1.36, "grad_norm": 4.609196985753529, "learning_rate": 2.471119196698491e-06, "loss": 0.1865, "step": 9979 }, { "epoch": 1.36, "grad_norm": 4.047852982864884, "learning_rate": 2.470168998171561e-06, "loss": 0.1544, "step": 9980 }, { "epoch": 1.36, "grad_norm": 3.456599351291212, "learning_rate": 2.4692189224315784e-06, "loss": 0.1496, "step": 9981 }, { "epoch": 1.36, "grad_norm": 3.4551958634805313, "learning_rate": 2.4682689695246557e-06, "loss": 0.1685, "step": 9982 }, { "epoch": 1.36, "grad_norm": 3.4552264046810253, "learning_rate": 2.4673191394968997e-06, "loss": 0.1621, "step": 9983 }, { "epoch": 1.36, "grad_norm": 3.2434356287464996, "learning_rate": 2.466369432394411e-06, "loss": 0.1569, "step": 9984 }, { "epoch": 1.36, "grad_norm": 4.027408922419382, "learning_rate": 2.4654198482632842e-06, "loss": 0.1694, "step": 9985 }, { "epoch": 1.36, "grad_norm": 3.7680515778695582, "learning_rate": 2.4644703871496077e-06, "loss": 0.1821, "step": 9986 }, { "epoch": 1.36, "grad_norm": 2.9313563625917025, "learning_rate": 2.4635210490994648e-06, "loss": 0.1561, "step": 9987 }, { "epoch": 1.36, "grad_norm": 3.2061190126537054, "learning_rate": 2.4625718341589293e-06, "loss": 0.1359, "step": 9988 }, { "epoch": 1.36, "grad_norm": 4.086989472820221, "learning_rate": 2.461622742374079e-06, "loss": 0.1471, "step": 9989 }, { "epoch": 1.36, "grad_norm": 3.5837101082297025, "learning_rate": 2.4606737737909696e-06, "loss": 0.1258, "step": 9990 }, { "epoch": 1.36, "grad_norm": 3.183838432825203, "learning_rate": 2.459724928455669e-06, "loss": 0.1469, "step": 9991 }, { "epoch": 1.36, "grad_norm": 2.682440363660665, "learning_rate": 2.458776206414221e-06, "loss": 0.1263, "step": 9992 }, { "epoch": 1.36, "grad_norm": 3.405656506419253, "learning_rate": 2.457827607712679e-06, "loss": 0.1715, "step": 9993 }, { "epoch": 1.36, "grad_norm": 3.8158433234380227, "learning_rate": 2.456879132397082e-06, "loss": 0.1656, "step": 9994 }, { "epoch": 1.36, "grad_norm": 3.1051414147309204, "learning_rate": 2.455930780513464e-06, "loss": 0.1288, "step": 9995 }, { "epoch": 1.36, "grad_norm": 3.626810875928087, "learning_rate": 2.4549825521078544e-06, "loss": 0.1292, "step": 9996 }, { "epoch": 1.36, "grad_norm": 3.3971531877838355, "learning_rate": 2.4540344472262766e-06, "loss": 0.1726, "step": 9997 }, { "epoch": 1.36, "grad_norm": 4.129435038568005, "learning_rate": 2.4530864659147467e-06, "loss": 0.1826, "step": 9998 }, { "epoch": 1.36, "grad_norm": 3.6807465395175045, "learning_rate": 2.452138608219276e-06, "loss": 0.2162, "step": 9999 }, { "epoch": 1.36, "grad_norm": 3.21762409216414, "learning_rate": 2.451190874185869e-06, "loss": 0.1712, "step": 10000 }, { "epoch": 1.36, "grad_norm": 4.21824880610322, "learning_rate": 2.450243263860525e-06, "loss": 0.171, "step": 10001 }, { "epoch": 1.36, "grad_norm": 3.827549022302126, "learning_rate": 2.4492957772892345e-06, "loss": 0.1614, "step": 10002 }, { "epoch": 1.36, "grad_norm": 3.1089797713374083, "learning_rate": 2.4483484145179903e-06, "loss": 0.1396, "step": 10003 }, { "epoch": 1.36, "grad_norm": 4.3741048253384145, "learning_rate": 2.4474011755927657e-06, "loss": 0.1875, "step": 10004 }, { "epoch": 1.36, "grad_norm": 4.789136499904224, "learning_rate": 2.4464540605595428e-06, "loss": 0.1366, "step": 10005 }, { "epoch": 1.36, "grad_norm": 3.2655077300966595, "learning_rate": 2.4455070694642836e-06, "loss": 0.1535, "step": 10006 }, { "epoch": 1.36, "grad_norm": 3.885891143696378, "learning_rate": 2.4445602023529558e-06, "loss": 0.1875, "step": 10007 }, { "epoch": 1.36, "grad_norm": 3.2706386992314527, "learning_rate": 2.443613459271515e-06, "loss": 0.1745, "step": 10008 }, { "epoch": 1.36, "grad_norm": 3.5144463409229694, "learning_rate": 2.442666840265911e-06, "loss": 0.1409, "step": 10009 }, { "epoch": 1.36, "grad_norm": 3.5229127754132583, "learning_rate": 2.4417203453820892e-06, "loss": 0.162, "step": 10010 }, { "epoch": 1.36, "grad_norm": 3.721897206269361, "learning_rate": 2.4407739746659888e-06, "loss": 0.2094, "step": 10011 }, { "epoch": 1.36, "grad_norm": 3.3412568558361784, "learning_rate": 2.439827728163542e-06, "loss": 0.1937, "step": 10012 }, { "epoch": 1.36, "grad_norm": 2.8565428999095244, "learning_rate": 2.438881605920675e-06, "loss": 0.1188, "step": 10013 }, { "epoch": 1.36, "grad_norm": 3.189394045012762, "learning_rate": 2.4379356079833067e-06, "loss": 0.1875, "step": 10014 }, { "epoch": 1.36, "grad_norm": 4.4855487154273, "learning_rate": 2.436989734397358e-06, "loss": 0.1846, "step": 10015 }, { "epoch": 1.36, "grad_norm": 3.9526939416130036, "learning_rate": 2.436043985208732e-06, "loss": 0.1839, "step": 10016 }, { "epoch": 1.36, "grad_norm": 3.8525249601818117, "learning_rate": 2.4350983604633323e-06, "loss": 0.171, "step": 10017 }, { "epoch": 1.36, "grad_norm": 4.002017952645207, "learning_rate": 2.434152860207055e-06, "loss": 0.1644, "step": 10018 }, { "epoch": 1.36, "grad_norm": 3.0528098178931815, "learning_rate": 2.4332074844857923e-06, "loss": 0.1605, "step": 10019 }, { "epoch": 1.36, "grad_norm": 4.2485502537370445, "learning_rate": 2.4322622333454265e-06, "loss": 0.163, "step": 10020 }, { "epoch": 1.36, "grad_norm": 2.7598809808673574, "learning_rate": 2.431317106831836e-06, "loss": 0.1727, "step": 10021 }, { "epoch": 1.36, "grad_norm": 3.685500606361363, "learning_rate": 2.4303721049908973e-06, "loss": 0.1654, "step": 10022 }, { "epoch": 1.36, "grad_norm": 2.6255685756339497, "learning_rate": 2.42942722786847e-06, "loss": 0.1599, "step": 10023 }, { "epoch": 1.36, "grad_norm": 4.259959091273431, "learning_rate": 2.428482475510422e-06, "loss": 0.1807, "step": 10024 }, { "epoch": 1.36, "grad_norm": 3.3717873969642405, "learning_rate": 2.427537847962599e-06, "loss": 0.163, "step": 10025 }, { "epoch": 1.36, "grad_norm": 3.593613439353338, "learning_rate": 2.426593345270855e-06, "loss": 0.1769, "step": 10026 }, { "epoch": 1.36, "grad_norm": 2.849182270536071, "learning_rate": 2.425648967481031e-06, "loss": 0.1289, "step": 10027 }, { "epoch": 1.36, "grad_norm": 4.175756829126428, "learning_rate": 2.4247047146389616e-06, "loss": 0.1474, "step": 10028 }, { "epoch": 1.36, "grad_norm": 3.5366639025909774, "learning_rate": 2.4237605867904786e-06, "loss": 0.1557, "step": 10029 }, { "epoch": 1.36, "grad_norm": 2.475031827690541, "learning_rate": 2.4228165839814037e-06, "loss": 0.1516, "step": 10030 }, { "epoch": 1.36, "grad_norm": 4.238221745373112, "learning_rate": 2.421872706257556e-06, "loss": 0.1629, "step": 10031 }, { "epoch": 1.36, "grad_norm": 6.114343242610761, "learning_rate": 2.4209289536647467e-06, "loss": 0.2117, "step": 10032 }, { "epoch": 1.36, "grad_norm": 3.453652305625011, "learning_rate": 2.419985326248782e-06, "loss": 0.1896, "step": 10033 }, { "epoch": 1.36, "grad_norm": 3.5828295260565963, "learning_rate": 2.419041824055461e-06, "loss": 0.1618, "step": 10034 }, { "epoch": 1.37, "grad_norm": 6.652655447344687, "learning_rate": 2.4180984471305754e-06, "loss": 0.1315, "step": 10035 }, { "epoch": 1.37, "grad_norm": 4.054707268163672, "learning_rate": 2.417155195519918e-06, "loss": 0.1393, "step": 10036 }, { "epoch": 1.37, "grad_norm": 3.652990459965185, "learning_rate": 2.4162120692692623e-06, "loss": 0.1555, "step": 10037 }, { "epoch": 1.37, "grad_norm": 3.998872282568315, "learning_rate": 2.4152690684243923e-06, "loss": 0.1412, "step": 10038 }, { "epoch": 1.37, "grad_norm": 3.7359509738625483, "learning_rate": 2.414326193031069e-06, "loss": 0.1487, "step": 10039 }, { "epoch": 1.37, "grad_norm": 4.1663591495142285, "learning_rate": 2.4133834431350605e-06, "loss": 0.1491, "step": 10040 }, { "epoch": 1.37, "grad_norm": 3.9723753372276374, "learning_rate": 2.4124408187821223e-06, "loss": 0.1387, "step": 10041 }, { "epoch": 1.37, "grad_norm": 3.1952928965589855, "learning_rate": 2.4114983200180053e-06, "loss": 0.1411, "step": 10042 }, { "epoch": 1.37, "grad_norm": 3.013780132624332, "learning_rate": 2.410555946888455e-06, "loss": 0.1277, "step": 10043 }, { "epoch": 1.37, "grad_norm": 2.9457240703775303, "learning_rate": 2.409613699439208e-06, "loss": 0.1317, "step": 10044 }, { "epoch": 1.37, "grad_norm": 6.976365797383033, "learning_rate": 2.4086715777159995e-06, "loss": 0.1909, "step": 10045 }, { "epoch": 1.37, "grad_norm": 4.215145854665284, "learning_rate": 2.4077295817645543e-06, "loss": 0.1986, "step": 10046 }, { "epoch": 1.37, "grad_norm": 4.670168672063735, "learning_rate": 2.406787711630591e-06, "loss": 0.151, "step": 10047 }, { "epoch": 1.37, "grad_norm": 3.03110338619435, "learning_rate": 2.40584596735983e-06, "loss": 0.1628, "step": 10048 }, { "epoch": 1.37, "grad_norm": 3.667872124985993, "learning_rate": 2.404904348997972e-06, "loss": 0.177, "step": 10049 }, { "epoch": 1.37, "grad_norm": 3.2379816808456545, "learning_rate": 2.403962856590726e-06, "loss": 0.1586, "step": 10050 }, { "epoch": 1.37, "grad_norm": 4.233743805015168, "learning_rate": 2.4030214901837807e-06, "loss": 0.1549, "step": 10051 }, { "epoch": 1.37, "grad_norm": 3.4312385330325648, "learning_rate": 2.4020802498228333e-06, "loss": 0.1789, "step": 10052 }, { "epoch": 1.37, "grad_norm": 4.599201867038829, "learning_rate": 2.4011391355535605e-06, "loss": 0.2003, "step": 10053 }, { "epoch": 1.37, "grad_norm": 4.38283836180049, "learning_rate": 2.400198147421645e-06, "loss": 0.205, "step": 10054 }, { "epoch": 1.37, "grad_norm": 4.702263455021335, "learning_rate": 2.399257285472756e-06, "loss": 0.1478, "step": 10055 }, { "epoch": 1.37, "grad_norm": 3.935795896795663, "learning_rate": 2.39831654975256e-06, "loss": 0.186, "step": 10056 }, { "epoch": 1.37, "grad_norm": 4.417224761945141, "learning_rate": 2.3973759403067175e-06, "loss": 0.1825, "step": 10057 }, { "epoch": 1.37, "grad_norm": 3.168267110721807, "learning_rate": 2.3964354571808756e-06, "loss": 0.2073, "step": 10058 }, { "epoch": 1.37, "grad_norm": 2.5926084516813637, "learning_rate": 2.395495100420687e-06, "loss": 0.1683, "step": 10059 }, { "epoch": 1.37, "grad_norm": 3.7176695792968815, "learning_rate": 2.3945548700717914e-06, "loss": 0.1716, "step": 10060 }, { "epoch": 1.37, "grad_norm": 4.351801179812524, "learning_rate": 2.393614766179822e-06, "loss": 0.1588, "step": 10061 }, { "epoch": 1.37, "grad_norm": 2.648618281272365, "learning_rate": 2.3926747887904084e-06, "loss": 0.1394, "step": 10062 }, { "epoch": 1.37, "grad_norm": 3.4582343079631714, "learning_rate": 2.3917349379491727e-06, "loss": 0.1441, "step": 10063 }, { "epoch": 1.37, "grad_norm": 3.472949435558887, "learning_rate": 2.3907952137017316e-06, "loss": 0.1627, "step": 10064 }, { "epoch": 1.37, "grad_norm": 3.5364971023271314, "learning_rate": 2.3898556160936947e-06, "loss": 0.1398, "step": 10065 }, { "epoch": 1.37, "grad_norm": 3.1462613840409324, "learning_rate": 2.3889161451706646e-06, "loss": 0.1692, "step": 10066 }, { "epoch": 1.37, "grad_norm": 2.7743745118130816, "learning_rate": 2.3879768009782434e-06, "loss": 0.123, "step": 10067 }, { "epoch": 1.37, "grad_norm": 3.4291946492415497, "learning_rate": 2.3870375835620168e-06, "loss": 0.1377, "step": 10068 }, { "epoch": 1.37, "grad_norm": 3.9763948345374023, "learning_rate": 2.3860984929675775e-06, "loss": 0.1643, "step": 10069 }, { "epoch": 1.37, "grad_norm": 3.08842411913925, "learning_rate": 2.3851595292404965e-06, "loss": 0.1796, "step": 10070 }, { "epoch": 1.37, "grad_norm": 3.761618052031581, "learning_rate": 2.3842206924263554e-06, "loss": 0.1365, "step": 10071 }, { "epoch": 1.37, "grad_norm": 3.4940291646827584, "learning_rate": 2.3832819825707136e-06, "loss": 0.15, "step": 10072 }, { "epoch": 1.37, "grad_norm": 3.854850434879807, "learning_rate": 2.3823433997191374e-06, "loss": 0.1413, "step": 10073 }, { "epoch": 1.37, "grad_norm": 4.4815317534742425, "learning_rate": 2.381404943917179e-06, "loss": 0.1742, "step": 10074 }, { "epoch": 1.37, "grad_norm": 3.241269769139351, "learning_rate": 2.380466615210389e-06, "loss": 0.153, "step": 10075 }, { "epoch": 1.37, "grad_norm": 3.9067357520109187, "learning_rate": 2.379528413644307e-06, "loss": 0.1427, "step": 10076 }, { "epoch": 1.37, "grad_norm": 3.3209982227542683, "learning_rate": 2.3785903392644714e-06, "loss": 0.1724, "step": 10077 }, { "epoch": 1.37, "grad_norm": 3.471366150908971, "learning_rate": 2.3776523921164113e-06, "loss": 0.139, "step": 10078 }, { "epoch": 1.37, "grad_norm": 4.4974135312795465, "learning_rate": 2.376714572245651e-06, "loss": 0.1472, "step": 10079 }, { "epoch": 1.37, "grad_norm": 2.9313259978141626, "learning_rate": 2.3757768796977056e-06, "loss": 0.1734, "step": 10080 }, { "epoch": 1.37, "grad_norm": 4.059281594781233, "learning_rate": 2.3748393145180926e-06, "loss": 0.1933, "step": 10081 }, { "epoch": 1.37, "grad_norm": 2.853577271848897, "learning_rate": 2.37390187675231e-06, "loss": 0.1275, "step": 10082 }, { "epoch": 1.37, "grad_norm": 4.1755466141526005, "learning_rate": 2.3729645664458637e-06, "loss": 0.1784, "step": 10083 }, { "epoch": 1.37, "grad_norm": 3.412931260682345, "learning_rate": 2.3720273836442404e-06, "loss": 0.1477, "step": 10084 }, { "epoch": 1.37, "grad_norm": 4.249129785637104, "learning_rate": 2.3710903283929315e-06, "loss": 0.1736, "step": 10085 }, { "epoch": 1.37, "grad_norm": 3.357698297569351, "learning_rate": 2.370153400737416e-06, "loss": 0.1612, "step": 10086 }, { "epoch": 1.37, "grad_norm": 3.6443653083249257, "learning_rate": 2.3692166007231686e-06, "loss": 0.1442, "step": 10087 }, { "epoch": 1.37, "grad_norm": 2.6617255992921454, "learning_rate": 2.368279928395657e-06, "loss": 0.1214, "step": 10088 }, { "epoch": 1.37, "grad_norm": 5.493917532615942, "learning_rate": 2.3673433838003433e-06, "loss": 0.1442, "step": 10089 }, { "epoch": 1.37, "grad_norm": 3.012211830226271, "learning_rate": 2.3664069669826836e-06, "loss": 0.1547, "step": 10090 }, { "epoch": 1.37, "grad_norm": 4.352159163020035, "learning_rate": 2.365470677988127e-06, "loss": 0.1773, "step": 10091 }, { "epoch": 1.37, "grad_norm": 3.644706109539433, "learning_rate": 2.364534516862117e-06, "loss": 0.1787, "step": 10092 }, { "epoch": 1.37, "grad_norm": 4.3620238922615675, "learning_rate": 2.3635984836500912e-06, "loss": 0.1942, "step": 10093 }, { "epoch": 1.37, "grad_norm": 3.9573184377339845, "learning_rate": 2.362662578397478e-06, "loss": 0.1839, "step": 10094 }, { "epoch": 1.37, "grad_norm": 3.714086129234584, "learning_rate": 2.3617268011497087e-06, "loss": 0.165, "step": 10095 }, { "epoch": 1.37, "grad_norm": 3.4084395606958586, "learning_rate": 2.3607911519521947e-06, "loss": 0.1776, "step": 10096 }, { "epoch": 1.37, "grad_norm": 3.4124554946974865, "learning_rate": 2.359855630850352e-06, "loss": 0.1439, "step": 10097 }, { "epoch": 1.37, "grad_norm": 3.560535903500575, "learning_rate": 2.3589202378895854e-06, "loss": 0.1567, "step": 10098 }, { "epoch": 1.37, "grad_norm": 4.05734788419096, "learning_rate": 2.3579849731152936e-06, "loss": 0.1751, "step": 10099 }, { "epoch": 1.37, "grad_norm": 2.8689968707738265, "learning_rate": 2.3570498365728755e-06, "loss": 0.1239, "step": 10100 }, { "epoch": 1.37, "grad_norm": 3.5483804255413585, "learning_rate": 2.356114828307711e-06, "loss": 0.1977, "step": 10101 }, { "epoch": 1.37, "grad_norm": 2.8759966054158634, "learning_rate": 2.3551799483651894e-06, "loss": 0.1553, "step": 10102 }, { "epoch": 1.37, "grad_norm": 4.708635695692045, "learning_rate": 2.3542451967906776e-06, "loss": 0.1892, "step": 10103 }, { "epoch": 1.37, "grad_norm": 2.864481687952826, "learning_rate": 2.3533105736295527e-06, "loss": 0.1275, "step": 10104 }, { "epoch": 1.37, "grad_norm": 3.588183799700471, "learning_rate": 2.352376078927168e-06, "loss": 0.1561, "step": 10105 }, { "epoch": 1.37, "grad_norm": 2.6815932655645005, "learning_rate": 2.3514417127288873e-06, "loss": 0.1306, "step": 10106 }, { "epoch": 1.37, "grad_norm": 4.344999103838945, "learning_rate": 2.3505074750800585e-06, "loss": 0.1745, "step": 10107 }, { "epoch": 1.37, "grad_norm": 3.9466995319610256, "learning_rate": 2.349573366026024e-06, "loss": 0.1809, "step": 10108 }, { "epoch": 1.38, "grad_norm": 3.2160705138077708, "learning_rate": 2.3486393856121222e-06, "loss": 0.1842, "step": 10109 }, { "epoch": 1.38, "grad_norm": 2.9103468507955994, "learning_rate": 2.3477055338836845e-06, "loss": 0.1151, "step": 10110 }, { "epoch": 1.38, "grad_norm": 3.5490199727220966, "learning_rate": 2.3467718108860364e-06, "loss": 0.1528, "step": 10111 }, { "epoch": 1.38, "grad_norm": 2.4829677917270323, "learning_rate": 2.3458382166644967e-06, "loss": 0.12, "step": 10112 }, { "epoch": 1.38, "grad_norm": 3.6189327411877925, "learning_rate": 2.3449047512643753e-06, "loss": 0.1368, "step": 10113 }, { "epoch": 1.38, "grad_norm": 3.3753259312029895, "learning_rate": 2.3439714147309845e-06, "loss": 0.1412, "step": 10114 }, { "epoch": 1.38, "grad_norm": 3.4575778140221916, "learning_rate": 2.343038207109617e-06, "loss": 0.1722, "step": 10115 }, { "epoch": 1.38, "grad_norm": 2.9710246862037306, "learning_rate": 2.3421051284455744e-06, "loss": 0.1673, "step": 10116 }, { "epoch": 1.38, "grad_norm": 3.206620351568983, "learning_rate": 2.3411721787841363e-06, "loss": 0.169, "step": 10117 }, { "epoch": 1.38, "grad_norm": 3.2909684961572117, "learning_rate": 2.3402393581705894e-06, "loss": 0.1728, "step": 10118 }, { "epoch": 1.38, "grad_norm": 5.688390915976633, "learning_rate": 2.339306666650208e-06, "loss": 0.1871, "step": 10119 }, { "epoch": 1.38, "grad_norm": 4.421742252587976, "learning_rate": 2.3383741042682596e-06, "loss": 0.168, "step": 10120 }, { "epoch": 1.38, "grad_norm": 3.9886043528695296, "learning_rate": 2.3374416710700076e-06, "loss": 0.1538, "step": 10121 }, { "epoch": 1.38, "grad_norm": 3.896127682864408, "learning_rate": 2.3365093671007078e-06, "loss": 0.1803, "step": 10122 }, { "epoch": 1.38, "grad_norm": 4.330235317848253, "learning_rate": 2.3355771924056098e-06, "loss": 0.1439, "step": 10123 }, { "epoch": 1.38, "grad_norm": 3.758125899258881, "learning_rate": 2.334645147029958e-06, "loss": 0.1359, "step": 10124 }, { "epoch": 1.38, "grad_norm": 4.4147181878299735, "learning_rate": 2.3337132310189897e-06, "loss": 0.1706, "step": 10125 }, { "epoch": 1.38, "grad_norm": 3.6708854289600232, "learning_rate": 2.3327814444179357e-06, "loss": 0.1928, "step": 10126 }, { "epoch": 1.38, "grad_norm": 4.081452163186364, "learning_rate": 2.3318497872720193e-06, "loss": 0.1627, "step": 10127 }, { "epoch": 1.38, "grad_norm": 3.6530699501916093, "learning_rate": 2.3309182596264646e-06, "loss": 0.1616, "step": 10128 }, { "epoch": 1.38, "grad_norm": 2.687534310853006, "learning_rate": 2.329986861526476e-06, "loss": 0.1781, "step": 10129 }, { "epoch": 1.38, "grad_norm": 2.865504895701407, "learning_rate": 2.3290555930172676e-06, "loss": 0.1237, "step": 10130 }, { "epoch": 1.38, "grad_norm": 3.601199745919849, "learning_rate": 2.3281244541440305e-06, "loss": 0.1983, "step": 10131 }, { "epoch": 1.38, "grad_norm": 3.5081033391903484, "learning_rate": 2.327193444951966e-06, "loss": 0.1503, "step": 10132 }, { "epoch": 1.38, "grad_norm": 3.719418221261387, "learning_rate": 2.3262625654862567e-06, "loss": 0.1708, "step": 10133 }, { "epoch": 1.38, "grad_norm": 3.960844236429093, "learning_rate": 2.325331815792085e-06, "loss": 0.2023, "step": 10134 }, { "epoch": 1.38, "grad_norm": 4.094332689673321, "learning_rate": 2.3244011959146253e-06, "loss": 0.2185, "step": 10135 }, { "epoch": 1.38, "grad_norm": 3.582593303081983, "learning_rate": 2.3234707058990452e-06, "loss": 0.159, "step": 10136 }, { "epoch": 1.38, "grad_norm": 4.359773345773118, "learning_rate": 2.322540345790508e-06, "loss": 0.1698, "step": 10137 }, { "epoch": 1.38, "grad_norm": 3.954672271438548, "learning_rate": 2.321610115634167e-06, "loss": 0.1821, "step": 10138 }, { "epoch": 1.38, "grad_norm": 2.9219522499739123, "learning_rate": 2.3206800154751737e-06, "loss": 0.1146, "step": 10139 }, { "epoch": 1.38, "grad_norm": 2.8710962789201626, "learning_rate": 2.3197500453586695e-06, "loss": 0.1443, "step": 10140 }, { "epoch": 1.38, "grad_norm": 2.8601435697656856, "learning_rate": 2.3188202053297928e-06, "loss": 0.14, "step": 10141 }, { "epoch": 1.38, "grad_norm": 3.502418672558727, "learning_rate": 2.3178904954336718e-06, "loss": 0.1556, "step": 10142 }, { "epoch": 1.38, "grad_norm": 2.8457071512343717, "learning_rate": 2.3169609157154326e-06, "loss": 0.1508, "step": 10143 }, { "epoch": 1.38, "grad_norm": 3.5226838673124656, "learning_rate": 2.3160314662201917e-06, "loss": 0.1582, "step": 10144 }, { "epoch": 1.38, "grad_norm": 2.5259714870957315, "learning_rate": 2.315102146993061e-06, "loss": 0.1317, "step": 10145 }, { "epoch": 1.38, "grad_norm": 3.3263413650622944, "learning_rate": 2.314172958079144e-06, "loss": 0.1712, "step": 10146 }, { "epoch": 1.38, "grad_norm": 4.810926509956917, "learning_rate": 2.313243899523544e-06, "loss": 0.2065, "step": 10147 }, { "epoch": 1.38, "grad_norm": 2.638151048937914, "learning_rate": 2.3123149713713474e-06, "loss": 0.0998, "step": 10148 }, { "epoch": 1.38, "grad_norm": 4.202591803678943, "learning_rate": 2.3113861736676463e-06, "loss": 0.166, "step": 10149 }, { "epoch": 1.38, "grad_norm": 3.560937185362248, "learning_rate": 2.3104575064575146e-06, "loss": 0.1405, "step": 10150 }, { "epoch": 1.38, "grad_norm": 3.7213150057976887, "learning_rate": 2.30952896978603e-06, "loss": 0.1698, "step": 10151 }, { "epoch": 1.38, "grad_norm": 3.444537236447893, "learning_rate": 2.3086005636982582e-06, "loss": 0.155, "step": 10152 }, { "epoch": 1.38, "grad_norm": 3.823442576400413, "learning_rate": 2.3076722882392603e-06, "loss": 0.161, "step": 10153 }, { "epoch": 1.38, "grad_norm": 4.254612382146766, "learning_rate": 2.3067441434540906e-06, "loss": 0.1378, "step": 10154 }, { "epoch": 1.38, "grad_norm": 3.6929674987258365, "learning_rate": 2.3058161293877966e-06, "loss": 0.1908, "step": 10155 }, { "epoch": 1.38, "grad_norm": 4.586293125299963, "learning_rate": 2.3048882460854212e-06, "loss": 0.2016, "step": 10156 }, { "epoch": 1.38, "grad_norm": 4.259897298258697, "learning_rate": 2.303960493591999e-06, "loss": 0.2153, "step": 10157 }, { "epoch": 1.38, "grad_norm": 2.9951146865005054, "learning_rate": 2.303032871952559e-06, "loss": 0.1648, "step": 10158 }, { "epoch": 1.38, "grad_norm": 3.2713757730189967, "learning_rate": 2.302105381212124e-06, "loss": 0.1657, "step": 10159 }, { "epoch": 1.38, "grad_norm": 3.1904211381700573, "learning_rate": 2.301178021415709e-06, "loss": 0.175, "step": 10160 }, { "epoch": 1.38, "grad_norm": 2.9382726107572115, "learning_rate": 2.3002507926083285e-06, "loss": 0.1518, "step": 10161 }, { "epoch": 1.38, "grad_norm": 3.348456682808077, "learning_rate": 2.29932369483498e-06, "loss": 0.1348, "step": 10162 }, { "epoch": 1.38, "grad_norm": 3.268111438372404, "learning_rate": 2.298396728140667e-06, "loss": 0.1993, "step": 10163 }, { "epoch": 1.38, "grad_norm": 2.8679028386371495, "learning_rate": 2.2974698925703743e-06, "loss": 0.1416, "step": 10164 }, { "epoch": 1.38, "grad_norm": 4.070540206856164, "learning_rate": 2.296543188169091e-06, "loss": 0.1658, "step": 10165 }, { "epoch": 1.38, "grad_norm": 3.7696412649643447, "learning_rate": 2.2956166149817934e-06, "loss": 0.1864, "step": 10166 }, { "epoch": 1.38, "grad_norm": 3.4103928272134385, "learning_rate": 2.2946901730534533e-06, "loss": 0.2003, "step": 10167 }, { "epoch": 1.38, "grad_norm": 2.664562956647685, "learning_rate": 2.293763862429037e-06, "loss": 0.1406, "step": 10168 }, { "epoch": 1.38, "grad_norm": 3.725851996800669, "learning_rate": 2.292837683153503e-06, "loss": 0.1729, "step": 10169 }, { "epoch": 1.38, "grad_norm": 4.028320693084621, "learning_rate": 2.291911635271804e-06, "loss": 0.1632, "step": 10170 }, { "epoch": 1.38, "grad_norm": 3.178781185462472, "learning_rate": 2.2909857188288864e-06, "loss": 0.1442, "step": 10171 }, { "epoch": 1.38, "grad_norm": 2.8550278737467054, "learning_rate": 2.29005993386969e-06, "loss": 0.1212, "step": 10172 }, { "epoch": 1.38, "grad_norm": 3.83302441898706, "learning_rate": 2.2891342804391485e-06, "loss": 0.127, "step": 10173 }, { "epoch": 1.38, "grad_norm": 6.744210041771076, "learning_rate": 2.288208758582187e-06, "loss": 0.2068, "step": 10174 }, { "epoch": 1.38, "grad_norm": 4.132946380624031, "learning_rate": 2.2872833683437322e-06, "loss": 0.1471, "step": 10175 }, { "epoch": 1.38, "grad_norm": 3.2109399602517326, "learning_rate": 2.286358109768693e-06, "loss": 0.1754, "step": 10176 }, { "epoch": 1.38, "grad_norm": 3.9162301287612644, "learning_rate": 2.285432982901979e-06, "loss": 0.1728, "step": 10177 }, { "epoch": 1.38, "grad_norm": 3.3873634423733736, "learning_rate": 2.2845079877884913e-06, "loss": 0.1465, "step": 10178 }, { "epoch": 1.38, "grad_norm": 3.8074703290197194, "learning_rate": 2.2835831244731245e-06, "loss": 0.1947, "step": 10179 }, { "epoch": 1.38, "grad_norm": 3.016221548287441, "learning_rate": 2.2826583930007716e-06, "loss": 0.1783, "step": 10180 }, { "epoch": 1.38, "grad_norm": 3.6133231902011596, "learning_rate": 2.281733793416309e-06, "loss": 0.1666, "step": 10181 }, { "epoch": 1.39, "grad_norm": 3.786008652553784, "learning_rate": 2.2808093257646184e-06, "loss": 0.1798, "step": 10182 }, { "epoch": 1.39, "grad_norm": 2.8895696731905307, "learning_rate": 2.279884990090564e-06, "loss": 0.1244, "step": 10183 }, { "epoch": 1.39, "grad_norm": 4.125216744009946, "learning_rate": 2.2789607864390133e-06, "loss": 0.147, "step": 10184 }, { "epoch": 1.39, "grad_norm": 3.413613042249828, "learning_rate": 2.2780367148548214e-06, "loss": 0.1689, "step": 10185 }, { "epoch": 1.39, "grad_norm": 2.739676725744518, "learning_rate": 2.2771127753828395e-06, "loss": 0.161, "step": 10186 }, { "epoch": 1.39, "grad_norm": 3.7987929517549754, "learning_rate": 2.2761889680679106e-06, "loss": 0.1355, "step": 10187 }, { "epoch": 1.39, "grad_norm": 6.070746644844315, "learning_rate": 2.2752652929548725e-06, "loss": 0.1436, "step": 10188 }, { "epoch": 1.39, "grad_norm": 3.045299807985369, "learning_rate": 2.274341750088557e-06, "loss": 0.168, "step": 10189 }, { "epoch": 1.39, "grad_norm": 3.7307587543499077, "learning_rate": 2.2734183395137876e-06, "loss": 0.1988, "step": 10190 }, { "epoch": 1.39, "grad_norm": 3.439953012900898, "learning_rate": 2.2724950612753844e-06, "loss": 0.1638, "step": 10191 }, { "epoch": 1.39, "grad_norm": 5.048014716816041, "learning_rate": 2.271571915418157e-06, "loss": 0.1701, "step": 10192 }, { "epoch": 1.39, "grad_norm": 3.747974175432096, "learning_rate": 2.270648901986911e-06, "loss": 0.1562, "step": 10193 }, { "epoch": 1.39, "grad_norm": 3.7909470756112773, "learning_rate": 2.2697260210264506e-06, "loss": 0.1453, "step": 10194 }, { "epoch": 1.39, "grad_norm": 3.251327963630224, "learning_rate": 2.26880327258156e-06, "loss": 0.2041, "step": 10195 }, { "epoch": 1.39, "grad_norm": 4.063232527684521, "learning_rate": 2.2678806566970336e-06, "loss": 0.1827, "step": 10196 }, { "epoch": 1.39, "grad_norm": 3.011937546619738, "learning_rate": 2.266958173417644e-06, "loss": 0.1392, "step": 10197 }, { "epoch": 1.39, "grad_norm": 3.5631745931090513, "learning_rate": 2.266035822788169e-06, "loss": 0.1411, "step": 10198 }, { "epoch": 1.39, "grad_norm": 5.6898985918818195, "learning_rate": 2.2651136048533746e-06, "loss": 0.165, "step": 10199 }, { "epoch": 1.39, "grad_norm": 3.1848001781909825, "learning_rate": 2.264191519658021e-06, "loss": 0.1335, "step": 10200 }, { "epoch": 1.39, "grad_norm": 3.719258289260156, "learning_rate": 2.263269567246862e-06, "loss": 0.1787, "step": 10201 }, { "epoch": 1.39, "grad_norm": 2.9886219772871834, "learning_rate": 2.2623477476646447e-06, "loss": 0.1122, "step": 10202 }, { "epoch": 1.39, "grad_norm": 4.276862588688735, "learning_rate": 2.261426060956112e-06, "loss": 0.1577, "step": 10203 }, { "epoch": 1.39, "grad_norm": 3.240491089333173, "learning_rate": 2.2605045071659966e-06, "loss": 0.159, "step": 10204 }, { "epoch": 1.39, "grad_norm": 2.739542229431738, "learning_rate": 2.2595830863390273e-06, "loss": 0.1522, "step": 10205 }, { "epoch": 1.39, "grad_norm": 3.2927157347896663, "learning_rate": 2.2586617985199255e-06, "loss": 0.1464, "step": 10206 }, { "epoch": 1.39, "grad_norm": 3.4283480609691646, "learning_rate": 2.2577406437534055e-06, "loss": 0.1659, "step": 10207 }, { "epoch": 1.39, "grad_norm": 2.5790923673525588, "learning_rate": 2.2568196220841815e-06, "loss": 0.1243, "step": 10208 }, { "epoch": 1.39, "grad_norm": 3.5560881033136376, "learning_rate": 2.2558987335569477e-06, "loss": 0.1645, "step": 10209 }, { "epoch": 1.39, "grad_norm": 3.7691925038743417, "learning_rate": 2.2549779782164084e-06, "loss": 0.2069, "step": 10210 }, { "epoch": 1.39, "grad_norm": 2.9286750763176297, "learning_rate": 2.254057356107245e-06, "loss": 0.1576, "step": 10211 }, { "epoch": 1.39, "grad_norm": 2.9452872275959523, "learning_rate": 2.253136867274146e-06, "loss": 0.1286, "step": 10212 }, { "epoch": 1.39, "grad_norm": 3.4548419524184912, "learning_rate": 2.252216511761786e-06, "loss": 0.1187, "step": 10213 }, { "epoch": 1.39, "grad_norm": 3.5344895739862934, "learning_rate": 2.251296289614836e-06, "loss": 0.1743, "step": 10214 }, { "epoch": 1.39, "grad_norm": 3.692090561577162, "learning_rate": 2.2503762008779583e-06, "loss": 0.1902, "step": 10215 }, { "epoch": 1.39, "grad_norm": 2.9999360832017574, "learning_rate": 2.249456245595811e-06, "loss": 0.1568, "step": 10216 }, { "epoch": 1.39, "grad_norm": 3.5277847618703873, "learning_rate": 2.2485364238130435e-06, "loss": 0.1265, "step": 10217 }, { "epoch": 1.39, "grad_norm": 2.7367878292283407, "learning_rate": 2.247616735574301e-06, "loss": 0.1078, "step": 10218 }, { "epoch": 1.39, "grad_norm": 2.858344266187088, "learning_rate": 2.246697180924221e-06, "loss": 0.1267, "step": 10219 }, { "epoch": 1.39, "grad_norm": 3.8253772586956125, "learning_rate": 2.2457777599074345e-06, "loss": 0.1641, "step": 10220 }, { "epoch": 1.39, "grad_norm": 3.3397232725592474, "learning_rate": 2.244858472568566e-06, "loss": 0.1652, "step": 10221 }, { "epoch": 1.39, "grad_norm": 3.1968850632567025, "learning_rate": 2.243939318952234e-06, "loss": 0.1375, "step": 10222 }, { "epoch": 1.39, "grad_norm": 4.090453309478784, "learning_rate": 2.2430202991030496e-06, "loss": 0.1865, "step": 10223 }, { "epoch": 1.39, "grad_norm": 3.0216098102478064, "learning_rate": 2.2421014130656193e-06, "loss": 0.1708, "step": 10224 }, { "epoch": 1.39, "grad_norm": 3.9041488806645375, "learning_rate": 2.2411826608845404e-06, "loss": 0.1465, "step": 10225 }, { "epoch": 1.39, "grad_norm": 4.154021894291103, "learning_rate": 2.2402640426044035e-06, "loss": 0.1427, "step": 10226 }, { "epoch": 1.39, "grad_norm": 4.240144981299135, "learning_rate": 2.239345558269801e-06, "loss": 0.1763, "step": 10227 }, { "epoch": 1.39, "grad_norm": 2.917521660267296, "learning_rate": 2.238427207925304e-06, "loss": 0.1802, "step": 10228 }, { "epoch": 1.39, "grad_norm": 2.9716558150564465, "learning_rate": 2.2375089916154917e-06, "loss": 0.1293, "step": 10229 }, { "epoch": 1.39, "grad_norm": 3.1317603310079654, "learning_rate": 2.2365909093849247e-06, "loss": 0.1587, "step": 10230 }, { "epoch": 1.39, "grad_norm": 2.7606662482594286, "learning_rate": 2.235672961278167e-06, "loss": 0.1331, "step": 10231 }, { "epoch": 1.39, "grad_norm": 2.6072131244419494, "learning_rate": 2.23475514733977e-06, "loss": 0.1413, "step": 10232 }, { "epoch": 1.39, "grad_norm": 3.135930100088554, "learning_rate": 2.233837467614281e-06, "loss": 0.1515, "step": 10233 }, { "epoch": 1.39, "grad_norm": 5.397312432232146, "learning_rate": 2.2329199221462393e-06, "loss": 0.18, "step": 10234 }, { "epoch": 1.39, "grad_norm": 3.9102849000788042, "learning_rate": 2.2320025109801797e-06, "loss": 0.153, "step": 10235 }, { "epoch": 1.39, "grad_norm": 3.194310269568162, "learning_rate": 2.231085234160628e-06, "loss": 0.1959, "step": 10236 }, { "epoch": 1.39, "grad_norm": 3.236318616808481, "learning_rate": 2.230168091732106e-06, "loss": 0.1621, "step": 10237 }, { "epoch": 1.39, "grad_norm": 3.1791302612952244, "learning_rate": 2.229251083739127e-06, "loss": 0.1764, "step": 10238 }, { "epoch": 1.39, "grad_norm": 3.162508807515659, "learning_rate": 2.228334210226199e-06, "loss": 0.1215, "step": 10239 }, { "epoch": 1.39, "grad_norm": 3.9368839916458276, "learning_rate": 2.227417471237821e-06, "loss": 0.1686, "step": 10240 }, { "epoch": 1.39, "grad_norm": 3.285371206162886, "learning_rate": 2.2265008668184928e-06, "loss": 0.1486, "step": 10241 }, { "epoch": 1.39, "grad_norm": 4.462991006111914, "learning_rate": 2.2255843970126957e-06, "loss": 0.1851, "step": 10242 }, { "epoch": 1.39, "grad_norm": 3.866585027515821, "learning_rate": 2.224668061864918e-06, "loss": 0.1252, "step": 10243 }, { "epoch": 1.39, "grad_norm": 3.610542557327451, "learning_rate": 2.2237518614196273e-06, "loss": 0.1787, "step": 10244 }, { "epoch": 1.39, "grad_norm": 3.0755688625626267, "learning_rate": 2.2228357957212977e-06, "loss": 0.1597, "step": 10245 }, { "epoch": 1.39, "grad_norm": 4.549284640097434, "learning_rate": 2.2219198648143886e-06, "loss": 0.1329, "step": 10246 }, { "epoch": 1.39, "grad_norm": 3.435180570551568, "learning_rate": 2.221004068743356e-06, "loss": 0.1537, "step": 10247 }, { "epoch": 1.39, "grad_norm": 3.0832824878595892, "learning_rate": 2.220088407552648e-06, "loss": 0.1844, "step": 10248 }, { "epoch": 1.39, "grad_norm": 3.1713881762170186, "learning_rate": 2.2191728812867076e-06, "loss": 0.1506, "step": 10249 }, { "epoch": 1.39, "grad_norm": 3.581413245613751, "learning_rate": 2.21825748998997e-06, "loss": 0.1559, "step": 10250 }, { "epoch": 1.39, "grad_norm": 3.5256705013816028, "learning_rate": 2.217342233706864e-06, "loss": 0.1206, "step": 10251 }, { "epoch": 1.39, "grad_norm": 3.5774655912672744, "learning_rate": 2.2164271124818103e-06, "loss": 0.1393, "step": 10252 }, { "epoch": 1.39, "grad_norm": 3.3522578903612437, "learning_rate": 2.2155121263592313e-06, "loss": 0.1554, "step": 10253 }, { "epoch": 1.39, "grad_norm": 4.792416364724764, "learning_rate": 2.2145972753835275e-06, "loss": 0.2244, "step": 10254 }, { "epoch": 1.39, "grad_norm": 2.475691523702315, "learning_rate": 2.213682559599111e-06, "loss": 0.126, "step": 10255 }, { "epoch": 1.4, "grad_norm": 3.497001543883717, "learning_rate": 2.2127679790503716e-06, "loss": 0.1129, "step": 10256 }, { "epoch": 1.4, "grad_norm": 4.312081457257822, "learning_rate": 2.2118535337817003e-06, "loss": 0.1469, "step": 10257 }, { "epoch": 1.4, "grad_norm": 3.4329652274030895, "learning_rate": 2.2109392238374818e-06, "loss": 0.1784, "step": 10258 }, { "epoch": 1.4, "grad_norm": 3.4621193431077755, "learning_rate": 2.2100250492620895e-06, "loss": 0.1587, "step": 10259 }, { "epoch": 1.4, "grad_norm": 3.6842419279661405, "learning_rate": 2.2091110100998996e-06, "loss": 0.1421, "step": 10260 }, { "epoch": 1.4, "grad_norm": 3.437082550789841, "learning_rate": 2.2081971063952677e-06, "loss": 0.1711, "step": 10261 }, { "epoch": 1.4, "grad_norm": 3.8698870469695774, "learning_rate": 2.207283338192559e-06, "loss": 0.1898, "step": 10262 }, { "epoch": 1.4, "grad_norm": 3.866654819397559, "learning_rate": 2.2063697055361154e-06, "loss": 0.1286, "step": 10263 }, { "epoch": 1.4, "grad_norm": 3.4998530162958152, "learning_rate": 2.2054562084702874e-06, "loss": 0.19, "step": 10264 }, { "epoch": 1.4, "grad_norm": 3.1089441317023616, "learning_rate": 2.2045428470394085e-06, "loss": 0.1709, "step": 10265 }, { "epoch": 1.4, "grad_norm": 3.6694154280861877, "learning_rate": 2.203629621287811e-06, "loss": 0.1412, "step": 10266 }, { "epoch": 1.4, "grad_norm": 3.249722251240141, "learning_rate": 2.2027165312598185e-06, "loss": 0.1769, "step": 10267 }, { "epoch": 1.4, "grad_norm": 3.55149044585548, "learning_rate": 2.2018035769997476e-06, "loss": 0.1582, "step": 10268 }, { "epoch": 1.4, "grad_norm": 4.925574627382154, "learning_rate": 2.2008907585519094e-06, "loss": 0.2058, "step": 10269 }, { "epoch": 1.4, "grad_norm": 3.0159566904826067, "learning_rate": 2.1999780759606094e-06, "loss": 0.17, "step": 10270 }, { "epoch": 1.4, "grad_norm": 3.894987527288844, "learning_rate": 2.1990655292701434e-06, "loss": 0.1193, "step": 10271 }, { "epoch": 1.4, "grad_norm": 3.9698316332187114, "learning_rate": 2.1981531185248034e-06, "loss": 0.1833, "step": 10272 }, { "epoch": 1.4, "grad_norm": 3.0262018820748517, "learning_rate": 2.197240843768872e-06, "loss": 0.1382, "step": 10273 }, { "epoch": 1.4, "grad_norm": 4.4509811473652645, "learning_rate": 2.196328705046632e-06, "loss": 0.1885, "step": 10274 }, { "epoch": 1.4, "grad_norm": 2.7873666515236524, "learning_rate": 2.1954167024023474e-06, "loss": 0.1457, "step": 10275 }, { "epoch": 1.4, "grad_norm": 3.3536661555515797, "learning_rate": 2.194504835880291e-06, "loss": 0.1544, "step": 10276 }, { "epoch": 1.4, "grad_norm": 4.302873312057079, "learning_rate": 2.1935931055247127e-06, "loss": 0.1865, "step": 10277 }, { "epoch": 1.4, "grad_norm": 3.1591899953128655, "learning_rate": 2.19268151137987e-06, "loss": 0.1565, "step": 10278 }, { "epoch": 1.4, "grad_norm": 3.9604025681057506, "learning_rate": 2.1917700534900048e-06, "loss": 0.1716, "step": 10279 }, { "epoch": 1.4, "grad_norm": 3.3590842207235663, "learning_rate": 2.1908587318993563e-06, "loss": 0.1667, "step": 10280 }, { "epoch": 1.4, "grad_norm": 3.3712988991436332, "learning_rate": 2.189947546652156e-06, "loss": 0.1427, "step": 10281 }, { "epoch": 1.4, "grad_norm": 3.072740793137623, "learning_rate": 2.1890364977926283e-06, "loss": 0.1504, "step": 10282 }, { "epoch": 1.4, "grad_norm": 4.89923161469996, "learning_rate": 2.1881255853649914e-06, "loss": 0.2109, "step": 10283 }, { "epoch": 1.4, "grad_norm": 3.410141065132569, "learning_rate": 2.1872148094134578e-06, "loss": 0.1965, "step": 10284 }, { "epoch": 1.4, "grad_norm": 4.06126484301442, "learning_rate": 2.1863041699822303e-06, "loss": 0.1541, "step": 10285 }, { "epoch": 1.4, "grad_norm": 3.1072353240996176, "learning_rate": 2.185393667115513e-06, "loss": 0.108, "step": 10286 }, { "epoch": 1.4, "grad_norm": 3.68016578672066, "learning_rate": 2.18448330085749e-06, "loss": 0.1433, "step": 10287 }, { "epoch": 1.4, "grad_norm": 3.217372217179036, "learning_rate": 2.1835730712523544e-06, "loss": 0.1077, "step": 10288 }, { "epoch": 1.4, "grad_norm": 3.81055154538921, "learning_rate": 2.182662978344278e-06, "loss": 0.1528, "step": 10289 }, { "epoch": 1.4, "grad_norm": 2.659506500511951, "learning_rate": 2.181753022177438e-06, "loss": 0.1438, "step": 10290 }, { "epoch": 1.4, "grad_norm": 3.717446965135612, "learning_rate": 2.1808432027959943e-06, "loss": 0.1375, "step": 10291 }, { "epoch": 1.4, "grad_norm": 5.331502625383341, "learning_rate": 2.1799335202441104e-06, "loss": 0.213, "step": 10292 }, { "epoch": 1.4, "grad_norm": 4.216341607637338, "learning_rate": 2.1790239745659363e-06, "loss": 0.1566, "step": 10293 }, { "epoch": 1.4, "grad_norm": 4.700450194085904, "learning_rate": 2.1781145658056174e-06, "loss": 0.2001, "step": 10294 }, { "epoch": 1.4, "grad_norm": 4.531694066189468, "learning_rate": 2.177205294007293e-06, "loss": 0.1806, "step": 10295 }, { "epoch": 1.4, "grad_norm": 3.9732775214626814, "learning_rate": 2.1762961592150942e-06, "loss": 0.113, "step": 10296 }, { "epoch": 1.4, "grad_norm": 3.92006887032468, "learning_rate": 2.1753871614731474e-06, "loss": 0.1919, "step": 10297 }, { "epoch": 1.4, "grad_norm": 3.4825565319611105, "learning_rate": 2.174478300825571e-06, "loss": 0.1584, "step": 10298 }, { "epoch": 1.4, "grad_norm": 3.408270817523527, "learning_rate": 2.1735695773164772e-06, "loss": 0.1379, "step": 10299 }, { "epoch": 1.4, "grad_norm": 3.915048554522951, "learning_rate": 2.172660990989971e-06, "loss": 0.1585, "step": 10300 }, { "epoch": 1.4, "grad_norm": 4.287473589035414, "learning_rate": 2.171752541890152e-06, "loss": 0.1694, "step": 10301 }, { "epoch": 1.4, "grad_norm": 4.225182726031753, "learning_rate": 2.1708442300611115e-06, "loss": 0.1246, "step": 10302 }, { "epoch": 1.4, "grad_norm": 3.2392641647917992, "learning_rate": 2.1699360555469357e-06, "loss": 0.1793, "step": 10303 }, { "epoch": 1.4, "grad_norm": 3.8278070577931516, "learning_rate": 2.1690280183917016e-06, "loss": 0.1838, "step": 10304 }, { "epoch": 1.4, "grad_norm": 3.70999733857578, "learning_rate": 2.168120118639486e-06, "loss": 0.1538, "step": 10305 }, { "epoch": 1.4, "grad_norm": 4.503787906166448, "learning_rate": 2.1672123563343477e-06, "loss": 0.165, "step": 10306 }, { "epoch": 1.4, "grad_norm": 3.7718972772421067, "learning_rate": 2.1663047315203533e-06, "loss": 0.1647, "step": 10307 }, { "epoch": 1.4, "grad_norm": 2.883666667789323, "learning_rate": 2.1653972442415465e-06, "loss": 0.1376, "step": 10308 }, { "epoch": 1.4, "grad_norm": 3.5773013774176556, "learning_rate": 2.1644898945419814e-06, "loss": 0.1502, "step": 10309 }, { "epoch": 1.4, "grad_norm": 4.113785864776491, "learning_rate": 2.163582682465688e-06, "loss": 0.1464, "step": 10310 }, { "epoch": 1.4, "grad_norm": 4.002650393854683, "learning_rate": 2.162675608056705e-06, "loss": 0.176, "step": 10311 }, { "epoch": 1.4, "grad_norm": 3.5801647946801984, "learning_rate": 2.1617686713590557e-06, "loss": 0.122, "step": 10312 }, { "epoch": 1.4, "grad_norm": 3.981556801700743, "learning_rate": 2.1608618724167585e-06, "loss": 0.1725, "step": 10313 }, { "epoch": 1.4, "grad_norm": 3.0105945009849115, "learning_rate": 2.1599552112738258e-06, "loss": 0.1417, "step": 10314 }, { "epoch": 1.4, "grad_norm": 3.319514596273458, "learning_rate": 2.1590486879742633e-06, "loss": 0.1343, "step": 10315 }, { "epoch": 1.4, "grad_norm": 3.8631458559120357, "learning_rate": 2.1581423025620695e-06, "loss": 0.1953, "step": 10316 }, { "epoch": 1.4, "grad_norm": 2.4714979129128642, "learning_rate": 2.1572360550812354e-06, "loss": 0.1678, "step": 10317 }, { "epoch": 1.4, "grad_norm": 3.605085306725153, "learning_rate": 2.1563299455757458e-06, "loss": 0.2325, "step": 10318 }, { "epoch": 1.4, "grad_norm": 3.9779116681342575, "learning_rate": 2.1554239740895844e-06, "loss": 0.1813, "step": 10319 }, { "epoch": 1.4, "grad_norm": 3.2227021526868334, "learning_rate": 2.1545181406667153e-06, "loss": 0.1988, "step": 10320 }, { "epoch": 1.4, "grad_norm": 4.0684865047986225, "learning_rate": 2.1536124453511114e-06, "loss": 0.1741, "step": 10321 }, { "epoch": 1.4, "grad_norm": 3.423317054314657, "learning_rate": 2.1527068881867243e-06, "loss": 0.1715, "step": 10322 }, { "epoch": 1.4, "grad_norm": 3.7377191700130172, "learning_rate": 2.151801469217512e-06, "loss": 0.1624, "step": 10323 }, { "epoch": 1.4, "grad_norm": 3.586455959280157, "learning_rate": 2.1508961884874134e-06, "loss": 0.1997, "step": 10324 }, { "epoch": 1.4, "grad_norm": 2.24472655908616, "learning_rate": 2.149991046040372e-06, "loss": 0.1364, "step": 10325 }, { "epoch": 1.4, "grad_norm": 4.808978097875207, "learning_rate": 2.149086041920317e-06, "loss": 0.185, "step": 10326 }, { "epoch": 1.4, "grad_norm": 3.767317649536043, "learning_rate": 2.148181176171174e-06, "loss": 0.1874, "step": 10327 }, { "epoch": 1.4, "grad_norm": 3.891086195497383, "learning_rate": 2.147276448836861e-06, "loss": 0.1439, "step": 10328 }, { "epoch": 1.41, "grad_norm": 2.927083839461603, "learning_rate": 2.1463718599612896e-06, "loss": 0.2009, "step": 10329 }, { "epoch": 1.41, "grad_norm": 3.691088987238251, "learning_rate": 2.145467409588365e-06, "loss": 0.1588, "step": 10330 }, { "epoch": 1.41, "grad_norm": 4.226816056781735, "learning_rate": 2.144563097761984e-06, "loss": 0.1725, "step": 10331 }, { "epoch": 1.41, "grad_norm": 3.7644649590273875, "learning_rate": 2.1436589245260375e-06, "loss": 0.165, "step": 10332 }, { "epoch": 1.41, "grad_norm": 2.802208678476957, "learning_rate": 2.1427548899244155e-06, "loss": 0.1797, "step": 10333 }, { "epoch": 1.41, "grad_norm": 3.8065019041108163, "learning_rate": 2.1418509940009873e-06, "loss": 0.1721, "step": 10334 }, { "epoch": 1.41, "grad_norm": 3.2922879132876415, "learning_rate": 2.1409472367996325e-06, "loss": 0.1816, "step": 10335 }, { "epoch": 1.41, "grad_norm": 3.281258648910614, "learning_rate": 2.1400436183642105e-06, "loss": 0.1556, "step": 10336 }, { "epoch": 1.41, "grad_norm": 5.012363890406769, "learning_rate": 2.1391401387385773e-06, "loss": 0.2167, "step": 10337 }, { "epoch": 1.41, "grad_norm": 3.6302687839724435, "learning_rate": 2.138236797966591e-06, "loss": 0.197, "step": 10338 }, { "epoch": 1.41, "grad_norm": 2.947349683414655, "learning_rate": 2.137333596092087e-06, "loss": 0.1451, "step": 10339 }, { "epoch": 1.41, "grad_norm": 3.888575855163271, "learning_rate": 2.136430533158912e-06, "loss": 0.1609, "step": 10340 }, { "epoch": 1.41, "grad_norm": 3.3607941363267297, "learning_rate": 2.1355276092108873e-06, "loss": 0.1267, "step": 10341 }, { "epoch": 1.41, "grad_norm": 3.547009122348815, "learning_rate": 2.134624824291846e-06, "loss": 0.1399, "step": 10342 }, { "epoch": 1.41, "grad_norm": 3.1495857329630987, "learning_rate": 2.1337221784455963e-06, "loss": 0.1818, "step": 10343 }, { "epoch": 1.41, "grad_norm": 3.449613003984543, "learning_rate": 2.132819671715956e-06, "loss": 0.1849, "step": 10344 }, { "epoch": 1.41, "grad_norm": 3.28246892720828, "learning_rate": 2.1319173041467255e-06, "loss": 0.1623, "step": 10345 }, { "epoch": 1.41, "grad_norm": 3.7341128404984363, "learning_rate": 2.1310150757817027e-06, "loss": 0.1384, "step": 10346 }, { "epoch": 1.41, "grad_norm": 3.3360790807714813, "learning_rate": 2.1301129866646774e-06, "loss": 0.1421, "step": 10347 }, { "epoch": 1.41, "grad_norm": 3.5816509085554125, "learning_rate": 2.1292110368394335e-06, "loss": 0.1628, "step": 10348 }, { "epoch": 1.41, "grad_norm": 3.4564850528560878, "learning_rate": 2.1283092263497473e-06, "loss": 0.2047, "step": 10349 }, { "epoch": 1.41, "grad_norm": 3.3727738619182017, "learning_rate": 2.1274075552393886e-06, "loss": 0.1199, "step": 10350 }, { "epoch": 1.41, "grad_norm": 4.18501527419207, "learning_rate": 2.126506023552119e-06, "loss": 0.2005, "step": 10351 }, { "epoch": 1.41, "grad_norm": 3.1901330568139517, "learning_rate": 2.1256046313317002e-06, "loss": 0.1566, "step": 10352 }, { "epoch": 1.41, "grad_norm": 3.7796807270459505, "learning_rate": 2.124703378621875e-06, "loss": 0.1461, "step": 10353 }, { "epoch": 1.41, "grad_norm": 3.6922803644192186, "learning_rate": 2.1238022654663937e-06, "loss": 0.1184, "step": 10354 }, { "epoch": 1.41, "grad_norm": 4.191068885129777, "learning_rate": 2.1229012919089843e-06, "loss": 0.1458, "step": 10355 }, { "epoch": 1.41, "grad_norm": 3.321228480767164, "learning_rate": 2.1220004579933816e-06, "loss": 0.1345, "step": 10356 }, { "epoch": 1.41, "grad_norm": 3.732977905682245, "learning_rate": 2.1210997637633067e-06, "loss": 0.1546, "step": 10357 }, { "epoch": 1.41, "grad_norm": 3.152592218837495, "learning_rate": 2.1201992092624756e-06, "loss": 0.1616, "step": 10358 }, { "epoch": 1.41, "grad_norm": 4.392782209384546, "learning_rate": 2.119298794534597e-06, "loss": 0.1666, "step": 10359 }, { "epoch": 1.41, "grad_norm": 3.8925915905286743, "learning_rate": 2.1183985196233736e-06, "loss": 0.1764, "step": 10360 }, { "epoch": 1.41, "grad_norm": 4.1173101740869615, "learning_rate": 2.1174983845725e-06, "loss": 0.1674, "step": 10361 }, { "epoch": 1.41, "grad_norm": 3.4661858717909793, "learning_rate": 2.1165983894256647e-06, "loss": 0.1566, "step": 10362 }, { "epoch": 1.41, "grad_norm": 3.8562712786861795, "learning_rate": 2.1156985342265507e-06, "loss": 0.1631, "step": 10363 }, { "epoch": 1.41, "grad_norm": 4.549853045767169, "learning_rate": 2.114798819018832e-06, "loss": 0.1722, "step": 10364 }, { "epoch": 1.41, "grad_norm": 4.366571302166467, "learning_rate": 2.1138992438461754e-06, "loss": 0.1412, "step": 10365 }, { "epoch": 1.41, "grad_norm": 3.6488846777393946, "learning_rate": 2.112999808752248e-06, "loss": 0.1678, "step": 10366 }, { "epoch": 1.41, "grad_norm": 3.6302800898008822, "learning_rate": 2.1121005137806964e-06, "loss": 0.1878, "step": 10367 }, { "epoch": 1.41, "grad_norm": 2.256972379017358, "learning_rate": 2.1112013589751766e-06, "loss": 0.1394, "step": 10368 }, { "epoch": 1.41, "grad_norm": 4.455263539533883, "learning_rate": 2.1103023443793225e-06, "loss": 0.1555, "step": 10369 }, { "epoch": 1.41, "grad_norm": 3.2324425517918756, "learning_rate": 2.1094034700367733e-06, "loss": 0.1087, "step": 10370 }, { "epoch": 1.41, "grad_norm": 4.458676076366461, "learning_rate": 2.108504735991154e-06, "loss": 0.1468, "step": 10371 }, { "epoch": 1.41, "grad_norm": 3.748080758775941, "learning_rate": 2.1076061422860862e-06, "loss": 0.1713, "step": 10372 }, { "epoch": 1.41, "grad_norm": 2.891392074224546, "learning_rate": 2.1067076889651835e-06, "loss": 0.1512, "step": 10373 }, { "epoch": 1.41, "grad_norm": 4.102275253844707, "learning_rate": 2.105809376072053e-06, "loss": 0.1679, "step": 10374 }, { "epoch": 1.41, "grad_norm": 3.643998062762759, "learning_rate": 2.1049112036502943e-06, "loss": 0.1712, "step": 10375 }, { "epoch": 1.41, "grad_norm": 3.1165806714455084, "learning_rate": 2.1040131717435015e-06, "loss": 0.16, "step": 10376 }, { "epoch": 1.41, "grad_norm": 3.84420586653276, "learning_rate": 2.1031152803952605e-06, "loss": 0.1492, "step": 10377 }, { "epoch": 1.41, "grad_norm": 3.32854073149718, "learning_rate": 2.1022175296491516e-06, "loss": 0.1339, "step": 10378 }, { "epoch": 1.41, "grad_norm": 4.067106545592501, "learning_rate": 2.101319919548747e-06, "loss": 0.1798, "step": 10379 }, { "epoch": 1.41, "grad_norm": 2.9115406814185563, "learning_rate": 2.100422450137614e-06, "loss": 0.1506, "step": 10380 }, { "epoch": 1.41, "grad_norm": 3.420978482426976, "learning_rate": 2.09952512145931e-06, "loss": 0.187, "step": 10381 }, { "epoch": 1.41, "grad_norm": 4.145132193048948, "learning_rate": 2.098627933557389e-06, "loss": 0.1896, "step": 10382 }, { "epoch": 1.41, "grad_norm": 2.6930420994770334, "learning_rate": 2.097730886475396e-06, "loss": 0.1316, "step": 10383 }, { "epoch": 1.41, "grad_norm": 3.0749554913949706, "learning_rate": 2.096833980256868e-06, "loss": 0.1387, "step": 10384 }, { "epoch": 1.41, "grad_norm": 3.5421406457773195, "learning_rate": 2.0959372149453424e-06, "loss": 0.1244, "step": 10385 }, { "epoch": 1.41, "grad_norm": 4.692354389923114, "learning_rate": 2.095040590584337e-06, "loss": 0.158, "step": 10386 }, { "epoch": 1.41, "grad_norm": 3.572096116432262, "learning_rate": 2.0941441072173766e-06, "loss": 0.1732, "step": 10387 }, { "epoch": 1.41, "grad_norm": 3.3148455430463093, "learning_rate": 2.093247764887967e-06, "loss": 0.1772, "step": 10388 }, { "epoch": 1.41, "grad_norm": 3.5519696094618554, "learning_rate": 2.0923515636396164e-06, "loss": 0.1267, "step": 10389 }, { "epoch": 1.41, "grad_norm": 4.0906014360845875, "learning_rate": 2.0914555035158217e-06, "loss": 0.1976, "step": 10390 }, { "epoch": 1.41, "grad_norm": 3.3519272024114355, "learning_rate": 2.0905595845600733e-06, "loss": 0.183, "step": 10391 }, { "epoch": 1.41, "grad_norm": 3.7542822131262645, "learning_rate": 2.089663806815856e-06, "loss": 0.1521, "step": 10392 }, { "epoch": 1.41, "grad_norm": 2.8635770114852606, "learning_rate": 2.0887681703266453e-06, "loss": 0.1492, "step": 10393 }, { "epoch": 1.41, "grad_norm": 3.7054367739185397, "learning_rate": 2.087872675135913e-06, "loss": 0.1584, "step": 10394 }, { "epoch": 1.41, "grad_norm": 4.3734546988695016, "learning_rate": 2.086977321287122e-06, "loss": 0.1379, "step": 10395 }, { "epoch": 1.41, "grad_norm": 3.4364859642633627, "learning_rate": 2.0860821088237295e-06, "loss": 0.1562, "step": 10396 }, { "epoch": 1.41, "grad_norm": 3.1327530787068287, "learning_rate": 2.085187037789184e-06, "loss": 0.1225, "step": 10397 }, { "epoch": 1.41, "grad_norm": 3.5369290434874214, "learning_rate": 2.0842921082269268e-06, "loss": 0.1646, "step": 10398 }, { "epoch": 1.41, "grad_norm": 3.4414406425045434, "learning_rate": 2.0833973201803996e-06, "loss": 0.1552, "step": 10399 }, { "epoch": 1.41, "grad_norm": 3.5687811661328714, "learning_rate": 2.082502673693025e-06, "loss": 0.1664, "step": 10400 }, { "epoch": 1.41, "grad_norm": 3.6919468039763608, "learning_rate": 2.0816081688082313e-06, "loss": 0.1803, "step": 10401 }, { "epoch": 1.41, "grad_norm": 2.8776576704073347, "learning_rate": 2.080713805569427e-06, "loss": 0.1227, "step": 10402 }, { "epoch": 1.42, "grad_norm": 3.2804551740028995, "learning_rate": 2.079819584020026e-06, "loss": 0.1772, "step": 10403 }, { "epoch": 1.42, "grad_norm": 3.347748470819383, "learning_rate": 2.0789255042034284e-06, "loss": 0.1224, "step": 10404 }, { "epoch": 1.42, "grad_norm": 2.9252781182544765, "learning_rate": 2.078031566163029e-06, "loss": 0.1707, "step": 10405 }, { "epoch": 1.42, "grad_norm": 3.308058654800676, "learning_rate": 2.0771377699422146e-06, "loss": 0.1175, "step": 10406 }, { "epoch": 1.42, "grad_norm": 3.0249648988270574, "learning_rate": 2.0762441155843678e-06, "loss": 0.1458, "step": 10407 }, { "epoch": 1.42, "grad_norm": 3.029803348311652, "learning_rate": 2.0753506031328615e-06, "loss": 0.1305, "step": 10408 }, { "epoch": 1.42, "grad_norm": 3.1672999472850543, "learning_rate": 2.074457232631063e-06, "loss": 0.1659, "step": 10409 }, { "epoch": 1.42, "grad_norm": 2.7719663687524814, "learning_rate": 2.0735640041223325e-06, "loss": 0.1415, "step": 10410 }, { "epoch": 1.42, "grad_norm": 2.6866008989095005, "learning_rate": 2.0726709176500244e-06, "loss": 0.1552, "step": 10411 }, { "epoch": 1.42, "grad_norm": 2.6203141324705985, "learning_rate": 2.071777973257482e-06, "loss": 0.1617, "step": 10412 }, { "epoch": 1.42, "grad_norm": 2.89532991015486, "learning_rate": 2.070885170988052e-06, "loss": 0.1881, "step": 10413 }, { "epoch": 1.42, "grad_norm": 3.626626402883535, "learning_rate": 2.069992510885058e-06, "loss": 0.1622, "step": 10414 }, { "epoch": 1.42, "grad_norm": 4.3259143867186305, "learning_rate": 2.0690999929918347e-06, "loss": 0.1669, "step": 10415 }, { "epoch": 1.42, "grad_norm": 4.049285194918697, "learning_rate": 2.068207617351695e-06, "loss": 0.1548, "step": 10416 }, { "epoch": 1.42, "grad_norm": 3.4826665290957326, "learning_rate": 2.0673153840079502e-06, "loss": 0.1486, "step": 10417 }, { "epoch": 1.42, "grad_norm": 3.4564785773360285, "learning_rate": 2.066423293003912e-06, "loss": 0.1297, "step": 10418 }, { "epoch": 1.42, "grad_norm": 3.6248256293311427, "learning_rate": 2.06553134438287e-06, "loss": 0.1237, "step": 10419 }, { "epoch": 1.42, "grad_norm": 2.754701984390951, "learning_rate": 2.0646395381881245e-06, "loss": 0.147, "step": 10420 }, { "epoch": 1.42, "grad_norm": 3.662170547638231, "learning_rate": 2.0637478744629514e-06, "loss": 0.1689, "step": 10421 }, { "epoch": 1.42, "grad_norm": 4.1859392714156485, "learning_rate": 2.0628563532506334e-06, "loss": 0.2079, "step": 10422 }, { "epoch": 1.42, "grad_norm": 4.398223045663828, "learning_rate": 2.0619649745944402e-06, "loss": 0.1768, "step": 10423 }, { "epoch": 1.42, "grad_norm": 5.3346513197660546, "learning_rate": 2.061073738537635e-06, "loss": 0.1235, "step": 10424 }, { "epoch": 1.42, "grad_norm": 2.9397730685546324, "learning_rate": 2.060182645123474e-06, "loss": 0.1173, "step": 10425 }, { "epoch": 1.42, "grad_norm": 3.1761917078266686, "learning_rate": 2.059291694395208e-06, "loss": 0.1655, "step": 10426 }, { "epoch": 1.42, "grad_norm": 3.086342521436611, "learning_rate": 2.058400886396079e-06, "loss": 0.1649, "step": 10427 }, { "epoch": 1.42, "grad_norm": 4.126940508073597, "learning_rate": 2.057510221169323e-06, "loss": 0.1303, "step": 10428 }, { "epoch": 1.42, "grad_norm": 3.185939433686913, "learning_rate": 2.0566196987581695e-06, "loss": 0.1407, "step": 10429 }, { "epoch": 1.42, "grad_norm": 4.790372725379036, "learning_rate": 2.0557293192058402e-06, "loss": 0.1844, "step": 10430 }, { "epoch": 1.42, "grad_norm": 3.575532460874273, "learning_rate": 2.0548390825555485e-06, "loss": 0.1915, "step": 10431 }, { "epoch": 1.42, "grad_norm": 4.168189706773542, "learning_rate": 2.053948988850508e-06, "loss": 0.1727, "step": 10432 }, { "epoch": 1.42, "grad_norm": 3.0586351221246786, "learning_rate": 2.053059038133913e-06, "loss": 0.1284, "step": 10433 }, { "epoch": 1.42, "grad_norm": 4.500252696087136, "learning_rate": 2.0521692304489647e-06, "loss": 0.1789, "step": 10434 }, { "epoch": 1.42, "grad_norm": 5.372444202865968, "learning_rate": 2.0512795658388435e-06, "loss": 0.1916, "step": 10435 }, { "epoch": 1.42, "grad_norm": 3.299452787518368, "learning_rate": 2.0503900443467352e-06, "loss": 0.126, "step": 10436 }, { "epoch": 1.42, "grad_norm": 3.930921977905028, "learning_rate": 2.0495006660158113e-06, "loss": 0.1703, "step": 10437 }, { "epoch": 1.42, "grad_norm": 2.998523626487883, "learning_rate": 2.048611430889238e-06, "loss": 0.1574, "step": 10438 }, { "epoch": 1.42, "grad_norm": 3.378533102679439, "learning_rate": 2.047722339010176e-06, "loss": 0.1491, "step": 10439 }, { "epoch": 1.42, "grad_norm": 3.688548230195376, "learning_rate": 2.0468333904217765e-06, "loss": 0.1525, "step": 10440 }, { "epoch": 1.42, "grad_norm": 4.645856891936382, "learning_rate": 2.0459445851671865e-06, "loss": 0.1735, "step": 10441 }, { "epoch": 1.42, "grad_norm": 3.840853041290077, "learning_rate": 2.045055923289544e-06, "loss": 0.1448, "step": 10442 }, { "epoch": 1.42, "grad_norm": 3.4247744460946086, "learning_rate": 2.044167404831981e-06, "loss": 0.1405, "step": 10443 }, { "epoch": 1.42, "grad_norm": 3.3339675389031673, "learning_rate": 2.043279029837622e-06, "loss": 0.1308, "step": 10444 }, { "epoch": 1.42, "grad_norm": 4.183345467714165, "learning_rate": 2.042390798349584e-06, "loss": 0.139, "step": 10445 }, { "epoch": 1.42, "grad_norm": 3.6987737580869697, "learning_rate": 2.041502710410982e-06, "loss": 0.1506, "step": 10446 }, { "epoch": 1.42, "grad_norm": 2.884674947445184, "learning_rate": 2.040614766064913e-06, "loss": 0.1795, "step": 10447 }, { "epoch": 1.42, "grad_norm": 3.434496787980605, "learning_rate": 2.039726965354482e-06, "loss": 0.1392, "step": 10448 }, { "epoch": 1.42, "grad_norm": 3.337610584658908, "learning_rate": 2.038839308322771e-06, "loss": 0.1545, "step": 10449 }, { "epoch": 1.42, "grad_norm": 4.876597817316629, "learning_rate": 2.0379517950128685e-06, "loss": 0.1574, "step": 10450 }, { "epoch": 1.42, "grad_norm": 4.424450535391286, "learning_rate": 2.037064425467849e-06, "loss": 0.1826, "step": 10451 }, { "epoch": 1.42, "grad_norm": 4.398319673828026, "learning_rate": 2.036177199730781e-06, "loss": 0.1932, "step": 10452 }, { "epoch": 1.42, "grad_norm": 2.8986454409240094, "learning_rate": 2.035290117844727e-06, "loss": 0.1918, "step": 10453 }, { "epoch": 1.42, "grad_norm": 3.5371871480053443, "learning_rate": 2.0344031798527425e-06, "loss": 0.1548, "step": 10454 }, { "epoch": 1.42, "grad_norm": 3.0570019896770226, "learning_rate": 2.0335163857978747e-06, "loss": 0.141, "step": 10455 }, { "epoch": 1.42, "grad_norm": 3.6841041082941763, "learning_rate": 2.032629735723165e-06, "loss": 0.1703, "step": 10456 }, { "epoch": 1.42, "grad_norm": 5.042793082996306, "learning_rate": 2.0317432296716475e-06, "loss": 0.1907, "step": 10457 }, { "epoch": 1.42, "grad_norm": 4.3180119761840094, "learning_rate": 2.0308568676863498e-06, "loss": 0.1544, "step": 10458 }, { "epoch": 1.42, "grad_norm": 2.994018327596987, "learning_rate": 2.029970649810291e-06, "loss": 0.148, "step": 10459 }, { "epoch": 1.42, "grad_norm": 3.518921695903864, "learning_rate": 2.0290845760864856e-06, "loss": 0.1683, "step": 10460 }, { "epoch": 1.42, "grad_norm": 4.059530134019969, "learning_rate": 2.0281986465579388e-06, "loss": 0.184, "step": 10461 }, { "epoch": 1.42, "grad_norm": 3.3954906165116157, "learning_rate": 2.0273128612676506e-06, "loss": 0.1423, "step": 10462 }, { "epoch": 1.42, "grad_norm": 3.0438965142534378, "learning_rate": 2.026427220258612e-06, "loss": 0.1126, "step": 10463 }, { "epoch": 1.42, "grad_norm": 3.4982620447439685, "learning_rate": 2.0255417235738074e-06, "loss": 0.1836, "step": 10464 }, { "epoch": 1.42, "grad_norm": 3.8894874598518427, "learning_rate": 2.0246563712562195e-06, "loss": 0.2159, "step": 10465 }, { "epoch": 1.42, "grad_norm": 4.246371287993773, "learning_rate": 2.023771163348812e-06, "loss": 0.1463, "step": 10466 }, { "epoch": 1.42, "grad_norm": 3.3406441876451827, "learning_rate": 2.0228860998945577e-06, "loss": 0.1838, "step": 10467 }, { "epoch": 1.42, "grad_norm": 3.354596682222416, "learning_rate": 2.0220011809364054e-06, "loss": 0.1447, "step": 10468 }, { "epoch": 1.42, "grad_norm": 3.913538438830017, "learning_rate": 2.021116406517311e-06, "loss": 0.1429, "step": 10469 }, { "epoch": 1.42, "grad_norm": 3.4068326327455973, "learning_rate": 2.0202317766802155e-06, "loss": 0.1505, "step": 10470 }, { "epoch": 1.42, "grad_norm": 2.5835348914366625, "learning_rate": 2.019347291468055e-06, "loss": 0.1403, "step": 10471 }, { "epoch": 1.42, "grad_norm": 3.298007574315884, "learning_rate": 2.0184629509237583e-06, "loss": 0.1886, "step": 10472 }, { "epoch": 1.42, "grad_norm": 2.689131875124843, "learning_rate": 2.017578755090249e-06, "loss": 0.1372, "step": 10473 }, { "epoch": 1.42, "grad_norm": 2.9559548996564127, "learning_rate": 2.0166947040104396e-06, "loss": 0.1521, "step": 10474 }, { "epoch": 1.42, "grad_norm": 2.8779883940939217, "learning_rate": 2.01581079772724e-06, "loss": 0.1679, "step": 10475 }, { "epoch": 1.43, "grad_norm": 3.1941238354382917, "learning_rate": 2.0149270362835504e-06, "loss": 0.1977, "step": 10476 }, { "epoch": 1.43, "grad_norm": 4.7395488945858535, "learning_rate": 2.0140434197222647e-06, "loss": 0.2075, "step": 10477 }, { "epoch": 1.43, "grad_norm": 3.7106928669241164, "learning_rate": 2.013159948086268e-06, "loss": 0.1755, "step": 10478 }, { "epoch": 1.43, "grad_norm": 3.8255503753296076, "learning_rate": 2.0122766214184457e-06, "loss": 0.1822, "step": 10479 }, { "epoch": 1.43, "grad_norm": 3.877567546866184, "learning_rate": 2.0113934397616634e-06, "loss": 0.1602, "step": 10480 }, { "epoch": 1.43, "grad_norm": 3.866947784433186, "learning_rate": 2.0105104031587947e-06, "loss": 0.1565, "step": 10481 }, { "epoch": 1.43, "grad_norm": 3.4172973824146684, "learning_rate": 2.00962751165269e-06, "loss": 0.1394, "step": 10482 }, { "epoch": 1.43, "grad_norm": 3.456533864792826, "learning_rate": 2.008744765286207e-06, "loss": 0.1638, "step": 10483 }, { "epoch": 1.43, "grad_norm": 3.6396284616181673, "learning_rate": 2.0078621641021884e-06, "loss": 0.1362, "step": 10484 }, { "epoch": 1.43, "grad_norm": 4.5895525787954385, "learning_rate": 2.006979708143472e-06, "loss": 0.1723, "step": 10485 }, { "epoch": 1.43, "grad_norm": 3.3571777062464294, "learning_rate": 2.0060973974528873e-06, "loss": 0.1704, "step": 10486 }, { "epoch": 1.43, "grad_norm": 2.566316636691437, "learning_rate": 2.00521523207326e-06, "loss": 0.1597, "step": 10487 }, { "epoch": 1.43, "grad_norm": 3.1393089405717527, "learning_rate": 2.0043332120474044e-06, "loss": 0.14, "step": 10488 }, { "epoch": 1.43, "grad_norm": 3.040357321332443, "learning_rate": 2.0034513374181314e-06, "loss": 0.1638, "step": 10489 }, { "epoch": 1.43, "grad_norm": 3.2359465848567552, "learning_rate": 2.0025696082282427e-06, "loss": 0.1314, "step": 10490 }, { "epoch": 1.43, "grad_norm": 3.0387413346166703, "learning_rate": 2.0016880245205335e-06, "loss": 0.1456, "step": 10491 }, { "epoch": 1.43, "grad_norm": 3.0502367561157184, "learning_rate": 2.0008065863377903e-06, "loss": 0.1877, "step": 10492 }, { "epoch": 1.43, "grad_norm": 4.615627180862988, "learning_rate": 1.9999252937227993e-06, "loss": 0.1488, "step": 10493 }, { "epoch": 1.43, "grad_norm": 3.157304331806603, "learning_rate": 1.999044146718328e-06, "loss": 0.1371, "step": 10494 }, { "epoch": 1.43, "grad_norm": 3.335267563558094, "learning_rate": 1.998163145367151e-06, "loss": 0.1473, "step": 10495 }, { "epoch": 1.43, "grad_norm": 3.6824267492139784, "learning_rate": 1.9972822897120196e-06, "loss": 0.1473, "step": 10496 }, { "epoch": 1.43, "grad_norm": 2.91236070873506, "learning_rate": 1.996401579795693e-06, "loss": 0.199, "step": 10497 }, { "epoch": 1.43, "grad_norm": 6.4718687506037025, "learning_rate": 1.9955210156609177e-06, "loss": 0.1931, "step": 10498 }, { "epoch": 1.43, "grad_norm": 3.058166519066109, "learning_rate": 1.9946405973504257e-06, "loss": 0.165, "step": 10499 }, { "epoch": 1.43, "grad_norm": 3.8460369147206106, "learning_rate": 1.993760324906957e-06, "loss": 0.1881, "step": 10500 }, { "epoch": 1.43, "grad_norm": 2.4385858489375183, "learning_rate": 1.992880198373228e-06, "loss": 0.1374, "step": 10501 }, { "epoch": 1.43, "grad_norm": 3.6395549829156546, "learning_rate": 1.9920002177919622e-06, "loss": 0.1691, "step": 10502 }, { "epoch": 1.43, "grad_norm": 3.907872631344083, "learning_rate": 1.9911203832058686e-06, "loss": 0.1718, "step": 10503 }, { "epoch": 1.43, "grad_norm": 3.214891135098667, "learning_rate": 1.990240694657649e-06, "loss": 0.1834, "step": 10504 }, { "epoch": 1.43, "grad_norm": 3.5019071374369166, "learning_rate": 1.9893611521900014e-06, "loss": 0.1638, "step": 10505 }, { "epoch": 1.43, "grad_norm": 3.9513777147071236, "learning_rate": 1.988481755845614e-06, "loss": 0.1517, "step": 10506 }, { "epoch": 1.43, "grad_norm": 3.321105285279852, "learning_rate": 1.987602505667169e-06, "loss": 0.1381, "step": 10507 }, { "epoch": 1.43, "grad_norm": 3.5301563384688945, "learning_rate": 1.9867234016973407e-06, "loss": 0.1461, "step": 10508 }, { "epoch": 1.43, "grad_norm": 3.9185217177936353, "learning_rate": 1.9858444439787983e-06, "loss": 0.1594, "step": 10509 }, { "epoch": 1.43, "grad_norm": 3.995165957542971, "learning_rate": 1.984965632554202e-06, "loss": 0.1577, "step": 10510 }, { "epoch": 1.43, "grad_norm": 3.8049972017207225, "learning_rate": 1.984086967466204e-06, "loss": 0.1767, "step": 10511 }, { "epoch": 1.43, "grad_norm": 3.0457855261890403, "learning_rate": 1.983208448757455e-06, "loss": 0.1412, "step": 10512 }, { "epoch": 1.43, "grad_norm": 3.320413238296263, "learning_rate": 1.982330076470589e-06, "loss": 0.1647, "step": 10513 }, { "epoch": 1.43, "grad_norm": 2.6004389325733146, "learning_rate": 1.9814518506482437e-06, "loss": 0.1665, "step": 10514 }, { "epoch": 1.43, "grad_norm": 4.039666815754888, "learning_rate": 1.9805737713330385e-06, "loss": 0.2351, "step": 10515 }, { "epoch": 1.43, "grad_norm": 4.466455150279747, "learning_rate": 1.979695838567597e-06, "loss": 0.1671, "step": 10516 }, { "epoch": 1.43, "grad_norm": 4.267653153721533, "learning_rate": 1.978818052394528e-06, "loss": 0.1301, "step": 10517 }, { "epoch": 1.43, "grad_norm": 4.647983481579305, "learning_rate": 1.977940412856435e-06, "loss": 0.1557, "step": 10518 }, { "epoch": 1.43, "grad_norm": 4.16478469672467, "learning_rate": 1.977062919995915e-06, "loss": 0.1548, "step": 10519 }, { "epoch": 1.43, "grad_norm": 4.739339710344098, "learning_rate": 1.976185573855558e-06, "loss": 0.1928, "step": 10520 }, { "epoch": 1.43, "grad_norm": 3.617144070711437, "learning_rate": 1.975308374477947e-06, "loss": 0.1685, "step": 10521 }, { "epoch": 1.43, "grad_norm": 3.854375368614331, "learning_rate": 1.974431321905656e-06, "loss": 0.1764, "step": 10522 }, { "epoch": 1.43, "grad_norm": 2.986038553796051, "learning_rate": 1.9735544161812526e-06, "loss": 0.1546, "step": 10523 }, { "epoch": 1.43, "grad_norm": 5.131567133055447, "learning_rate": 1.9726776573473033e-06, "loss": 0.1702, "step": 10524 }, { "epoch": 1.43, "grad_norm": 4.166683741237145, "learning_rate": 1.971801045446355e-06, "loss": 0.176, "step": 10525 }, { "epoch": 1.43, "grad_norm": 3.630823746537456, "learning_rate": 1.970924580520961e-06, "loss": 0.165, "step": 10526 }, { "epoch": 1.43, "grad_norm": 4.987216021705762, "learning_rate": 1.9700482626136548e-06, "loss": 0.1837, "step": 10527 }, { "epoch": 1.43, "grad_norm": 5.848149555435679, "learning_rate": 1.9691720917669754e-06, "loss": 0.1928, "step": 10528 }, { "epoch": 1.43, "grad_norm": 3.2959645658751504, "learning_rate": 1.968296068023442e-06, "loss": 0.166, "step": 10529 }, { "epoch": 1.43, "grad_norm": 3.39407702658891, "learning_rate": 1.9674201914255782e-06, "loss": 0.1865, "step": 10530 }, { "epoch": 1.43, "grad_norm": 3.730776457188892, "learning_rate": 1.966544462015893e-06, "loss": 0.1499, "step": 10531 }, { "epoch": 1.43, "grad_norm": 3.368066030941584, "learning_rate": 1.9656688798368905e-06, "loss": 0.2004, "step": 10532 }, { "epoch": 1.43, "grad_norm": 4.0631710542210175, "learning_rate": 1.964793444931068e-06, "loss": 0.1572, "step": 10533 }, { "epoch": 1.43, "grad_norm": 4.330486713336361, "learning_rate": 1.9639181573409143e-06, "loss": 0.1568, "step": 10534 }, { "epoch": 1.43, "grad_norm": 4.718791051215653, "learning_rate": 1.963043017108914e-06, "loss": 0.2048, "step": 10535 }, { "epoch": 1.43, "grad_norm": 3.2449868803276747, "learning_rate": 1.962168024277541e-06, "loss": 0.1382, "step": 10536 }, { "epoch": 1.43, "grad_norm": 4.508968079678757, "learning_rate": 1.9612931788892637e-06, "loss": 0.1683, "step": 10537 }, { "epoch": 1.43, "grad_norm": 3.1917979253068043, "learning_rate": 1.960418480986544e-06, "loss": 0.1565, "step": 10538 }, { "epoch": 1.43, "grad_norm": 4.640726106835178, "learning_rate": 1.959543930611835e-06, "loss": 0.1877, "step": 10539 }, { "epoch": 1.43, "grad_norm": 2.8241122477604077, "learning_rate": 1.9586695278075846e-06, "loss": 0.1282, "step": 10540 }, { "epoch": 1.43, "grad_norm": 2.8462096666588055, "learning_rate": 1.957795272616232e-06, "loss": 0.1358, "step": 10541 }, { "epoch": 1.43, "grad_norm": 3.820318010251468, "learning_rate": 1.956921165080208e-06, "loss": 0.1626, "step": 10542 }, { "epoch": 1.43, "grad_norm": 3.8923713396885415, "learning_rate": 1.9560472052419427e-06, "loss": 0.2116, "step": 10543 }, { "epoch": 1.43, "grad_norm": 3.410829511322439, "learning_rate": 1.9551733931438477e-06, "loss": 0.1701, "step": 10544 }, { "epoch": 1.43, "grad_norm": 5.018227728807921, "learning_rate": 1.954299728828342e-06, "loss": 0.1517, "step": 10545 }, { "epoch": 1.43, "grad_norm": 4.221020412022939, "learning_rate": 1.9534262123378205e-06, "loss": 0.1686, "step": 10546 }, { "epoch": 1.43, "grad_norm": 4.7536638501988735, "learning_rate": 1.9525528437146886e-06, "loss": 0.143, "step": 10547 }, { "epoch": 1.43, "grad_norm": 3.2987303658107923, "learning_rate": 1.9516796230013275e-06, "loss": 0.1528, "step": 10548 }, { "epoch": 1.43, "grad_norm": 3.8723375787721306, "learning_rate": 1.9508065502401256e-06, "loss": 0.1452, "step": 10549 }, { "epoch": 1.44, "grad_norm": 3.8230230625180908, "learning_rate": 1.9499336254734563e-06, "loss": 0.1572, "step": 10550 }, { "epoch": 1.44, "grad_norm": 4.088831843680159, "learning_rate": 1.9490608487436873e-06, "loss": 0.1379, "step": 10551 }, { "epoch": 1.44, "grad_norm": 4.039684881734551, "learning_rate": 1.9481882200931794e-06, "loss": 0.1694, "step": 10552 }, { "epoch": 1.44, "grad_norm": 3.8717408095966044, "learning_rate": 1.947315739564286e-06, "loss": 0.1837, "step": 10553 }, { "epoch": 1.44, "grad_norm": 3.7573708326867776, "learning_rate": 1.9464434071993536e-06, "loss": 0.1669, "step": 10554 }, { "epoch": 1.44, "grad_norm": 3.8267155003409608, "learning_rate": 1.945571223040722e-06, "loss": 0.1552, "step": 10555 }, { "epoch": 1.44, "grad_norm": 5.086683843289245, "learning_rate": 1.9446991871307204e-06, "loss": 0.1684, "step": 10556 }, { "epoch": 1.44, "grad_norm": 3.083186508806233, "learning_rate": 1.94382729951168e-06, "loss": 0.1223, "step": 10557 }, { "epoch": 1.44, "grad_norm": 3.2846470213004073, "learning_rate": 1.9429555602259096e-06, "loss": 0.2008, "step": 10558 }, { "epoch": 1.44, "grad_norm": 3.889560344882086, "learning_rate": 1.9420839693157288e-06, "loss": 0.1481, "step": 10559 }, { "epoch": 1.44, "grad_norm": 3.0345287951191597, "learning_rate": 1.9412125268234326e-06, "loss": 0.1566, "step": 10560 }, { "epoch": 1.44, "grad_norm": 2.83593317427809, "learning_rate": 1.9403412327913244e-06, "loss": 0.1911, "step": 10561 }, { "epoch": 1.44, "grad_norm": 2.900300274027987, "learning_rate": 1.9394700872616856e-06, "loss": 0.164, "step": 10562 }, { "epoch": 1.44, "grad_norm": 3.0558370200230534, "learning_rate": 1.938599090276804e-06, "loss": 0.1391, "step": 10563 }, { "epoch": 1.44, "grad_norm": 3.361632185707993, "learning_rate": 1.937728241878951e-06, "loss": 0.1627, "step": 10564 }, { "epoch": 1.44, "grad_norm": 3.2145986119666468, "learning_rate": 1.936857542110394e-06, "loss": 0.1406, "step": 10565 }, { "epoch": 1.44, "grad_norm": 3.2523614097218725, "learning_rate": 1.935986991013394e-06, "loss": 0.1762, "step": 10566 }, { "epoch": 1.44, "grad_norm": 2.7875434055596795, "learning_rate": 1.9351165886302026e-06, "loss": 0.1487, "step": 10567 }, { "epoch": 1.44, "grad_norm": 3.227665895245711, "learning_rate": 1.934246335003066e-06, "loss": 0.1496, "step": 10568 }, { "epoch": 1.44, "grad_norm": 3.660755219682904, "learning_rate": 1.933376230174222e-06, "loss": 0.163, "step": 10569 }, { "epoch": 1.44, "grad_norm": 4.571846076130588, "learning_rate": 1.9325062741859002e-06, "loss": 0.2179, "step": 10570 }, { "epoch": 1.44, "grad_norm": 3.4196996542071294, "learning_rate": 1.93163646708033e-06, "loss": 0.1676, "step": 10571 }, { "epoch": 1.44, "grad_norm": 4.086641073033784, "learning_rate": 1.9307668088997206e-06, "loss": 0.1709, "step": 10572 }, { "epoch": 1.44, "grad_norm": 3.6975353444335446, "learning_rate": 1.929897299686289e-06, "loss": 0.1586, "step": 10573 }, { "epoch": 1.44, "grad_norm": 3.1364912905552713, "learning_rate": 1.9290279394822297e-06, "loss": 0.1577, "step": 10574 }, { "epoch": 1.44, "grad_norm": 3.5732474838472017, "learning_rate": 1.928158728329743e-06, "loss": 0.1476, "step": 10575 }, { "epoch": 1.44, "grad_norm": 3.9928476504930672, "learning_rate": 1.927289666271015e-06, "loss": 0.1116, "step": 10576 }, { "epoch": 1.44, "grad_norm": 3.4799292637312926, "learning_rate": 1.9264207533482264e-06, "loss": 0.1655, "step": 10577 }, { "epoch": 1.44, "grad_norm": 6.689707023956427, "learning_rate": 1.9255519896035517e-06, "loss": 0.1424, "step": 10578 }, { "epoch": 1.44, "grad_norm": 4.2499427653389, "learning_rate": 1.9246833750791526e-06, "loss": 0.1628, "step": 10579 }, { "epoch": 1.44, "grad_norm": 4.082021810524778, "learning_rate": 1.923814909817194e-06, "loss": 0.1595, "step": 10580 }, { "epoch": 1.44, "grad_norm": 3.595926000673415, "learning_rate": 1.9229465938598204e-06, "loss": 0.1346, "step": 10581 }, { "epoch": 1.44, "grad_norm": 3.1225250627154773, "learning_rate": 1.922078427249181e-06, "loss": 0.1277, "step": 10582 }, { "epoch": 1.44, "grad_norm": 3.597612534394076, "learning_rate": 1.921210410027412e-06, "loss": 0.1545, "step": 10583 }, { "epoch": 1.44, "grad_norm": 2.535509421706599, "learning_rate": 1.9203425422366432e-06, "loss": 0.1259, "step": 10584 }, { "epoch": 1.44, "grad_norm": 5.733060941744238, "learning_rate": 1.9194748239189963e-06, "loss": 0.1782, "step": 10585 }, { "epoch": 1.44, "grad_norm": 3.0919704697589934, "learning_rate": 1.9186072551165868e-06, "loss": 0.1793, "step": 10586 }, { "epoch": 1.44, "grad_norm": 3.140987140417504, "learning_rate": 1.917739835871523e-06, "loss": 0.1537, "step": 10587 }, { "epoch": 1.44, "grad_norm": 3.3367055358112503, "learning_rate": 1.9168725662259054e-06, "loss": 0.1561, "step": 10588 }, { "epoch": 1.44, "grad_norm": 2.883552446961325, "learning_rate": 1.9160054462218257e-06, "loss": 0.1265, "step": 10589 }, { "epoch": 1.44, "grad_norm": 3.19768788084304, "learning_rate": 1.915138475901375e-06, "loss": 0.1501, "step": 10590 }, { "epoch": 1.44, "grad_norm": 2.9967625557831634, "learning_rate": 1.9142716553066266e-06, "loss": 0.1708, "step": 10591 }, { "epoch": 1.44, "grad_norm": 3.8206852438183336, "learning_rate": 1.9134049844796583e-06, "loss": 0.1486, "step": 10592 }, { "epoch": 1.44, "grad_norm": 3.1335808696442085, "learning_rate": 1.912538463462527e-06, "loss": 0.1172, "step": 10593 }, { "epoch": 1.44, "grad_norm": 2.4064858675178398, "learning_rate": 1.9116720922972957e-06, "loss": 0.1266, "step": 10594 }, { "epoch": 1.44, "grad_norm": 3.14535148139901, "learning_rate": 1.9108058710260126e-06, "loss": 0.1583, "step": 10595 }, { "epoch": 1.44, "grad_norm": 3.5787709612151626, "learning_rate": 1.90993979969072e-06, "loss": 0.1344, "step": 10596 }, { "epoch": 1.44, "grad_norm": 3.253671878429446, "learning_rate": 1.9090738783334535e-06, "loss": 0.1594, "step": 10597 }, { "epoch": 1.44, "grad_norm": 2.679473398415882, "learning_rate": 1.9082081069962405e-06, "loss": 0.1241, "step": 10598 }, { "epoch": 1.44, "grad_norm": 3.705191252414115, "learning_rate": 1.9073424857211026e-06, "loss": 0.1506, "step": 10599 }, { "epoch": 1.44, "grad_norm": 3.3356924464672244, "learning_rate": 1.9064770145500534e-06, "loss": 0.1275, "step": 10600 }, { "epoch": 1.44, "grad_norm": 4.08326496280693, "learning_rate": 1.905611693525098e-06, "loss": 0.1487, "step": 10601 }, { "epoch": 1.44, "grad_norm": 3.240859222289073, "learning_rate": 1.904746522688236e-06, "loss": 0.1588, "step": 10602 }, { "epoch": 1.44, "grad_norm": 3.9082150907098065, "learning_rate": 1.9038815020814572e-06, "loss": 0.1524, "step": 10603 }, { "epoch": 1.44, "grad_norm": 2.8264148943395866, "learning_rate": 1.9030166317467514e-06, "loss": 0.1337, "step": 10604 }, { "epoch": 1.44, "grad_norm": 3.050201703825674, "learning_rate": 1.9021519117260878e-06, "loss": 0.1428, "step": 10605 }, { "epoch": 1.44, "grad_norm": 3.903829504490811, "learning_rate": 1.9012873420614441e-06, "loss": 0.157, "step": 10606 }, { "epoch": 1.44, "grad_norm": 4.66334923961528, "learning_rate": 1.9004229227947752e-06, "loss": 0.2054, "step": 10607 }, { "epoch": 1.44, "grad_norm": 3.593819104105935, "learning_rate": 1.8995586539680422e-06, "loss": 0.1354, "step": 10608 }, { "epoch": 1.44, "grad_norm": 3.1142311610131057, "learning_rate": 1.89869453562319e-06, "loss": 0.1217, "step": 10609 }, { "epoch": 1.44, "grad_norm": 3.9816747963750516, "learning_rate": 1.8978305678021598e-06, "loss": 0.1242, "step": 10610 }, { "epoch": 1.44, "grad_norm": 3.873068565355812, "learning_rate": 1.8969667505468852e-06, "loss": 0.1463, "step": 10611 }, { "epoch": 1.44, "grad_norm": 4.133157554296324, "learning_rate": 1.896103083899291e-06, "loss": 0.1608, "step": 10612 }, { "epoch": 1.44, "grad_norm": 2.887832609712894, "learning_rate": 1.895239567901297e-06, "loss": 0.1415, "step": 10613 }, { "epoch": 1.44, "grad_norm": 3.7050965073475473, "learning_rate": 1.8943762025948142e-06, "loss": 0.1486, "step": 10614 }, { "epoch": 1.44, "grad_norm": 4.5810134656168096, "learning_rate": 1.8935129880217461e-06, "loss": 0.1746, "step": 10615 }, { "epoch": 1.44, "grad_norm": 3.267839888383298, "learning_rate": 1.8926499242239893e-06, "loss": 0.1713, "step": 10616 }, { "epoch": 1.44, "grad_norm": 3.1356596631143137, "learning_rate": 1.891787011243434e-06, "loss": 0.1384, "step": 10617 }, { "epoch": 1.44, "grad_norm": 2.666196987212578, "learning_rate": 1.8909242491219615e-06, "loss": 0.0968, "step": 10618 }, { "epoch": 1.44, "grad_norm": 2.9496794433539484, "learning_rate": 1.890061637901447e-06, "loss": 0.1788, "step": 10619 }, { "epoch": 1.44, "grad_norm": 2.762653022513704, "learning_rate": 1.8891991776237573e-06, "loss": 0.1226, "step": 10620 }, { "epoch": 1.44, "grad_norm": 4.119772600987624, "learning_rate": 1.8883368683307523e-06, "loss": 0.1233, "step": 10621 }, { "epoch": 1.44, "grad_norm": 4.334465312327587, "learning_rate": 1.8874747100642844e-06, "loss": 0.1713, "step": 10622 }, { "epoch": 1.45, "grad_norm": 2.8628171299080782, "learning_rate": 1.886612702866203e-06, "loss": 0.1309, "step": 10623 }, { "epoch": 1.45, "grad_norm": 3.3462458786433897, "learning_rate": 1.8857508467783392e-06, "loss": 0.1297, "step": 10624 }, { "epoch": 1.45, "grad_norm": 3.43569532746882, "learning_rate": 1.8848891418425308e-06, "loss": 0.1248, "step": 10625 }, { "epoch": 1.45, "grad_norm": 3.210859412349111, "learning_rate": 1.884027588100595e-06, "loss": 0.1301, "step": 10626 }, { "epoch": 1.45, "grad_norm": 3.92079893176052, "learning_rate": 1.8831661855943517e-06, "loss": 0.1506, "step": 10627 }, { "epoch": 1.45, "grad_norm": 3.0475145978220954, "learning_rate": 1.8823049343656091e-06, "loss": 0.1594, "step": 10628 }, { "epoch": 1.45, "grad_norm": 3.0989824068558223, "learning_rate": 1.881443834456168e-06, "loss": 0.1581, "step": 10629 }, { "epoch": 1.45, "grad_norm": 3.6770183780052013, "learning_rate": 1.8805828859078223e-06, "loss": 0.1915, "step": 10630 }, { "epoch": 1.45, "grad_norm": 3.2589069390258154, "learning_rate": 1.8797220887623596e-06, "loss": 0.1292, "step": 10631 }, { "epoch": 1.45, "grad_norm": 3.6105859759315404, "learning_rate": 1.8788614430615582e-06, "loss": 0.1665, "step": 10632 }, { "epoch": 1.45, "grad_norm": 4.060919111906626, "learning_rate": 1.8780009488471902e-06, "loss": 0.1467, "step": 10633 }, { "epoch": 1.45, "grad_norm": 4.023005529677817, "learning_rate": 1.8771406061610202e-06, "loss": 0.1607, "step": 10634 }, { "epoch": 1.45, "grad_norm": 5.217583655358869, "learning_rate": 1.8762804150448067e-06, "loss": 0.1494, "step": 10635 }, { "epoch": 1.45, "grad_norm": 3.4402622247164683, "learning_rate": 1.875420375540296e-06, "loss": 0.1562, "step": 10636 }, { "epoch": 1.45, "grad_norm": 2.7066275660063677, "learning_rate": 1.8745604876892376e-06, "loss": 0.1371, "step": 10637 }, { "epoch": 1.45, "grad_norm": 3.37034326003493, "learning_rate": 1.8737007515333589e-06, "loss": 0.1476, "step": 10638 }, { "epoch": 1.45, "grad_norm": 3.0879068502737352, "learning_rate": 1.8728411671143942e-06, "loss": 0.1438, "step": 10639 }, { "epoch": 1.45, "grad_norm": 2.8005451445426206, "learning_rate": 1.8719817344740577e-06, "loss": 0.1483, "step": 10640 }, { "epoch": 1.45, "grad_norm": 3.9747243504155056, "learning_rate": 1.8711224536540678e-06, "loss": 0.16, "step": 10641 }, { "epoch": 1.45, "grad_norm": 3.5738536155858522, "learning_rate": 1.8702633246961282e-06, "loss": 0.1979, "step": 10642 }, { "epoch": 1.45, "grad_norm": 3.0922736595639058, "learning_rate": 1.869404347641937e-06, "loss": 0.133, "step": 10643 }, { "epoch": 1.45, "grad_norm": 5.075580002893104, "learning_rate": 1.868545522533186e-06, "loss": 0.1726, "step": 10644 }, { "epoch": 1.45, "grad_norm": 4.591514001386731, "learning_rate": 1.8676868494115585e-06, "loss": 0.2137, "step": 10645 }, { "epoch": 1.45, "grad_norm": 2.779443408295293, "learning_rate": 1.86682832831873e-06, "loss": 0.1363, "step": 10646 }, { "epoch": 1.45, "grad_norm": 3.3608782468508416, "learning_rate": 1.8659699592963705e-06, "loss": 0.1601, "step": 10647 }, { "epoch": 1.45, "grad_norm": 4.733646363321449, "learning_rate": 1.865111742386141e-06, "loss": 0.197, "step": 10648 }, { "epoch": 1.45, "grad_norm": 3.1551675337070826, "learning_rate": 1.8642536776296954e-06, "loss": 0.189, "step": 10649 }, { "epoch": 1.45, "grad_norm": 4.523410471291502, "learning_rate": 1.8633957650686785e-06, "loss": 0.1637, "step": 10650 }, { "epoch": 1.45, "grad_norm": 3.1118096764239622, "learning_rate": 1.8625380047447355e-06, "loss": 0.1824, "step": 10651 }, { "epoch": 1.45, "grad_norm": 4.487262977608849, "learning_rate": 1.8616803966994912e-06, "loss": 0.217, "step": 10652 }, { "epoch": 1.45, "grad_norm": 3.0892934154355416, "learning_rate": 1.8608229409745764e-06, "loss": 0.1406, "step": 10653 }, { "epoch": 1.45, "grad_norm": 2.9754137139894765, "learning_rate": 1.8599656376116026e-06, "loss": 0.1401, "step": 10654 }, { "epoch": 1.45, "grad_norm": 3.689530379023412, "learning_rate": 1.8591084866521836e-06, "loss": 0.1562, "step": 10655 }, { "epoch": 1.45, "grad_norm": 3.849917175358422, "learning_rate": 1.8582514881379205e-06, "loss": 0.1627, "step": 10656 }, { "epoch": 1.45, "grad_norm": 4.020825744166589, "learning_rate": 1.8573946421104082e-06, "loss": 0.161, "step": 10657 }, { "epoch": 1.45, "grad_norm": 4.175689832894727, "learning_rate": 1.856537948611236e-06, "loss": 0.1567, "step": 10658 }, { "epoch": 1.45, "grad_norm": 3.5163904400068176, "learning_rate": 1.8556814076819796e-06, "loss": 0.1671, "step": 10659 }, { "epoch": 1.45, "grad_norm": 2.977150273797502, "learning_rate": 1.8548250193642153e-06, "loss": 0.1528, "step": 10660 }, { "epoch": 1.45, "grad_norm": 3.8103570659197166, "learning_rate": 1.8539687836995085e-06, "loss": 0.1686, "step": 10661 }, { "epoch": 1.45, "grad_norm": 3.3098193341719413, "learning_rate": 1.8531127007294159e-06, "loss": 0.1738, "step": 10662 }, { "epoch": 1.45, "grad_norm": 4.275941892988211, "learning_rate": 1.852256770495488e-06, "loss": 0.2012, "step": 10663 }, { "epoch": 1.45, "grad_norm": 3.491083887521044, "learning_rate": 1.851400993039269e-06, "loss": 0.1684, "step": 10664 }, { "epoch": 1.45, "grad_norm": 3.4240298170388246, "learning_rate": 1.8505453684022934e-06, "loss": 0.165, "step": 10665 }, { "epoch": 1.45, "grad_norm": 3.533800370771588, "learning_rate": 1.8496898966260906e-06, "loss": 0.1164, "step": 10666 }, { "epoch": 1.45, "grad_norm": 2.995969534264149, "learning_rate": 1.8488345777521804e-06, "loss": 0.13, "step": 10667 }, { "epoch": 1.45, "grad_norm": 5.907863373956716, "learning_rate": 1.8479794118220773e-06, "loss": 0.142, "step": 10668 }, { "epoch": 1.45, "grad_norm": 3.021371040321006, "learning_rate": 1.8471243988772846e-06, "loss": 0.1275, "step": 10669 }, { "epoch": 1.45, "grad_norm": 3.023872125895446, "learning_rate": 1.8462695389593066e-06, "loss": 0.1251, "step": 10670 }, { "epoch": 1.45, "grad_norm": 5.102046204736282, "learning_rate": 1.8454148321096272e-06, "loss": 0.1901, "step": 10671 }, { "epoch": 1.45, "grad_norm": 3.6219603236471327, "learning_rate": 1.8445602783697375e-06, "loss": 0.2035, "step": 10672 }, { "epoch": 1.45, "grad_norm": 3.7767089981000748, "learning_rate": 1.8437058777811067e-06, "loss": 0.1875, "step": 10673 }, { "epoch": 1.45, "grad_norm": 2.7932203455125966, "learning_rate": 1.842851630385209e-06, "loss": 0.1656, "step": 10674 }, { "epoch": 1.45, "grad_norm": 3.127941536081405, "learning_rate": 1.8419975362235043e-06, "loss": 0.1152, "step": 10675 }, { "epoch": 1.45, "grad_norm": 2.7699246733957272, "learning_rate": 1.8411435953374457e-06, "loss": 0.1334, "step": 10676 }, { "epoch": 1.45, "grad_norm": 3.164031367573796, "learning_rate": 1.8402898077684806e-06, "loss": 0.1092, "step": 10677 }, { "epoch": 1.45, "grad_norm": 8.257972833460187, "learning_rate": 1.8394361735580473e-06, "loss": 0.1649, "step": 10678 }, { "epoch": 1.45, "grad_norm": 3.1316007606697926, "learning_rate": 1.8385826927475785e-06, "loss": 0.1406, "step": 10679 }, { "epoch": 1.45, "grad_norm": 3.8633445059696903, "learning_rate": 1.8377293653784983e-06, "loss": 0.1746, "step": 10680 }, { "epoch": 1.45, "grad_norm": 3.5290635700260222, "learning_rate": 1.8368761914922228e-06, "loss": 0.1914, "step": 10681 }, { "epoch": 1.45, "grad_norm": 4.786900225713866, "learning_rate": 1.8360231711301618e-06, "loss": 0.206, "step": 10682 }, { "epoch": 1.45, "grad_norm": 3.4295168481972285, "learning_rate": 1.835170304333715e-06, "loss": 0.1767, "step": 10683 }, { "epoch": 1.45, "grad_norm": 4.201094037286364, "learning_rate": 1.8343175911442822e-06, "loss": 0.1354, "step": 10684 }, { "epoch": 1.45, "grad_norm": 2.822651705834643, "learning_rate": 1.8334650316032444e-06, "loss": 0.1164, "step": 10685 }, { "epoch": 1.45, "grad_norm": 2.596703618728015, "learning_rate": 1.8326126257519865e-06, "loss": 0.1145, "step": 10686 }, { "epoch": 1.45, "grad_norm": 3.3209863035030316, "learning_rate": 1.8317603736318746e-06, "loss": 0.1162, "step": 10687 }, { "epoch": 1.45, "grad_norm": 4.392123537314922, "learning_rate": 1.8309082752842782e-06, "loss": 0.1749, "step": 10688 }, { "epoch": 1.45, "grad_norm": 4.694758527666899, "learning_rate": 1.8300563307505532e-06, "loss": 0.1624, "step": 10689 }, { "epoch": 1.45, "grad_norm": 4.18732425736974, "learning_rate": 1.8292045400720488e-06, "loss": 0.1592, "step": 10690 }, { "epoch": 1.45, "grad_norm": 4.468518727220975, "learning_rate": 1.8283529032901065e-06, "loss": 0.1283, "step": 10691 }, { "epoch": 1.45, "grad_norm": 3.6226747082987334, "learning_rate": 1.8275014204460623e-06, "loss": 0.174, "step": 10692 }, { "epoch": 1.45, "grad_norm": 2.7507825762049607, "learning_rate": 1.826650091581243e-06, "loss": 0.1632, "step": 10693 }, { "epoch": 1.45, "grad_norm": 3.4351042012854287, "learning_rate": 1.8257989167369678e-06, "loss": 0.1406, "step": 10694 }, { "epoch": 1.45, "grad_norm": 3.7009244682325892, "learning_rate": 1.8249478959545503e-06, "loss": 0.1282, "step": 10695 }, { "epoch": 1.45, "grad_norm": 3.1316139329128627, "learning_rate": 1.8240970292752941e-06, "loss": 0.1735, "step": 10696 }, { "epoch": 1.46, "grad_norm": 3.827800951014132, "learning_rate": 1.8232463167404968e-06, "loss": 0.1704, "step": 10697 }, { "epoch": 1.46, "grad_norm": 3.643170520870292, "learning_rate": 1.8223957583914486e-06, "loss": 0.1234, "step": 10698 }, { "epoch": 1.46, "grad_norm": 3.0771112008129022, "learning_rate": 1.8215453542694317e-06, "loss": 0.1436, "step": 10699 }, { "epoch": 1.46, "grad_norm": 3.0871169689382185, "learning_rate": 1.8206951044157212e-06, "loss": 0.1566, "step": 10700 }, { "epoch": 1.46, "grad_norm": 3.9767513454458974, "learning_rate": 1.8198450088715841e-06, "loss": 0.2069, "step": 10701 }, { "epoch": 1.46, "grad_norm": 3.4795244837771917, "learning_rate": 1.818995067678279e-06, "loss": 0.1649, "step": 10702 }, { "epoch": 1.46, "grad_norm": 4.644032174240447, "learning_rate": 1.8181452808770638e-06, "loss": 0.2201, "step": 10703 }, { "epoch": 1.46, "grad_norm": 3.933602364404387, "learning_rate": 1.817295648509176e-06, "loss": 0.1749, "step": 10704 }, { "epoch": 1.46, "grad_norm": 3.5122669818145185, "learning_rate": 1.8164461706158597e-06, "loss": 0.1136, "step": 10705 }, { "epoch": 1.46, "grad_norm": 4.298604309143023, "learning_rate": 1.8155968472383385e-06, "loss": 0.1519, "step": 10706 }, { "epoch": 1.46, "grad_norm": 3.840431738587388, "learning_rate": 1.8147476784178398e-06, "loss": 0.1713, "step": 10707 }, { "epoch": 1.46, "grad_norm": 3.0339355166823445, "learning_rate": 1.8138986641955769e-06, "loss": 0.1438, "step": 10708 }, { "epoch": 1.46, "grad_norm": 3.544980495487839, "learning_rate": 1.813049804612757e-06, "loss": 0.165, "step": 10709 }, { "epoch": 1.46, "grad_norm": 3.7421125655287035, "learning_rate": 1.8122010997105804e-06, "loss": 0.1513, "step": 10710 }, { "epoch": 1.46, "grad_norm": 3.187948166954572, "learning_rate": 1.8113525495302393e-06, "loss": 0.1641, "step": 10711 }, { "epoch": 1.46, "grad_norm": 3.505860656304012, "learning_rate": 1.8105041541129187e-06, "loss": 0.1736, "step": 10712 }, { "epoch": 1.46, "grad_norm": 4.553500399110386, "learning_rate": 1.8096559134997959e-06, "loss": 0.1935, "step": 10713 }, { "epoch": 1.46, "grad_norm": 2.700106452851446, "learning_rate": 1.8088078277320409e-06, "loss": 0.153, "step": 10714 }, { "epoch": 1.46, "grad_norm": 3.6910875995452983, "learning_rate": 1.807959896850816e-06, "loss": 0.1646, "step": 10715 }, { "epoch": 1.46, "grad_norm": 4.238543855386155, "learning_rate": 1.8071121208972747e-06, "loss": 0.1575, "step": 10716 }, { "epoch": 1.46, "grad_norm": 3.0057254121609485, "learning_rate": 1.8062644999125694e-06, "loss": 0.1452, "step": 10717 }, { "epoch": 1.46, "grad_norm": 3.051876554906111, "learning_rate": 1.8054170339378325e-06, "loss": 0.1314, "step": 10718 }, { "epoch": 1.46, "grad_norm": 6.348703668367942, "learning_rate": 1.8045697230142035e-06, "loss": 0.1594, "step": 10719 }, { "epoch": 1.46, "grad_norm": 2.2609742075216968, "learning_rate": 1.8037225671828e-06, "loss": 0.1149, "step": 10720 }, { "epoch": 1.46, "grad_norm": 3.5032838590883535, "learning_rate": 1.8028755664847453e-06, "loss": 0.1685, "step": 10721 }, { "epoch": 1.46, "grad_norm": 3.0171158959914663, "learning_rate": 1.8020287209611464e-06, "loss": 0.1769, "step": 10722 }, { "epoch": 1.46, "grad_norm": 3.0202016633419615, "learning_rate": 1.8011820306531063e-06, "loss": 0.168, "step": 10723 }, { "epoch": 1.46, "grad_norm": 3.2882496322963592, "learning_rate": 1.800335495601719e-06, "loss": 0.1231, "step": 10724 }, { "epoch": 1.46, "grad_norm": 4.015744644064115, "learning_rate": 1.7994891158480721e-06, "loss": 0.1968, "step": 10725 }, { "epoch": 1.46, "grad_norm": 2.414372238750229, "learning_rate": 1.7986428914332443e-06, "loss": 0.1306, "step": 10726 }, { "epoch": 1.46, "grad_norm": 3.2855850647208733, "learning_rate": 1.7977968223983089e-06, "loss": 0.1546, "step": 10727 }, { "epoch": 1.46, "grad_norm": 3.707173920705449, "learning_rate": 1.7969509087843296e-06, "loss": 0.1981, "step": 10728 }, { "epoch": 1.46, "grad_norm": 3.9244590268185506, "learning_rate": 1.7961051506323636e-06, "loss": 0.1661, "step": 10729 }, { "epoch": 1.46, "grad_norm": 4.529322300210389, "learning_rate": 1.7952595479834583e-06, "loss": 0.1866, "step": 10730 }, { "epoch": 1.46, "grad_norm": 3.0884573092531165, "learning_rate": 1.7944141008786608e-06, "loss": 0.1913, "step": 10731 }, { "epoch": 1.46, "grad_norm": 4.181393380962532, "learning_rate": 1.7935688093589987e-06, "loss": 0.2018, "step": 10732 }, { "epoch": 1.46, "grad_norm": 5.851814309505634, "learning_rate": 1.7927236734655056e-06, "loss": 0.1571, "step": 10733 }, { "epoch": 1.46, "grad_norm": 5.388595745048138, "learning_rate": 1.7918786932391945e-06, "loss": 0.2046, "step": 10734 }, { "epoch": 1.46, "grad_norm": 3.3410892856899057, "learning_rate": 1.791033868721081e-06, "loss": 0.1661, "step": 10735 }, { "epoch": 1.46, "grad_norm": 2.8592527890500263, "learning_rate": 1.790189199952168e-06, "loss": 0.1433, "step": 10736 }, { "epoch": 1.46, "grad_norm": 3.129201097645521, "learning_rate": 1.789344686973452e-06, "loss": 0.154, "step": 10737 }, { "epoch": 1.46, "grad_norm": 3.552142212856844, "learning_rate": 1.7885003298259235e-06, "loss": 0.1286, "step": 10738 }, { "epoch": 1.46, "grad_norm": 4.116344353111143, "learning_rate": 1.7876561285505595e-06, "loss": 0.1706, "step": 10739 }, { "epoch": 1.46, "grad_norm": 2.9657263122129676, "learning_rate": 1.7868120831883385e-06, "loss": 0.121, "step": 10740 }, { "epoch": 1.46, "grad_norm": 3.565861950948242, "learning_rate": 1.7859681937802241e-06, "loss": 0.1916, "step": 10741 }, { "epoch": 1.46, "grad_norm": 2.5478295801156454, "learning_rate": 1.785124460367177e-06, "loss": 0.1206, "step": 10742 }, { "epoch": 1.46, "grad_norm": 3.238068214651476, "learning_rate": 1.7842808829901471e-06, "loss": 0.1723, "step": 10743 }, { "epoch": 1.46, "grad_norm": 3.299242124495815, "learning_rate": 1.7834374616900773e-06, "loss": 0.1658, "step": 10744 }, { "epoch": 1.46, "grad_norm": 4.318146313385549, "learning_rate": 1.7825941965079053e-06, "loss": 0.1926, "step": 10745 }, { "epoch": 1.46, "grad_norm": 5.712747751523906, "learning_rate": 1.7817510874845585e-06, "loss": 0.204, "step": 10746 }, { "epoch": 1.46, "grad_norm": 3.5242663322656798, "learning_rate": 1.7809081346609574e-06, "loss": 0.1666, "step": 10747 }, { "epoch": 1.46, "grad_norm": 3.755767382885962, "learning_rate": 1.7800653380780163e-06, "loss": 0.1554, "step": 10748 }, { "epoch": 1.46, "grad_norm": 3.7764150701147168, "learning_rate": 1.779222697776638e-06, "loss": 0.1774, "step": 10749 }, { "epoch": 1.46, "grad_norm": 3.723630421784046, "learning_rate": 1.7783802137977268e-06, "loss": 0.1365, "step": 10750 }, { "epoch": 1.46, "grad_norm": 3.331256903071485, "learning_rate": 1.7775378861821658e-06, "loss": 0.1699, "step": 10751 }, { "epoch": 1.46, "grad_norm": 3.657991998370184, "learning_rate": 1.7766957149708442e-06, "loss": 0.1873, "step": 10752 }, { "epoch": 1.46, "grad_norm": 3.61432389041986, "learning_rate": 1.7758537002046317e-06, "loss": 0.1589, "step": 10753 }, { "epoch": 1.46, "grad_norm": 2.610569304584023, "learning_rate": 1.7750118419244001e-06, "loss": 0.1269, "step": 10754 }, { "epoch": 1.46, "grad_norm": 3.562662500667158, "learning_rate": 1.7741701401710081e-06, "loss": 0.1386, "step": 10755 }, { "epoch": 1.46, "grad_norm": 3.91955348299847, "learning_rate": 1.7733285949853085e-06, "loss": 0.1958, "step": 10756 }, { "epoch": 1.46, "grad_norm": 3.395018297204016, "learning_rate": 1.7724872064081461e-06, "loss": 0.1898, "step": 10757 }, { "epoch": 1.46, "grad_norm": 3.9556725441767933, "learning_rate": 1.7716459744803583e-06, "loss": 0.1531, "step": 10758 }, { "epoch": 1.46, "grad_norm": 4.449172694228859, "learning_rate": 1.7708048992427746e-06, "loss": 0.1449, "step": 10759 }, { "epoch": 1.46, "grad_norm": 3.2634787133896452, "learning_rate": 1.7699639807362173e-06, "loss": 0.1268, "step": 10760 }, { "epoch": 1.46, "grad_norm": 3.8415422705572575, "learning_rate": 1.769123219001499e-06, "loss": 0.1482, "step": 10761 }, { "epoch": 1.46, "grad_norm": 3.653950968160411, "learning_rate": 1.768282614079432e-06, "loss": 0.1767, "step": 10762 }, { "epoch": 1.46, "grad_norm": 3.03308307365866, "learning_rate": 1.7674421660108089e-06, "loss": 0.0862, "step": 10763 }, { "epoch": 1.46, "grad_norm": 3.2395426482238294, "learning_rate": 1.7666018748364272e-06, "loss": 0.1398, "step": 10764 }, { "epoch": 1.46, "grad_norm": 3.0962247892350203, "learning_rate": 1.765761740597065e-06, "loss": 0.1555, "step": 10765 }, { "epoch": 1.46, "grad_norm": 3.7567055255880955, "learning_rate": 1.7649217633335058e-06, "loss": 0.181, "step": 10766 }, { "epoch": 1.46, "grad_norm": 3.361600727813622, "learning_rate": 1.7640819430865113e-06, "loss": 0.151, "step": 10767 }, { "epoch": 1.46, "grad_norm": 3.602693778371949, "learning_rate": 1.7632422798968474e-06, "loss": 0.1749, "step": 10768 }, { "epoch": 1.46, "grad_norm": 2.6690659453048764, "learning_rate": 1.7624027738052669e-06, "loss": 0.177, "step": 10769 }, { "epoch": 1.47, "grad_norm": 11.984855317706518, "learning_rate": 1.7615634248525144e-06, "loss": 0.1626, "step": 10770 }, { "epoch": 1.47, "grad_norm": 3.7101993334602557, "learning_rate": 1.7607242330793296e-06, "loss": 0.1811, "step": 10771 }, { "epoch": 1.47, "grad_norm": 4.271456969583413, "learning_rate": 1.7598851985264426e-06, "loss": 0.1422, "step": 10772 }, { "epoch": 1.47, "grad_norm": 4.36505970351692, "learning_rate": 1.759046321234576e-06, "loss": 0.1869, "step": 10773 }, { "epoch": 1.47, "grad_norm": 2.996098132076132, "learning_rate": 1.7582076012444466e-06, "loss": 0.1487, "step": 10774 }, { "epoch": 1.47, "grad_norm": 3.5286748626402744, "learning_rate": 1.757369038596759e-06, "loss": 0.1557, "step": 10775 }, { "epoch": 1.47, "grad_norm": 3.3857230299404937, "learning_rate": 1.7565306333322191e-06, "loss": 0.1435, "step": 10776 }, { "epoch": 1.47, "grad_norm": 3.3922537284692127, "learning_rate": 1.7556923854915148e-06, "loss": 0.1632, "step": 10777 }, { "epoch": 1.47, "grad_norm": 3.2720805333374368, "learning_rate": 1.7548542951153318e-06, "loss": 0.1526, "step": 10778 }, { "epoch": 1.47, "grad_norm": 3.9214679655738913, "learning_rate": 1.754016362244348e-06, "loss": 0.1446, "step": 10779 }, { "epoch": 1.47, "grad_norm": 5.747851817281868, "learning_rate": 1.753178586919231e-06, "loss": 0.1594, "step": 10780 }, { "epoch": 1.47, "grad_norm": 3.4137883845786474, "learning_rate": 1.7523409691806482e-06, "loss": 0.1688, "step": 10781 }, { "epoch": 1.47, "grad_norm": 4.375783558649258, "learning_rate": 1.7515035090692466e-06, "loss": 0.1748, "step": 10782 }, { "epoch": 1.47, "grad_norm": 3.4998771032871576, "learning_rate": 1.7506662066256802e-06, "loss": 0.1402, "step": 10783 }, { "epoch": 1.47, "grad_norm": 3.313093014586104, "learning_rate": 1.7498290618905806e-06, "loss": 0.1513, "step": 10784 }, { "epoch": 1.47, "grad_norm": 4.444158163047377, "learning_rate": 1.7489920749045862e-06, "loss": 0.185, "step": 10785 }, { "epoch": 1.47, "grad_norm": 2.9645593478533065, "learning_rate": 1.748155245708314e-06, "loss": 0.1862, "step": 10786 }, { "epoch": 1.47, "grad_norm": 3.04576286357898, "learning_rate": 1.7473185743423853e-06, "loss": 0.1463, "step": 10787 }, { "epoch": 1.47, "grad_norm": 3.1140522533821517, "learning_rate": 1.7464820608474066e-06, "loss": 0.1406, "step": 10788 }, { "epoch": 1.47, "grad_norm": 4.082821519237283, "learning_rate": 1.745645705263978e-06, "loss": 0.1498, "step": 10789 }, { "epoch": 1.47, "grad_norm": 2.804608856575295, "learning_rate": 1.7448095076326937e-06, "loss": 0.1261, "step": 10790 }, { "epoch": 1.47, "grad_norm": 3.63789617680005, "learning_rate": 1.7439734679941384e-06, "loss": 0.1435, "step": 10791 }, { "epoch": 1.47, "grad_norm": 3.6137587531991024, "learning_rate": 1.74313758638889e-06, "loss": 0.1399, "step": 10792 }, { "epoch": 1.47, "grad_norm": 3.0523163098415456, "learning_rate": 1.7423018628575184e-06, "loss": 0.1446, "step": 10793 }, { "epoch": 1.47, "grad_norm": 3.9206404815839604, "learning_rate": 1.7414662974405838e-06, "loss": 0.1683, "step": 10794 }, { "epoch": 1.47, "grad_norm": 2.6553443465522912, "learning_rate": 1.7406308901786472e-06, "loss": 0.1502, "step": 10795 }, { "epoch": 1.47, "grad_norm": 2.480103846029944, "learning_rate": 1.739795641112248e-06, "loss": 0.1224, "step": 10796 }, { "epoch": 1.47, "grad_norm": 3.6185193336293047, "learning_rate": 1.7389605502819324e-06, "loss": 0.1869, "step": 10797 }, { "epoch": 1.47, "grad_norm": 3.6093124644794456, "learning_rate": 1.7381256177282251e-06, "loss": 0.1632, "step": 10798 }, { "epoch": 1.47, "grad_norm": 3.6200125642693486, "learning_rate": 1.7372908434916575e-06, "loss": 0.1642, "step": 10799 }, { "epoch": 1.47, "grad_norm": 4.3164786980027205, "learning_rate": 1.7364562276127383e-06, "loss": 0.1708, "step": 10800 }, { "epoch": 1.47, "grad_norm": 3.70658717882397, "learning_rate": 1.7356217701319823e-06, "loss": 0.1904, "step": 10801 }, { "epoch": 1.47, "grad_norm": 3.665785377387766, "learning_rate": 1.734787471089887e-06, "loss": 0.1736, "step": 10802 }, { "epoch": 1.47, "grad_norm": 3.5910321149346482, "learning_rate": 1.7339533305269474e-06, "loss": 0.1577, "step": 10803 }, { "epoch": 1.47, "grad_norm": 2.6423292445054223, "learning_rate": 1.7331193484836485e-06, "loss": 0.1633, "step": 10804 }, { "epoch": 1.47, "grad_norm": 4.88608788405332, "learning_rate": 1.7322855250004678e-06, "loss": 0.2153, "step": 10805 }, { "epoch": 1.47, "grad_norm": 2.8605959820361146, "learning_rate": 1.7314518601178754e-06, "loss": 0.1212, "step": 10806 }, { "epoch": 1.47, "grad_norm": 3.6452704602111785, "learning_rate": 1.730618353876334e-06, "loss": 0.1458, "step": 10807 }, { "epoch": 1.47, "grad_norm": 2.7719723941003913, "learning_rate": 1.729785006316297e-06, "loss": 0.1836, "step": 10808 }, { "epoch": 1.47, "grad_norm": 2.6751207774292753, "learning_rate": 1.7289518174782154e-06, "loss": 0.1544, "step": 10809 }, { "epoch": 1.47, "grad_norm": 4.111345019391761, "learning_rate": 1.7281187874025235e-06, "loss": 0.1836, "step": 10810 }, { "epoch": 1.47, "grad_norm": 3.1438494033242326, "learning_rate": 1.727285916129658e-06, "loss": 0.1289, "step": 10811 }, { "epoch": 1.47, "grad_norm": 3.3497217318481547, "learning_rate": 1.726453203700037e-06, "loss": 0.1671, "step": 10812 }, { "epoch": 1.47, "grad_norm": 4.0425458489578165, "learning_rate": 1.7256206501540818e-06, "loss": 0.1855, "step": 10813 }, { "epoch": 1.47, "grad_norm": 4.054870950657972, "learning_rate": 1.7247882555321987e-06, "loss": 0.1605, "step": 10814 }, { "epoch": 1.47, "grad_norm": 4.504507961376386, "learning_rate": 1.723956019874789e-06, "loss": 0.1374, "step": 10815 }, { "epoch": 1.47, "grad_norm": 3.954981604820402, "learning_rate": 1.7231239432222458e-06, "loss": 0.1656, "step": 10816 }, { "epoch": 1.47, "grad_norm": 3.139832678239403, "learning_rate": 1.7222920256149544e-06, "loss": 0.1129, "step": 10817 }, { "epoch": 1.47, "grad_norm": 3.7151935950609665, "learning_rate": 1.7214602670932934e-06, "loss": 0.1602, "step": 10818 }, { "epoch": 1.47, "grad_norm": 5.169654355458423, "learning_rate": 1.7206286676976286e-06, "loss": 0.1516, "step": 10819 }, { "epoch": 1.47, "grad_norm": 3.8563185904548507, "learning_rate": 1.7197972274683273e-06, "loss": 0.1548, "step": 10820 }, { "epoch": 1.47, "grad_norm": 3.1857881106136556, "learning_rate": 1.7189659464457415e-06, "loss": 0.1621, "step": 10821 }, { "epoch": 1.47, "grad_norm": 4.3440906766272915, "learning_rate": 1.7181348246702184e-06, "loss": 0.1208, "step": 10822 }, { "epoch": 1.47, "grad_norm": 3.0577046698286674, "learning_rate": 1.717303862182097e-06, "loss": 0.1477, "step": 10823 }, { "epoch": 1.47, "grad_norm": 3.702096042058431, "learning_rate": 1.7164730590217088e-06, "loss": 0.1821, "step": 10824 }, { "epoch": 1.47, "grad_norm": 3.6498107900642807, "learning_rate": 1.715642415229377e-06, "loss": 0.1417, "step": 10825 }, { "epoch": 1.47, "grad_norm": 4.064403821946654, "learning_rate": 1.714811930845418e-06, "loss": 0.2034, "step": 10826 }, { "epoch": 1.47, "grad_norm": 4.577256816616531, "learning_rate": 1.7139816059101372e-06, "loss": 0.1714, "step": 10827 }, { "epoch": 1.47, "grad_norm": 3.0619044736788785, "learning_rate": 1.7131514404638405e-06, "loss": 0.1243, "step": 10828 }, { "epoch": 1.47, "grad_norm": 5.695302380410993, "learning_rate": 1.7123214345468141e-06, "loss": 0.1592, "step": 10829 }, { "epoch": 1.47, "grad_norm": 3.3482059201847205, "learning_rate": 1.7114915881993499e-06, "loss": 0.212, "step": 10830 }, { "epoch": 1.47, "grad_norm": 2.9525203560581383, "learning_rate": 1.710661901461717e-06, "loss": 0.1132, "step": 10831 }, { "epoch": 1.47, "grad_norm": 3.247087722512132, "learning_rate": 1.7098323743741906e-06, "loss": 0.1669, "step": 10832 }, { "epoch": 1.47, "grad_norm": 3.393451331149627, "learning_rate": 1.7090030069770309e-06, "loss": 0.1901, "step": 10833 }, { "epoch": 1.47, "grad_norm": 3.659559956432487, "learning_rate": 1.7081737993104913e-06, "loss": 0.1764, "step": 10834 }, { "epoch": 1.47, "grad_norm": 4.81277835624758, "learning_rate": 1.7073447514148178e-06, "loss": 0.1724, "step": 10835 }, { "epoch": 1.47, "grad_norm": 2.9682158535216203, "learning_rate": 1.7065158633302492e-06, "loss": 0.1412, "step": 10836 }, { "epoch": 1.47, "grad_norm": 4.800361263987534, "learning_rate": 1.705687135097016e-06, "loss": 0.1405, "step": 10837 }, { "epoch": 1.47, "grad_norm": 4.780495518195317, "learning_rate": 1.7048585667553414e-06, "loss": 0.1798, "step": 10838 }, { "epoch": 1.47, "grad_norm": 4.101615647823138, "learning_rate": 1.7040301583454393e-06, "loss": 0.1935, "step": 10839 }, { "epoch": 1.47, "grad_norm": 4.03298402309791, "learning_rate": 1.7032019099075186e-06, "loss": 0.1657, "step": 10840 }, { "epoch": 1.47, "grad_norm": 3.07901834145951, "learning_rate": 1.7023738214817758e-06, "loss": 0.1224, "step": 10841 }, { "epoch": 1.47, "grad_norm": 4.793249953032223, "learning_rate": 1.7015458931084084e-06, "loss": 0.2048, "step": 10842 }, { "epoch": 1.47, "grad_norm": 3.0781434486450636, "learning_rate": 1.700718124827594e-06, "loss": 0.1645, "step": 10843 }, { "epoch": 1.48, "grad_norm": 2.86876666938668, "learning_rate": 1.699890516679515e-06, "loss": 0.1468, "step": 10844 }, { "epoch": 1.48, "grad_norm": 2.9427554228410515, "learning_rate": 1.699063068704333e-06, "loss": 0.154, "step": 10845 }, { "epoch": 1.48, "grad_norm": 3.679909143905149, "learning_rate": 1.6982357809422146e-06, "loss": 0.1439, "step": 10846 }, { "epoch": 1.48, "grad_norm": 3.4871396677791373, "learning_rate": 1.69740865343331e-06, "loss": 0.1821, "step": 10847 }, { "epoch": 1.48, "grad_norm": 3.1409099052673466, "learning_rate": 1.6965816862177659e-06, "loss": 0.13, "step": 10848 }, { "epoch": 1.48, "grad_norm": 3.101668012248872, "learning_rate": 1.6957548793357182e-06, "loss": 0.1599, "step": 10849 }, { "epoch": 1.48, "grad_norm": 3.968476793716654, "learning_rate": 1.6949282328272964e-06, "loss": 0.1313, "step": 10850 }, { "epoch": 1.48, "grad_norm": 5.864474729148197, "learning_rate": 1.6941017467326238e-06, "loss": 0.1687, "step": 10851 }, { "epoch": 1.48, "grad_norm": 2.4112727311885616, "learning_rate": 1.6932754210918133e-06, "loss": 0.1531, "step": 10852 }, { "epoch": 1.48, "grad_norm": 3.010899157804818, "learning_rate": 1.6924492559449713e-06, "loss": 0.1432, "step": 10853 }, { "epoch": 1.48, "grad_norm": 3.032463638969639, "learning_rate": 1.6916232513321961e-06, "loss": 0.1265, "step": 10854 }, { "epoch": 1.48, "grad_norm": 3.238701808644781, "learning_rate": 1.6907974072935774e-06, "loss": 0.1731, "step": 10855 }, { "epoch": 1.48, "grad_norm": 3.569474025388782, "learning_rate": 1.689971723869202e-06, "loss": 0.1425, "step": 10856 }, { "epoch": 1.48, "grad_norm": 3.1424907933700914, "learning_rate": 1.689146201099141e-06, "loss": 0.1358, "step": 10857 }, { "epoch": 1.48, "grad_norm": 4.052306384168595, "learning_rate": 1.688320839023463e-06, "loss": 0.1495, "step": 10858 }, { "epoch": 1.48, "grad_norm": 3.421729595615316, "learning_rate": 1.6874956376822271e-06, "loss": 0.1549, "step": 10859 }, { "epoch": 1.48, "grad_norm": 3.1158073286776546, "learning_rate": 1.6866705971154835e-06, "loss": 0.1282, "step": 10860 }, { "epoch": 1.48, "grad_norm": 3.493387861708228, "learning_rate": 1.685845717363281e-06, "loss": 0.1381, "step": 10861 }, { "epoch": 1.48, "grad_norm": 3.5472502787499582, "learning_rate": 1.6850209984656497e-06, "loss": 0.1267, "step": 10862 }, { "epoch": 1.48, "grad_norm": 3.604237528689855, "learning_rate": 1.6841964404626233e-06, "loss": 0.1456, "step": 10863 }, { "epoch": 1.48, "grad_norm": 3.0003441916441447, "learning_rate": 1.6833720433942163e-06, "loss": 0.1571, "step": 10864 }, { "epoch": 1.48, "grad_norm": 3.411502553166849, "learning_rate": 1.6825478073004465e-06, "loss": 0.1284, "step": 10865 }, { "epoch": 1.48, "grad_norm": 4.480838263220635, "learning_rate": 1.6817237322213164e-06, "loss": 0.1721, "step": 10866 }, { "epoch": 1.48, "grad_norm": 2.646133989493695, "learning_rate": 1.6808998181968238e-06, "loss": 0.1492, "step": 10867 }, { "epoch": 1.48, "grad_norm": 2.9501890343548975, "learning_rate": 1.6800760652669574e-06, "loss": 0.1448, "step": 10868 }, { "epoch": 1.48, "grad_norm": 2.852922363709526, "learning_rate": 1.6792524734716987e-06, "loss": 0.1553, "step": 10869 }, { "epoch": 1.48, "grad_norm": 4.217655531094378, "learning_rate": 1.6784290428510214e-06, "loss": 0.1711, "step": 10870 }, { "epoch": 1.48, "grad_norm": 3.5512351950247267, "learning_rate": 1.6776057734448908e-06, "loss": 0.1479, "step": 10871 }, { "epoch": 1.48, "grad_norm": 3.4223270273600246, "learning_rate": 1.6767826652932651e-06, "loss": 0.154, "step": 10872 }, { "epoch": 1.48, "grad_norm": 2.70486987178097, "learning_rate": 1.6759597184360942e-06, "loss": 0.1381, "step": 10873 }, { "epoch": 1.48, "grad_norm": 3.011807549144217, "learning_rate": 1.6751369329133188e-06, "loss": 0.1713, "step": 10874 }, { "epoch": 1.48, "grad_norm": 2.9521689596012757, "learning_rate": 1.6743143087648784e-06, "loss": 0.1282, "step": 10875 }, { "epoch": 1.48, "grad_norm": 4.391580661454278, "learning_rate": 1.6734918460306927e-06, "loss": 0.1769, "step": 10876 }, { "epoch": 1.48, "grad_norm": 2.96726449259889, "learning_rate": 1.6726695447506873e-06, "loss": 0.1469, "step": 10877 }, { "epoch": 1.48, "grad_norm": 2.891015615670138, "learning_rate": 1.6718474049647665e-06, "loss": 0.124, "step": 10878 }, { "epoch": 1.48, "grad_norm": 4.023742469216612, "learning_rate": 1.6710254267128378e-06, "loss": 0.1616, "step": 10879 }, { "epoch": 1.48, "grad_norm": 4.4412036241742125, "learning_rate": 1.6702036100347958e-06, "loss": 0.1621, "step": 10880 }, { "epoch": 1.48, "grad_norm": 3.786550587722309, "learning_rate": 1.669381954970527e-06, "loss": 0.145, "step": 10881 }, { "epoch": 1.48, "grad_norm": 4.967130107295284, "learning_rate": 1.6685604615599117e-06, "loss": 0.1568, "step": 10882 }, { "epoch": 1.48, "grad_norm": 4.419079954442158, "learning_rate": 1.6677391298428208e-06, "loss": 0.1594, "step": 10883 }, { "epoch": 1.48, "grad_norm": 3.2251954560007112, "learning_rate": 1.6669179598591183e-06, "loss": 0.1924, "step": 10884 }, { "epoch": 1.48, "grad_norm": 5.744215855787245, "learning_rate": 1.6660969516486609e-06, "loss": 0.174, "step": 10885 }, { "epoch": 1.48, "grad_norm": 3.3110448751559143, "learning_rate": 1.665276105251296e-06, "loss": 0.179, "step": 10886 }, { "epoch": 1.48, "grad_norm": 3.87261680060735, "learning_rate": 1.6644554207068642e-06, "loss": 0.1539, "step": 10887 }, { "epoch": 1.48, "grad_norm": 4.0314667517762315, "learning_rate": 1.6636348980551965e-06, "loss": 0.1855, "step": 10888 }, { "epoch": 1.48, "grad_norm": 3.1915491328396275, "learning_rate": 1.662814537336122e-06, "loss": 0.1587, "step": 10889 }, { "epoch": 1.48, "grad_norm": 4.486303396963563, "learning_rate": 1.661994338589451e-06, "loss": 0.1992, "step": 10890 }, { "epoch": 1.48, "grad_norm": 3.555219531906129, "learning_rate": 1.6611743018549992e-06, "loss": 0.191, "step": 10891 }, { "epoch": 1.48, "grad_norm": 3.593315626572119, "learning_rate": 1.6603544271725607e-06, "loss": 0.1379, "step": 10892 }, { "epoch": 1.48, "grad_norm": 3.095200359416293, "learning_rate": 1.6595347145819335e-06, "loss": 0.1715, "step": 10893 }, { "epoch": 1.48, "grad_norm": 4.198960518185445, "learning_rate": 1.6587151641229016e-06, "loss": 0.1186, "step": 10894 }, { "epoch": 1.48, "grad_norm": 4.255848014930984, "learning_rate": 1.657895775835242e-06, "loss": 0.1555, "step": 10895 }, { "epoch": 1.48, "grad_norm": 3.4770440122682564, "learning_rate": 1.6570765497587248e-06, "loss": 0.1606, "step": 10896 }, { "epoch": 1.48, "grad_norm": 3.3232857080938687, "learning_rate": 1.656257485933111e-06, "loss": 0.137, "step": 10897 }, { "epoch": 1.48, "grad_norm": 4.573071188322104, "learning_rate": 1.655438584398155e-06, "loss": 0.1768, "step": 10898 }, { "epoch": 1.48, "grad_norm": 4.035681203763185, "learning_rate": 1.6546198451936024e-06, "loss": 0.1786, "step": 10899 }, { "epoch": 1.48, "grad_norm": 3.9951574972828188, "learning_rate": 1.6538012683591908e-06, "loss": 0.1426, "step": 10900 }, { "epoch": 1.48, "grad_norm": 4.322018131710561, "learning_rate": 1.652982853934651e-06, "loss": 0.1554, "step": 10901 }, { "epoch": 1.48, "grad_norm": 4.292321556933536, "learning_rate": 1.652164601959705e-06, "loss": 0.174, "step": 10902 }, { "epoch": 1.48, "grad_norm": 3.542868137988411, "learning_rate": 1.651346512474067e-06, "loss": 0.1054, "step": 10903 }, { "epoch": 1.48, "grad_norm": 2.9633930574072, "learning_rate": 1.6505285855174435e-06, "loss": 0.1326, "step": 10904 }, { "epoch": 1.48, "grad_norm": 3.099611000493238, "learning_rate": 1.6497108211295338e-06, "loss": 0.1448, "step": 10905 }, { "epoch": 1.48, "grad_norm": 3.0868428525327003, "learning_rate": 1.648893219350028e-06, "loss": 0.1629, "step": 10906 }, { "epoch": 1.48, "grad_norm": 2.9560212940444934, "learning_rate": 1.648075780218607e-06, "loss": 0.1548, "step": 10907 }, { "epoch": 1.48, "grad_norm": 3.1919467472942857, "learning_rate": 1.647258503774951e-06, "loss": 0.1419, "step": 10908 }, { "epoch": 1.48, "grad_norm": 2.744746102693665, "learning_rate": 1.6464413900587212e-06, "loss": 0.1389, "step": 10909 }, { "epoch": 1.48, "grad_norm": 2.7386616768745267, "learning_rate": 1.6456244391095815e-06, "loss": 0.1147, "step": 10910 }, { "epoch": 1.48, "grad_norm": 3.3756044086334045, "learning_rate": 1.6448076509671783e-06, "loss": 0.1411, "step": 10911 }, { "epoch": 1.48, "grad_norm": 4.921681032985672, "learning_rate": 1.6439910256711595e-06, "loss": 0.1936, "step": 10912 }, { "epoch": 1.48, "grad_norm": 3.4077875884544286, "learning_rate": 1.6431745632611578e-06, "loss": 0.1275, "step": 10913 }, { "epoch": 1.48, "grad_norm": 3.117101006667899, "learning_rate": 1.6423582637768021e-06, "loss": 0.1354, "step": 10914 }, { "epoch": 1.48, "grad_norm": 2.990436758699797, "learning_rate": 1.6415421272577114e-06, "loss": 0.1555, "step": 10915 }, { "epoch": 1.48, "grad_norm": 3.0635949056350116, "learning_rate": 1.640726153743497e-06, "loss": 0.1856, "step": 10916 }, { "epoch": 1.49, "grad_norm": 3.1304749146402413, "learning_rate": 1.6399103432737635e-06, "loss": 0.1772, "step": 10917 }, { "epoch": 1.49, "grad_norm": 2.9988780991179356, "learning_rate": 1.639094695888106e-06, "loss": 0.1562, "step": 10918 }, { "epoch": 1.49, "grad_norm": 2.700480000297729, "learning_rate": 1.6382792116261131e-06, "loss": 0.1353, "step": 10919 }, { "epoch": 1.49, "grad_norm": 3.410786792986946, "learning_rate": 1.6374638905273643e-06, "loss": 0.1538, "step": 10920 }, { "epoch": 1.49, "grad_norm": 3.3667681822308184, "learning_rate": 1.6366487326314305e-06, "loss": 0.1815, "step": 10921 }, { "epoch": 1.49, "grad_norm": 3.6086615980307064, "learning_rate": 1.635833737977881e-06, "loss": 0.1642, "step": 10922 }, { "epoch": 1.49, "grad_norm": 3.039176672982941, "learning_rate": 1.6350189066062649e-06, "loss": 0.1527, "step": 10923 }, { "epoch": 1.49, "grad_norm": 3.191477578387162, "learning_rate": 1.634204238556138e-06, "loss": 0.1647, "step": 10924 }, { "epoch": 1.49, "grad_norm": 4.993473420862739, "learning_rate": 1.633389733867033e-06, "loss": 0.1686, "step": 10925 }, { "epoch": 1.49, "grad_norm": 3.981429742189772, "learning_rate": 1.6325753925784883e-06, "loss": 0.1722, "step": 10926 }, { "epoch": 1.49, "grad_norm": 3.1358948080601814, "learning_rate": 1.631761214730026e-06, "loss": 0.1648, "step": 10927 }, { "epoch": 1.49, "grad_norm": 3.36448489058016, "learning_rate": 1.6309472003611632e-06, "loss": 0.1501, "step": 10928 }, { "epoch": 1.49, "grad_norm": 3.230535334354074, "learning_rate": 1.630133349511409e-06, "loss": 0.1583, "step": 10929 }, { "epoch": 1.49, "grad_norm": 2.814607491332308, "learning_rate": 1.6293196622202635e-06, "loss": 0.1349, "step": 10930 }, { "epoch": 1.49, "grad_norm": 3.2934761117299667, "learning_rate": 1.62850613852722e-06, "loss": 0.1856, "step": 10931 }, { "epoch": 1.49, "grad_norm": 3.061143210300325, "learning_rate": 1.6276927784717628e-06, "loss": 0.1345, "step": 10932 }, { "epoch": 1.49, "grad_norm": 4.019406917917494, "learning_rate": 1.6268795820933692e-06, "loss": 0.1885, "step": 10933 }, { "epoch": 1.49, "grad_norm": 5.162457343346542, "learning_rate": 1.6260665494315087e-06, "loss": 0.158, "step": 10934 }, { "epoch": 1.49, "grad_norm": 4.187881203362657, "learning_rate": 1.6252536805256397e-06, "loss": 0.1398, "step": 10935 }, { "epoch": 1.49, "grad_norm": 4.326680247546346, "learning_rate": 1.6244409754152212e-06, "loss": 0.2211, "step": 10936 }, { "epoch": 1.49, "grad_norm": 4.1272953441381475, "learning_rate": 1.623628434139693e-06, "loss": 0.1854, "step": 10937 }, { "epoch": 1.49, "grad_norm": 2.5139682843524076, "learning_rate": 1.6228160567384933e-06, "loss": 0.1465, "step": 10938 }, { "epoch": 1.49, "grad_norm": 5.578656885465312, "learning_rate": 1.6220038432510527e-06, "loss": 0.1764, "step": 10939 }, { "epoch": 1.49, "grad_norm": 2.9346171815779334, "learning_rate": 1.6211917937167898e-06, "loss": 0.1374, "step": 10940 }, { "epoch": 1.49, "grad_norm": 4.925192883900253, "learning_rate": 1.6203799081751232e-06, "loss": 0.1845, "step": 10941 }, { "epoch": 1.49, "grad_norm": 3.225144961555045, "learning_rate": 1.6195681866654517e-06, "loss": 0.1304, "step": 10942 }, { "epoch": 1.49, "grad_norm": 3.657640424642823, "learning_rate": 1.6187566292271795e-06, "loss": 0.1295, "step": 10943 }, { "epoch": 1.49, "grad_norm": 3.2496813788014687, "learning_rate": 1.6179452358996888e-06, "loss": 0.1656, "step": 10944 }, { "epoch": 1.49, "grad_norm": 3.439033918697151, "learning_rate": 1.6171340067223673e-06, "loss": 0.1367, "step": 10945 }, { "epoch": 1.49, "grad_norm": 3.1137448836717763, "learning_rate": 1.6163229417345855e-06, "loss": 0.1456, "step": 10946 }, { "epoch": 1.49, "grad_norm": 2.7227817528724616, "learning_rate": 1.6155120409757096e-06, "loss": 0.1304, "step": 10947 }, { "epoch": 1.49, "grad_norm": 3.5648810292487796, "learning_rate": 1.6147013044850973e-06, "loss": 0.1756, "step": 10948 }, { "epoch": 1.49, "grad_norm": 3.4581326083002017, "learning_rate": 1.6138907323020974e-06, "loss": 0.139, "step": 10949 }, { "epoch": 1.49, "grad_norm": 3.7529297031725655, "learning_rate": 1.613080324466053e-06, "loss": 0.1522, "step": 10950 }, { "epoch": 1.49, "grad_norm": 3.662044664743728, "learning_rate": 1.6122700810162967e-06, "loss": 0.1398, "step": 10951 }, { "epoch": 1.49, "grad_norm": 4.016881938045542, "learning_rate": 1.6114600019921538e-06, "loss": 0.1329, "step": 10952 }, { "epoch": 1.49, "grad_norm": 4.24157570845627, "learning_rate": 1.6106500874329433e-06, "loss": 0.1925, "step": 10953 }, { "epoch": 1.49, "grad_norm": 3.3517977206529146, "learning_rate": 1.6098403373779725e-06, "loss": 0.1975, "step": 10954 }, { "epoch": 1.49, "grad_norm": 4.685029967457141, "learning_rate": 1.6090307518665478e-06, "loss": 0.1722, "step": 10955 }, { "epoch": 1.49, "grad_norm": 2.836292395680732, "learning_rate": 1.6082213309379568e-06, "loss": 0.1383, "step": 10956 }, { "epoch": 1.49, "grad_norm": 3.2777118415037503, "learning_rate": 1.6074120746314915e-06, "loss": 0.1824, "step": 10957 }, { "epoch": 1.49, "grad_norm": 3.2177397711644913, "learning_rate": 1.6066029829864239e-06, "loss": 0.1816, "step": 10958 }, { "epoch": 1.49, "grad_norm": 3.389059779680246, "learning_rate": 1.6057940560420276e-06, "loss": 0.1277, "step": 10959 }, { "epoch": 1.49, "grad_norm": 4.4306334009414705, "learning_rate": 1.6049852938375631e-06, "loss": 0.158, "step": 10960 }, { "epoch": 1.49, "grad_norm": 2.9110150831097723, "learning_rate": 1.6041766964122846e-06, "loss": 0.1231, "step": 10961 }, { "epoch": 1.49, "grad_norm": 3.688255548791616, "learning_rate": 1.6033682638054376e-06, "loss": 0.1179, "step": 10962 }, { "epoch": 1.49, "grad_norm": 4.086355910060287, "learning_rate": 1.60255999605626e-06, "loss": 0.1686, "step": 10963 }, { "epoch": 1.49, "grad_norm": 2.713901991236544, "learning_rate": 1.6017518932039816e-06, "loss": 0.1564, "step": 10964 }, { "epoch": 1.49, "grad_norm": 3.4738186534644253, "learning_rate": 1.6009439552878242e-06, "loss": 0.1678, "step": 10965 }, { "epoch": 1.49, "grad_norm": 2.96708345334705, "learning_rate": 1.6001361823470007e-06, "loss": 0.1334, "step": 10966 }, { "epoch": 1.49, "grad_norm": 3.769200659530289, "learning_rate": 1.5993285744207183e-06, "loss": 0.1631, "step": 10967 }, { "epoch": 1.49, "grad_norm": 3.463878604458854, "learning_rate": 1.5985211315481725e-06, "loss": 0.1805, "step": 10968 }, { "epoch": 1.49, "grad_norm": 4.30085443538408, "learning_rate": 1.5977138537685576e-06, "loss": 0.1348, "step": 10969 }, { "epoch": 1.49, "grad_norm": 3.5551119748576827, "learning_rate": 1.5969067411210492e-06, "loss": 0.1369, "step": 10970 }, { "epoch": 1.49, "grad_norm": 3.8054585445031197, "learning_rate": 1.5960997936448274e-06, "loss": 0.1869, "step": 10971 }, { "epoch": 1.49, "grad_norm": 5.544696025720577, "learning_rate": 1.5952930113790516e-06, "loss": 0.1855, "step": 10972 }, { "epoch": 1.49, "grad_norm": 3.40502059097326, "learning_rate": 1.5944863943628847e-06, "loss": 0.1433, "step": 10973 }, { "epoch": 1.49, "grad_norm": 2.6232666643186398, "learning_rate": 1.5936799426354733e-06, "loss": 0.1338, "step": 10974 }, { "epoch": 1.49, "grad_norm": 2.9441471135798416, "learning_rate": 1.5928736562359603e-06, "loss": 0.1302, "step": 10975 }, { "epoch": 1.49, "grad_norm": 4.151642903393627, "learning_rate": 1.5920675352034792e-06, "loss": 0.1946, "step": 10976 }, { "epoch": 1.49, "grad_norm": 3.660799435350368, "learning_rate": 1.5912615795771557e-06, "loss": 0.1609, "step": 10977 }, { "epoch": 1.49, "grad_norm": 2.9504478410111874, "learning_rate": 1.590455789396107e-06, "loss": 0.1412, "step": 10978 }, { "epoch": 1.49, "grad_norm": 3.8890616570968843, "learning_rate": 1.5896501646994422e-06, "loss": 0.1527, "step": 10979 }, { "epoch": 1.49, "grad_norm": 3.2564392609929964, "learning_rate": 1.5888447055262641e-06, "loss": 0.1286, "step": 10980 }, { "epoch": 1.49, "grad_norm": 2.930193123467231, "learning_rate": 1.5880394119156644e-06, "loss": 0.1575, "step": 10981 }, { "epoch": 1.49, "grad_norm": 3.4236593214583637, "learning_rate": 1.5872342839067305e-06, "loss": 0.1521, "step": 10982 }, { "epoch": 1.49, "grad_norm": 3.278771765158989, "learning_rate": 1.5864293215385384e-06, "loss": 0.1611, "step": 10983 }, { "epoch": 1.49, "grad_norm": 3.471580593656817, "learning_rate": 1.5856245248501578e-06, "loss": 0.1739, "step": 10984 }, { "epoch": 1.49, "grad_norm": 2.651899629504026, "learning_rate": 1.5848198938806503e-06, "loss": 0.1472, "step": 10985 }, { "epoch": 1.49, "grad_norm": 4.257002466149615, "learning_rate": 1.584015428669069e-06, "loss": 0.1412, "step": 10986 }, { "epoch": 1.49, "grad_norm": 2.9571299355715506, "learning_rate": 1.5832111292544571e-06, "loss": 0.1708, "step": 10987 }, { "epoch": 1.49, "grad_norm": 2.3926240385868938, "learning_rate": 1.5824069956758575e-06, "loss": 0.1551, "step": 10988 }, { "epoch": 1.49, "grad_norm": 4.24275426037109, "learning_rate": 1.581603027972292e-06, "loss": 0.1388, "step": 10989 }, { "epoch": 1.49, "grad_norm": 2.9397608930402925, "learning_rate": 1.5807992261827888e-06, "loss": 0.1598, "step": 10990 }, { "epoch": 1.5, "grad_norm": 3.059971689587447, "learning_rate": 1.5799955903463543e-06, "loss": 0.1752, "step": 10991 }, { "epoch": 1.5, "grad_norm": 2.9158542316097122, "learning_rate": 1.5791921205019984e-06, "loss": 0.1465, "step": 10992 }, { "epoch": 1.5, "grad_norm": 4.142643977486776, "learning_rate": 1.5783888166887156e-06, "loss": 0.1767, "step": 10993 }, { "epoch": 1.5, "grad_norm": 2.8951746821391335, "learning_rate": 1.5775856789454957e-06, "loss": 0.1231, "step": 10994 }, { "epoch": 1.5, "grad_norm": 3.2589061224269327, "learning_rate": 1.576782707311319e-06, "loss": 0.1396, "step": 10995 }, { "epoch": 1.5, "grad_norm": 3.459385640805196, "learning_rate": 1.5759799018251586e-06, "loss": 0.1261, "step": 10996 }, { "epoch": 1.5, "grad_norm": 4.509497111705006, "learning_rate": 1.5751772625259787e-06, "loss": 0.1523, "step": 10997 }, { "epoch": 1.5, "grad_norm": 2.632556505064492, "learning_rate": 1.5743747894527366e-06, "loss": 0.1321, "step": 10998 }, { "epoch": 1.5, "grad_norm": 2.827026583121626, "learning_rate": 1.5735724826443788e-06, "loss": 0.1139, "step": 10999 }, { "epoch": 1.5, "grad_norm": 2.91503558264945, "learning_rate": 1.5727703421398504e-06, "loss": 0.1388, "step": 11000 }, { "epoch": 1.5, "grad_norm": 3.235800947041348, "learning_rate": 1.5719683679780773e-06, "loss": 0.1276, "step": 11001 }, { "epoch": 1.5, "grad_norm": 3.4064150822317196, "learning_rate": 1.571166560197991e-06, "loss": 0.1503, "step": 11002 }, { "epoch": 1.5, "grad_norm": 3.1224859332724204, "learning_rate": 1.5703649188385006e-06, "loss": 0.1604, "step": 11003 }, { "epoch": 1.5, "grad_norm": 3.4198188746103075, "learning_rate": 1.569563443938521e-06, "loss": 0.1656, "step": 11004 }, { "epoch": 1.5, "grad_norm": 3.6474412846858484, "learning_rate": 1.568762135536946e-06, "loss": 0.1845, "step": 11005 }, { "epoch": 1.5, "grad_norm": 3.7447617862631013, "learning_rate": 1.567960993672672e-06, "loss": 0.2019, "step": 11006 }, { "epoch": 1.5, "grad_norm": 3.058260512141662, "learning_rate": 1.567160018384582e-06, "loss": 0.1507, "step": 11007 }, { "epoch": 1.5, "grad_norm": 3.361403251777211, "learning_rate": 1.566359209711551e-06, "loss": 0.1602, "step": 11008 }, { "epoch": 1.5, "grad_norm": 2.5936218963809, "learning_rate": 1.5655585676924474e-06, "loss": 0.1184, "step": 11009 }, { "epoch": 1.5, "grad_norm": 3.1879608924790013, "learning_rate": 1.5647580923661305e-06, "loss": 0.1568, "step": 11010 }, { "epoch": 1.5, "grad_norm": 3.9473821029106944, "learning_rate": 1.5639577837714514e-06, "loss": 0.1378, "step": 11011 }, { "epoch": 1.5, "grad_norm": 4.543093401905084, "learning_rate": 1.563157641947255e-06, "loss": 0.1455, "step": 11012 }, { "epoch": 1.5, "grad_norm": 3.991750155950475, "learning_rate": 1.5623576669323743e-06, "loss": 0.1829, "step": 11013 }, { "epoch": 1.5, "grad_norm": 5.0911964132214225, "learning_rate": 1.561557858765641e-06, "loss": 0.1796, "step": 11014 }, { "epoch": 1.5, "grad_norm": 4.134378503075597, "learning_rate": 1.5607582174858682e-06, "loss": 0.182, "step": 11015 }, { "epoch": 1.5, "grad_norm": 3.2496403134094463, "learning_rate": 1.5599587431318735e-06, "loss": 0.1696, "step": 11016 }, { "epoch": 1.5, "grad_norm": 4.477061051852084, "learning_rate": 1.5591594357424555e-06, "loss": 0.1462, "step": 11017 }, { "epoch": 1.5, "grad_norm": 3.8489227645872126, "learning_rate": 1.558360295356408e-06, "loss": 0.1254, "step": 11018 }, { "epoch": 1.5, "grad_norm": 4.137249166382448, "learning_rate": 1.5575613220125236e-06, "loss": 0.1684, "step": 11019 }, { "epoch": 1.5, "grad_norm": 2.8554000995640587, "learning_rate": 1.5567625157495741e-06, "loss": 0.1117, "step": 11020 }, { "epoch": 1.5, "grad_norm": 4.993821137219496, "learning_rate": 1.5559638766063367e-06, "loss": 0.1627, "step": 11021 }, { "epoch": 1.5, "grad_norm": 3.045479420976812, "learning_rate": 1.555165404621567e-06, "loss": 0.2039, "step": 11022 }, { "epoch": 1.5, "grad_norm": 3.8809029375201383, "learning_rate": 1.5543670998340266e-06, "loss": 0.1653, "step": 11023 }, { "epoch": 1.5, "grad_norm": 3.8638273339239837, "learning_rate": 1.5535689622824547e-06, "loss": 0.1581, "step": 11024 }, { "epoch": 1.5, "grad_norm": 4.855640533842102, "learning_rate": 1.5527709920055938e-06, "loss": 0.1777, "step": 11025 }, { "epoch": 1.5, "grad_norm": 4.603762814247172, "learning_rate": 1.5519731890421729e-06, "loss": 0.1483, "step": 11026 }, { "epoch": 1.5, "grad_norm": 3.800492035401361, "learning_rate": 1.5511755534309143e-06, "loss": 0.187, "step": 11027 }, { "epoch": 1.5, "grad_norm": 3.375601289775866, "learning_rate": 1.550378085210531e-06, "loss": 0.1633, "step": 11028 }, { "epoch": 1.5, "grad_norm": 2.980154790163018, "learning_rate": 1.5495807844197287e-06, "loss": 0.1381, "step": 11029 }, { "epoch": 1.5, "grad_norm": 3.6277179586465116, "learning_rate": 1.5487836510972049e-06, "loss": 0.1388, "step": 11030 }, { "epoch": 1.5, "grad_norm": 4.364119996333819, "learning_rate": 1.5479866852816494e-06, "loss": 0.1472, "step": 11031 }, { "epoch": 1.5, "grad_norm": 3.862447966666889, "learning_rate": 1.5471898870117414e-06, "loss": 0.1401, "step": 11032 }, { "epoch": 1.5, "grad_norm": 3.611047248078567, "learning_rate": 1.5463932563261592e-06, "loss": 0.1657, "step": 11033 }, { "epoch": 1.5, "grad_norm": 4.664150998397348, "learning_rate": 1.5455967932635606e-06, "loss": 0.1919, "step": 11034 }, { "epoch": 1.5, "grad_norm": 3.392771895355186, "learning_rate": 1.5448004978626097e-06, "loss": 0.1944, "step": 11035 }, { "epoch": 1.5, "grad_norm": 3.767665519142083, "learning_rate": 1.5440043701619484e-06, "loss": 0.1623, "step": 11036 }, { "epoch": 1.5, "grad_norm": 3.2778325992291566, "learning_rate": 1.5432084102002243e-06, "loss": 0.1397, "step": 11037 }, { "epoch": 1.5, "grad_norm": 3.0336920598510417, "learning_rate": 1.542412618016062e-06, "loss": 0.103, "step": 11038 }, { "epoch": 1.5, "grad_norm": 2.5270725842731205, "learning_rate": 1.5416169936480917e-06, "loss": 0.1129, "step": 11039 }, { "epoch": 1.5, "grad_norm": 2.9615284688122783, "learning_rate": 1.5408215371349277e-06, "loss": 0.141, "step": 11040 }, { "epoch": 1.5, "grad_norm": 3.054105742885219, "learning_rate": 1.5400262485151784e-06, "loss": 0.1479, "step": 11041 }, { "epoch": 1.5, "grad_norm": 3.239017673144057, "learning_rate": 1.539231127827443e-06, "loss": 0.2038, "step": 11042 }, { "epoch": 1.5, "grad_norm": 6.404236399730672, "learning_rate": 1.5384361751103127e-06, "loss": 0.1787, "step": 11043 }, { "epoch": 1.5, "grad_norm": 3.167608451922604, "learning_rate": 1.5376413904023723e-06, "loss": 0.1769, "step": 11044 }, { "epoch": 1.5, "grad_norm": 3.969185723325949, "learning_rate": 1.5368467737421965e-06, "loss": 0.157, "step": 11045 }, { "epoch": 1.5, "grad_norm": 2.9931315203797286, "learning_rate": 1.5360523251683511e-06, "loss": 0.1499, "step": 11046 }, { "epoch": 1.5, "grad_norm": 3.8398398071319564, "learning_rate": 1.5352580447194e-06, "loss": 0.1892, "step": 11047 }, { "epoch": 1.5, "grad_norm": 4.4580526816815995, "learning_rate": 1.5344639324338866e-06, "loss": 0.1721, "step": 11048 }, { "epoch": 1.5, "grad_norm": 2.7085376810000112, "learning_rate": 1.533669988350362e-06, "loss": 0.1395, "step": 11049 }, { "epoch": 1.5, "grad_norm": 5.185814322540972, "learning_rate": 1.5328762125073531e-06, "loss": 0.161, "step": 11050 }, { "epoch": 1.5, "grad_norm": 3.363829993077783, "learning_rate": 1.5320826049433913e-06, "loss": 0.1644, "step": 11051 }, { "epoch": 1.5, "grad_norm": 3.269259391670938, "learning_rate": 1.5312891656969936e-06, "loss": 0.1306, "step": 11052 }, { "epoch": 1.5, "grad_norm": 2.7334229957047183, "learning_rate": 1.5304958948066694e-06, "loss": 0.1235, "step": 11053 }, { "epoch": 1.5, "grad_norm": 2.8055234007106176, "learning_rate": 1.5297027923109213e-06, "loss": 0.1201, "step": 11054 }, { "epoch": 1.5, "grad_norm": 4.43534398153329, "learning_rate": 1.5289098582482432e-06, "loss": 0.1484, "step": 11055 }, { "epoch": 1.5, "grad_norm": 3.932351418089837, "learning_rate": 1.5281170926571199e-06, "loss": 0.1594, "step": 11056 }, { "epoch": 1.5, "grad_norm": 3.0791438468960566, "learning_rate": 1.5273244955760286e-06, "loss": 0.1555, "step": 11057 }, { "epoch": 1.5, "grad_norm": 3.9350692283572815, "learning_rate": 1.5265320670434397e-06, "loss": 0.1875, "step": 11058 }, { "epoch": 1.5, "grad_norm": 4.27606322176683, "learning_rate": 1.525739807097813e-06, "loss": 0.173, "step": 11059 }, { "epoch": 1.5, "grad_norm": 2.9126133828703016, "learning_rate": 1.524947715777602e-06, "loss": 0.1596, "step": 11060 }, { "epoch": 1.5, "grad_norm": 4.268982215333995, "learning_rate": 1.5241557931212515e-06, "loss": 0.2079, "step": 11061 }, { "epoch": 1.5, "grad_norm": 2.9764245845412334, "learning_rate": 1.5233640391671973e-06, "loss": 0.1412, "step": 11062 }, { "epoch": 1.5, "grad_norm": 2.7149765277027167, "learning_rate": 1.5225724539538677e-06, "loss": 0.1438, "step": 11063 }, { "epoch": 1.5, "grad_norm": 3.517779414378465, "learning_rate": 1.521781037519684e-06, "loss": 0.2018, "step": 11064 }, { "epoch": 1.51, "grad_norm": 3.683115363834051, "learning_rate": 1.5209897899030551e-06, "loss": 0.1589, "step": 11065 }, { "epoch": 1.51, "grad_norm": 3.434663458751528, "learning_rate": 1.5201987111423904e-06, "loss": 0.1643, "step": 11066 }, { "epoch": 1.51, "grad_norm": 3.5133709759310032, "learning_rate": 1.5194078012760781e-06, "loss": 0.1562, "step": 11067 }, { "epoch": 1.51, "grad_norm": 3.470085854415452, "learning_rate": 1.5186170603425132e-06, "loss": 0.1732, "step": 11068 }, { "epoch": 1.51, "grad_norm": 3.5105588848775455, "learning_rate": 1.517826488380067e-06, "loss": 0.1642, "step": 11069 }, { "epoch": 1.51, "grad_norm": 3.801349264198287, "learning_rate": 1.517036085427116e-06, "loss": 0.1584, "step": 11070 }, { "epoch": 1.51, "grad_norm": 3.213525395311131, "learning_rate": 1.5162458515220213e-06, "loss": 0.171, "step": 11071 }, { "epoch": 1.51, "grad_norm": 4.150424103228312, "learning_rate": 1.5154557867031378e-06, "loss": 0.1506, "step": 11072 }, { "epoch": 1.51, "grad_norm": 3.694460898050962, "learning_rate": 1.5146658910088108e-06, "loss": 0.1583, "step": 11073 }, { "epoch": 1.51, "grad_norm": 4.2363394603426405, "learning_rate": 1.5138761644773786e-06, "loss": 0.1654, "step": 11074 }, { "epoch": 1.51, "grad_norm": 2.969401340362524, "learning_rate": 1.5130866071471717e-06, "loss": 0.1358, "step": 11075 }, { "epoch": 1.51, "grad_norm": 2.8294423463330105, "learning_rate": 1.5122972190565116e-06, "loss": 0.1501, "step": 11076 }, { "epoch": 1.51, "grad_norm": 3.380487177599558, "learning_rate": 1.511508000243711e-06, "loss": 0.1444, "step": 11077 }, { "epoch": 1.51, "grad_norm": 3.6161640368496095, "learning_rate": 1.510718950747076e-06, "loss": 0.1281, "step": 11078 }, { "epoch": 1.51, "grad_norm": 3.8855469429763367, "learning_rate": 1.509930070604902e-06, "loss": 0.1715, "step": 11079 }, { "epoch": 1.51, "grad_norm": 3.625926134888867, "learning_rate": 1.5091413598554816e-06, "loss": 0.1747, "step": 11080 }, { "epoch": 1.51, "grad_norm": 3.2103728574051797, "learning_rate": 1.5083528185370904e-06, "loss": 0.1444, "step": 11081 }, { "epoch": 1.51, "grad_norm": 2.654345536759291, "learning_rate": 1.5075644466880063e-06, "loss": 0.1158, "step": 11082 }, { "epoch": 1.51, "grad_norm": 3.943370831321145, "learning_rate": 1.5067762443464868e-06, "loss": 0.1612, "step": 11083 }, { "epoch": 1.51, "grad_norm": 2.7510372973019397, "learning_rate": 1.5059882115507935e-06, "loss": 0.1461, "step": 11084 }, { "epoch": 1.51, "grad_norm": 2.369306632379338, "learning_rate": 1.5052003483391715e-06, "loss": 0.1548, "step": 11085 }, { "epoch": 1.51, "grad_norm": 4.517079245237413, "learning_rate": 1.5044126547498606e-06, "loss": 0.1454, "step": 11086 }, { "epoch": 1.51, "grad_norm": 3.5417394770145005, "learning_rate": 1.5036251308210926e-06, "loss": 0.1105, "step": 11087 }, { "epoch": 1.51, "grad_norm": 2.78137863581435, "learning_rate": 1.5028377765910895e-06, "loss": 0.1352, "step": 11088 }, { "epoch": 1.51, "grad_norm": 3.242850139629226, "learning_rate": 1.5020505920980666e-06, "loss": 0.1337, "step": 11089 }, { "epoch": 1.51, "grad_norm": 3.6372948440084762, "learning_rate": 1.5012635773802304e-06, "loss": 0.1361, "step": 11090 }, { "epoch": 1.51, "grad_norm": 2.921330663929891, "learning_rate": 1.500476732475779e-06, "loss": 0.1553, "step": 11091 }, { "epoch": 1.51, "grad_norm": 3.75908423168503, "learning_rate": 1.4996900574229022e-06, "loss": 0.129, "step": 11092 }, { "epoch": 1.51, "grad_norm": 5.034584671626574, "learning_rate": 1.4989035522597805e-06, "loss": 0.1541, "step": 11093 }, { "epoch": 1.51, "grad_norm": 3.2362230940247727, "learning_rate": 1.4981172170245923e-06, "loss": 0.1716, "step": 11094 }, { "epoch": 1.51, "grad_norm": 3.5375015824355542, "learning_rate": 1.4973310517554961e-06, "loss": 0.1474, "step": 11095 }, { "epoch": 1.51, "grad_norm": 3.6058988776209375, "learning_rate": 1.4965450564906559e-06, "loss": 0.1694, "step": 11096 }, { "epoch": 1.51, "grad_norm": 2.8073953694498606, "learning_rate": 1.4957592312682157e-06, "loss": 0.1511, "step": 11097 }, { "epoch": 1.51, "grad_norm": 5.518444159282519, "learning_rate": 1.494973576126315e-06, "loss": 0.1772, "step": 11098 }, { "epoch": 1.51, "grad_norm": 3.9040711699606105, "learning_rate": 1.4941880911030915e-06, "loss": 0.107, "step": 11099 }, { "epoch": 1.51, "grad_norm": 3.8551975592501826, "learning_rate": 1.4934027762366632e-06, "loss": 0.1428, "step": 11100 }, { "epoch": 1.51, "grad_norm": 3.5373626506001044, "learning_rate": 1.4926176315651518e-06, "loss": 0.1951, "step": 11101 }, { "epoch": 1.51, "grad_norm": 4.83545641649403, "learning_rate": 1.4918326571266584e-06, "loss": 0.161, "step": 11102 }, { "epoch": 1.51, "grad_norm": 3.3323386341407013, "learning_rate": 1.491047852959287e-06, "loss": 0.1454, "step": 11103 }, { "epoch": 1.51, "grad_norm": 3.8029711812557525, "learning_rate": 1.4902632191011273e-06, "loss": 0.1737, "step": 11104 }, { "epoch": 1.51, "grad_norm": 4.2090217996497294, "learning_rate": 1.4894787555902617e-06, "loss": 0.1854, "step": 11105 }, { "epoch": 1.51, "grad_norm": 3.064883332927932, "learning_rate": 1.4886944624647647e-06, "loss": 0.1547, "step": 11106 }, { "epoch": 1.51, "grad_norm": 3.802632727705982, "learning_rate": 1.4879103397627027e-06, "loss": 0.1467, "step": 11107 }, { "epoch": 1.51, "grad_norm": 3.074856467399909, "learning_rate": 1.4871263875221332e-06, "loss": 0.1538, "step": 11108 }, { "epoch": 1.51, "grad_norm": 3.149881949986279, "learning_rate": 1.486342605781106e-06, "loss": 0.1977, "step": 11109 }, { "epoch": 1.51, "grad_norm": 2.8028765716783894, "learning_rate": 1.4855589945776622e-06, "loss": 0.1685, "step": 11110 }, { "epoch": 1.51, "grad_norm": 3.972835223135853, "learning_rate": 1.4847755539498353e-06, "loss": 0.1391, "step": 11111 }, { "epoch": 1.51, "grad_norm": 2.4082179554662755, "learning_rate": 1.4839922839356484e-06, "loss": 0.1197, "step": 11112 }, { "epoch": 1.51, "grad_norm": 5.342908670856374, "learning_rate": 1.4832091845731223e-06, "loss": 0.1281, "step": 11113 }, { "epoch": 1.51, "grad_norm": 3.671113786648763, "learning_rate": 1.4824262559002595e-06, "loss": 0.167, "step": 11114 }, { "epoch": 1.51, "grad_norm": 3.313688286147175, "learning_rate": 1.4816434979550666e-06, "loss": 0.1438, "step": 11115 }, { "epoch": 1.51, "grad_norm": 4.0606644776683645, "learning_rate": 1.4808609107755272e-06, "loss": 0.2015, "step": 11116 }, { "epoch": 1.51, "grad_norm": 2.604132343208886, "learning_rate": 1.4800784943996316e-06, "loss": 0.1375, "step": 11117 }, { "epoch": 1.51, "grad_norm": 2.921340144065705, "learning_rate": 1.4792962488653516e-06, "loss": 0.1656, "step": 11118 }, { "epoch": 1.51, "grad_norm": 3.670131208401261, "learning_rate": 1.4785141742106546e-06, "loss": 0.1928, "step": 11119 }, { "epoch": 1.51, "grad_norm": 3.1245652489987834, "learning_rate": 1.4777322704734993e-06, "loss": 0.1577, "step": 11120 }, { "epoch": 1.51, "grad_norm": 3.0040474906300076, "learning_rate": 1.4769505376918358e-06, "loss": 0.1601, "step": 11121 }, { "epoch": 1.51, "grad_norm": 5.451061327434814, "learning_rate": 1.4761689759036058e-06, "loss": 0.1842, "step": 11122 }, { "epoch": 1.51, "grad_norm": 3.9139363375696377, "learning_rate": 1.475387585146743e-06, "loss": 0.1637, "step": 11123 }, { "epoch": 1.51, "grad_norm": 4.39862596000663, "learning_rate": 1.4746063654591725e-06, "loss": 0.099, "step": 11124 }, { "epoch": 1.51, "grad_norm": 4.341582232016106, "learning_rate": 1.4738253168788118e-06, "loss": 0.1898, "step": 11125 }, { "epoch": 1.51, "grad_norm": 4.053384523596827, "learning_rate": 1.4730444394435677e-06, "loss": 0.1394, "step": 11126 }, { "epoch": 1.51, "grad_norm": 2.8734300458703577, "learning_rate": 1.4722637331913447e-06, "loss": 0.1239, "step": 11127 }, { "epoch": 1.51, "grad_norm": 3.087559052629666, "learning_rate": 1.4714831981600298e-06, "loss": 0.1386, "step": 11128 }, { "epoch": 1.51, "grad_norm": 11.767000090900241, "learning_rate": 1.4707028343875123e-06, "loss": 0.1865, "step": 11129 }, { "epoch": 1.51, "grad_norm": 3.544738045366772, "learning_rate": 1.4699226419116613e-06, "loss": 0.1464, "step": 11130 }, { "epoch": 1.51, "grad_norm": 4.691196843323566, "learning_rate": 1.469142620770349e-06, "loss": 0.1389, "step": 11131 }, { "epoch": 1.51, "grad_norm": 4.0201633077509475, "learning_rate": 1.4683627710014325e-06, "loss": 0.1684, "step": 11132 }, { "epoch": 1.51, "grad_norm": 3.3643732449803614, "learning_rate": 1.467583092642762e-06, "loss": 0.1614, "step": 11133 }, { "epoch": 1.51, "grad_norm": 3.4420529356327956, "learning_rate": 1.4668035857321804e-06, "loss": 0.1505, "step": 11134 }, { "epoch": 1.51, "grad_norm": 2.9197004388817027, "learning_rate": 1.4660242503075211e-06, "loss": 0.1264, "step": 11135 }, { "epoch": 1.51, "grad_norm": 3.7186951788257123, "learning_rate": 1.4652450864066093e-06, "loss": 0.1739, "step": 11136 }, { "epoch": 1.51, "grad_norm": 5.559056616523638, "learning_rate": 1.4644660940672628e-06, "loss": 0.1749, "step": 11137 }, { "epoch": 1.52, "grad_norm": 3.613454855802014, "learning_rate": 1.46368727332729e-06, "loss": 0.1689, "step": 11138 }, { "epoch": 1.52, "grad_norm": 3.3815543409079636, "learning_rate": 1.4629086242244923e-06, "loss": 0.1678, "step": 11139 }, { "epoch": 1.52, "grad_norm": 2.9280611971212407, "learning_rate": 1.4621301467966608e-06, "loss": 0.1496, "step": 11140 }, { "epoch": 1.52, "grad_norm": 4.240792283855773, "learning_rate": 1.4613518410815803e-06, "loss": 0.1481, "step": 11141 }, { "epoch": 1.52, "grad_norm": 3.125980167747509, "learning_rate": 1.4605737071170257e-06, "loss": 0.1315, "step": 11142 }, { "epoch": 1.52, "grad_norm": 3.6205421263598287, "learning_rate": 1.4597957449407646e-06, "loss": 0.1891, "step": 11143 }, { "epoch": 1.52, "grad_norm": 2.843535243905169, "learning_rate": 1.4590179545905559e-06, "loss": 0.1415, "step": 11144 }, { "epoch": 1.52, "grad_norm": 2.9037552325628493, "learning_rate": 1.4582403361041487e-06, "loss": 0.1406, "step": 11145 }, { "epoch": 1.52, "grad_norm": 5.066718544993852, "learning_rate": 1.4574628895192894e-06, "loss": 0.182, "step": 11146 }, { "epoch": 1.52, "grad_norm": 2.4838079531607953, "learning_rate": 1.4566856148737057e-06, "loss": 0.123, "step": 11147 }, { "epoch": 1.52, "grad_norm": 2.749177019324375, "learning_rate": 1.4559085122051296e-06, "loss": 0.1306, "step": 11148 }, { "epoch": 1.52, "grad_norm": 3.1734634309789596, "learning_rate": 1.455131581551273e-06, "loss": 0.139, "step": 11149 }, { "epoch": 1.52, "grad_norm": 3.1083952492783213, "learning_rate": 1.454354822949847e-06, "loss": 0.1492, "step": 11150 }, { "epoch": 1.52, "grad_norm": 3.600943964235098, "learning_rate": 1.4535782364385526e-06, "loss": 0.1392, "step": 11151 }, { "epoch": 1.52, "grad_norm": 3.8150431961149747, "learning_rate": 1.452801822055081e-06, "loss": 0.1485, "step": 11152 }, { "epoch": 1.52, "grad_norm": 3.8370268513842887, "learning_rate": 1.4520255798371163e-06, "loss": 0.1487, "step": 11153 }, { "epoch": 1.52, "grad_norm": 5.030820567456568, "learning_rate": 1.451249509822334e-06, "loss": 0.1636, "step": 11154 }, { "epoch": 1.52, "grad_norm": 3.9666311090326882, "learning_rate": 1.4504736120484009e-06, "loss": 0.151, "step": 11155 }, { "epoch": 1.52, "grad_norm": 3.0969409958690677, "learning_rate": 1.4496978865529753e-06, "loss": 0.1678, "step": 11156 }, { "epoch": 1.52, "grad_norm": 3.8552489291916765, "learning_rate": 1.4489223333737084e-06, "loss": 0.1948, "step": 11157 }, { "epoch": 1.52, "grad_norm": 2.9256937096097606, "learning_rate": 1.4481469525482417e-06, "loss": 0.0981, "step": 11158 }, { "epoch": 1.52, "grad_norm": 3.3164249645967305, "learning_rate": 1.447371744114207e-06, "loss": 0.1567, "step": 11159 }, { "epoch": 1.52, "grad_norm": 3.32420328944316, "learning_rate": 1.4465967081092346e-06, "loss": 0.1472, "step": 11160 }, { "epoch": 1.52, "grad_norm": 3.3905703505991904, "learning_rate": 1.445821844570935e-06, "loss": 0.1611, "step": 11161 }, { "epoch": 1.52, "grad_norm": 3.161909617670975, "learning_rate": 1.4450471535369225e-06, "loss": 0.1152, "step": 11162 }, { "epoch": 1.52, "grad_norm": 3.3285016480088228, "learning_rate": 1.4442726350447922e-06, "loss": 0.1505, "step": 11163 }, { "epoch": 1.52, "grad_norm": 5.004473306342941, "learning_rate": 1.4434982891321391e-06, "loss": 0.1722, "step": 11164 }, { "epoch": 1.52, "grad_norm": 2.672742626173905, "learning_rate": 1.4427241158365457e-06, "loss": 0.1676, "step": 11165 }, { "epoch": 1.52, "grad_norm": 3.7635024738548872, "learning_rate": 1.4419501151955872e-06, "loss": 0.1752, "step": 11166 }, { "epoch": 1.52, "grad_norm": 3.0338980220002085, "learning_rate": 1.44117628724683e-06, "loss": 0.1184, "step": 11167 }, { "epoch": 1.52, "grad_norm": 3.232269135345657, "learning_rate": 1.4404026320278318e-06, "loss": 0.1594, "step": 11168 }, { "epoch": 1.52, "grad_norm": 3.262463181694164, "learning_rate": 1.4396291495761433e-06, "loss": 0.1389, "step": 11169 }, { "epoch": 1.52, "grad_norm": 2.9385380124582228, "learning_rate": 1.4388558399293056e-06, "loss": 0.1293, "step": 11170 }, { "epoch": 1.52, "grad_norm": 5.20510308394796, "learning_rate": 1.4380827031248517e-06, "loss": 0.1584, "step": 11171 }, { "epoch": 1.52, "grad_norm": 3.6677744112947366, "learning_rate": 1.437309739200306e-06, "loss": 0.1706, "step": 11172 }, { "epoch": 1.52, "grad_norm": 2.5371362948162233, "learning_rate": 1.436536948193184e-06, "loss": 0.1884, "step": 11173 }, { "epoch": 1.52, "grad_norm": 4.180074868221318, "learning_rate": 1.4357643301409974e-06, "loss": 0.1927, "step": 11174 }, { "epoch": 1.52, "grad_norm": 3.3434666171250274, "learning_rate": 1.43499188508124e-06, "loss": 0.1532, "step": 11175 }, { "epoch": 1.52, "grad_norm": 3.098439791758076, "learning_rate": 1.4342196130514097e-06, "loss": 0.1734, "step": 11176 }, { "epoch": 1.52, "grad_norm": 3.046184828637215, "learning_rate": 1.4334475140889813e-06, "loss": 0.1489, "step": 11177 }, { "epoch": 1.52, "grad_norm": 4.439995251654928, "learning_rate": 1.432675588231436e-06, "loss": 0.1495, "step": 11178 }, { "epoch": 1.52, "grad_norm": 3.148236030657936, "learning_rate": 1.4319038355162374e-06, "loss": 0.1783, "step": 11179 }, { "epoch": 1.52, "grad_norm": 3.1887493086742213, "learning_rate": 1.43113225598084e-06, "loss": 0.171, "step": 11180 }, { "epoch": 1.52, "grad_norm": 4.572753673715984, "learning_rate": 1.4303608496626985e-06, "loss": 0.1477, "step": 11181 }, { "epoch": 1.52, "grad_norm": 3.7033390495503973, "learning_rate": 1.4295896165992473e-06, "loss": 0.1571, "step": 11182 }, { "epoch": 1.52, "grad_norm": 3.2496691688902044, "learning_rate": 1.4288185568279238e-06, "loss": 0.1547, "step": 11183 }, { "epoch": 1.52, "grad_norm": 3.960778294378105, "learning_rate": 1.42804767038615e-06, "loss": 0.1625, "step": 11184 }, { "epoch": 1.52, "grad_norm": 3.572498930801652, "learning_rate": 1.4272769573113416e-06, "loss": 0.1453, "step": 11185 }, { "epoch": 1.52, "grad_norm": 3.364127366321456, "learning_rate": 1.4265064176409048e-06, "loss": 0.1667, "step": 11186 }, { "epoch": 1.52, "grad_norm": 3.154800190046072, "learning_rate": 1.4257360514122393e-06, "loss": 0.1728, "step": 11187 }, { "epoch": 1.52, "grad_norm": 4.238228924614525, "learning_rate": 1.4249658586627346e-06, "loss": 0.1248, "step": 11188 }, { "epoch": 1.52, "grad_norm": 5.440426329998469, "learning_rate": 1.4241958394297734e-06, "loss": 0.1814, "step": 11189 }, { "epoch": 1.52, "grad_norm": 3.5619685055520307, "learning_rate": 1.4234259937507278e-06, "loss": 0.1769, "step": 11190 }, { "epoch": 1.52, "grad_norm": 4.262973414372797, "learning_rate": 1.4226563216629636e-06, "loss": 0.1849, "step": 11191 }, { "epoch": 1.52, "grad_norm": 3.0143721380524853, "learning_rate": 1.4218868232038351e-06, "loss": 0.1274, "step": 11192 }, { "epoch": 1.52, "grad_norm": 3.6641639498434206, "learning_rate": 1.4211174984106962e-06, "loss": 0.1583, "step": 11193 }, { "epoch": 1.52, "grad_norm": 4.603419040769151, "learning_rate": 1.4203483473208795e-06, "loss": 0.1708, "step": 11194 }, { "epoch": 1.52, "grad_norm": 4.3296366365628645, "learning_rate": 1.4195793699717225e-06, "loss": 0.1641, "step": 11195 }, { "epoch": 1.52, "grad_norm": 3.884625226739769, "learning_rate": 1.4188105664005418e-06, "loss": 0.1755, "step": 11196 }, { "epoch": 1.52, "grad_norm": 3.236115615321463, "learning_rate": 1.4180419366446568e-06, "loss": 0.1618, "step": 11197 }, { "epoch": 1.52, "grad_norm": 3.6698379474148575, "learning_rate": 1.4172734807413712e-06, "loss": 0.1304, "step": 11198 }, { "epoch": 1.52, "grad_norm": 3.7770143893963373, "learning_rate": 1.4165051987279832e-06, "loss": 0.1448, "step": 11199 }, { "epoch": 1.52, "grad_norm": 3.808284470807653, "learning_rate": 1.4157370906417818e-06, "loss": 0.1861, "step": 11200 }, { "epoch": 1.52, "grad_norm": 2.7028166220101224, "learning_rate": 1.4149691565200468e-06, "loss": 0.1282, "step": 11201 }, { "epoch": 1.52, "grad_norm": 3.170868454966819, "learning_rate": 1.4142013964000513e-06, "loss": 0.1157, "step": 11202 }, { "epoch": 1.52, "grad_norm": 4.573213987610933, "learning_rate": 1.4134338103190582e-06, "loss": 0.1616, "step": 11203 }, { "epoch": 1.52, "grad_norm": 4.6206391409078025, "learning_rate": 1.4126663983143235e-06, "loss": 0.1986, "step": 11204 }, { "epoch": 1.52, "grad_norm": 3.5454885377704435, "learning_rate": 1.4118991604230941e-06, "loss": 0.1611, "step": 11205 }, { "epoch": 1.52, "grad_norm": 3.929763107537382, "learning_rate": 1.411132096682606e-06, "loss": 0.2014, "step": 11206 }, { "epoch": 1.52, "grad_norm": 3.96729864982111, "learning_rate": 1.4103652071300945e-06, "loss": 0.2005, "step": 11207 }, { "epoch": 1.52, "grad_norm": 3.7609076962014583, "learning_rate": 1.4095984918027744e-06, "loss": 0.1803, "step": 11208 }, { "epoch": 1.52, "grad_norm": 3.2574566900589526, "learning_rate": 1.4088319507378657e-06, "loss": 0.1525, "step": 11209 }, { "epoch": 1.52, "grad_norm": 3.530960963874701, "learning_rate": 1.4080655839725655e-06, "loss": 0.1629, "step": 11210 }, { "epoch": 1.52, "grad_norm": 3.6500041334523354, "learning_rate": 1.4072993915440752e-06, "loss": 0.1579, "step": 11211 }, { "epoch": 1.53, "grad_norm": 2.86110984574354, "learning_rate": 1.4065333734895815e-06, "loss": 0.1726, "step": 11212 }, { "epoch": 1.53, "grad_norm": 5.055248029034828, "learning_rate": 1.4057675298462626e-06, "loss": 0.1429, "step": 11213 }, { "epoch": 1.53, "grad_norm": 3.732718371888354, "learning_rate": 1.40500186065129e-06, "loss": 0.1592, "step": 11214 }, { "epoch": 1.53, "grad_norm": 2.8305691111539604, "learning_rate": 1.4042363659418252e-06, "loss": 0.1138, "step": 11215 }, { "epoch": 1.53, "grad_norm": 4.181571561677018, "learning_rate": 1.4034710457550222e-06, "loss": 0.1661, "step": 11216 }, { "epoch": 1.53, "grad_norm": 3.297502099356471, "learning_rate": 1.4027059001280269e-06, "loss": 0.1616, "step": 11217 }, { "epoch": 1.53, "grad_norm": 2.6203004425042353, "learning_rate": 1.4019409290979752e-06, "loss": 0.1587, "step": 11218 }, { "epoch": 1.53, "grad_norm": 4.008955903717917, "learning_rate": 1.4011761327019962e-06, "loss": 0.1451, "step": 11219 }, { "epoch": 1.53, "grad_norm": 2.7116732787386995, "learning_rate": 1.4004115109772092e-06, "loss": 0.1575, "step": 11220 }, { "epoch": 1.53, "grad_norm": 3.199438937713125, "learning_rate": 1.3996470639607258e-06, "loss": 0.1433, "step": 11221 }, { "epoch": 1.53, "grad_norm": 4.303645054421193, "learning_rate": 1.3988827916896491e-06, "loss": 0.1691, "step": 11222 }, { "epoch": 1.53, "grad_norm": 4.1005689773041, "learning_rate": 1.3981186942010732e-06, "loss": 0.1458, "step": 11223 }, { "epoch": 1.53, "grad_norm": 3.4504202004970432, "learning_rate": 1.3973547715320846e-06, "loss": 0.152, "step": 11224 }, { "epoch": 1.53, "grad_norm": 3.9495718548997854, "learning_rate": 1.3965910237197589e-06, "loss": 0.1616, "step": 11225 }, { "epoch": 1.53, "grad_norm": 2.7372143420870896, "learning_rate": 1.3958274508011698e-06, "loss": 0.1136, "step": 11226 }, { "epoch": 1.53, "grad_norm": 3.2457118843345056, "learning_rate": 1.3950640528133713e-06, "loss": 0.1622, "step": 11227 }, { "epoch": 1.53, "grad_norm": 2.4822177272307115, "learning_rate": 1.3943008297934218e-06, "loss": 0.1358, "step": 11228 }, { "epoch": 1.53, "grad_norm": 2.8181138709436966, "learning_rate": 1.3935377817783585e-06, "loss": 0.1446, "step": 11229 }, { "epoch": 1.53, "grad_norm": 2.808250766641343, "learning_rate": 1.3927749088052218e-06, "loss": 0.1391, "step": 11230 }, { "epoch": 1.53, "grad_norm": 3.5969183898927564, "learning_rate": 1.3920122109110356e-06, "loss": 0.1546, "step": 11231 }, { "epoch": 1.53, "grad_norm": 3.8023152192246092, "learning_rate": 1.3912496881328185e-06, "loss": 0.1239, "step": 11232 }, { "epoch": 1.53, "grad_norm": 3.969571354095034, "learning_rate": 1.3904873405075797e-06, "loss": 0.1603, "step": 11233 }, { "epoch": 1.53, "grad_norm": 3.6543803268702315, "learning_rate": 1.3897251680723212e-06, "loss": 0.1968, "step": 11234 }, { "epoch": 1.53, "grad_norm": 3.5146669459659137, "learning_rate": 1.388963170864034e-06, "loss": 0.1663, "step": 11235 }, { "epoch": 1.53, "grad_norm": 3.86921729300209, "learning_rate": 1.3882013489197032e-06, "loss": 0.1176, "step": 11236 }, { "epoch": 1.53, "grad_norm": 4.209639374935026, "learning_rate": 1.3874397022763024e-06, "loss": 0.1674, "step": 11237 }, { "epoch": 1.53, "grad_norm": 3.7996349275451218, "learning_rate": 1.3866782309708033e-06, "loss": 0.1979, "step": 11238 }, { "epoch": 1.53, "grad_norm": 5.990820583137229, "learning_rate": 1.3859169350401581e-06, "loss": 0.1855, "step": 11239 }, { "epoch": 1.53, "grad_norm": 3.127307107528038, "learning_rate": 1.3851558145213228e-06, "loss": 0.1416, "step": 11240 }, { "epoch": 1.53, "grad_norm": 4.797079688640208, "learning_rate": 1.3843948694512339e-06, "loss": 0.174, "step": 11241 }, { "epoch": 1.53, "grad_norm": 2.979553625724201, "learning_rate": 1.3836340998668284e-06, "loss": 0.1127, "step": 11242 }, { "epoch": 1.53, "grad_norm": 2.557742757685496, "learning_rate": 1.3828735058050263e-06, "loss": 0.1345, "step": 11243 }, { "epoch": 1.53, "grad_norm": 3.31740127786584, "learning_rate": 1.3821130873027478e-06, "loss": 0.1792, "step": 11244 }, { "epoch": 1.53, "grad_norm": 4.2205189690234075, "learning_rate": 1.381352844396898e-06, "loss": 0.1335, "step": 11245 }, { "epoch": 1.53, "grad_norm": 3.7464022019043033, "learning_rate": 1.3805927771243772e-06, "loss": 0.1701, "step": 11246 }, { "epoch": 1.53, "grad_norm": 2.865803732913981, "learning_rate": 1.379832885522074e-06, "loss": 0.1054, "step": 11247 }, { "epoch": 1.53, "grad_norm": 2.8645282061532455, "learning_rate": 1.3790731696268711e-06, "loss": 0.1555, "step": 11248 }, { "epoch": 1.53, "grad_norm": 3.300631176506963, "learning_rate": 1.3783136294756422e-06, "loss": 0.1262, "step": 11249 }, { "epoch": 1.53, "grad_norm": 4.928836467868089, "learning_rate": 1.3775542651052514e-06, "loss": 0.1847, "step": 11250 }, { "epoch": 1.53, "grad_norm": 2.9245755341439885, "learning_rate": 1.3767950765525535e-06, "loss": 0.1653, "step": 11251 }, { "epoch": 1.53, "grad_norm": 2.629348631359129, "learning_rate": 1.3760360638544012e-06, "loss": 0.1536, "step": 11252 }, { "epoch": 1.53, "grad_norm": 4.017467580949737, "learning_rate": 1.3752772270476266e-06, "loss": 0.1454, "step": 11253 }, { "epoch": 1.53, "grad_norm": 3.1613040743579064, "learning_rate": 1.374518566169068e-06, "loss": 0.1392, "step": 11254 }, { "epoch": 1.53, "grad_norm": 3.69864510790523, "learning_rate": 1.3737600812555397e-06, "loss": 0.1809, "step": 11255 }, { "epoch": 1.53, "grad_norm": 2.4129916516165477, "learning_rate": 1.3730017723438627e-06, "loss": 0.1173, "step": 11256 }, { "epoch": 1.53, "grad_norm": 3.5386181786711375, "learning_rate": 1.3722436394708349e-06, "loss": 0.151, "step": 11257 }, { "epoch": 1.53, "grad_norm": 4.008935339416854, "learning_rate": 1.3714856826732576e-06, "loss": 0.1424, "step": 11258 }, { "epoch": 1.53, "grad_norm": 4.9541107123349635, "learning_rate": 1.3707279019879195e-06, "loss": 0.1433, "step": 11259 }, { "epoch": 1.53, "grad_norm": 4.335483346832603, "learning_rate": 1.3699702974515939e-06, "loss": 0.1375, "step": 11260 }, { "epoch": 1.53, "grad_norm": 3.0700854046315764, "learning_rate": 1.3692128691010592e-06, "loss": 0.1299, "step": 11261 }, { "epoch": 1.53, "grad_norm": 3.7621124188189907, "learning_rate": 1.3684556169730706e-06, "loss": 0.1418, "step": 11262 }, { "epoch": 1.53, "grad_norm": 3.103544121120586, "learning_rate": 1.367698541104387e-06, "loss": 0.1701, "step": 11263 }, { "epoch": 1.53, "grad_norm": 2.6839187916216445, "learning_rate": 1.3669416415317516e-06, "loss": 0.0988, "step": 11264 }, { "epoch": 1.53, "grad_norm": 8.007566119953204, "learning_rate": 1.3661849182919008e-06, "loss": 0.1145, "step": 11265 }, { "epoch": 1.53, "grad_norm": 3.6273204473886613, "learning_rate": 1.3654283714215632e-06, "loss": 0.1426, "step": 11266 }, { "epoch": 1.53, "grad_norm": 3.3646557935581476, "learning_rate": 1.3646720009574582e-06, "loss": 0.1543, "step": 11267 }, { "epoch": 1.53, "grad_norm": 4.077277433744782, "learning_rate": 1.3639158069362968e-06, "loss": 0.1763, "step": 11268 }, { "epoch": 1.53, "grad_norm": 3.030157541997935, "learning_rate": 1.3631597893947807e-06, "loss": 0.1472, "step": 11269 }, { "epoch": 1.53, "grad_norm": 3.334356687087981, "learning_rate": 1.3624039483696028e-06, "loss": 0.1355, "step": 11270 }, { "epoch": 1.53, "grad_norm": 3.023297102121821, "learning_rate": 1.3616482838974527e-06, "loss": 0.1225, "step": 11271 }, { "epoch": 1.53, "grad_norm": 2.8983779937097256, "learning_rate": 1.3608927960150008e-06, "loss": 0.1487, "step": 11272 }, { "epoch": 1.53, "grad_norm": 3.7498616594188086, "learning_rate": 1.3601374847589215e-06, "loss": 0.1771, "step": 11273 }, { "epoch": 1.53, "grad_norm": 2.678880478306771, "learning_rate": 1.3593823501658687e-06, "loss": 0.135, "step": 11274 }, { "epoch": 1.53, "grad_norm": 3.7816623170968358, "learning_rate": 1.358627392272498e-06, "loss": 0.1622, "step": 11275 }, { "epoch": 1.53, "grad_norm": 3.4859456212875135, "learning_rate": 1.3578726111154472e-06, "loss": 0.1788, "step": 11276 }, { "epoch": 1.53, "grad_norm": 3.416012722527176, "learning_rate": 1.3571180067313539e-06, "loss": 0.1201, "step": 11277 }, { "epoch": 1.53, "grad_norm": 3.612194998494174, "learning_rate": 1.3563635791568415e-06, "loss": 0.1296, "step": 11278 }, { "epoch": 1.53, "grad_norm": 4.472116167590342, "learning_rate": 1.355609328428527e-06, "loss": 0.1478, "step": 11279 }, { "epoch": 1.53, "grad_norm": 2.8699863547426827, "learning_rate": 1.3548552545830185e-06, "loss": 0.1132, "step": 11280 }, { "epoch": 1.53, "grad_norm": 3.4715971799340593, "learning_rate": 1.3541013576569156e-06, "loss": 0.161, "step": 11281 }, { "epoch": 1.53, "grad_norm": 3.561553885479942, "learning_rate": 1.3533476376868088e-06, "loss": 0.1811, "step": 11282 }, { "epoch": 1.53, "grad_norm": 3.8929514085748522, "learning_rate": 1.3525940947092808e-06, "loss": 0.1578, "step": 11283 }, { "epoch": 1.53, "grad_norm": 3.453720293402085, "learning_rate": 1.351840728760903e-06, "loss": 0.1656, "step": 11284 }, { "epoch": 1.54, "grad_norm": 3.0118580197236606, "learning_rate": 1.3510875398782458e-06, "loss": 0.1541, "step": 11285 }, { "epoch": 1.54, "grad_norm": 2.3571648574981485, "learning_rate": 1.350334528097859e-06, "loss": 0.1042, "step": 11286 }, { "epoch": 1.54, "grad_norm": 3.2985431665374843, "learning_rate": 1.3495816934562976e-06, "loss": 0.1245, "step": 11287 }, { "epoch": 1.54, "grad_norm": 3.2897062977825193, "learning_rate": 1.3488290359900935e-06, "loss": 0.1189, "step": 11288 }, { "epoch": 1.54, "grad_norm": 2.813303706106686, "learning_rate": 1.3480765557357833e-06, "loss": 0.1498, "step": 11289 }, { "epoch": 1.54, "grad_norm": 3.609628698732128, "learning_rate": 1.3473242527298863e-06, "loss": 0.1827, "step": 11290 }, { "epoch": 1.54, "grad_norm": 4.145713283538471, "learning_rate": 1.3465721270089167e-06, "loss": 0.1815, "step": 11291 }, { "epoch": 1.54, "grad_norm": 3.2574775692478912, "learning_rate": 1.3458201786093795e-06, "loss": 0.1368, "step": 11292 }, { "epoch": 1.54, "grad_norm": 3.841233637037483, "learning_rate": 1.3450684075677705e-06, "loss": 0.1519, "step": 11293 }, { "epoch": 1.54, "grad_norm": 2.7338301855680003, "learning_rate": 1.3443168139205782e-06, "loss": 0.1361, "step": 11294 }, { "epoch": 1.54, "grad_norm": 3.720479322327217, "learning_rate": 1.343565397704281e-06, "loss": 0.1616, "step": 11295 }, { "epoch": 1.54, "grad_norm": 5.502444123181744, "learning_rate": 1.3428141589553496e-06, "loss": 0.188, "step": 11296 }, { "epoch": 1.54, "grad_norm": 2.897594013378021, "learning_rate": 1.3420630977102455e-06, "loss": 0.1526, "step": 11297 }, { "epoch": 1.54, "grad_norm": 2.3748455092519536, "learning_rate": 1.3413122140054219e-06, "loss": 0.1457, "step": 11298 }, { "epoch": 1.54, "grad_norm": 3.5831698843139317, "learning_rate": 1.3405615078773243e-06, "loss": 0.1471, "step": 11299 }, { "epoch": 1.54, "grad_norm": 2.993660094984994, "learning_rate": 1.3398109793623882e-06, "loss": 0.1309, "step": 11300 }, { "epoch": 1.54, "grad_norm": 3.2400894317626348, "learning_rate": 1.3390606284970403e-06, "loss": 0.1581, "step": 11301 }, { "epoch": 1.54, "grad_norm": 3.414824276566338, "learning_rate": 1.3383104553177001e-06, "loss": 0.1469, "step": 11302 }, { "epoch": 1.54, "grad_norm": 3.3886312855393306, "learning_rate": 1.337560459860776e-06, "loss": 0.1199, "step": 11303 }, { "epoch": 1.54, "grad_norm": 3.6781429576636757, "learning_rate": 1.3368106421626747e-06, "loss": 0.1689, "step": 11304 }, { "epoch": 1.54, "grad_norm": 3.9015961234431766, "learning_rate": 1.336061002259782e-06, "loss": 0.1751, "step": 11305 }, { "epoch": 1.54, "grad_norm": 3.0253455791901196, "learning_rate": 1.3353115401884886e-06, "loss": 0.1557, "step": 11306 }, { "epoch": 1.54, "grad_norm": 3.3405429497171393, "learning_rate": 1.334562255985164e-06, "loss": 0.1592, "step": 11307 }, { "epoch": 1.54, "grad_norm": 3.3294434111847853, "learning_rate": 1.3338131496861801e-06, "loss": 0.1477, "step": 11308 }, { "epoch": 1.54, "grad_norm": 4.6201171221965796, "learning_rate": 1.3330642213278933e-06, "loss": 0.2164, "step": 11309 }, { "epoch": 1.54, "grad_norm": 3.342005365545904, "learning_rate": 1.3323154709466528e-06, "loss": 0.1667, "step": 11310 }, { "epoch": 1.54, "grad_norm": 3.8202283850117063, "learning_rate": 1.331566898578801e-06, "loss": 0.1576, "step": 11311 }, { "epoch": 1.54, "grad_norm": 3.313343914706897, "learning_rate": 1.3308185042606698e-06, "loss": 0.1631, "step": 11312 }, { "epoch": 1.54, "grad_norm": 2.9779041962873136, "learning_rate": 1.3300702880285827e-06, "loss": 0.1615, "step": 11313 }, { "epoch": 1.54, "grad_norm": 3.268148366028191, "learning_rate": 1.3293222499188553e-06, "loss": 0.1766, "step": 11314 }, { "epoch": 1.54, "grad_norm": 2.62260787677078, "learning_rate": 1.3285743899677934e-06, "loss": 0.1115, "step": 11315 }, { "epoch": 1.54, "grad_norm": 3.501566882437138, "learning_rate": 1.3278267082116953e-06, "loss": 0.1483, "step": 11316 }, { "epoch": 1.54, "grad_norm": 3.6618608245665003, "learning_rate": 1.3270792046868486e-06, "loss": 0.1581, "step": 11317 }, { "epoch": 1.54, "grad_norm": 3.3751223394310883, "learning_rate": 1.3263318794295383e-06, "loss": 0.1075, "step": 11318 }, { "epoch": 1.54, "grad_norm": 3.7740091135810214, "learning_rate": 1.3255847324760308e-06, "loss": 0.1562, "step": 11319 }, { "epoch": 1.54, "grad_norm": 3.649755137935581, "learning_rate": 1.3248377638625947e-06, "loss": 0.1419, "step": 11320 }, { "epoch": 1.54, "grad_norm": 4.861903251629606, "learning_rate": 1.3240909736254791e-06, "loss": 0.1489, "step": 11321 }, { "epoch": 1.54, "grad_norm": 3.9458726526870884, "learning_rate": 1.323344361800934e-06, "loss": 0.1881, "step": 11322 }, { "epoch": 1.54, "grad_norm": 2.989537338435451, "learning_rate": 1.3225979284251955e-06, "loss": 0.1187, "step": 11323 }, { "epoch": 1.54, "grad_norm": 4.24720940452207, "learning_rate": 1.3218516735344921e-06, "loss": 0.1973, "step": 11324 }, { "epoch": 1.54, "grad_norm": 3.3236147730066725, "learning_rate": 1.3211055971650438e-06, "loss": 0.11, "step": 11325 }, { "epoch": 1.54, "grad_norm": 3.038811099957701, "learning_rate": 1.3203596993530615e-06, "loss": 0.1619, "step": 11326 }, { "epoch": 1.54, "grad_norm": 2.442545329341686, "learning_rate": 1.3196139801347485e-06, "loss": 0.1213, "step": 11327 }, { "epoch": 1.54, "grad_norm": 3.158950494439533, "learning_rate": 1.3188684395462987e-06, "loss": 0.1672, "step": 11328 }, { "epoch": 1.54, "grad_norm": 3.2994116511787963, "learning_rate": 1.3181230776238962e-06, "loss": 0.1488, "step": 11329 }, { "epoch": 1.54, "grad_norm": 3.307174662246653, "learning_rate": 1.3173778944037191e-06, "loss": 0.1551, "step": 11330 }, { "epoch": 1.54, "grad_norm": 3.579277215190219, "learning_rate": 1.3166328899219333e-06, "loss": 0.1696, "step": 11331 }, { "epoch": 1.54, "grad_norm": 3.6786398498264523, "learning_rate": 1.3158880642147026e-06, "loss": 0.1556, "step": 11332 }, { "epoch": 1.54, "grad_norm": 3.3390146515857144, "learning_rate": 1.3151434173181716e-06, "loss": 0.1579, "step": 11333 }, { "epoch": 1.54, "grad_norm": 2.888890808633312, "learning_rate": 1.3143989492684878e-06, "loss": 0.1109, "step": 11334 }, { "epoch": 1.54, "grad_norm": 5.21603740921447, "learning_rate": 1.3136546601017785e-06, "loss": 0.1667, "step": 11335 }, { "epoch": 1.54, "grad_norm": 2.9991140096509334, "learning_rate": 1.3129105498541732e-06, "loss": 0.1488, "step": 11336 }, { "epoch": 1.54, "grad_norm": 3.479884619945585, "learning_rate": 1.3121666185617859e-06, "loss": 0.1278, "step": 11337 }, { "epoch": 1.54, "grad_norm": 3.3260734465631265, "learning_rate": 1.3114228662607241e-06, "loss": 0.1153, "step": 11338 }, { "epoch": 1.54, "grad_norm": 3.9974304572350774, "learning_rate": 1.3106792929870882e-06, "loss": 0.1799, "step": 11339 }, { "epoch": 1.54, "grad_norm": 4.26613246159661, "learning_rate": 1.3099358987769623e-06, "loss": 0.141, "step": 11340 }, { "epoch": 1.54, "grad_norm": 2.799102066852094, "learning_rate": 1.3091926836664325e-06, "loss": 0.1669, "step": 11341 }, { "epoch": 1.54, "grad_norm": 3.1224708682080022, "learning_rate": 1.3084496476915698e-06, "loss": 0.1698, "step": 11342 }, { "epoch": 1.54, "grad_norm": 4.189652149011901, "learning_rate": 1.3077067908884383e-06, "loss": 0.1677, "step": 11343 }, { "epoch": 1.54, "grad_norm": 5.038082367792856, "learning_rate": 1.3069641132930928e-06, "loss": 0.1245, "step": 11344 }, { "epoch": 1.54, "grad_norm": 3.2161658250905045, "learning_rate": 1.3062216149415791e-06, "loss": 0.1364, "step": 11345 }, { "epoch": 1.54, "grad_norm": 4.138570660627713, "learning_rate": 1.3054792958699352e-06, "loss": 0.1643, "step": 11346 }, { "epoch": 1.54, "grad_norm": 3.8217195940772934, "learning_rate": 1.3047371561141903e-06, "loss": 0.186, "step": 11347 }, { "epoch": 1.54, "grad_norm": 3.870045619454045, "learning_rate": 1.3039951957103641e-06, "loss": 0.1752, "step": 11348 }, { "epoch": 1.54, "grad_norm": 3.2447746823703776, "learning_rate": 1.3032534146944686e-06, "loss": 0.1252, "step": 11349 }, { "epoch": 1.54, "grad_norm": 3.0649623604327947, "learning_rate": 1.302511813102505e-06, "loss": 0.1683, "step": 11350 }, { "epoch": 1.54, "grad_norm": 5.1974470366176435, "learning_rate": 1.3017703909704716e-06, "loss": 0.1494, "step": 11351 }, { "epoch": 1.54, "grad_norm": 3.9286081035644127, "learning_rate": 1.3010291483343478e-06, "loss": 0.1314, "step": 11352 }, { "epoch": 1.54, "grad_norm": 3.0684639218593146, "learning_rate": 1.300288085230117e-06, "loss": 0.1097, "step": 11353 }, { "epoch": 1.54, "grad_norm": 2.680595439382768, "learning_rate": 1.2995472016937405e-06, "loss": 0.1242, "step": 11354 }, { "epoch": 1.54, "grad_norm": 4.048204062293534, "learning_rate": 1.298806497761182e-06, "loss": 0.2157, "step": 11355 }, { "epoch": 1.54, "grad_norm": 3.1347244927968085, "learning_rate": 1.2980659734683908e-06, "loss": 0.1113, "step": 11356 }, { "epoch": 1.54, "grad_norm": 3.8489188094423876, "learning_rate": 1.2973256288513086e-06, "loss": 0.1714, "step": 11357 }, { "epoch": 1.54, "grad_norm": 3.494037470626549, "learning_rate": 1.296585463945869e-06, "loss": 0.1318, "step": 11358 }, { "epoch": 1.55, "grad_norm": 3.2552029074053186, "learning_rate": 1.2958454787879953e-06, "loss": 0.185, "step": 11359 }, { "epoch": 1.55, "grad_norm": 3.4258193142850883, "learning_rate": 1.295105673413604e-06, "loss": 0.2, "step": 11360 }, { "epoch": 1.55, "grad_norm": 3.263366220329499, "learning_rate": 1.2943660478586017e-06, "loss": 0.1623, "step": 11361 }, { "epoch": 1.55, "grad_norm": 4.079486291049739, "learning_rate": 1.2936266021588872e-06, "loss": 0.1619, "step": 11362 }, { "epoch": 1.55, "grad_norm": 5.405025804641463, "learning_rate": 1.292887336350349e-06, "loss": 0.1703, "step": 11363 }, { "epoch": 1.55, "grad_norm": 2.9370336038778544, "learning_rate": 1.2921482504688664e-06, "loss": 0.1391, "step": 11364 }, { "epoch": 1.55, "grad_norm": 3.000645847534019, "learning_rate": 1.2914093445503167e-06, "loss": 0.1546, "step": 11365 }, { "epoch": 1.55, "grad_norm": 3.2799164965711767, "learning_rate": 1.2906706186305569e-06, "loss": 0.1466, "step": 11366 }, { "epoch": 1.55, "grad_norm": 3.2249293917427027, "learning_rate": 1.2899320727454472e-06, "loss": 0.1543, "step": 11367 }, { "epoch": 1.55, "grad_norm": 4.294095603336031, "learning_rate": 1.2891937069308275e-06, "loss": 0.1568, "step": 11368 }, { "epoch": 1.55, "grad_norm": 3.6316240636134727, "learning_rate": 1.2884555212225396e-06, "loss": 0.1869, "step": 11369 }, { "epoch": 1.55, "grad_norm": 2.600113935043354, "learning_rate": 1.2877175156564097e-06, "loss": 0.142, "step": 11370 }, { "epoch": 1.55, "grad_norm": 3.0314385176673784, "learning_rate": 1.2869796902682584e-06, "loss": 0.1184, "step": 11371 }, { "epoch": 1.55, "grad_norm": 2.913734050649313, "learning_rate": 1.2862420450938955e-06, "loss": 0.1381, "step": 11372 }, { "epoch": 1.55, "grad_norm": 2.92263999011059, "learning_rate": 1.2855045801691236e-06, "loss": 0.16, "step": 11373 }, { "epoch": 1.55, "grad_norm": 3.9063032673593847, "learning_rate": 1.2847672955297357e-06, "loss": 0.1537, "step": 11374 }, { "epoch": 1.55, "grad_norm": 3.73541570209467, "learning_rate": 1.284030191211517e-06, "loss": 0.1076, "step": 11375 }, { "epoch": 1.55, "grad_norm": 3.4471882098009683, "learning_rate": 1.2832932672502424e-06, "loss": 0.1665, "step": 11376 }, { "epoch": 1.55, "grad_norm": 3.3511460400069772, "learning_rate": 1.28255652368168e-06, "loss": 0.1424, "step": 11377 }, { "epoch": 1.55, "grad_norm": 3.826747481299563, "learning_rate": 1.2818199605415877e-06, "loss": 0.1551, "step": 11378 }, { "epoch": 1.55, "grad_norm": 4.832844224551598, "learning_rate": 1.2810835778657144e-06, "loss": 0.1673, "step": 11379 }, { "epoch": 1.55, "grad_norm": 3.9154330917859514, "learning_rate": 1.2803473756898017e-06, "loss": 0.1687, "step": 11380 }, { "epoch": 1.55, "grad_norm": 4.046924281726547, "learning_rate": 1.2796113540495814e-06, "loss": 0.1395, "step": 11381 }, { "epoch": 1.55, "grad_norm": 3.521309087148544, "learning_rate": 1.2788755129807767e-06, "loss": 0.1463, "step": 11382 }, { "epoch": 1.55, "grad_norm": 2.904616136669821, "learning_rate": 1.2781398525191008e-06, "loss": 0.1975, "step": 11383 }, { "epoch": 1.55, "grad_norm": 3.6175370767172867, "learning_rate": 1.2774043727002638e-06, "loss": 0.1235, "step": 11384 }, { "epoch": 1.55, "grad_norm": 3.35130502104058, "learning_rate": 1.2766690735599569e-06, "loss": 0.1669, "step": 11385 }, { "epoch": 1.55, "grad_norm": 3.4649667877128434, "learning_rate": 1.2759339551338733e-06, "loss": 0.1594, "step": 11386 }, { "epoch": 1.55, "grad_norm": 3.6969060740765287, "learning_rate": 1.2751990174576883e-06, "loss": 0.1152, "step": 11387 }, { "epoch": 1.55, "grad_norm": 4.666524810239635, "learning_rate": 1.2744642605670754e-06, "loss": 0.1599, "step": 11388 }, { "epoch": 1.55, "grad_norm": 3.5489746627380576, "learning_rate": 1.2737296844976959e-06, "loss": 0.1861, "step": 11389 }, { "epoch": 1.55, "grad_norm": 3.105037454972097, "learning_rate": 1.272995289285202e-06, "loss": 0.138, "step": 11390 }, { "epoch": 1.55, "grad_norm": 3.562716662524749, "learning_rate": 1.2722610749652397e-06, "loss": 0.1251, "step": 11391 }, { "epoch": 1.55, "grad_norm": 3.2363777332115706, "learning_rate": 1.2715270415734425e-06, "loss": 0.1586, "step": 11392 }, { "epoch": 1.55, "grad_norm": 3.817390781400511, "learning_rate": 1.2707931891454389e-06, "loss": 0.1597, "step": 11393 }, { "epoch": 1.55, "grad_norm": 4.370877285896312, "learning_rate": 1.2700595177168462e-06, "loss": 0.1863, "step": 11394 }, { "epoch": 1.55, "grad_norm": 2.8954791086386673, "learning_rate": 1.2693260273232732e-06, "loss": 0.1817, "step": 11395 }, { "epoch": 1.55, "grad_norm": 2.809839521403623, "learning_rate": 1.2685927180003205e-06, "loss": 0.1631, "step": 11396 }, { "epoch": 1.55, "grad_norm": 4.235388521386146, "learning_rate": 1.2678595897835788e-06, "loss": 0.1696, "step": 11397 }, { "epoch": 1.55, "grad_norm": 3.8902563070188867, "learning_rate": 1.2671266427086354e-06, "loss": 0.2126, "step": 11398 }, { "epoch": 1.55, "grad_norm": 3.3135166326842795, "learning_rate": 1.2663938768110572e-06, "loss": 0.1349, "step": 11399 }, { "epoch": 1.55, "grad_norm": 4.005823139636446, "learning_rate": 1.2656612921264171e-06, "loss": 0.1722, "step": 11400 }, { "epoch": 1.55, "grad_norm": 3.325949097260997, "learning_rate": 1.2649288886902644e-06, "loss": 0.1894, "step": 11401 }, { "epoch": 1.55, "grad_norm": 3.7298888250516953, "learning_rate": 1.2641966665381517e-06, "loss": 0.1507, "step": 11402 }, { "epoch": 1.55, "grad_norm": 3.5885904935748076, "learning_rate": 1.263464625705616e-06, "loss": 0.1769, "step": 11403 }, { "epoch": 1.55, "grad_norm": 3.094258510116258, "learning_rate": 1.262732766228188e-06, "loss": 0.1458, "step": 11404 }, { "epoch": 1.55, "grad_norm": 3.6903427378662608, "learning_rate": 1.2620010881413885e-06, "loss": 0.1501, "step": 11405 }, { "epoch": 1.55, "grad_norm": 3.4788220146303903, "learning_rate": 1.26126959148073e-06, "loss": 0.1179, "step": 11406 }, { "epoch": 1.55, "grad_norm": 4.713020804331854, "learning_rate": 1.2605382762817164e-06, "loss": 0.1759, "step": 11407 }, { "epoch": 1.55, "grad_norm": 5.595620425178424, "learning_rate": 1.259807142579842e-06, "loss": 0.166, "step": 11408 }, { "epoch": 1.55, "grad_norm": 3.854047328228978, "learning_rate": 1.2590761904105937e-06, "loss": 0.1588, "step": 11409 }, { "epoch": 1.55, "grad_norm": 3.53911082241916, "learning_rate": 1.2583454198094475e-06, "loss": 0.1156, "step": 11410 }, { "epoch": 1.55, "grad_norm": 3.5542028750368964, "learning_rate": 1.2576148308118718e-06, "loss": 0.1846, "step": 11411 }, { "epoch": 1.55, "grad_norm": 3.665785484894035, "learning_rate": 1.2568844234533294e-06, "loss": 0.2041, "step": 11412 }, { "epoch": 1.55, "grad_norm": 3.9772248731822186, "learning_rate": 1.256154197769266e-06, "loss": 0.1977, "step": 11413 }, { "epoch": 1.55, "grad_norm": 3.655642214802246, "learning_rate": 1.255424153795129e-06, "loss": 0.148, "step": 11414 }, { "epoch": 1.55, "grad_norm": 3.3830524252910132, "learning_rate": 1.2546942915663457e-06, "loss": 0.1231, "step": 11415 }, { "epoch": 1.55, "grad_norm": 3.794815313487445, "learning_rate": 1.2539646111183452e-06, "loss": 0.1646, "step": 11416 }, { "epoch": 1.55, "grad_norm": 4.1717492426685565, "learning_rate": 1.253235112486541e-06, "loss": 0.1346, "step": 11417 }, { "epoch": 1.55, "grad_norm": 4.483540619450742, "learning_rate": 1.2525057957063397e-06, "loss": 0.1389, "step": 11418 }, { "epoch": 1.55, "grad_norm": 3.9106031289094165, "learning_rate": 1.2517766608131415e-06, "loss": 0.1506, "step": 11419 }, { "epoch": 1.55, "grad_norm": 3.6491334132548436, "learning_rate": 1.2510477078423306e-06, "loss": 0.1171, "step": 11420 }, { "epoch": 1.55, "grad_norm": 4.26329101335444, "learning_rate": 1.2503189368292912e-06, "loss": 0.1608, "step": 11421 }, { "epoch": 1.55, "grad_norm": 3.0194958621920573, "learning_rate": 1.249590347809393e-06, "loss": 0.1014, "step": 11422 }, { "epoch": 1.55, "grad_norm": 2.8658855342213707, "learning_rate": 1.2488619408179996e-06, "loss": 0.1296, "step": 11423 }, { "epoch": 1.55, "grad_norm": 3.3065899537381105, "learning_rate": 1.248133715890464e-06, "loss": 0.1725, "step": 11424 }, { "epoch": 1.55, "grad_norm": 3.5608919499044274, "learning_rate": 1.247405673062131e-06, "loss": 0.1734, "step": 11425 }, { "epoch": 1.55, "grad_norm": 3.7532830921504776, "learning_rate": 1.246677812368337e-06, "loss": 0.1367, "step": 11426 }, { "epoch": 1.55, "grad_norm": 3.1499152491236235, "learning_rate": 1.2459501338444085e-06, "loss": 0.1347, "step": 11427 }, { "epoch": 1.55, "grad_norm": 3.4662552253209706, "learning_rate": 1.2452226375256648e-06, "loss": 0.1478, "step": 11428 }, { "epoch": 1.55, "grad_norm": 3.8732879480624307, "learning_rate": 1.2444953234474144e-06, "loss": 0.1787, "step": 11429 }, { "epoch": 1.55, "grad_norm": 2.9561226036141997, "learning_rate": 1.2437681916449573e-06, "loss": 0.1712, "step": 11430 }, { "epoch": 1.55, "grad_norm": 3.3607526986624063, "learning_rate": 1.2430412421535892e-06, "loss": 0.1695, "step": 11431 }, { "epoch": 1.56, "grad_norm": 3.2796525305949173, "learning_rate": 1.2423144750085875e-06, "loss": 0.1292, "step": 11432 }, { "epoch": 1.56, "grad_norm": 3.7816895586127117, "learning_rate": 1.2415878902452321e-06, "loss": 0.1843, "step": 11433 }, { "epoch": 1.56, "grad_norm": 3.380001006243708, "learning_rate": 1.2408614878987818e-06, "loss": 0.1871, "step": 11434 }, { "epoch": 1.56, "grad_norm": 4.03849168222316, "learning_rate": 1.2401352680044982e-06, "loss": 0.1554, "step": 11435 }, { "epoch": 1.56, "grad_norm": 3.6616775722444603, "learning_rate": 1.2394092305976274e-06, "loss": 0.1156, "step": 11436 }, { "epoch": 1.56, "grad_norm": 3.893130450242388, "learning_rate": 1.2386833757134076e-06, "loss": 0.1592, "step": 11437 }, { "epoch": 1.56, "grad_norm": 2.6192987028962555, "learning_rate": 1.2379577033870683e-06, "loss": 0.1454, "step": 11438 }, { "epoch": 1.56, "grad_norm": 2.68251038317545, "learning_rate": 1.2372322136538311e-06, "loss": 0.1288, "step": 11439 }, { "epoch": 1.56, "grad_norm": 3.2885310092153794, "learning_rate": 1.2365069065489083e-06, "loss": 0.1831, "step": 11440 }, { "epoch": 1.56, "grad_norm": 3.6654543181150503, "learning_rate": 1.2357817821075025e-06, "loss": 0.1599, "step": 11441 }, { "epoch": 1.56, "grad_norm": 2.9511500217583517, "learning_rate": 1.2350568403648088e-06, "loss": 0.159, "step": 11442 }, { "epoch": 1.56, "grad_norm": 2.7930726557775487, "learning_rate": 1.2343320813560118e-06, "loss": 0.1187, "step": 11443 }, { "epoch": 1.56, "grad_norm": 4.275606137430864, "learning_rate": 1.2336075051162872e-06, "loss": 0.1439, "step": 11444 }, { "epoch": 1.56, "grad_norm": 2.8567965363479395, "learning_rate": 1.232883111680807e-06, "loss": 0.1332, "step": 11445 }, { "epoch": 1.56, "grad_norm": 3.289663251879871, "learning_rate": 1.2321589010847246e-06, "loss": 0.1677, "step": 11446 }, { "epoch": 1.56, "grad_norm": 2.9958951758831502, "learning_rate": 1.2314348733631958e-06, "loss": 0.1141, "step": 11447 }, { "epoch": 1.56, "grad_norm": 4.699358942694786, "learning_rate": 1.2307110285513552e-06, "loss": 0.1808, "step": 11448 }, { "epoch": 1.56, "grad_norm": 3.0539141062805006, "learning_rate": 1.2299873666843404e-06, "loss": 0.1683, "step": 11449 }, { "epoch": 1.56, "grad_norm": 3.417908128013327, "learning_rate": 1.2292638877972723e-06, "loss": 0.1633, "step": 11450 }, { "epoch": 1.56, "grad_norm": 3.4832562033859302, "learning_rate": 1.228540591925267e-06, "loss": 0.1445, "step": 11451 }, { "epoch": 1.56, "grad_norm": 2.964762127821855, "learning_rate": 1.2278174791034281e-06, "loss": 0.1223, "step": 11452 }, { "epoch": 1.56, "grad_norm": 3.4928812229402815, "learning_rate": 1.2270945493668535e-06, "loss": 0.1439, "step": 11453 }, { "epoch": 1.56, "grad_norm": 3.503520992131795, "learning_rate": 1.226371802750631e-06, "loss": 0.1872, "step": 11454 }, { "epoch": 1.56, "grad_norm": 4.456188866487882, "learning_rate": 1.22564923928984e-06, "loss": 0.17, "step": 11455 }, { "epoch": 1.56, "grad_norm": 4.853067823491656, "learning_rate": 1.2249268590195479e-06, "loss": 0.152, "step": 11456 }, { "epoch": 1.56, "grad_norm": 3.446895750422973, "learning_rate": 1.224204661974821e-06, "loss": 0.1697, "step": 11457 }, { "epoch": 1.56, "grad_norm": 3.9716099355135652, "learning_rate": 1.2234826481907075e-06, "loss": 0.1828, "step": 11458 }, { "epoch": 1.56, "grad_norm": 3.8048229645511404, "learning_rate": 1.2227608177022522e-06, "loss": 0.14, "step": 11459 }, { "epoch": 1.56, "grad_norm": 4.210753977046517, "learning_rate": 1.2220391705444884e-06, "loss": 0.1099, "step": 11460 }, { "epoch": 1.56, "grad_norm": 2.988335480880473, "learning_rate": 1.2213177067524428e-06, "loss": 0.124, "step": 11461 }, { "epoch": 1.56, "grad_norm": 4.25229553255579, "learning_rate": 1.2205964263611325e-06, "loss": 0.1494, "step": 11462 }, { "epoch": 1.56, "grad_norm": 4.574469831214055, "learning_rate": 1.2198753294055627e-06, "loss": 0.1835, "step": 11463 }, { "epoch": 1.56, "grad_norm": 3.2909457458825493, "learning_rate": 1.2191544159207376e-06, "loss": 0.1329, "step": 11464 }, { "epoch": 1.56, "grad_norm": 3.978135908593723, "learning_rate": 1.2184336859416408e-06, "loss": 0.1792, "step": 11465 }, { "epoch": 1.56, "grad_norm": 3.4488244126297896, "learning_rate": 1.2177131395032592e-06, "loss": 0.1261, "step": 11466 }, { "epoch": 1.56, "grad_norm": 3.580508660106955, "learning_rate": 1.2169927766405598e-06, "loss": 0.1112, "step": 11467 }, { "epoch": 1.56, "grad_norm": 3.0153174818732014, "learning_rate": 1.2162725973885092e-06, "loss": 0.0836, "step": 11468 }, { "epoch": 1.56, "grad_norm": 3.5096781794089775, "learning_rate": 1.2155526017820612e-06, "loss": 0.1266, "step": 11469 }, { "epoch": 1.56, "grad_norm": 4.1297427094088714, "learning_rate": 1.2148327898561612e-06, "loss": 0.1676, "step": 11470 }, { "epoch": 1.56, "grad_norm": 4.5137996886874, "learning_rate": 1.2141131616457452e-06, "loss": 0.1704, "step": 11471 }, { "epoch": 1.56, "grad_norm": 4.670479562211925, "learning_rate": 1.2133937171857406e-06, "loss": 0.1221, "step": 11472 }, { "epoch": 1.56, "grad_norm": 3.2579101145950333, "learning_rate": 1.2126744565110676e-06, "loss": 0.1991, "step": 11473 }, { "epoch": 1.56, "grad_norm": 3.876546007166804, "learning_rate": 1.2119553796566342e-06, "loss": 0.2007, "step": 11474 }, { "epoch": 1.56, "grad_norm": 3.6525858757992027, "learning_rate": 1.2112364866573407e-06, "loss": 0.1642, "step": 11475 }, { "epoch": 1.56, "grad_norm": 3.7164675798446876, "learning_rate": 1.2105177775480831e-06, "loss": 0.1673, "step": 11476 }, { "epoch": 1.56, "grad_norm": 3.7772614855928763, "learning_rate": 1.2097992523637387e-06, "loss": 0.1795, "step": 11477 }, { "epoch": 1.56, "grad_norm": 4.34957800195986, "learning_rate": 1.209080911139187e-06, "loss": 0.1492, "step": 11478 }, { "epoch": 1.56, "grad_norm": 3.9510389363355713, "learning_rate": 1.2083627539092884e-06, "loss": 0.1544, "step": 11479 }, { "epoch": 1.56, "grad_norm": 3.5882537074342276, "learning_rate": 1.2076447807089042e-06, "loss": 0.1346, "step": 11480 }, { "epoch": 1.56, "grad_norm": 2.8870133667749256, "learning_rate": 1.2069269915728755e-06, "loss": 0.1634, "step": 11481 }, { "epoch": 1.56, "grad_norm": 4.312980822005069, "learning_rate": 1.2062093865360458e-06, "loss": 0.1626, "step": 11482 }, { "epoch": 1.56, "grad_norm": 2.6243657171212926, "learning_rate": 1.205491965633242e-06, "loss": 0.1503, "step": 11483 }, { "epoch": 1.56, "grad_norm": 2.940744578327449, "learning_rate": 1.204774728899286e-06, "loss": 0.1195, "step": 11484 }, { "epoch": 1.56, "grad_norm": 4.078774900749607, "learning_rate": 1.2040576763689881e-06, "loss": 0.197, "step": 11485 }, { "epoch": 1.56, "grad_norm": 3.533905777732073, "learning_rate": 1.203340808077152e-06, "loss": 0.1723, "step": 11486 }, { "epoch": 1.56, "grad_norm": 3.3847010590974356, "learning_rate": 1.2026241240585702e-06, "loss": 0.1476, "step": 11487 }, { "epoch": 1.56, "grad_norm": 4.226861004698985, "learning_rate": 1.2019076243480283e-06, "loss": 0.1668, "step": 11488 }, { "epoch": 1.56, "grad_norm": 5.388576098559297, "learning_rate": 1.2011913089803e-06, "loss": 0.1654, "step": 11489 }, { "epoch": 1.56, "grad_norm": 2.817791080513135, "learning_rate": 1.200475177990157e-06, "loss": 0.15, "step": 11490 }, { "epoch": 1.56, "grad_norm": 2.7229104290988815, "learning_rate": 1.1997592314123506e-06, "loss": 0.1384, "step": 11491 }, { "epoch": 1.56, "grad_norm": 4.104442995448661, "learning_rate": 1.1990434692816367e-06, "loss": 0.1677, "step": 11492 }, { "epoch": 1.56, "grad_norm": 3.748901304943192, "learning_rate": 1.1983278916327485e-06, "loss": 0.1458, "step": 11493 }, { "epoch": 1.56, "grad_norm": 3.4404353513845596, "learning_rate": 1.1976124985004235e-06, "loss": 0.1486, "step": 11494 }, { "epoch": 1.56, "grad_norm": 3.1877342322541375, "learning_rate": 1.1968972899193771e-06, "loss": 0.1422, "step": 11495 }, { "epoch": 1.56, "grad_norm": 3.7814574610951674, "learning_rate": 1.1961822659243273e-06, "loss": 0.1519, "step": 11496 }, { "epoch": 1.56, "grad_norm": 3.9595539010690497, "learning_rate": 1.1954674265499773e-06, "loss": 0.1618, "step": 11497 }, { "epoch": 1.56, "grad_norm": 3.4713058664853977, "learning_rate": 1.194752771831021e-06, "loss": 0.1654, "step": 11498 }, { "epoch": 1.56, "grad_norm": 3.5328900401522563, "learning_rate": 1.1940383018021468e-06, "loss": 0.1339, "step": 11499 }, { "epoch": 1.56, "grad_norm": 5.356183852125207, "learning_rate": 1.193324016498028e-06, "loss": 0.1706, "step": 11500 }, { "epoch": 1.56, "grad_norm": 2.8662383146117016, "learning_rate": 1.1926099159533367e-06, "loss": 0.1138, "step": 11501 }, { "epoch": 1.56, "grad_norm": 3.3784278498379243, "learning_rate": 1.1918960002027308e-06, "loss": 0.1532, "step": 11502 }, { "epoch": 1.56, "grad_norm": 4.96820284010826, "learning_rate": 1.1911822692808605e-06, "loss": 0.1595, "step": 11503 }, { "epoch": 1.56, "grad_norm": 3.1893004348426226, "learning_rate": 1.1904687232223678e-06, "loss": 0.1626, "step": 11504 }, { "epoch": 1.56, "grad_norm": 3.335267727135409, "learning_rate": 1.1897553620618845e-06, "loss": 0.1874, "step": 11505 }, { "epoch": 1.57, "grad_norm": 3.297375289349136, "learning_rate": 1.189042185834034e-06, "loss": 0.1773, "step": 11506 }, { "epoch": 1.57, "grad_norm": 2.930517662612655, "learning_rate": 1.1883291945734315e-06, "loss": 0.159, "step": 11507 }, { "epoch": 1.57, "grad_norm": 3.2906642953652376, "learning_rate": 1.1876163883146808e-06, "loss": 0.1439, "step": 11508 }, { "epoch": 1.57, "grad_norm": 3.8521948360027753, "learning_rate": 1.1869037670923817e-06, "loss": 0.1453, "step": 11509 }, { "epoch": 1.57, "grad_norm": 3.3044242250529665, "learning_rate": 1.1861913309411172e-06, "loss": 0.1552, "step": 11510 }, { "epoch": 1.57, "grad_norm": 3.219755308012506, "learning_rate": 1.1854790798954713e-06, "loss": 0.1704, "step": 11511 }, { "epoch": 1.57, "grad_norm": 3.4802969641092107, "learning_rate": 1.1847670139900074e-06, "loss": 0.1551, "step": 11512 }, { "epoch": 1.57, "grad_norm": 3.1956809154177117, "learning_rate": 1.184055133259292e-06, "loss": 0.1466, "step": 11513 }, { "epoch": 1.57, "grad_norm": 3.0117979588786596, "learning_rate": 1.183343437737871e-06, "loss": 0.1212, "step": 11514 }, { "epoch": 1.57, "grad_norm": 4.842170667648115, "learning_rate": 1.1826319274602915e-06, "loss": 0.1121, "step": 11515 }, { "epoch": 1.57, "grad_norm": 4.953772786313231, "learning_rate": 1.1819206024610858e-06, "loss": 0.1253, "step": 11516 }, { "epoch": 1.57, "grad_norm": 3.4602167911133384, "learning_rate": 1.1812094627747777e-06, "loss": 0.1579, "step": 11517 }, { "epoch": 1.57, "grad_norm": 5.416956690134462, "learning_rate": 1.1804985084358832e-06, "loss": 0.1402, "step": 11518 }, { "epoch": 1.57, "grad_norm": 3.5516764975801665, "learning_rate": 1.1797877394789093e-06, "loss": 0.1363, "step": 11519 }, { "epoch": 1.57, "grad_norm": 3.484156184978576, "learning_rate": 1.1790771559383536e-06, "loss": 0.1345, "step": 11520 }, { "epoch": 1.57, "grad_norm": 3.2736208038905583, "learning_rate": 1.1783667578487035e-06, "loss": 0.1489, "step": 11521 }, { "epoch": 1.57, "grad_norm": 5.134065012971367, "learning_rate": 1.1776565452444389e-06, "loss": 0.1513, "step": 11522 }, { "epoch": 1.57, "grad_norm": 3.3938883886438616, "learning_rate": 1.1769465181600336e-06, "loss": 0.1503, "step": 11523 }, { "epoch": 1.57, "grad_norm": 3.1439843757861037, "learning_rate": 1.1762366766299442e-06, "loss": 0.1261, "step": 11524 }, { "epoch": 1.57, "grad_norm": 3.3524503830267376, "learning_rate": 1.1755270206886289e-06, "loss": 0.1317, "step": 11525 }, { "epoch": 1.57, "grad_norm": 3.520779037201345, "learning_rate": 1.1748175503705245e-06, "loss": 0.159, "step": 11526 }, { "epoch": 1.57, "grad_norm": 4.021857676176453, "learning_rate": 1.174108265710071e-06, "loss": 0.198, "step": 11527 }, { "epoch": 1.57, "grad_norm": 2.790902994382263, "learning_rate": 1.1733991667416928e-06, "loss": 0.1592, "step": 11528 }, { "epoch": 1.57, "grad_norm": 2.7283287917981567, "learning_rate": 1.1726902534998059e-06, "loss": 0.1573, "step": 11529 }, { "epoch": 1.57, "grad_norm": 3.699896954544652, "learning_rate": 1.1719815260188171e-06, "loss": 0.1343, "step": 11530 }, { "epoch": 1.57, "grad_norm": 3.4767589436878095, "learning_rate": 1.1712729843331266e-06, "loss": 0.1732, "step": 11531 }, { "epoch": 1.57, "grad_norm": 3.6853558489002194, "learning_rate": 1.1705646284771227e-06, "loss": 0.1733, "step": 11532 }, { "epoch": 1.57, "grad_norm": 2.962850229147835, "learning_rate": 1.1698564584851862e-06, "loss": 0.1779, "step": 11533 }, { "epoch": 1.57, "grad_norm": 4.0241589074742485, "learning_rate": 1.1691484743916886e-06, "loss": 0.195, "step": 11534 }, { "epoch": 1.57, "grad_norm": 3.1841241556169937, "learning_rate": 1.1684406762309925e-06, "loss": 0.1737, "step": 11535 }, { "epoch": 1.57, "grad_norm": 3.734295903976324, "learning_rate": 1.1677330640374496e-06, "loss": 0.137, "step": 11536 }, { "epoch": 1.57, "grad_norm": 3.698209653188048, "learning_rate": 1.1670256378454093e-06, "loss": 0.1429, "step": 11537 }, { "epoch": 1.57, "grad_norm": 2.8522074331166394, "learning_rate": 1.1663183976892018e-06, "loss": 0.1683, "step": 11538 }, { "epoch": 1.57, "grad_norm": 2.828809560227206, "learning_rate": 1.1656113436031553e-06, "loss": 0.1655, "step": 11539 }, { "epoch": 1.57, "grad_norm": 2.604477397375025, "learning_rate": 1.1649044756215872e-06, "loss": 0.1369, "step": 11540 }, { "epoch": 1.57, "grad_norm": 4.495922304653907, "learning_rate": 1.164197793778804e-06, "loss": 0.1419, "step": 11541 }, { "epoch": 1.57, "grad_norm": 3.0884134117333324, "learning_rate": 1.1634912981091096e-06, "loss": 0.1872, "step": 11542 }, { "epoch": 1.57, "grad_norm": 3.72002403294532, "learning_rate": 1.1627849886467885e-06, "loss": 0.1459, "step": 11543 }, { "epoch": 1.57, "grad_norm": 2.865149438215055, "learning_rate": 1.1620788654261272e-06, "loss": 0.1536, "step": 11544 }, { "epoch": 1.57, "grad_norm": 3.7488265568311605, "learning_rate": 1.1613729284813924e-06, "loss": 0.1041, "step": 11545 }, { "epoch": 1.57, "grad_norm": 4.247313155745198, "learning_rate": 1.1606671778468515e-06, "loss": 0.1237, "step": 11546 }, { "epoch": 1.57, "grad_norm": 3.4604155513274697, "learning_rate": 1.159961613556757e-06, "loss": 0.1314, "step": 11547 }, { "epoch": 1.57, "grad_norm": 3.0992608671511856, "learning_rate": 1.159256235645354e-06, "loss": 0.1634, "step": 11548 }, { "epoch": 1.57, "grad_norm": 4.0087535359101825, "learning_rate": 1.1585510441468784e-06, "loss": 0.1676, "step": 11549 }, { "epoch": 1.57, "grad_norm": 3.878348207858476, "learning_rate": 1.1578460390955576e-06, "loss": 0.1731, "step": 11550 }, { "epoch": 1.57, "grad_norm": 3.6853414633010115, "learning_rate": 1.1571412205256089e-06, "loss": 0.1764, "step": 11551 }, { "epoch": 1.57, "grad_norm": 3.2537737317294164, "learning_rate": 1.1564365884712409e-06, "loss": 0.1425, "step": 11552 }, { "epoch": 1.57, "grad_norm": 3.5562541777956085, "learning_rate": 1.155732142966654e-06, "loss": 0.1424, "step": 11553 }, { "epoch": 1.57, "grad_norm": 3.5482896895390024, "learning_rate": 1.1550278840460382e-06, "loss": 0.1629, "step": 11554 }, { "epoch": 1.57, "grad_norm": 3.347582444279871, "learning_rate": 1.1543238117435746e-06, "loss": 0.1603, "step": 11555 }, { "epoch": 1.57, "grad_norm": 3.534329772792004, "learning_rate": 1.1536199260934395e-06, "loss": 0.1621, "step": 11556 }, { "epoch": 1.57, "grad_norm": 3.0688270626920264, "learning_rate": 1.1529162271297912e-06, "loss": 0.1174, "step": 11557 }, { "epoch": 1.57, "grad_norm": 7.432558082333218, "learning_rate": 1.152212714886789e-06, "loss": 0.1599, "step": 11558 }, { "epoch": 1.57, "grad_norm": 3.5743311792944037, "learning_rate": 1.1515093893985734e-06, "loss": 0.1932, "step": 11559 }, { "epoch": 1.57, "grad_norm": 3.2882458337049583, "learning_rate": 1.1508062506992844e-06, "loss": 0.1374, "step": 11560 }, { "epoch": 1.57, "grad_norm": 3.617250592404738, "learning_rate": 1.1501032988230487e-06, "loss": 0.1924, "step": 11561 }, { "epoch": 1.57, "grad_norm": 4.803750630221656, "learning_rate": 1.1494005338039839e-06, "loss": 0.1675, "step": 11562 }, { "epoch": 1.57, "grad_norm": 3.6214138637234177, "learning_rate": 1.1486979556761991e-06, "loss": 0.1518, "step": 11563 }, { "epoch": 1.57, "grad_norm": 4.407004416762557, "learning_rate": 1.1479955644737945e-06, "loss": 0.1826, "step": 11564 }, { "epoch": 1.57, "grad_norm": 3.718625325097088, "learning_rate": 1.1472933602308612e-06, "loss": 0.159, "step": 11565 }, { "epoch": 1.57, "grad_norm": 3.072788887310808, "learning_rate": 1.1465913429814813e-06, "loss": 0.1706, "step": 11566 }, { "epoch": 1.57, "grad_norm": 3.942053626953227, "learning_rate": 1.1458895127597275e-06, "loss": 0.1795, "step": 11567 }, { "epoch": 1.57, "grad_norm": 3.466615256208736, "learning_rate": 1.1451878695996631e-06, "loss": 0.1966, "step": 11568 }, { "epoch": 1.57, "grad_norm": 2.732015199865271, "learning_rate": 1.1444864135353424e-06, "loss": 0.1552, "step": 11569 }, { "epoch": 1.57, "grad_norm": 3.787571661467617, "learning_rate": 1.143785144600814e-06, "loss": 0.1868, "step": 11570 }, { "epoch": 1.57, "grad_norm": 3.594278738356912, "learning_rate": 1.1430840628301093e-06, "loss": 0.158, "step": 11571 }, { "epoch": 1.57, "grad_norm": 3.3934218871475146, "learning_rate": 1.1423831682572623e-06, "loss": 0.1767, "step": 11572 }, { "epoch": 1.57, "grad_norm": 3.4679877341130108, "learning_rate": 1.1416824609162842e-06, "loss": 0.151, "step": 11573 }, { "epoch": 1.57, "grad_norm": 7.521980202403136, "learning_rate": 1.1409819408411898e-06, "loss": 0.144, "step": 11574 }, { "epoch": 1.57, "grad_norm": 3.4570069916320088, "learning_rate": 1.1402816080659768e-06, "loss": 0.1341, "step": 11575 }, { "epoch": 1.57, "grad_norm": 3.0131327167391864, "learning_rate": 1.1395814626246372e-06, "loss": 0.1233, "step": 11576 }, { "epoch": 1.57, "grad_norm": 6.680916738730264, "learning_rate": 1.1388815045511525e-06, "loss": 0.1632, "step": 11577 }, { "epoch": 1.57, "grad_norm": 3.7804554489877376, "learning_rate": 1.1381817338794954e-06, "loss": 0.15, "step": 11578 }, { "epoch": 1.58, "grad_norm": 5.167229983471247, "learning_rate": 1.13748215064363e-06, "loss": 0.1699, "step": 11579 }, { "epoch": 1.58, "grad_norm": 3.6445513511906484, "learning_rate": 1.1367827548775117e-06, "loss": 0.1679, "step": 11580 }, { "epoch": 1.58, "grad_norm": 2.9389215219383167, "learning_rate": 1.1360835466150849e-06, "loss": 0.1133, "step": 11581 }, { "epoch": 1.58, "grad_norm": 3.6828789198695655, "learning_rate": 1.1353845258902867e-06, "loss": 0.1261, "step": 11582 }, { "epoch": 1.58, "grad_norm": 4.184991856964836, "learning_rate": 1.1346856927370448e-06, "loss": 0.1789, "step": 11583 }, { "epoch": 1.58, "grad_norm": 3.0797125112062327, "learning_rate": 1.1339870471892771e-06, "loss": 0.1618, "step": 11584 }, { "epoch": 1.58, "grad_norm": 3.7834012578810836, "learning_rate": 1.133288589280893e-06, "loss": 0.1291, "step": 11585 }, { "epoch": 1.58, "grad_norm": 3.2055762269289434, "learning_rate": 1.1325903190457922e-06, "loss": 0.1267, "step": 11586 }, { "epoch": 1.58, "grad_norm": 3.0713312761255995, "learning_rate": 1.131892236517866e-06, "loss": 0.1291, "step": 11587 }, { "epoch": 1.58, "grad_norm": 3.446705466427996, "learning_rate": 1.1311943417309951e-06, "loss": 0.1743, "step": 11588 }, { "epoch": 1.58, "grad_norm": 3.76557138019603, "learning_rate": 1.130496634719056e-06, "loss": 0.164, "step": 11589 }, { "epoch": 1.58, "grad_norm": 2.92225915665389, "learning_rate": 1.129799115515907e-06, "loss": 0.142, "step": 11590 }, { "epoch": 1.58, "grad_norm": 3.0388805130577974, "learning_rate": 1.1291017841554081e-06, "loss": 0.1245, "step": 11591 }, { "epoch": 1.58, "grad_norm": 4.229818754577493, "learning_rate": 1.1284046406713994e-06, "loss": 0.1452, "step": 11592 }, { "epoch": 1.58, "grad_norm": 2.960511123936671, "learning_rate": 1.1277076850977214e-06, "loss": 0.1166, "step": 11593 }, { "epoch": 1.58, "grad_norm": 3.560067543972837, "learning_rate": 1.1270109174681992e-06, "loss": 0.1591, "step": 11594 }, { "epoch": 1.58, "grad_norm": 2.962578973656311, "learning_rate": 1.126314337816652e-06, "loss": 0.1437, "step": 11595 }, { "epoch": 1.58, "grad_norm": 3.7866723626518697, "learning_rate": 1.125617946176888e-06, "loss": 0.174, "step": 11596 }, { "epoch": 1.58, "grad_norm": 3.518804012820878, "learning_rate": 1.1249217425827063e-06, "loss": 0.1647, "step": 11597 }, { "epoch": 1.58, "grad_norm": 2.431008617636405, "learning_rate": 1.124225727067899e-06, "loss": 0.1332, "step": 11598 }, { "epoch": 1.58, "grad_norm": 4.7201557338631694, "learning_rate": 1.1235298996662474e-06, "loss": 0.2021, "step": 11599 }, { "epoch": 1.58, "grad_norm": 3.5384549789042303, "learning_rate": 1.1228342604115227e-06, "loss": 0.1621, "step": 11600 }, { "epoch": 1.58, "grad_norm": 3.1853119857969463, "learning_rate": 1.1221388093374896e-06, "loss": 0.1816, "step": 11601 }, { "epoch": 1.58, "grad_norm": 3.079795422721796, "learning_rate": 1.1214435464779006e-06, "loss": 0.1599, "step": 11602 }, { "epoch": 1.58, "grad_norm": 3.3790771336720034, "learning_rate": 1.120748471866504e-06, "loss": 0.149, "step": 11603 }, { "epoch": 1.58, "grad_norm": 3.0147582305768355, "learning_rate": 1.120053585537031e-06, "loss": 0.1517, "step": 11604 }, { "epoch": 1.58, "grad_norm": 4.721165120940972, "learning_rate": 1.1193588875232136e-06, "loss": 0.1633, "step": 11605 }, { "epoch": 1.58, "grad_norm": 3.1674754233851194, "learning_rate": 1.1186643778587642e-06, "loss": 0.1428, "step": 11606 }, { "epoch": 1.58, "grad_norm": 3.6271142097545996, "learning_rate": 1.117970056577395e-06, "loss": 0.1111, "step": 11607 }, { "epoch": 1.58, "grad_norm": 7.7607722875580105, "learning_rate": 1.117275923712804e-06, "loss": 0.1588, "step": 11608 }, { "epoch": 1.58, "grad_norm": 4.053950208554313, "learning_rate": 1.116581979298681e-06, "loss": 0.1812, "step": 11609 }, { "epoch": 1.58, "grad_norm": 4.7388684454537255, "learning_rate": 1.1158882233687079e-06, "loss": 0.1153, "step": 11610 }, { "epoch": 1.58, "grad_norm": 3.52941265736272, "learning_rate": 1.1151946559565563e-06, "loss": 0.152, "step": 11611 }, { "epoch": 1.58, "grad_norm": 4.100588458463484, "learning_rate": 1.1145012770958885e-06, "loss": 0.1982, "step": 11612 }, { "epoch": 1.58, "grad_norm": 2.546658527825313, "learning_rate": 1.1138080868203583e-06, "loss": 0.1281, "step": 11613 }, { "epoch": 1.58, "grad_norm": 3.5465481142175106, "learning_rate": 1.1131150851636107e-06, "loss": 0.1396, "step": 11614 }, { "epoch": 1.58, "grad_norm": 3.7041089042001363, "learning_rate": 1.1124222721592797e-06, "loss": 0.1982, "step": 11615 }, { "epoch": 1.58, "grad_norm": 2.933913416610386, "learning_rate": 1.1117296478409916e-06, "loss": 0.1411, "step": 11616 }, { "epoch": 1.58, "grad_norm": 4.654110369654281, "learning_rate": 1.1110372122423663e-06, "loss": 0.1318, "step": 11617 }, { "epoch": 1.58, "grad_norm": 3.3760125308974787, "learning_rate": 1.1103449653970077e-06, "loss": 0.1512, "step": 11618 }, { "epoch": 1.58, "grad_norm": 3.5879961102262934, "learning_rate": 1.1096529073385164e-06, "loss": 0.2064, "step": 11619 }, { "epoch": 1.58, "grad_norm": 3.189793421838549, "learning_rate": 1.1089610381004812e-06, "loss": 0.1101, "step": 11620 }, { "epoch": 1.58, "grad_norm": 2.943344689936395, "learning_rate": 1.1082693577164816e-06, "loss": 0.1462, "step": 11621 }, { "epoch": 1.58, "grad_norm": 4.267924967393479, "learning_rate": 1.107577866220092e-06, "loss": 0.1526, "step": 11622 }, { "epoch": 1.58, "grad_norm": 4.780600325007542, "learning_rate": 1.1068865636448701e-06, "loss": 0.1841, "step": 11623 }, { "epoch": 1.58, "grad_norm": 3.447263380166839, "learning_rate": 1.1061954500243733e-06, "loss": 0.1862, "step": 11624 }, { "epoch": 1.58, "grad_norm": 3.1050071723962356, "learning_rate": 1.1055045253921399e-06, "loss": 0.1627, "step": 11625 }, { "epoch": 1.58, "grad_norm": 3.624757503611801, "learning_rate": 1.1048137897817084e-06, "loss": 0.1971, "step": 11626 }, { "epoch": 1.58, "grad_norm": 3.379900309565241, "learning_rate": 1.104123243226603e-06, "loss": 0.1384, "step": 11627 }, { "epoch": 1.58, "grad_norm": 5.152389871895521, "learning_rate": 1.1034328857603398e-06, "loss": 0.1801, "step": 11628 }, { "epoch": 1.58, "grad_norm": 3.1614565411586497, "learning_rate": 1.1027427174164256e-06, "loss": 0.1497, "step": 11629 }, { "epoch": 1.58, "grad_norm": 2.7116169722603805, "learning_rate": 1.1020527382283585e-06, "loss": 0.1525, "step": 11630 }, { "epoch": 1.58, "grad_norm": 2.983843781059517, "learning_rate": 1.1013629482296268e-06, "loss": 0.1805, "step": 11631 }, { "epoch": 1.58, "grad_norm": 4.301840231007286, "learning_rate": 1.1006733474537095e-06, "loss": 0.1469, "step": 11632 }, { "epoch": 1.58, "grad_norm": 3.2243451982883506, "learning_rate": 1.099983935934077e-06, "loss": 0.1827, "step": 11633 }, { "epoch": 1.58, "grad_norm": 4.839291518043323, "learning_rate": 1.099294713704191e-06, "loss": 0.1677, "step": 11634 }, { "epoch": 1.58, "grad_norm": 3.0885643204861677, "learning_rate": 1.098605680797501e-06, "loss": 0.1641, "step": 11635 }, { "epoch": 1.58, "grad_norm": 3.081183658061783, "learning_rate": 1.0979168372474546e-06, "loss": 0.1596, "step": 11636 }, { "epoch": 1.58, "grad_norm": 3.215952892989924, "learning_rate": 1.0972281830874794e-06, "loss": 0.133, "step": 11637 }, { "epoch": 1.58, "grad_norm": 4.420028669689378, "learning_rate": 1.0965397183510052e-06, "loss": 0.1748, "step": 11638 }, { "epoch": 1.58, "grad_norm": 4.74051191220594, "learning_rate": 1.0958514430714412e-06, "loss": 0.1941, "step": 11639 }, { "epoch": 1.58, "grad_norm": 2.6570050683223245, "learning_rate": 1.095163357282198e-06, "loss": 0.1057, "step": 11640 }, { "epoch": 1.58, "grad_norm": 3.3283687456146067, "learning_rate": 1.0944754610166703e-06, "loss": 0.1691, "step": 11641 }, { "epoch": 1.58, "grad_norm": 3.201884777945128, "learning_rate": 1.0937877543082464e-06, "loss": 0.1504, "step": 11642 }, { "epoch": 1.58, "grad_norm": 3.3648701780471324, "learning_rate": 1.0931002371903038e-06, "loss": 0.1689, "step": 11643 }, { "epoch": 1.58, "grad_norm": 3.227744081500641, "learning_rate": 1.0924129096962122e-06, "loss": 0.1266, "step": 11644 }, { "epoch": 1.58, "grad_norm": 2.8077093779542306, "learning_rate": 1.0917257718593304e-06, "loss": 0.1448, "step": 11645 }, { "epoch": 1.58, "grad_norm": 2.7095101440727283, "learning_rate": 1.0910388237130104e-06, "loss": 0.1298, "step": 11646 }, { "epoch": 1.58, "grad_norm": 2.7885785874760827, "learning_rate": 1.090352065290593e-06, "loss": 0.1466, "step": 11647 }, { "epoch": 1.58, "grad_norm": 3.342397003384032, "learning_rate": 1.0896654966254105e-06, "loss": 0.1072, "step": 11648 }, { "epoch": 1.58, "grad_norm": 3.051917725825546, "learning_rate": 1.0889791177507846e-06, "loss": 0.1531, "step": 11649 }, { "epoch": 1.58, "grad_norm": 3.992413210419185, "learning_rate": 1.0882929287000338e-06, "loss": 0.1566, "step": 11650 }, { "epoch": 1.58, "grad_norm": 3.7124560204918287, "learning_rate": 1.0876069295064556e-06, "loss": 0.1784, "step": 11651 }, { "epoch": 1.58, "grad_norm": 3.113637750679344, "learning_rate": 1.086921120203353e-06, "loss": 0.1448, "step": 11652 }, { "epoch": 1.59, "grad_norm": 3.9695239285181243, "learning_rate": 1.086235500824005e-06, "loss": 0.1931, "step": 11653 }, { "epoch": 1.59, "grad_norm": 13.877488673831646, "learning_rate": 1.0855500714016936e-06, "loss": 0.177, "step": 11654 }, { "epoch": 1.59, "grad_norm": 2.8687483817333974, "learning_rate": 1.0848648319696852e-06, "loss": 0.1644, "step": 11655 }, { "epoch": 1.59, "grad_norm": 3.364311404648304, "learning_rate": 1.0841797825612377e-06, "loss": 0.1468, "step": 11656 }, { "epoch": 1.59, "grad_norm": 4.984198156203792, "learning_rate": 1.0834949232096008e-06, "loss": 0.1783, "step": 11657 }, { "epoch": 1.59, "grad_norm": 3.4876064477769884, "learning_rate": 1.082810253948015e-06, "loss": 0.1638, "step": 11658 }, { "epoch": 1.59, "grad_norm": 2.28623818502867, "learning_rate": 1.0821257748097102e-06, "loss": 0.1333, "step": 11659 }, { "epoch": 1.59, "grad_norm": 3.194830603173622, "learning_rate": 1.0814414858279088e-06, "loss": 0.1706, "step": 11660 }, { "epoch": 1.59, "grad_norm": 4.222076251120252, "learning_rate": 1.080757387035823e-06, "loss": 0.1675, "step": 11661 }, { "epoch": 1.59, "grad_norm": 3.655512988702261, "learning_rate": 1.0800734784666556e-06, "loss": 0.1557, "step": 11662 }, { "epoch": 1.59, "grad_norm": 2.756083099166261, "learning_rate": 1.0793897601536013e-06, "loss": 0.1456, "step": 11663 }, { "epoch": 1.59, "grad_norm": 3.292219704956488, "learning_rate": 1.0787062321298441e-06, "loss": 0.1064, "step": 11664 }, { "epoch": 1.59, "grad_norm": 3.5051323176347546, "learning_rate": 1.0780228944285598e-06, "loss": 0.143, "step": 11665 }, { "epoch": 1.59, "grad_norm": 2.7762984743541907, "learning_rate": 1.0773397470829145e-06, "loss": 0.1275, "step": 11666 }, { "epoch": 1.59, "grad_norm": 2.725900352909047, "learning_rate": 1.076656790126065e-06, "loss": 0.1364, "step": 11667 }, { "epoch": 1.59, "grad_norm": 3.0573363487882763, "learning_rate": 1.0759740235911581e-06, "loss": 0.1564, "step": 11668 }, { "epoch": 1.59, "grad_norm": 3.026750569350093, "learning_rate": 1.0752914475113363e-06, "loss": 0.1345, "step": 11669 }, { "epoch": 1.59, "grad_norm": 3.888439911788861, "learning_rate": 1.074609061919723e-06, "loss": 0.1611, "step": 11670 }, { "epoch": 1.59, "grad_norm": 4.032199728269799, "learning_rate": 1.0739268668494445e-06, "loss": 0.1768, "step": 11671 }, { "epoch": 1.59, "grad_norm": 3.675820492440374, "learning_rate": 1.0732448623336057e-06, "loss": 0.1373, "step": 11672 }, { "epoch": 1.59, "grad_norm": 3.6428778617808604, "learning_rate": 1.0725630484053113e-06, "loss": 0.1426, "step": 11673 }, { "epoch": 1.59, "grad_norm": 3.917884704659286, "learning_rate": 1.0718814250976533e-06, "loss": 0.1516, "step": 11674 }, { "epoch": 1.59, "grad_norm": 4.514220386220672, "learning_rate": 1.071199992443715e-06, "loss": 0.1436, "step": 11675 }, { "epoch": 1.59, "grad_norm": 4.426526087798236, "learning_rate": 1.0705187504765691e-06, "loss": 0.1593, "step": 11676 }, { "epoch": 1.59, "grad_norm": 3.1917669580325914, "learning_rate": 1.0698376992292808e-06, "loss": 0.1791, "step": 11677 }, { "epoch": 1.59, "grad_norm": 3.8876576287469655, "learning_rate": 1.069156838734905e-06, "loss": 0.1806, "step": 11678 }, { "epoch": 1.59, "grad_norm": 3.534360427321941, "learning_rate": 1.0684761690264879e-06, "loss": 0.1431, "step": 11679 }, { "epoch": 1.59, "grad_norm": 2.9069804508944794, "learning_rate": 1.0677956901370662e-06, "loss": 0.1472, "step": 11680 }, { "epoch": 1.59, "grad_norm": 3.627458922515531, "learning_rate": 1.0671154020996677e-06, "loss": 0.1855, "step": 11681 }, { "epoch": 1.59, "grad_norm": 2.9151154448290684, "learning_rate": 1.0664353049473085e-06, "loss": 0.1394, "step": 11682 }, { "epoch": 1.59, "grad_norm": 4.729166594596, "learning_rate": 1.065755398713002e-06, "loss": 0.2139, "step": 11683 }, { "epoch": 1.59, "grad_norm": 4.29767415223861, "learning_rate": 1.0650756834297426e-06, "loss": 0.159, "step": 11684 }, { "epoch": 1.59, "grad_norm": 4.018111727070552, "learning_rate": 1.0643961591305264e-06, "loss": 0.1569, "step": 11685 }, { "epoch": 1.59, "grad_norm": 4.762062974143223, "learning_rate": 1.0637168258483283e-06, "loss": 0.1645, "step": 11686 }, { "epoch": 1.59, "grad_norm": 3.1074881911238195, "learning_rate": 1.0630376836161248e-06, "loss": 0.1417, "step": 11687 }, { "epoch": 1.59, "grad_norm": 4.212962026928392, "learning_rate": 1.0623587324668771e-06, "loss": 0.1966, "step": 11688 }, { "epoch": 1.59, "grad_norm": 3.197007903071443, "learning_rate": 1.0616799724335387e-06, "loss": 0.17, "step": 11689 }, { "epoch": 1.59, "grad_norm": 3.1512672554860246, "learning_rate": 1.0610014035490529e-06, "loss": 0.1628, "step": 11690 }, { "epoch": 1.59, "grad_norm": 5.21945141887966, "learning_rate": 1.0603230258463553e-06, "loss": 0.1759, "step": 11691 }, { "epoch": 1.59, "grad_norm": 3.4339737797490537, "learning_rate": 1.0596448393583709e-06, "loss": 0.1608, "step": 11692 }, { "epoch": 1.59, "grad_norm": 3.0000670215120353, "learning_rate": 1.0589668441180156e-06, "loss": 0.1337, "step": 11693 }, { "epoch": 1.59, "grad_norm": 4.338990674870896, "learning_rate": 1.0582890401581963e-06, "loss": 0.1444, "step": 11694 }, { "epoch": 1.59, "grad_norm": 5.69430906246798, "learning_rate": 1.0576114275118132e-06, "loss": 0.1741, "step": 11695 }, { "epoch": 1.59, "grad_norm": 3.821535907228663, "learning_rate": 1.0569340062117505e-06, "loss": 0.2004, "step": 11696 }, { "epoch": 1.59, "grad_norm": 2.4170120984981343, "learning_rate": 1.0562567762908915e-06, "loss": 0.1339, "step": 11697 }, { "epoch": 1.59, "grad_norm": 5.354821526224748, "learning_rate": 1.0555797377821032e-06, "loss": 0.1764, "step": 11698 }, { "epoch": 1.59, "grad_norm": 3.899015889813059, "learning_rate": 1.0549028907182458e-06, "loss": 0.2069, "step": 11699 }, { "epoch": 1.59, "grad_norm": 3.2092014280017254, "learning_rate": 1.0542262351321719e-06, "loss": 0.1288, "step": 11700 }, { "epoch": 1.59, "grad_norm": 3.291557218780829, "learning_rate": 1.0535497710567215e-06, "loss": 0.1444, "step": 11701 }, { "epoch": 1.59, "grad_norm": 4.2668560037602905, "learning_rate": 1.052873498524732e-06, "loss": 0.1984, "step": 11702 }, { "epoch": 1.59, "grad_norm": 2.8299105336275687, "learning_rate": 1.0521974175690197e-06, "loss": 0.1606, "step": 11703 }, { "epoch": 1.59, "grad_norm": 2.934414533268933, "learning_rate": 1.0515215282224056e-06, "loss": 0.1316, "step": 11704 }, { "epoch": 1.59, "grad_norm": 3.0926496867219617, "learning_rate": 1.0508458305176883e-06, "loss": 0.1499, "step": 11705 }, { "epoch": 1.59, "grad_norm": 3.1158985363417515, "learning_rate": 1.050170324487667e-06, "loss": 0.1602, "step": 11706 }, { "epoch": 1.59, "grad_norm": 4.414817697475844, "learning_rate": 1.0494950101651274e-06, "loss": 0.1529, "step": 11707 }, { "epoch": 1.59, "grad_norm": 3.493118560287074, "learning_rate": 1.0488198875828453e-06, "loss": 0.1594, "step": 11708 }, { "epoch": 1.59, "grad_norm": 3.3427231986654986, "learning_rate": 1.048144956773589e-06, "loss": 0.1492, "step": 11709 }, { "epoch": 1.59, "grad_norm": 2.6189215977253553, "learning_rate": 1.0474702177701162e-06, "loss": 0.1154, "step": 11710 }, { "epoch": 1.59, "grad_norm": 2.5713205731466475, "learning_rate": 1.0467956706051763e-06, "loss": 0.1128, "step": 11711 }, { "epoch": 1.59, "grad_norm": 2.7668843136203463, "learning_rate": 1.046121315311508e-06, "loss": 0.1441, "step": 11712 }, { "epoch": 1.59, "grad_norm": 2.9677592502013788, "learning_rate": 1.0454471519218406e-06, "loss": 0.1736, "step": 11713 }, { "epoch": 1.59, "grad_norm": 3.128581480796159, "learning_rate": 1.0447731804688993e-06, "loss": 0.1505, "step": 11714 }, { "epoch": 1.59, "grad_norm": 4.545775019614531, "learning_rate": 1.0440994009853905e-06, "loss": 0.1913, "step": 11715 }, { "epoch": 1.59, "grad_norm": 2.721023068660727, "learning_rate": 1.0434258135040214e-06, "loss": 0.1092, "step": 11716 }, { "epoch": 1.59, "grad_norm": 3.187440569315027, "learning_rate": 1.04275241805748e-06, "loss": 0.1669, "step": 11717 }, { "epoch": 1.59, "grad_norm": 3.113906205584259, "learning_rate": 1.0420792146784553e-06, "loss": 0.1491, "step": 11718 }, { "epoch": 1.59, "grad_norm": 3.0682583645137202, "learning_rate": 1.0414062033996158e-06, "loss": 0.1396, "step": 11719 }, { "epoch": 1.59, "grad_norm": 3.463516015658642, "learning_rate": 1.0407333842536315e-06, "loss": 0.1771, "step": 11720 }, { "epoch": 1.59, "grad_norm": 3.170990474677365, "learning_rate": 1.040060757273156e-06, "loss": 0.1664, "step": 11721 }, { "epoch": 1.59, "grad_norm": 2.619176669706518, "learning_rate": 1.0393883224908358e-06, "loss": 0.1262, "step": 11722 }, { "epoch": 1.59, "grad_norm": 4.3107287111301815, "learning_rate": 1.0387160799393075e-06, "loss": 0.2379, "step": 11723 }, { "epoch": 1.59, "grad_norm": 3.498065699685379, "learning_rate": 1.0380440296512002e-06, "loss": 0.1438, "step": 11724 }, { "epoch": 1.59, "grad_norm": 3.1137056921305004, "learning_rate": 1.0373721716591311e-06, "loss": 0.1523, "step": 11725 }, { "epoch": 1.6, "grad_norm": 3.4239161270298335, "learning_rate": 1.0367005059957097e-06, "loss": 0.1777, "step": 11726 }, { "epoch": 1.6, "grad_norm": 3.23232259119021, "learning_rate": 1.036029032693534e-06, "loss": 0.1263, "step": 11727 }, { "epoch": 1.6, "grad_norm": 3.2611821697292487, "learning_rate": 1.0353577517851987e-06, "loss": 0.1452, "step": 11728 }, { "epoch": 1.6, "grad_norm": 5.053881361047564, "learning_rate": 1.0346866633032793e-06, "loss": 0.1361, "step": 11729 }, { "epoch": 1.6, "grad_norm": 2.2704573022381385, "learning_rate": 1.034015767280353e-06, "loss": 0.1238, "step": 11730 }, { "epoch": 1.6, "grad_norm": 3.7806756795048333, "learning_rate": 1.0333450637489772e-06, "loss": 0.1468, "step": 11731 }, { "epoch": 1.6, "grad_norm": 3.817666870881421, "learning_rate": 1.0326745527417098e-06, "loss": 0.2049, "step": 11732 }, { "epoch": 1.6, "grad_norm": 2.602470629370998, "learning_rate": 1.0320042342910892e-06, "loss": 0.1365, "step": 11733 }, { "epoch": 1.6, "grad_norm": 3.1942590147979617, "learning_rate": 1.0313341084296535e-06, "loss": 0.1024, "step": 11734 }, { "epoch": 1.6, "grad_norm": 3.184541891023217, "learning_rate": 1.0306641751899265e-06, "loss": 0.1493, "step": 11735 }, { "epoch": 1.6, "grad_norm": 2.519805735535672, "learning_rate": 1.0299944346044243e-06, "loss": 0.1099, "step": 11736 }, { "epoch": 1.6, "grad_norm": 2.8272447507339185, "learning_rate": 1.0293248867056527e-06, "loss": 0.1279, "step": 11737 }, { "epoch": 1.6, "grad_norm": 3.843110843813212, "learning_rate": 1.0286555315261088e-06, "loss": 0.1593, "step": 11738 }, { "epoch": 1.6, "grad_norm": 3.0003168497476893, "learning_rate": 1.02798636909828e-06, "loss": 0.1448, "step": 11739 }, { "epoch": 1.6, "grad_norm": 3.595743585789739, "learning_rate": 1.0273173994546447e-06, "loss": 0.1563, "step": 11740 }, { "epoch": 1.6, "grad_norm": 2.638696651175242, "learning_rate": 1.0266486226276718e-06, "loss": 0.1254, "step": 11741 }, { "epoch": 1.6, "grad_norm": 4.379814291129417, "learning_rate": 1.0259800386498204e-06, "loss": 0.143, "step": 11742 }, { "epoch": 1.6, "grad_norm": 3.901644874701981, "learning_rate": 1.0253116475535413e-06, "loss": 0.1857, "step": 11743 }, { "epoch": 1.6, "grad_norm": 3.6594812290269725, "learning_rate": 1.0246434493712747e-06, "loss": 0.1685, "step": 11744 }, { "epoch": 1.6, "grad_norm": 4.8546221252422335, "learning_rate": 1.0239754441354528e-06, "loss": 0.1346, "step": 11745 }, { "epoch": 1.6, "grad_norm": 3.490060614388403, "learning_rate": 1.0233076318784952e-06, "loss": 0.132, "step": 11746 }, { "epoch": 1.6, "grad_norm": 4.012393537795301, "learning_rate": 1.022640012632819e-06, "loss": 0.1377, "step": 11747 }, { "epoch": 1.6, "grad_norm": 2.645694554481313, "learning_rate": 1.0219725864308228e-06, "loss": 0.1272, "step": 11748 }, { "epoch": 1.6, "grad_norm": 3.333576405557465, "learning_rate": 1.0213053533049055e-06, "loss": 0.1835, "step": 11749 }, { "epoch": 1.6, "grad_norm": 2.7599945849771044, "learning_rate": 1.020638313287446e-06, "loss": 0.1559, "step": 11750 }, { "epoch": 1.6, "grad_norm": 3.669984246066098, "learning_rate": 1.0199714664108252e-06, "loss": 0.1387, "step": 11751 }, { "epoch": 1.6, "grad_norm": 3.9971530350943194, "learning_rate": 1.0193048127074034e-06, "loss": 0.1985, "step": 11752 }, { "epoch": 1.6, "grad_norm": 2.962674035520401, "learning_rate": 1.0186383522095416e-06, "loss": 0.1209, "step": 11753 }, { "epoch": 1.6, "grad_norm": 4.107067189887731, "learning_rate": 1.017972084949585e-06, "loss": 0.1755, "step": 11754 }, { "epoch": 1.6, "grad_norm": 2.9952785587199915, "learning_rate": 1.0173060109598715e-06, "loss": 0.1403, "step": 11755 }, { "epoch": 1.6, "grad_norm": 4.7209111225807785, "learning_rate": 1.0166401302727296e-06, "loss": 0.1758, "step": 11756 }, { "epoch": 1.6, "grad_norm": 3.7443378897590724, "learning_rate": 1.0159744429204776e-06, "loss": 0.1446, "step": 11757 }, { "epoch": 1.6, "grad_norm": 3.0103937058822203, "learning_rate": 1.0153089489354256e-06, "loss": 0.0947, "step": 11758 }, { "epoch": 1.6, "grad_norm": 3.8048959297623837, "learning_rate": 1.0146436483498735e-06, "loss": 0.1587, "step": 11759 }, { "epoch": 1.6, "grad_norm": 3.5178202403400425, "learning_rate": 1.0139785411961111e-06, "loss": 0.1299, "step": 11760 }, { "epoch": 1.6, "grad_norm": 3.7777930737160315, "learning_rate": 1.0133136275064237e-06, "loss": 0.1653, "step": 11761 }, { "epoch": 1.6, "grad_norm": 3.087898880663911, "learning_rate": 1.0126489073130779e-06, "loss": 0.1582, "step": 11762 }, { "epoch": 1.6, "grad_norm": 3.2450776438656352, "learning_rate": 1.0119843806483416e-06, "loss": 0.1681, "step": 11763 }, { "epoch": 1.6, "grad_norm": 3.549393736712011, "learning_rate": 1.0113200475444623e-06, "loss": 0.1606, "step": 11764 }, { "epoch": 1.6, "grad_norm": 2.9545496768887456, "learning_rate": 1.010655908033688e-06, "loss": 0.148, "step": 11765 }, { "epoch": 1.6, "grad_norm": 3.4375007038391545, "learning_rate": 1.009991962148253e-06, "loss": 0.1501, "step": 11766 }, { "epoch": 1.6, "grad_norm": 2.9520768540714837, "learning_rate": 1.0093282099203805e-06, "loss": 0.1646, "step": 11767 }, { "epoch": 1.6, "grad_norm": 4.012865518332819, "learning_rate": 1.0086646513822873e-06, "loss": 0.1576, "step": 11768 }, { "epoch": 1.6, "grad_norm": 4.669811381525851, "learning_rate": 1.0080012865661798e-06, "loss": 0.1795, "step": 11769 }, { "epoch": 1.6, "grad_norm": 3.5644349358308376, "learning_rate": 1.007338115504254e-06, "loss": 0.1531, "step": 11770 }, { "epoch": 1.6, "grad_norm": 3.1611592769040797, "learning_rate": 1.0066751382286977e-06, "loss": 0.1449, "step": 11771 }, { "epoch": 1.6, "grad_norm": 2.8594602114611676, "learning_rate": 1.0060123547716888e-06, "loss": 0.1402, "step": 11772 }, { "epoch": 1.6, "grad_norm": 2.9220980190330925, "learning_rate": 1.005349765165396e-06, "loss": 0.1257, "step": 11773 }, { "epoch": 1.6, "grad_norm": 5.503162667679869, "learning_rate": 1.0046873694419773e-06, "loss": 0.1347, "step": 11774 }, { "epoch": 1.6, "grad_norm": 3.719461653814907, "learning_rate": 1.0040251676335861e-06, "loss": 0.1702, "step": 11775 }, { "epoch": 1.6, "grad_norm": 3.467833598694447, "learning_rate": 1.0033631597723582e-06, "loss": 0.1337, "step": 11776 }, { "epoch": 1.6, "grad_norm": 4.388174205451656, "learning_rate": 1.0027013458904288e-06, "loss": 0.1432, "step": 11777 }, { "epoch": 1.6, "grad_norm": 3.949744526142175, "learning_rate": 1.0020397260199167e-06, "loss": 0.1547, "step": 11778 }, { "epoch": 1.6, "grad_norm": 3.7907813982711347, "learning_rate": 1.0013783001929333e-06, "loss": 0.1563, "step": 11779 }, { "epoch": 1.6, "grad_norm": 3.1879491685938923, "learning_rate": 1.0007170684415857e-06, "loss": 0.1599, "step": 11780 }, { "epoch": 1.6, "grad_norm": 3.4060189508463914, "learning_rate": 1.0000560307979613e-06, "loss": 0.1355, "step": 11781 }, { "epoch": 1.6, "grad_norm": 2.73131236175723, "learning_rate": 9.993951872941493e-07, "loss": 0.1286, "step": 11782 }, { "epoch": 1.6, "grad_norm": 3.894489155614743, "learning_rate": 9.987345379622198e-07, "loss": 0.1569, "step": 11783 }, { "epoch": 1.6, "grad_norm": 4.652073234526315, "learning_rate": 9.98074082834241e-07, "loss": 0.1653, "step": 11784 }, { "epoch": 1.6, "grad_norm": 3.6073811350358294, "learning_rate": 9.974138219422674e-07, "loss": 0.1663, "step": 11785 }, { "epoch": 1.6, "grad_norm": 3.4258208226668887, "learning_rate": 9.967537553183448e-07, "loss": 0.1348, "step": 11786 }, { "epoch": 1.6, "grad_norm": 4.300445113316875, "learning_rate": 9.960938829945104e-07, "loss": 0.1466, "step": 11787 }, { "epoch": 1.6, "grad_norm": 4.075673646843625, "learning_rate": 9.954342050027922e-07, "loss": 0.1714, "step": 11788 }, { "epoch": 1.6, "grad_norm": 2.733029645514497, "learning_rate": 9.947747213752062e-07, "loss": 0.1225, "step": 11789 }, { "epoch": 1.6, "grad_norm": 4.048289160658564, "learning_rate": 9.941154321437629e-07, "loss": 0.1469, "step": 11790 }, { "epoch": 1.6, "grad_norm": 3.483157853267603, "learning_rate": 9.934563373404598e-07, "loss": 0.1436, "step": 11791 }, { "epoch": 1.6, "grad_norm": 4.246609563344737, "learning_rate": 9.927974369972871e-07, "loss": 0.1901, "step": 11792 }, { "epoch": 1.6, "grad_norm": 4.050259374373742, "learning_rate": 9.92138731146224e-07, "loss": 0.1066, "step": 11793 }, { "epoch": 1.6, "grad_norm": 3.5538476274177015, "learning_rate": 9.914802198192442e-07, "loss": 0.1648, "step": 11794 }, { "epoch": 1.6, "grad_norm": 3.1288268015950016, "learning_rate": 9.908219030483046e-07, "loss": 0.1193, "step": 11795 }, { "epoch": 1.6, "grad_norm": 3.5078974793983413, "learning_rate": 9.901637808653619e-07, "loss": 0.1585, "step": 11796 }, { "epoch": 1.6, "grad_norm": 3.107000672027842, "learning_rate": 9.895058533023532e-07, "loss": 0.1342, "step": 11797 }, { "epoch": 1.6, "grad_norm": 3.8381432918834992, "learning_rate": 9.888481203912153e-07, "loss": 0.195, "step": 11798 }, { "epoch": 1.6, "grad_norm": 3.096664818397264, "learning_rate": 9.88190582163871e-07, "loss": 0.1574, "step": 11799 }, { "epoch": 1.61, "grad_norm": 3.0034438622076074, "learning_rate": 9.875332386522335e-07, "loss": 0.1858, "step": 11800 }, { "epoch": 1.61, "grad_norm": 3.5261973826983564, "learning_rate": 9.868760898882074e-07, "loss": 0.1728, "step": 11801 }, { "epoch": 1.61, "grad_norm": 3.560305206630835, "learning_rate": 9.862191359036883e-07, "loss": 0.1552, "step": 11802 }, { "epoch": 1.61, "grad_norm": 3.6205972670146314, "learning_rate": 9.855623767305621e-07, "loss": 0.1386, "step": 11803 }, { "epoch": 1.61, "grad_norm": 3.2361333264942993, "learning_rate": 9.849058124007044e-07, "loss": 0.1661, "step": 11804 }, { "epoch": 1.61, "grad_norm": 3.626816122867731, "learning_rate": 9.84249442945982e-07, "loss": 0.1572, "step": 11805 }, { "epoch": 1.61, "grad_norm": 3.18985276857649, "learning_rate": 9.835932683982524e-07, "loss": 0.1506, "step": 11806 }, { "epoch": 1.61, "grad_norm": 3.365374835854195, "learning_rate": 9.829372887893624e-07, "loss": 0.2062, "step": 11807 }, { "epoch": 1.61, "grad_norm": 3.7934312456743227, "learning_rate": 9.82281504151154e-07, "loss": 0.1485, "step": 11808 }, { "epoch": 1.61, "grad_norm": 2.5839692865007313, "learning_rate": 9.816259145154511e-07, "loss": 0.1464, "step": 11809 }, { "epoch": 1.61, "grad_norm": 3.074968727306342, "learning_rate": 9.809705199140778e-07, "loss": 0.1885, "step": 11810 }, { "epoch": 1.61, "grad_norm": 2.635803922675266, "learning_rate": 9.803153203788396e-07, "loss": 0.1274, "step": 11811 }, { "epoch": 1.61, "grad_norm": 3.3423962000707985, "learning_rate": 9.796603159415407e-07, "loss": 0.1718, "step": 11812 }, { "epoch": 1.61, "grad_norm": 3.340435145056753, "learning_rate": 9.79005506633971e-07, "loss": 0.1659, "step": 11813 }, { "epoch": 1.61, "grad_norm": 2.7228396442338814, "learning_rate": 9.783508924879115e-07, "loss": 0.1452, "step": 11814 }, { "epoch": 1.61, "grad_norm": 3.8365324410834396, "learning_rate": 9.776964735351347e-07, "loss": 0.1735, "step": 11815 }, { "epoch": 1.61, "grad_norm": 3.746643807402629, "learning_rate": 9.770422498074034e-07, "loss": 0.1942, "step": 11816 }, { "epoch": 1.61, "grad_norm": 4.221138527144054, "learning_rate": 9.763882213364705e-07, "loss": 0.1614, "step": 11817 }, { "epoch": 1.61, "grad_norm": 4.0229962644657355, "learning_rate": 9.757343881540798e-07, "loss": 0.1725, "step": 11818 }, { "epoch": 1.61, "grad_norm": 4.3976526143237935, "learning_rate": 9.75080750291965e-07, "loss": 0.2184, "step": 11819 }, { "epoch": 1.61, "grad_norm": 3.9258654580198016, "learning_rate": 9.744273077818523e-07, "loss": 0.1542, "step": 11820 }, { "epoch": 1.61, "grad_norm": 3.4703855363489846, "learning_rate": 9.737740606554553e-07, "loss": 0.1528, "step": 11821 }, { "epoch": 1.61, "grad_norm": 3.2729123539209546, "learning_rate": 9.731210089444803e-07, "loss": 0.1934, "step": 11822 }, { "epoch": 1.61, "grad_norm": 3.4319167681354603, "learning_rate": 9.72468152680624e-07, "loss": 0.1411, "step": 11823 }, { "epoch": 1.61, "grad_norm": 3.199018427234135, "learning_rate": 9.71815491895573e-07, "loss": 0.1315, "step": 11824 }, { "epoch": 1.61, "grad_norm": 3.296501205463607, "learning_rate": 9.711630266210042e-07, "loss": 0.1218, "step": 11825 }, { "epoch": 1.61, "grad_norm": 2.325277578998439, "learning_rate": 9.70510756888584e-07, "loss": 0.0814, "step": 11826 }, { "epoch": 1.61, "grad_norm": 4.067192786216612, "learning_rate": 9.69858682729976e-07, "loss": 0.17, "step": 11827 }, { "epoch": 1.61, "grad_norm": 3.462721594361021, "learning_rate": 9.69206804176822e-07, "loss": 0.1375, "step": 11828 }, { "epoch": 1.61, "grad_norm": 3.4500078723448624, "learning_rate": 9.68555121260768e-07, "loss": 0.1168, "step": 11829 }, { "epoch": 1.61, "grad_norm": 4.992779831395855, "learning_rate": 9.679036340134379e-07, "loss": 0.1428, "step": 11830 }, { "epoch": 1.61, "grad_norm": 3.4655503519951867, "learning_rate": 9.67252342466456e-07, "loss": 0.1268, "step": 11831 }, { "epoch": 1.61, "grad_norm": 2.9955199737979776, "learning_rate": 9.66601246651432e-07, "loss": 0.1485, "step": 11832 }, { "epoch": 1.61, "grad_norm": 2.737337100944285, "learning_rate": 9.65950346599967e-07, "loss": 0.1554, "step": 11833 }, { "epoch": 1.61, "grad_norm": 4.047408045976453, "learning_rate": 9.652996423436528e-07, "loss": 0.1463, "step": 11834 }, { "epoch": 1.61, "grad_norm": 3.6490173467905396, "learning_rate": 9.646491339140723e-07, "loss": 0.1954, "step": 11835 }, { "epoch": 1.61, "grad_norm": 4.70166883351455, "learning_rate": 9.639988213427981e-07, "loss": 0.1551, "step": 11836 }, { "epoch": 1.61, "grad_norm": 2.9691280340655144, "learning_rate": 9.633487046613932e-07, "loss": 0.1224, "step": 11837 }, { "epoch": 1.61, "grad_norm": 3.494469035588966, "learning_rate": 9.626987839014119e-07, "loss": 0.1298, "step": 11838 }, { "epoch": 1.61, "grad_norm": 3.65047911285531, "learning_rate": 9.620490590943983e-07, "loss": 0.1626, "step": 11839 }, { "epoch": 1.61, "grad_norm": 4.014697655723393, "learning_rate": 9.613995302718853e-07, "loss": 0.1797, "step": 11840 }, { "epoch": 1.61, "grad_norm": 3.4728582738354072, "learning_rate": 9.607501974654031e-07, "loss": 0.1606, "step": 11841 }, { "epoch": 1.61, "grad_norm": 3.268384097198977, "learning_rate": 9.60101060706462e-07, "loss": 0.1283, "step": 11842 }, { "epoch": 1.61, "grad_norm": 3.972031988562719, "learning_rate": 9.594521200265734e-07, "loss": 0.1413, "step": 11843 }, { "epoch": 1.61, "grad_norm": 4.267259713684057, "learning_rate": 9.588033754572285e-07, "loss": 0.1258, "step": 11844 }, { "epoch": 1.61, "grad_norm": 3.0813155408749036, "learning_rate": 9.58154827029919e-07, "loss": 0.1463, "step": 11845 }, { "epoch": 1.61, "grad_norm": 4.60798592452901, "learning_rate": 9.575064747761203e-07, "loss": 0.1609, "step": 11846 }, { "epoch": 1.61, "grad_norm": 3.2046237813798286, "learning_rate": 9.568583187273018e-07, "loss": 0.1495, "step": 11847 }, { "epoch": 1.61, "grad_norm": 3.244759855629351, "learning_rate": 9.56210358914922e-07, "loss": 0.146, "step": 11848 }, { "epoch": 1.61, "grad_norm": 3.505392354810609, "learning_rate": 9.555625953704288e-07, "loss": 0.1428, "step": 11849 }, { "epoch": 1.61, "grad_norm": 3.365210483266655, "learning_rate": 9.549150281252633e-07, "loss": 0.1619, "step": 11850 }, { "epoch": 1.61, "grad_norm": 2.6156965199354514, "learning_rate": 9.54267657210855e-07, "loss": 0.1527, "step": 11851 }, { "epoch": 1.61, "grad_norm": 3.4436235077291353, "learning_rate": 9.536204826586243e-07, "loss": 0.1516, "step": 11852 }, { "epoch": 1.61, "grad_norm": 3.7082613867671377, "learning_rate": 9.529735044999822e-07, "loss": 0.1446, "step": 11853 }, { "epoch": 1.61, "grad_norm": 4.941068415405052, "learning_rate": 9.52326722766329e-07, "loss": 0.144, "step": 11854 }, { "epoch": 1.61, "grad_norm": 3.6313578406079703, "learning_rate": 9.51680137489061e-07, "loss": 0.1544, "step": 11855 }, { "epoch": 1.61, "grad_norm": 3.1798038975681613, "learning_rate": 9.510337486995547e-07, "loss": 0.1759, "step": 11856 }, { "epoch": 1.61, "grad_norm": 3.4706249896463968, "learning_rate": 9.503875564291886e-07, "loss": 0.1464, "step": 11857 }, { "epoch": 1.61, "grad_norm": 3.3197883670986195, "learning_rate": 9.497415607093208e-07, "loss": 0.1703, "step": 11858 }, { "epoch": 1.61, "grad_norm": 2.9167616538403895, "learning_rate": 9.490957615713092e-07, "loss": 0.1462, "step": 11859 }, { "epoch": 1.61, "grad_norm": 3.5694966006759548, "learning_rate": 9.484501590464978e-07, "loss": 0.1471, "step": 11860 }, { "epoch": 1.61, "grad_norm": 3.1536230788888107, "learning_rate": 9.478047531662177e-07, "loss": 0.1716, "step": 11861 }, { "epoch": 1.61, "grad_norm": 3.3548514577376385, "learning_rate": 9.471595439617986e-07, "loss": 0.1613, "step": 11862 }, { "epoch": 1.61, "grad_norm": 3.470034509360167, "learning_rate": 9.465145314645519e-07, "loss": 0.1501, "step": 11863 }, { "epoch": 1.61, "grad_norm": 3.634328404785838, "learning_rate": 9.458697157057867e-07, "loss": 0.1134, "step": 11864 }, { "epoch": 1.61, "grad_norm": 3.396274282522174, "learning_rate": 9.452250967167986e-07, "loss": 0.1703, "step": 11865 }, { "epoch": 1.61, "grad_norm": 3.533194536052022, "learning_rate": 9.445806745288749e-07, "loss": 0.175, "step": 11866 }, { "epoch": 1.61, "grad_norm": 3.908095111410309, "learning_rate": 9.439364491732927e-07, "loss": 0.1753, "step": 11867 }, { "epoch": 1.61, "grad_norm": 3.239430409677717, "learning_rate": 9.432924206813199e-07, "loss": 0.1777, "step": 11868 }, { "epoch": 1.61, "grad_norm": 3.2824538162159165, "learning_rate": 9.426485890842151e-07, "loss": 0.1292, "step": 11869 }, { "epoch": 1.61, "grad_norm": 3.897493304268044, "learning_rate": 9.420049544132265e-07, "loss": 0.1282, "step": 11870 }, { "epoch": 1.61, "grad_norm": 3.6698007303810174, "learning_rate": 9.413615166995938e-07, "loss": 0.1561, "step": 11871 }, { "epoch": 1.61, "grad_norm": 4.330735055202941, "learning_rate": 9.407182759745464e-07, "loss": 0.1076, "step": 11872 }, { "epoch": 1.62, "grad_norm": 2.983009715900874, "learning_rate": 9.400752322693035e-07, "loss": 0.1052, "step": 11873 }, { "epoch": 1.62, "grad_norm": 3.4020970154844714, "learning_rate": 9.394323856150794e-07, "loss": 0.1568, "step": 11874 }, { "epoch": 1.62, "grad_norm": 3.36956460279026, "learning_rate": 9.387897360430698e-07, "loss": 0.1381, "step": 11875 }, { "epoch": 1.62, "grad_norm": 3.3575224730297033, "learning_rate": 9.381472835844707e-07, "loss": 0.1291, "step": 11876 }, { "epoch": 1.62, "grad_norm": 3.036288662397877, "learning_rate": 9.375050282704596e-07, "loss": 0.1868, "step": 11877 }, { "epoch": 1.62, "grad_norm": 3.848527604410763, "learning_rate": 9.368629701322124e-07, "loss": 0.1746, "step": 11878 }, { "epoch": 1.62, "grad_norm": 3.079675690142244, "learning_rate": 9.362211092008899e-07, "loss": 0.158, "step": 11879 }, { "epoch": 1.62, "grad_norm": 2.474954360043269, "learning_rate": 9.355794455076461e-07, "loss": 0.0967, "step": 11880 }, { "epoch": 1.62, "grad_norm": 3.967460312707011, "learning_rate": 9.349379790836243e-07, "loss": 0.1446, "step": 11881 }, { "epoch": 1.62, "grad_norm": 3.6746873390038903, "learning_rate": 9.342967099599587e-07, "loss": 0.1603, "step": 11882 }, { "epoch": 1.62, "grad_norm": 4.251039958393053, "learning_rate": 9.33655638167773e-07, "loss": 0.1809, "step": 11883 }, { "epoch": 1.62, "grad_norm": 2.7676865923995515, "learning_rate": 9.330147637381825e-07, "loss": 0.1286, "step": 11884 }, { "epoch": 1.62, "grad_norm": 2.5459847462119254, "learning_rate": 9.323740867022924e-07, "loss": 0.1382, "step": 11885 }, { "epoch": 1.62, "grad_norm": 3.5579486870674533, "learning_rate": 9.317336070911981e-07, "loss": 0.1655, "step": 11886 }, { "epoch": 1.62, "grad_norm": 3.2707669297947284, "learning_rate": 9.31093324935985e-07, "loss": 0.1442, "step": 11887 }, { "epoch": 1.62, "grad_norm": 2.8831277698243314, "learning_rate": 9.30453240267733e-07, "loss": 0.1298, "step": 11888 }, { "epoch": 1.62, "grad_norm": 3.109729935241201, "learning_rate": 9.298133531175035e-07, "loss": 0.1368, "step": 11889 }, { "epoch": 1.62, "grad_norm": 3.9713085496920395, "learning_rate": 9.291736635163601e-07, "loss": 0.18, "step": 11890 }, { "epoch": 1.62, "grad_norm": 3.6534183115693617, "learning_rate": 9.285341714953444e-07, "loss": 0.1772, "step": 11891 }, { "epoch": 1.62, "grad_norm": 3.1332139936086008, "learning_rate": 9.278948770854984e-07, "loss": 0.1703, "step": 11892 }, { "epoch": 1.62, "grad_norm": 4.232679380934516, "learning_rate": 9.272557803178501e-07, "loss": 0.1502, "step": 11893 }, { "epoch": 1.62, "grad_norm": 4.4463008515600455, "learning_rate": 9.266168812234183e-07, "loss": 0.1648, "step": 11894 }, { "epoch": 1.62, "grad_norm": 2.610500065694343, "learning_rate": 9.259781798332119e-07, "loss": 0.1421, "step": 11895 }, { "epoch": 1.62, "grad_norm": 3.065647472117331, "learning_rate": 9.253396761782308e-07, "loss": 0.1222, "step": 11896 }, { "epoch": 1.62, "grad_norm": 3.0918647626400224, "learning_rate": 9.247013702894653e-07, "loss": 0.1671, "step": 11897 }, { "epoch": 1.62, "grad_norm": 3.7533157443329923, "learning_rate": 9.240632621978957e-07, "loss": 0.1487, "step": 11898 }, { "epoch": 1.62, "grad_norm": 2.8159115666654495, "learning_rate": 9.234253519344938e-07, "loss": 0.1437, "step": 11899 }, { "epoch": 1.62, "grad_norm": 3.4436085947756907, "learning_rate": 9.227876395302204e-07, "loss": 0.1489, "step": 11900 }, { "epoch": 1.62, "grad_norm": 3.5296906658569496, "learning_rate": 9.22150125016027e-07, "loss": 0.1231, "step": 11901 }, { "epoch": 1.62, "grad_norm": 3.370217355680508, "learning_rate": 9.215128084228564e-07, "loss": 0.1939, "step": 11902 }, { "epoch": 1.62, "grad_norm": 4.137649662591839, "learning_rate": 9.208756897816402e-07, "loss": 0.1617, "step": 11903 }, { "epoch": 1.62, "grad_norm": 4.613049649840623, "learning_rate": 9.202387691233028e-07, "loss": 0.1717, "step": 11904 }, { "epoch": 1.62, "grad_norm": 3.930267058390605, "learning_rate": 9.196020464787564e-07, "loss": 0.1418, "step": 11905 }, { "epoch": 1.62, "grad_norm": 3.5993897674087925, "learning_rate": 9.189655218789034e-07, "loss": 0.1473, "step": 11906 }, { "epoch": 1.62, "grad_norm": 3.5549267083520455, "learning_rate": 9.183291953546425e-07, "loss": 0.1616, "step": 11907 }, { "epoch": 1.62, "grad_norm": 2.5744764675810905, "learning_rate": 9.176930669368528e-07, "loss": 0.124, "step": 11908 }, { "epoch": 1.62, "grad_norm": 4.0660389353201065, "learning_rate": 9.170571366564146e-07, "loss": 0.1587, "step": 11909 }, { "epoch": 1.62, "grad_norm": 4.52982578204771, "learning_rate": 9.164214045441872e-07, "loss": 0.2182, "step": 11910 }, { "epoch": 1.62, "grad_norm": 4.492488960176361, "learning_rate": 9.157858706310307e-07, "loss": 0.1208, "step": 11911 }, { "epoch": 1.62, "grad_norm": 5.573856977807578, "learning_rate": 9.151505349477901e-07, "loss": 0.1348, "step": 11912 }, { "epoch": 1.62, "grad_norm": 3.4525851588964254, "learning_rate": 9.145153975253013e-07, "loss": 0.172, "step": 11913 }, { "epoch": 1.62, "grad_norm": 4.350075392127486, "learning_rate": 9.138804583943917e-07, "loss": 0.1188, "step": 11914 }, { "epoch": 1.62, "grad_norm": 3.1750061935966323, "learning_rate": 9.132457175858773e-07, "loss": 0.1861, "step": 11915 }, { "epoch": 1.62, "grad_norm": 3.473146514231383, "learning_rate": 9.126111751305672e-07, "loss": 0.1334, "step": 11916 }, { "epoch": 1.62, "grad_norm": 2.8236987575685943, "learning_rate": 9.11976831059258e-07, "loss": 0.1496, "step": 11917 }, { "epoch": 1.62, "grad_norm": 3.601161399874635, "learning_rate": 9.113426854027385e-07, "loss": 0.1403, "step": 11918 }, { "epoch": 1.62, "grad_norm": 3.8758118823939545, "learning_rate": 9.107087381917878e-07, "loss": 0.1765, "step": 11919 }, { "epoch": 1.62, "grad_norm": 4.033217832784614, "learning_rate": 9.100749894571731e-07, "loss": 0.2021, "step": 11920 }, { "epoch": 1.62, "grad_norm": 3.048727710847809, "learning_rate": 9.094414392296574e-07, "loss": 0.1326, "step": 11921 }, { "epoch": 1.62, "grad_norm": 7.062315145785548, "learning_rate": 9.088080875399862e-07, "loss": 0.1412, "step": 11922 }, { "epoch": 1.62, "grad_norm": 3.3696737247125204, "learning_rate": 9.081749344189039e-07, "loss": 0.1214, "step": 11923 }, { "epoch": 1.62, "grad_norm": 4.277312497627542, "learning_rate": 9.075419798971369e-07, "loss": 0.1445, "step": 11924 }, { "epoch": 1.62, "grad_norm": 3.201234996723932, "learning_rate": 9.069092240054084e-07, "loss": 0.1665, "step": 11925 }, { "epoch": 1.62, "grad_norm": 4.548769113255712, "learning_rate": 9.062766667744294e-07, "loss": 0.1616, "step": 11926 }, { "epoch": 1.62, "grad_norm": 2.5875183548104728, "learning_rate": 9.056443082349015e-07, "loss": 0.1507, "step": 11927 }, { "epoch": 1.62, "grad_norm": 3.2997454878786305, "learning_rate": 9.050121484175161e-07, "loss": 0.1731, "step": 11928 }, { "epoch": 1.62, "grad_norm": 5.857725642764755, "learning_rate": 9.043801873529557e-07, "loss": 0.1346, "step": 11929 }, { "epoch": 1.62, "grad_norm": 4.7753592694992175, "learning_rate": 9.037484250718931e-07, "loss": 0.1729, "step": 11930 }, { "epoch": 1.62, "grad_norm": 3.5032424933365047, "learning_rate": 9.031168616049912e-07, "loss": 0.1129, "step": 11931 }, { "epoch": 1.62, "grad_norm": 3.2993253813009167, "learning_rate": 9.024854969829016e-07, "loss": 0.1666, "step": 11932 }, { "epoch": 1.62, "grad_norm": 3.2813293955745806, "learning_rate": 9.018543312362727e-07, "loss": 0.152, "step": 11933 }, { "epoch": 1.62, "grad_norm": 2.9909166282922914, "learning_rate": 9.012233643957324e-07, "loss": 0.1601, "step": 11934 }, { "epoch": 1.62, "grad_norm": 3.200609623111479, "learning_rate": 9.00592596491911e-07, "loss": 0.1605, "step": 11935 }, { "epoch": 1.62, "grad_norm": 3.6418146524074007, "learning_rate": 8.999620275554177e-07, "loss": 0.1443, "step": 11936 }, { "epoch": 1.62, "grad_norm": 2.909702967934381, "learning_rate": 8.993316576168626e-07, "loss": 0.1293, "step": 11937 }, { "epoch": 1.62, "grad_norm": 3.502324140834685, "learning_rate": 8.987014867068361e-07, "loss": 0.1362, "step": 11938 }, { "epoch": 1.62, "grad_norm": 3.670564103633884, "learning_rate": 8.980715148559282e-07, "loss": 0.1067, "step": 11939 }, { "epoch": 1.62, "grad_norm": 2.9700701662963933, "learning_rate": 8.974417420947146e-07, "loss": 0.1498, "step": 11940 }, { "epoch": 1.62, "grad_norm": 3.4614804979125227, "learning_rate": 8.968121684537578e-07, "loss": 0.1598, "step": 11941 }, { "epoch": 1.62, "grad_norm": 3.1805834131151514, "learning_rate": 8.961827939636198e-07, "loss": 0.1511, "step": 11942 }, { "epoch": 1.62, "grad_norm": 3.810101107698176, "learning_rate": 8.955536186548425e-07, "loss": 0.1494, "step": 11943 }, { "epoch": 1.62, "grad_norm": 3.529894161010929, "learning_rate": 8.949246425579672e-07, "loss": 0.1147, "step": 11944 }, { "epoch": 1.62, "grad_norm": 4.349435597499668, "learning_rate": 8.942958657035206e-07, "loss": 0.1716, "step": 11945 }, { "epoch": 1.62, "grad_norm": 2.5090652602562504, "learning_rate": 8.936672881220204e-07, "loss": 0.0909, "step": 11946 }, { "epoch": 1.63, "grad_norm": 2.795037115402011, "learning_rate": 8.930389098439751e-07, "loss": 0.1319, "step": 11947 }, { "epoch": 1.63, "grad_norm": 4.149895903097651, "learning_rate": 8.924107308998837e-07, "loss": 0.1676, "step": 11948 }, { "epoch": 1.63, "grad_norm": 3.1333125712609364, "learning_rate": 8.917827513202349e-07, "loss": 0.1327, "step": 11949 }, { "epoch": 1.63, "grad_norm": 4.896823957955043, "learning_rate": 8.911549711355083e-07, "loss": 0.1785, "step": 11950 }, { "epoch": 1.63, "grad_norm": 2.7368064174956395, "learning_rate": 8.905273903761724e-07, "loss": 0.1062, "step": 11951 }, { "epoch": 1.63, "grad_norm": 3.3067180254316795, "learning_rate": 8.899000090726905e-07, "loss": 0.1581, "step": 11952 }, { "epoch": 1.63, "grad_norm": 4.188746371269452, "learning_rate": 8.89272827255509e-07, "loss": 0.1564, "step": 11953 }, { "epoch": 1.63, "grad_norm": 3.590533581327611, "learning_rate": 8.886458449550728e-07, "loss": 0.1279, "step": 11954 }, { "epoch": 1.63, "grad_norm": 3.684673682162937, "learning_rate": 8.880190622018075e-07, "loss": 0.1454, "step": 11955 }, { "epoch": 1.63, "grad_norm": 3.69444632663642, "learning_rate": 8.873924790261407e-07, "loss": 0.1906, "step": 11956 }, { "epoch": 1.63, "grad_norm": 2.8730222332196327, "learning_rate": 8.867660954584773e-07, "loss": 0.1284, "step": 11957 }, { "epoch": 1.63, "grad_norm": 3.2648360576326088, "learning_rate": 8.861399115292246e-07, "loss": 0.146, "step": 11958 }, { "epoch": 1.63, "grad_norm": 3.1808802549214703, "learning_rate": 8.855139272687724e-07, "loss": 0.1403, "step": 11959 }, { "epoch": 1.63, "grad_norm": 2.6710387109333036, "learning_rate": 8.84888142707504e-07, "loss": 0.1273, "step": 11960 }, { "epoch": 1.63, "grad_norm": 12.346820672585396, "learning_rate": 8.842625578757918e-07, "loss": 0.2029, "step": 11961 }, { "epoch": 1.63, "grad_norm": 3.179461261614267, "learning_rate": 8.836371728039989e-07, "loss": 0.161, "step": 11962 }, { "epoch": 1.63, "grad_norm": 2.293545685631188, "learning_rate": 8.830119875224796e-07, "loss": 0.1057, "step": 11963 }, { "epoch": 1.63, "grad_norm": 2.8481188562786355, "learning_rate": 8.823870020615765e-07, "loss": 0.1698, "step": 11964 }, { "epoch": 1.63, "grad_norm": 2.923737064277252, "learning_rate": 8.817622164516232e-07, "loss": 0.145, "step": 11965 }, { "epoch": 1.63, "grad_norm": 3.1203933696203716, "learning_rate": 8.811376307229474e-07, "loss": 0.1511, "step": 11966 }, { "epoch": 1.63, "grad_norm": 3.71165282156753, "learning_rate": 8.80513244905859e-07, "loss": 0.1758, "step": 11967 }, { "epoch": 1.63, "grad_norm": 2.919597375081566, "learning_rate": 8.798890590306674e-07, "loss": 0.1154, "step": 11968 }, { "epoch": 1.63, "grad_norm": 3.5846198137744256, "learning_rate": 8.792650731276642e-07, "loss": 0.1211, "step": 11969 }, { "epoch": 1.63, "grad_norm": 3.8038291112447418, "learning_rate": 8.786412872271382e-07, "loss": 0.1599, "step": 11970 }, { "epoch": 1.63, "grad_norm": 3.283387203280692, "learning_rate": 8.780177013593621e-07, "loss": 0.1481, "step": 11971 }, { "epoch": 1.63, "grad_norm": 3.6755983696656123, "learning_rate": 8.773943155546044e-07, "loss": 0.1297, "step": 11972 }, { "epoch": 1.63, "grad_norm": 4.3560482056599845, "learning_rate": 8.767711298431203e-07, "loss": 0.1211, "step": 11973 }, { "epoch": 1.63, "grad_norm": 3.8921436480406557, "learning_rate": 8.761481442551573e-07, "loss": 0.1542, "step": 11974 }, { "epoch": 1.63, "grad_norm": 4.458699635516318, "learning_rate": 8.755253588209523e-07, "loss": 0.1386, "step": 11975 }, { "epoch": 1.63, "grad_norm": 2.72977496951034, "learning_rate": 8.749027735707316e-07, "loss": 0.102, "step": 11976 }, { "epoch": 1.63, "grad_norm": 2.9323949710231423, "learning_rate": 8.74280388534714e-07, "loss": 0.1188, "step": 11977 }, { "epoch": 1.63, "grad_norm": 2.5656508144729795, "learning_rate": 8.736582037431062e-07, "loss": 0.0986, "step": 11978 }, { "epoch": 1.63, "grad_norm": 3.510366691462618, "learning_rate": 8.730362192261078e-07, "loss": 0.128, "step": 11979 }, { "epoch": 1.63, "grad_norm": 3.8480280563684284, "learning_rate": 8.724144350139058e-07, "loss": 0.1568, "step": 11980 }, { "epoch": 1.63, "grad_norm": 3.4428089332941716, "learning_rate": 8.717928511366791e-07, "loss": 0.1474, "step": 11981 }, { "epoch": 1.63, "grad_norm": 4.1015445652684335, "learning_rate": 8.711714676245975e-07, "loss": 0.1546, "step": 11982 }, { "epoch": 1.63, "grad_norm": 3.273742487737728, "learning_rate": 8.705502845078201e-07, "loss": 0.1063, "step": 11983 }, { "epoch": 1.63, "grad_norm": 3.339670413975528, "learning_rate": 8.699293018164945e-07, "loss": 0.1796, "step": 11984 }, { "epoch": 1.63, "grad_norm": 2.9762899319079223, "learning_rate": 8.693085195807643e-07, "loss": 0.18, "step": 11985 }, { "epoch": 1.63, "grad_norm": 3.399735671416319, "learning_rate": 8.68687937830755e-07, "loss": 0.1628, "step": 11986 }, { "epoch": 1.63, "grad_norm": 3.5210995534040728, "learning_rate": 8.680675565965918e-07, "loss": 0.1575, "step": 11987 }, { "epoch": 1.63, "grad_norm": 3.007402282366884, "learning_rate": 8.674473759083801e-07, "loss": 0.125, "step": 11988 }, { "epoch": 1.63, "grad_norm": 2.8211585853977375, "learning_rate": 8.668273957962259e-07, "loss": 0.1435, "step": 11989 }, { "epoch": 1.63, "grad_norm": 2.6458567090155394, "learning_rate": 8.662076162902156e-07, "loss": 0.1605, "step": 11990 }, { "epoch": 1.63, "grad_norm": 3.4389037766216033, "learning_rate": 8.655880374204334e-07, "loss": 0.1686, "step": 11991 }, { "epoch": 1.63, "grad_norm": 3.3850700800510714, "learning_rate": 8.64968659216951e-07, "loss": 0.1598, "step": 11992 }, { "epoch": 1.63, "grad_norm": 3.60708434828967, "learning_rate": 8.643494817098297e-07, "loss": 0.1693, "step": 11993 }, { "epoch": 1.63, "grad_norm": 3.5305201391503127, "learning_rate": 8.637305049291211e-07, "loss": 0.2177, "step": 11994 }, { "epoch": 1.63, "grad_norm": 2.6591462259971315, "learning_rate": 8.631117289048685e-07, "loss": 0.1249, "step": 11995 }, { "epoch": 1.63, "grad_norm": 4.992391341053311, "learning_rate": 8.624931536671043e-07, "loss": 0.1564, "step": 11996 }, { "epoch": 1.63, "grad_norm": 2.220041100475878, "learning_rate": 8.618747792458515e-07, "loss": 0.1065, "step": 11997 }, { "epoch": 1.63, "grad_norm": 3.165639143079126, "learning_rate": 8.612566056711219e-07, "loss": 0.1716, "step": 11998 }, { "epoch": 1.63, "grad_norm": 3.5082807910094354, "learning_rate": 8.606386329729227e-07, "loss": 0.1231, "step": 11999 }, { "epoch": 1.63, "grad_norm": 3.3625412240219212, "learning_rate": 8.600208611812427e-07, "loss": 0.1818, "step": 12000 }, { "epoch": 1.63, "grad_norm": 4.494347916022194, "learning_rate": 8.594032903260707e-07, "loss": 0.1564, "step": 12001 }, { "epoch": 1.63, "grad_norm": 3.3241439904733006, "learning_rate": 8.58785920437376e-07, "loss": 0.1284, "step": 12002 }, { "epoch": 1.63, "grad_norm": 2.7194655784785904, "learning_rate": 8.58168751545127e-07, "loss": 0.1258, "step": 12003 }, { "epoch": 1.63, "grad_norm": 4.114338947088204, "learning_rate": 8.575517836792769e-07, "loss": 0.1526, "step": 12004 }, { "epoch": 1.63, "grad_norm": 4.270894777614122, "learning_rate": 8.569350168697705e-07, "loss": 0.1539, "step": 12005 }, { "epoch": 1.63, "grad_norm": 3.3657406430254264, "learning_rate": 8.563184511465428e-07, "loss": 0.1561, "step": 12006 }, { "epoch": 1.63, "grad_norm": 3.650794397885418, "learning_rate": 8.557020865395194e-07, "loss": 0.1783, "step": 12007 }, { "epoch": 1.63, "grad_norm": 4.282694519274898, "learning_rate": 8.550859230786168e-07, "loss": 0.1172, "step": 12008 }, { "epoch": 1.63, "grad_norm": 3.4526645606391604, "learning_rate": 8.544699607937396e-07, "loss": 0.1333, "step": 12009 }, { "epoch": 1.63, "grad_norm": 4.50895959069946, "learning_rate": 8.538541997147842e-07, "loss": 0.1374, "step": 12010 }, { "epoch": 1.63, "grad_norm": 3.9343469566626728, "learning_rate": 8.532386398716375e-07, "loss": 0.1354, "step": 12011 }, { "epoch": 1.63, "grad_norm": 3.177546468644806, "learning_rate": 8.526232812941748e-07, "loss": 0.1407, "step": 12012 }, { "epoch": 1.63, "grad_norm": 3.7325520055456343, "learning_rate": 8.520081240122663e-07, "loss": 0.125, "step": 12013 }, { "epoch": 1.63, "grad_norm": 3.237603068188424, "learning_rate": 8.513931680557641e-07, "loss": 0.1437, "step": 12014 }, { "epoch": 1.63, "grad_norm": 2.82194967188313, "learning_rate": 8.507784134545205e-07, "loss": 0.129, "step": 12015 }, { "epoch": 1.63, "grad_norm": 3.1023999139958045, "learning_rate": 8.501638602383677e-07, "loss": 0.1607, "step": 12016 }, { "epoch": 1.63, "grad_norm": 4.1411116070626175, "learning_rate": 8.49549508437138e-07, "loss": 0.17, "step": 12017 }, { "epoch": 1.63, "grad_norm": 3.319131665696529, "learning_rate": 8.489353580806476e-07, "loss": 0.1639, "step": 12018 }, { "epoch": 1.63, "grad_norm": 3.218119966956213, "learning_rate": 8.483214091987041e-07, "loss": 0.1508, "step": 12019 }, { "epoch": 1.64, "grad_norm": 3.0629005036539687, "learning_rate": 8.477076618211083e-07, "loss": 0.1189, "step": 12020 }, { "epoch": 1.64, "grad_norm": 3.302093184229378, "learning_rate": 8.470941159776441e-07, "loss": 0.1173, "step": 12021 }, { "epoch": 1.64, "grad_norm": 3.259432558674838, "learning_rate": 8.464807716980961e-07, "loss": 0.1558, "step": 12022 }, { "epoch": 1.64, "grad_norm": 3.0532203270940967, "learning_rate": 8.458676290122281e-07, "loss": 0.1501, "step": 12023 }, { "epoch": 1.64, "grad_norm": 2.9207314714958166, "learning_rate": 8.452546879498024e-07, "loss": 0.1348, "step": 12024 }, { "epoch": 1.64, "grad_norm": 3.6202654841000883, "learning_rate": 8.446419485405689e-07, "loss": 0.1681, "step": 12025 }, { "epoch": 1.64, "grad_norm": 3.8458982220584548, "learning_rate": 8.440294108142655e-07, "loss": 0.1838, "step": 12026 }, { "epoch": 1.64, "grad_norm": 3.7348894917068196, "learning_rate": 8.434170748006226e-07, "loss": 0.188, "step": 12027 }, { "epoch": 1.64, "grad_norm": 2.587111092691921, "learning_rate": 8.428049405293609e-07, "loss": 0.1466, "step": 12028 }, { "epoch": 1.64, "grad_norm": 5.284463968320747, "learning_rate": 8.421930080301905e-07, "loss": 0.1752, "step": 12029 }, { "epoch": 1.64, "grad_norm": 3.0551357641623276, "learning_rate": 8.415812773328119e-07, "loss": 0.1361, "step": 12030 }, { "epoch": 1.64, "grad_norm": 4.119291698971233, "learning_rate": 8.409697484669138e-07, "loss": 0.1452, "step": 12031 }, { "epoch": 1.64, "grad_norm": 3.507164013784019, "learning_rate": 8.403584214621823e-07, "loss": 0.1164, "step": 12032 }, { "epoch": 1.64, "grad_norm": 4.267507115062888, "learning_rate": 8.397472963482828e-07, "loss": 0.1286, "step": 12033 }, { "epoch": 1.64, "grad_norm": 4.266735119963468, "learning_rate": 8.391363731548813e-07, "loss": 0.1546, "step": 12034 }, { "epoch": 1.64, "grad_norm": 4.617575243595997, "learning_rate": 8.385256519116247e-07, "loss": 0.1719, "step": 12035 }, { "epoch": 1.64, "grad_norm": 2.8943247166361834, "learning_rate": 8.379151326481588e-07, "loss": 0.1493, "step": 12036 }, { "epoch": 1.64, "grad_norm": 2.4732759644456794, "learning_rate": 8.373048153941144e-07, "loss": 0.126, "step": 12037 }, { "epoch": 1.64, "grad_norm": 2.771901681365019, "learning_rate": 8.366947001791125e-07, "loss": 0.1546, "step": 12038 }, { "epoch": 1.64, "grad_norm": 3.391454446896551, "learning_rate": 8.360847870327665e-07, "loss": 0.1201, "step": 12039 }, { "epoch": 1.64, "grad_norm": 3.390319474097511, "learning_rate": 8.354750759846785e-07, "loss": 0.1524, "step": 12040 }, { "epoch": 1.64, "grad_norm": 3.933297825218861, "learning_rate": 8.348655670644412e-07, "loss": 0.1704, "step": 12041 }, { "epoch": 1.64, "grad_norm": 3.5948467896905956, "learning_rate": 8.34256260301638e-07, "loss": 0.1768, "step": 12042 }, { "epoch": 1.64, "grad_norm": 3.4786352409792354, "learning_rate": 8.336471557258408e-07, "loss": 0.165, "step": 12043 }, { "epoch": 1.64, "grad_norm": 3.4512320557635716, "learning_rate": 8.330382533666142e-07, "loss": 0.1363, "step": 12044 }, { "epoch": 1.64, "grad_norm": 3.681062239909991, "learning_rate": 8.324295532535098e-07, "loss": 0.1743, "step": 12045 }, { "epoch": 1.64, "grad_norm": 2.7688266785775735, "learning_rate": 8.318210554160755e-07, "loss": 0.1603, "step": 12046 }, { "epoch": 1.64, "grad_norm": 4.786391115136382, "learning_rate": 8.312127598838387e-07, "loss": 0.1059, "step": 12047 }, { "epoch": 1.64, "grad_norm": 3.218681629655367, "learning_rate": 8.306046666863305e-07, "loss": 0.1692, "step": 12048 }, { "epoch": 1.64, "grad_norm": 3.2010210503095506, "learning_rate": 8.29996775853058e-07, "loss": 0.1679, "step": 12049 }, { "epoch": 1.64, "grad_norm": 3.6258888776304827, "learning_rate": 8.293890874135313e-07, "loss": 0.1624, "step": 12050 }, { "epoch": 1.64, "grad_norm": 3.396332472715405, "learning_rate": 8.287816013972422e-07, "loss": 0.1606, "step": 12051 }, { "epoch": 1.64, "grad_norm": 2.494719487142252, "learning_rate": 8.281743178336754e-07, "loss": 0.1036, "step": 12052 }, { "epoch": 1.64, "grad_norm": 3.941840110388382, "learning_rate": 8.275672367523069e-07, "loss": 0.1399, "step": 12053 }, { "epoch": 1.64, "grad_norm": 2.4376163388280814, "learning_rate": 8.269603581826013e-07, "loss": 0.1332, "step": 12054 }, { "epoch": 1.64, "grad_norm": 3.402900886025659, "learning_rate": 8.263536821540136e-07, "loss": 0.1297, "step": 12055 }, { "epoch": 1.64, "grad_norm": 4.1433116240641805, "learning_rate": 8.257472086959889e-07, "loss": 0.1335, "step": 12056 }, { "epoch": 1.64, "grad_norm": 3.3326836984614294, "learning_rate": 8.251409378379638e-07, "loss": 0.1751, "step": 12057 }, { "epoch": 1.64, "grad_norm": 2.930463019864692, "learning_rate": 8.24534869609363e-07, "loss": 0.1407, "step": 12058 }, { "epoch": 1.64, "grad_norm": 3.6656307626061233, "learning_rate": 8.239290040396025e-07, "loss": 0.1485, "step": 12059 }, { "epoch": 1.64, "grad_norm": 3.707274171728495, "learning_rate": 8.233233411580894e-07, "loss": 0.1332, "step": 12060 }, { "epoch": 1.64, "grad_norm": 3.4554940315650624, "learning_rate": 8.227178809942188e-07, "loss": 0.1485, "step": 12061 }, { "epoch": 1.64, "grad_norm": 4.193080046961135, "learning_rate": 8.22112623577378e-07, "loss": 0.163, "step": 12062 }, { "epoch": 1.64, "grad_norm": 3.5685372849140644, "learning_rate": 8.215075689369429e-07, "loss": 0.1541, "step": 12063 }, { "epoch": 1.64, "grad_norm": 3.2470987641474998, "learning_rate": 8.209027171022788e-07, "loss": 0.1501, "step": 12064 }, { "epoch": 1.64, "grad_norm": 3.344000140019787, "learning_rate": 8.202980681027467e-07, "loss": 0.1725, "step": 12065 }, { "epoch": 1.64, "grad_norm": 3.739003062495581, "learning_rate": 8.196936219676888e-07, "loss": 0.1779, "step": 12066 }, { "epoch": 1.64, "grad_norm": 3.132417498212242, "learning_rate": 8.19089378726447e-07, "loss": 0.1326, "step": 12067 }, { "epoch": 1.64, "grad_norm": 3.4458289704771072, "learning_rate": 8.184853384083435e-07, "loss": 0.1468, "step": 12068 }, { "epoch": 1.64, "grad_norm": 3.317369417570236, "learning_rate": 8.178815010426999e-07, "loss": 0.1431, "step": 12069 }, { "epoch": 1.64, "grad_norm": 3.072696101415552, "learning_rate": 8.172778666588222e-07, "loss": 0.157, "step": 12070 }, { "epoch": 1.64, "grad_norm": 3.3828076291311335, "learning_rate": 8.166744352860085e-07, "loss": 0.1617, "step": 12071 }, { "epoch": 1.64, "grad_norm": 2.934922179890991, "learning_rate": 8.160712069535464e-07, "loss": 0.1302, "step": 12072 }, { "epoch": 1.64, "grad_norm": 3.2452792971565367, "learning_rate": 8.154681816907146e-07, "loss": 0.1234, "step": 12073 }, { "epoch": 1.64, "grad_norm": 3.06515751735167, "learning_rate": 8.148653595267802e-07, "loss": 0.1474, "step": 12074 }, { "epoch": 1.64, "grad_norm": 3.188567542546061, "learning_rate": 8.142627404910025e-07, "loss": 0.138, "step": 12075 }, { "epoch": 1.64, "grad_norm": 2.3593681958210975, "learning_rate": 8.136603246126301e-07, "loss": 0.1414, "step": 12076 }, { "epoch": 1.64, "grad_norm": 3.456043195191026, "learning_rate": 8.130581119209008e-07, "loss": 0.1652, "step": 12077 }, { "epoch": 1.64, "grad_norm": 2.3727879223779063, "learning_rate": 8.124561024450428e-07, "loss": 0.1451, "step": 12078 }, { "epoch": 1.64, "grad_norm": 2.930698141440025, "learning_rate": 8.118542962142784e-07, "loss": 0.1369, "step": 12079 }, { "epoch": 1.64, "grad_norm": 3.502514430856694, "learning_rate": 8.112526932578118e-07, "loss": 0.1639, "step": 12080 }, { "epoch": 1.64, "grad_norm": 8.128559284796678, "learning_rate": 8.106512936048472e-07, "loss": 0.1032, "step": 12081 }, { "epoch": 1.64, "grad_norm": 4.086129534585488, "learning_rate": 8.100500972845688e-07, "loss": 0.1826, "step": 12082 }, { "epoch": 1.64, "grad_norm": 2.9758353141155913, "learning_rate": 8.094491043261598e-07, "loss": 0.1408, "step": 12083 }, { "epoch": 1.64, "grad_norm": 3.624333188527765, "learning_rate": 8.088483147587889e-07, "loss": 0.1596, "step": 12084 }, { "epoch": 1.64, "grad_norm": 3.789685512081264, "learning_rate": 8.082477286116152e-07, "loss": 0.1728, "step": 12085 }, { "epoch": 1.64, "grad_norm": 3.034947530741655, "learning_rate": 8.07647345913789e-07, "loss": 0.152, "step": 12086 }, { "epoch": 1.64, "grad_norm": 3.139022940500255, "learning_rate": 8.070471666944496e-07, "loss": 0.133, "step": 12087 }, { "epoch": 1.64, "grad_norm": 3.541884751726306, "learning_rate": 8.064471909827282e-07, "loss": 0.1434, "step": 12088 }, { "epoch": 1.64, "grad_norm": 3.773635883962743, "learning_rate": 8.058474188077442e-07, "loss": 0.1384, "step": 12089 }, { "epoch": 1.64, "grad_norm": 4.099897066315308, "learning_rate": 8.052478501986077e-07, "loss": 0.1441, "step": 12090 }, { "epoch": 1.64, "grad_norm": 4.264985025750664, "learning_rate": 8.046484851844199e-07, "loss": 0.148, "step": 12091 }, { "epoch": 1.64, "grad_norm": 3.775993178276212, "learning_rate": 8.040493237942698e-07, "loss": 0.2006, "step": 12092 }, { "epoch": 1.64, "grad_norm": 3.7625001837837844, "learning_rate": 8.034503660572413e-07, "loss": 0.1413, "step": 12093 }, { "epoch": 1.65, "grad_norm": 3.3133997684260508, "learning_rate": 8.02851612002401e-07, "loss": 0.1232, "step": 12094 }, { "epoch": 1.65, "grad_norm": 4.164295021116762, "learning_rate": 8.022530616588142e-07, "loss": 0.1362, "step": 12095 }, { "epoch": 1.65, "grad_norm": 3.0710415414710552, "learning_rate": 8.016547150555271e-07, "loss": 0.149, "step": 12096 }, { "epoch": 1.65, "grad_norm": 3.2362449280794583, "learning_rate": 8.010565722215851e-07, "loss": 0.146, "step": 12097 }, { "epoch": 1.65, "grad_norm": 3.4012361916155496, "learning_rate": 8.004586331860176e-07, "loss": 0.1569, "step": 12098 }, { "epoch": 1.65, "grad_norm": 5.332001118067706, "learning_rate": 7.998608979778461e-07, "loss": 0.1758, "step": 12099 }, { "epoch": 1.65, "grad_norm": 2.893960155733278, "learning_rate": 7.992633666260829e-07, "loss": 0.1406, "step": 12100 }, { "epoch": 1.65, "grad_norm": 4.937249307683717, "learning_rate": 7.986660391597262e-07, "loss": 0.1339, "step": 12101 }, { "epoch": 1.65, "grad_norm": 3.809843216611558, "learning_rate": 7.98068915607772e-07, "loss": 0.1723, "step": 12102 }, { "epoch": 1.65, "grad_norm": 4.385444410735614, "learning_rate": 7.974719959992e-07, "loss": 0.1689, "step": 12103 }, { "epoch": 1.65, "grad_norm": 3.7015399395840554, "learning_rate": 7.96875280362982e-07, "loss": 0.1677, "step": 12104 }, { "epoch": 1.65, "grad_norm": 3.303397228295146, "learning_rate": 7.96278768728081e-07, "loss": 0.1397, "step": 12105 }, { "epoch": 1.65, "grad_norm": 2.870523977073603, "learning_rate": 7.956824611234476e-07, "loss": 0.1624, "step": 12106 }, { "epoch": 1.65, "grad_norm": 3.779809454915404, "learning_rate": 7.950863575780249e-07, "loss": 0.1821, "step": 12107 }, { "epoch": 1.65, "grad_norm": 3.12524417888176, "learning_rate": 7.944904581207452e-07, "loss": 0.1486, "step": 12108 }, { "epoch": 1.65, "grad_norm": 2.9990691302616086, "learning_rate": 7.938947627805305e-07, "loss": 0.1141, "step": 12109 }, { "epoch": 1.65, "grad_norm": 3.2480919826941044, "learning_rate": 7.932992715862936e-07, "loss": 0.1307, "step": 12110 }, { "epoch": 1.65, "grad_norm": 4.667713615088407, "learning_rate": 7.927039845669355e-07, "loss": 0.1308, "step": 12111 }, { "epoch": 1.65, "grad_norm": 3.563469763687883, "learning_rate": 7.921089017513522e-07, "loss": 0.138, "step": 12112 }, { "epoch": 1.65, "grad_norm": 3.0369477279169423, "learning_rate": 7.915140231684227e-07, "loss": 0.1247, "step": 12113 }, { "epoch": 1.65, "grad_norm": 5.175296230140154, "learning_rate": 7.909193488470235e-07, "loss": 0.165, "step": 12114 }, { "epoch": 1.65, "grad_norm": 2.6472014727901483, "learning_rate": 7.903248788160128e-07, "loss": 0.1586, "step": 12115 }, { "epoch": 1.65, "grad_norm": 3.2428998831852156, "learning_rate": 7.897306131042482e-07, "loss": 0.1474, "step": 12116 }, { "epoch": 1.65, "grad_norm": 2.978366291166798, "learning_rate": 7.891365517405702e-07, "loss": 0.1337, "step": 12117 }, { "epoch": 1.65, "grad_norm": 3.9387467228148343, "learning_rate": 7.885426947538128e-07, "loss": 0.181, "step": 12118 }, { "epoch": 1.65, "grad_norm": 4.914462066110964, "learning_rate": 7.879490421727987e-07, "loss": 0.1815, "step": 12119 }, { "epoch": 1.65, "grad_norm": 2.7265441854814507, "learning_rate": 7.873555940263422e-07, "loss": 0.162, "step": 12120 }, { "epoch": 1.65, "grad_norm": 4.16677476379527, "learning_rate": 7.867623503432453e-07, "loss": 0.1962, "step": 12121 }, { "epoch": 1.65, "grad_norm": 5.518593187841155, "learning_rate": 7.861693111523022e-07, "loss": 0.1489, "step": 12122 }, { "epoch": 1.65, "grad_norm": 4.358012651898852, "learning_rate": 7.855764764822965e-07, "loss": 0.1673, "step": 12123 }, { "epoch": 1.65, "grad_norm": 3.5898962016361757, "learning_rate": 7.849838463620018e-07, "loss": 0.1513, "step": 12124 }, { "epoch": 1.65, "grad_norm": 2.857401144993137, "learning_rate": 7.843914208201808e-07, "loss": 0.1127, "step": 12125 }, { "epoch": 1.65, "grad_norm": 3.1163175328955783, "learning_rate": 7.837991998855899e-07, "loss": 0.1097, "step": 12126 }, { "epoch": 1.65, "grad_norm": 4.3437140622585035, "learning_rate": 7.832071835869687e-07, "loss": 0.1601, "step": 12127 }, { "epoch": 1.65, "grad_norm": 2.6986014131200116, "learning_rate": 7.826153719530566e-07, "loss": 0.1112, "step": 12128 }, { "epoch": 1.65, "grad_norm": 4.317116503826214, "learning_rate": 7.820237650125711e-07, "loss": 0.1373, "step": 12129 }, { "epoch": 1.65, "grad_norm": 4.680646485921684, "learning_rate": 7.814323627942316e-07, "loss": 0.2037, "step": 12130 }, { "epoch": 1.65, "grad_norm": 2.2419610630480786, "learning_rate": 7.808411653267395e-07, "loss": 0.111, "step": 12131 }, { "epoch": 1.65, "grad_norm": 2.812121840009732, "learning_rate": 7.802501726387901e-07, "loss": 0.1512, "step": 12132 }, { "epoch": 1.65, "grad_norm": 2.8105241512966446, "learning_rate": 7.796593847590672e-07, "loss": 0.1143, "step": 12133 }, { "epoch": 1.65, "grad_norm": 2.8129012490912273, "learning_rate": 7.790688017162446e-07, "loss": 0.1234, "step": 12134 }, { "epoch": 1.65, "grad_norm": 3.410573108033673, "learning_rate": 7.78478423538987e-07, "loss": 0.1322, "step": 12135 }, { "epoch": 1.65, "grad_norm": 3.5170249156603637, "learning_rate": 7.778882502559487e-07, "loss": 0.1406, "step": 12136 }, { "epoch": 1.65, "grad_norm": 2.662238798063079, "learning_rate": 7.772982818957742e-07, "loss": 0.1355, "step": 12137 }, { "epoch": 1.65, "grad_norm": 4.373921762557258, "learning_rate": 7.767085184870982e-07, "loss": 0.1637, "step": 12138 }, { "epoch": 1.65, "grad_norm": 4.881767207638395, "learning_rate": 7.76118960058545e-07, "loss": 0.1746, "step": 12139 }, { "epoch": 1.65, "grad_norm": 3.1224041356020584, "learning_rate": 7.755296066387291e-07, "loss": 0.1398, "step": 12140 }, { "epoch": 1.65, "grad_norm": 3.8815623827425694, "learning_rate": 7.749404582562553e-07, "loss": 0.1576, "step": 12141 }, { "epoch": 1.65, "grad_norm": 2.9859349574961573, "learning_rate": 7.743515149397185e-07, "loss": 0.1184, "step": 12142 }, { "epoch": 1.65, "grad_norm": 4.136760415535105, "learning_rate": 7.737627767177031e-07, "loss": 0.1441, "step": 12143 }, { "epoch": 1.65, "grad_norm": 3.581665291694417, "learning_rate": 7.731742436187823e-07, "loss": 0.171, "step": 12144 }, { "epoch": 1.65, "grad_norm": 2.7674692288985074, "learning_rate": 7.725859156715255e-07, "loss": 0.1577, "step": 12145 }, { "epoch": 1.65, "grad_norm": 3.6068177589530688, "learning_rate": 7.719977929044819e-07, "loss": 0.1522, "step": 12146 }, { "epoch": 1.65, "grad_norm": 3.6877593978833345, "learning_rate": 7.714098753462018e-07, "loss": 0.1233, "step": 12147 }, { "epoch": 1.65, "grad_norm": 2.677998334529468, "learning_rate": 7.708221630252149e-07, "loss": 0.1282, "step": 12148 }, { "epoch": 1.65, "grad_norm": 3.8778215833522416, "learning_rate": 7.702346559700502e-07, "loss": 0.1725, "step": 12149 }, { "epoch": 1.65, "grad_norm": 3.6518417953717557, "learning_rate": 7.696473542092214e-07, "loss": 0.1332, "step": 12150 }, { "epoch": 1.65, "grad_norm": 2.75908541861086, "learning_rate": 7.690602577712336e-07, "loss": 0.1801, "step": 12151 }, { "epoch": 1.65, "grad_norm": 3.240733544815215, "learning_rate": 7.684733666845812e-07, "loss": 0.1192, "step": 12152 }, { "epoch": 1.65, "grad_norm": 4.23684139005518, "learning_rate": 7.678866809777507e-07, "loss": 0.1373, "step": 12153 }, { "epoch": 1.65, "grad_norm": 4.344623814758953, "learning_rate": 7.673002006792157e-07, "loss": 0.1668, "step": 12154 }, { "epoch": 1.65, "grad_norm": 3.3771241449815834, "learning_rate": 7.667139258174427e-07, "loss": 0.1467, "step": 12155 }, { "epoch": 1.65, "grad_norm": 3.821161815958391, "learning_rate": 7.661278564208862e-07, "loss": 0.1522, "step": 12156 }, { "epoch": 1.65, "grad_norm": 3.4444199349238516, "learning_rate": 7.655419925179919e-07, "loss": 0.1901, "step": 12157 }, { "epoch": 1.65, "grad_norm": 3.2940074915496176, "learning_rate": 7.649563341371935e-07, "loss": 0.1403, "step": 12158 }, { "epoch": 1.65, "grad_norm": 4.654851307841623, "learning_rate": 7.6437088130692e-07, "loss": 0.1675, "step": 12159 }, { "epoch": 1.65, "grad_norm": 3.8170603587374514, "learning_rate": 7.637856340555822e-07, "loss": 0.1959, "step": 12160 }, { "epoch": 1.65, "grad_norm": 3.615276675492505, "learning_rate": 7.632005924115898e-07, "loss": 0.1196, "step": 12161 }, { "epoch": 1.65, "grad_norm": 2.89539075710513, "learning_rate": 7.626157564033332e-07, "loss": 0.1488, "step": 12162 }, { "epoch": 1.65, "grad_norm": 3.359909319000141, "learning_rate": 7.620311260592022e-07, "loss": 0.149, "step": 12163 }, { "epoch": 1.65, "grad_norm": 3.4167309408650572, "learning_rate": 7.614467014075705e-07, "loss": 0.1398, "step": 12164 }, { "epoch": 1.65, "grad_norm": 3.682437170432618, "learning_rate": 7.608624824768035e-07, "loss": 0.1741, "step": 12165 }, { "epoch": 1.65, "grad_norm": 3.0196494805768475, "learning_rate": 7.602784692952564e-07, "loss": 0.1918, "step": 12166 }, { "epoch": 1.66, "grad_norm": 2.739218986712367, "learning_rate": 7.596946618912754e-07, "loss": 0.153, "step": 12167 }, { "epoch": 1.66, "grad_norm": 3.052551781671438, "learning_rate": 7.591110602931955e-07, "loss": 0.1263, "step": 12168 }, { "epoch": 1.66, "grad_norm": 3.1626948439072313, "learning_rate": 7.585276645293421e-07, "loss": 0.204, "step": 12169 }, { "epoch": 1.66, "grad_norm": 2.8643114847353277, "learning_rate": 7.579444746280296e-07, "loss": 0.1438, "step": 12170 }, { "epoch": 1.66, "grad_norm": 3.1367680000208185, "learning_rate": 7.57361490617567e-07, "loss": 0.1404, "step": 12171 }, { "epoch": 1.66, "grad_norm": 3.2760807888297054, "learning_rate": 7.567787125262449e-07, "loss": 0.1454, "step": 12172 }, { "epoch": 1.66, "grad_norm": 3.302656289120022, "learning_rate": 7.561961403823537e-07, "loss": 0.1736, "step": 12173 }, { "epoch": 1.66, "grad_norm": 3.7097412021490883, "learning_rate": 7.556137742141645e-07, "loss": 0.1222, "step": 12174 }, { "epoch": 1.66, "grad_norm": 4.493673038850372, "learning_rate": 7.55031614049947e-07, "loss": 0.1629, "step": 12175 }, { "epoch": 1.66, "grad_norm": 3.19756213802527, "learning_rate": 7.544496599179523e-07, "loss": 0.1249, "step": 12176 }, { "epoch": 1.66, "grad_norm": 3.4670301218211494, "learning_rate": 7.538679118464298e-07, "loss": 0.1603, "step": 12177 }, { "epoch": 1.66, "grad_norm": 3.8277035054087687, "learning_rate": 7.532863698636128e-07, "loss": 0.1857, "step": 12178 }, { "epoch": 1.66, "grad_norm": 3.892421618583075, "learning_rate": 7.527050339977277e-07, "loss": 0.1494, "step": 12179 }, { "epoch": 1.66, "grad_norm": 3.2542142335023083, "learning_rate": 7.521239042769906e-07, "loss": 0.1317, "step": 12180 }, { "epoch": 1.66, "grad_norm": 2.9339681489272413, "learning_rate": 7.515429807296043e-07, "loss": 0.128, "step": 12181 }, { "epoch": 1.66, "grad_norm": 4.252087777954482, "learning_rate": 7.509622633837671e-07, "loss": 0.1373, "step": 12182 }, { "epoch": 1.66, "grad_norm": 3.736193368500226, "learning_rate": 7.503817522676632e-07, "loss": 0.1404, "step": 12183 }, { "epoch": 1.66, "grad_norm": 3.136549811279323, "learning_rate": 7.498014474094689e-07, "loss": 0.1892, "step": 12184 }, { "epoch": 1.66, "grad_norm": 3.3674181280781657, "learning_rate": 7.49221348837349e-07, "loss": 0.1296, "step": 12185 }, { "epoch": 1.66, "grad_norm": 4.16741742020501, "learning_rate": 7.486414565794587e-07, "loss": 0.1804, "step": 12186 }, { "epoch": 1.66, "grad_norm": 3.8234746926790293, "learning_rate": 7.480617706639442e-07, "loss": 0.1232, "step": 12187 }, { "epoch": 1.66, "grad_norm": 3.3611838877075675, "learning_rate": 7.4748229111894e-07, "loss": 0.1842, "step": 12188 }, { "epoch": 1.66, "grad_norm": 3.1801397718834563, "learning_rate": 7.469030179725728e-07, "loss": 0.1519, "step": 12189 }, { "epoch": 1.66, "grad_norm": 3.4709969746127003, "learning_rate": 7.463239512529569e-07, "loss": 0.1421, "step": 12190 }, { "epoch": 1.66, "grad_norm": 4.281321509472141, "learning_rate": 7.457450909881969e-07, "loss": 0.18, "step": 12191 }, { "epoch": 1.66, "grad_norm": 4.265026264219528, "learning_rate": 7.451664372063916e-07, "loss": 0.1401, "step": 12192 }, { "epoch": 1.66, "grad_norm": 3.3936105748097822, "learning_rate": 7.445879899356212e-07, "loss": 0.1551, "step": 12193 }, { "epoch": 1.66, "grad_norm": 5.20027885292641, "learning_rate": 7.440097492039666e-07, "loss": 0.1644, "step": 12194 }, { "epoch": 1.66, "grad_norm": 3.0413636686520977, "learning_rate": 7.434317150394882e-07, "loss": 0.1489, "step": 12195 }, { "epoch": 1.66, "grad_norm": 3.2100482979341836, "learning_rate": 7.42853887470244e-07, "loss": 0.1209, "step": 12196 }, { "epoch": 1.66, "grad_norm": 2.651245497229634, "learning_rate": 7.422762665242788e-07, "loss": 0.0947, "step": 12197 }, { "epoch": 1.66, "grad_norm": 3.09306748303324, "learning_rate": 7.416988522296276e-07, "loss": 0.1613, "step": 12198 }, { "epoch": 1.66, "grad_norm": 2.4660551326289997, "learning_rate": 7.411216446143149e-07, "loss": 0.1429, "step": 12199 }, { "epoch": 1.66, "grad_norm": 4.239021841212384, "learning_rate": 7.405446437063574e-07, "loss": 0.158, "step": 12200 }, { "epoch": 1.66, "grad_norm": 3.081044841898764, "learning_rate": 7.399678495337586e-07, "loss": 0.1498, "step": 12201 }, { "epoch": 1.66, "grad_norm": 4.930605031862958, "learning_rate": 7.393912621245142e-07, "loss": 0.154, "step": 12202 }, { "epoch": 1.66, "grad_norm": 4.2567256076003686, "learning_rate": 7.388148815066082e-07, "loss": 0.1955, "step": 12203 }, { "epoch": 1.66, "grad_norm": 4.507410859431445, "learning_rate": 7.382387077080194e-07, "loss": 0.14, "step": 12204 }, { "epoch": 1.66, "grad_norm": 3.6080954174985385, "learning_rate": 7.376627407567071e-07, "loss": 0.1703, "step": 12205 }, { "epoch": 1.66, "grad_norm": 3.999762559080004, "learning_rate": 7.370869806806319e-07, "loss": 0.2033, "step": 12206 }, { "epoch": 1.66, "grad_norm": 4.345403327980609, "learning_rate": 7.365114275077334e-07, "loss": 0.1837, "step": 12207 }, { "epoch": 1.66, "grad_norm": 3.757005445437409, "learning_rate": 7.359360812659517e-07, "loss": 0.1341, "step": 12208 }, { "epoch": 1.66, "grad_norm": 3.2956203878408568, "learning_rate": 7.353609419832058e-07, "loss": 0.1417, "step": 12209 }, { "epoch": 1.66, "grad_norm": 3.4782601372224957, "learning_rate": 7.347860096874154e-07, "loss": 0.1495, "step": 12210 }, { "epoch": 1.66, "grad_norm": 2.684011892787553, "learning_rate": 7.342112844064825e-07, "loss": 0.1164, "step": 12211 }, { "epoch": 1.66, "grad_norm": 3.006133417827855, "learning_rate": 7.33636766168303e-07, "loss": 0.1158, "step": 12212 }, { "epoch": 1.66, "grad_norm": 3.3345883450279783, "learning_rate": 7.330624550007615e-07, "loss": 0.148, "step": 12213 }, { "epoch": 1.66, "grad_norm": 3.9076636268664116, "learning_rate": 7.324883509317315e-07, "loss": 0.1676, "step": 12214 }, { "epoch": 1.66, "grad_norm": 3.3860611166476735, "learning_rate": 7.319144539890782e-07, "loss": 0.1538, "step": 12215 }, { "epoch": 1.66, "grad_norm": 3.581598167036128, "learning_rate": 7.313407642006559e-07, "loss": 0.2025, "step": 12216 }, { "epoch": 1.66, "grad_norm": 2.946451765812086, "learning_rate": 7.307672815943084e-07, "loss": 0.1599, "step": 12217 }, { "epoch": 1.66, "grad_norm": 2.7622582853603777, "learning_rate": 7.301940061978724e-07, "loss": 0.1826, "step": 12218 }, { "epoch": 1.66, "grad_norm": 3.857696432057637, "learning_rate": 7.2962093803917e-07, "loss": 0.1739, "step": 12219 }, { "epoch": 1.66, "grad_norm": 2.960086039103781, "learning_rate": 7.290480771460157e-07, "loss": 0.1817, "step": 12220 }, { "epoch": 1.66, "grad_norm": 3.4770869221903693, "learning_rate": 7.284754235462132e-07, "loss": 0.1726, "step": 12221 }, { "epoch": 1.66, "grad_norm": 6.705248017104202, "learning_rate": 7.279029772675572e-07, "loss": 0.1274, "step": 12222 }, { "epoch": 1.66, "grad_norm": 4.698881322266504, "learning_rate": 7.273307383378336e-07, "loss": 0.1839, "step": 12223 }, { "epoch": 1.66, "grad_norm": 3.459445226676689, "learning_rate": 7.26758706784812e-07, "loss": 0.157, "step": 12224 }, { "epoch": 1.66, "grad_norm": 4.176939917740413, "learning_rate": 7.261868826362617e-07, "loss": 0.1659, "step": 12225 }, { "epoch": 1.66, "grad_norm": 3.8317200602640407, "learning_rate": 7.256152659199311e-07, "loss": 0.1285, "step": 12226 }, { "epoch": 1.66, "grad_norm": 2.711394174603836, "learning_rate": 7.250438566635692e-07, "loss": 0.1239, "step": 12227 }, { "epoch": 1.66, "grad_norm": 3.067102413815831, "learning_rate": 7.244726548949049e-07, "loss": 0.1376, "step": 12228 }, { "epoch": 1.66, "grad_norm": 3.1721187673320683, "learning_rate": 7.239016606416654e-07, "loss": 0.1144, "step": 12229 }, { "epoch": 1.66, "grad_norm": 3.346709109740758, "learning_rate": 7.233308739315625e-07, "loss": 0.1699, "step": 12230 }, { "epoch": 1.66, "grad_norm": 2.5418422496794517, "learning_rate": 7.227602947923007e-07, "loss": 0.128, "step": 12231 }, { "epoch": 1.66, "grad_norm": 2.7452691850464364, "learning_rate": 7.221899232515727e-07, "loss": 0.1574, "step": 12232 }, { "epoch": 1.66, "grad_norm": 2.963536007019985, "learning_rate": 7.216197593370616e-07, "loss": 0.1427, "step": 12233 }, { "epoch": 1.66, "grad_norm": 5.405206471331848, "learning_rate": 7.210498030764417e-07, "loss": 0.1908, "step": 12234 }, { "epoch": 1.66, "grad_norm": 3.3601594796219176, "learning_rate": 7.20480054497375e-07, "loss": 0.1893, "step": 12235 }, { "epoch": 1.66, "grad_norm": 4.619510625506145, "learning_rate": 7.199105136275147e-07, "loss": 0.1776, "step": 12236 }, { "epoch": 1.66, "grad_norm": 3.2926363965310044, "learning_rate": 7.193411804945061e-07, "loss": 0.166, "step": 12237 }, { "epoch": 1.66, "grad_norm": 2.768593258928393, "learning_rate": 7.18772055125978e-07, "loss": 0.1497, "step": 12238 }, { "epoch": 1.66, "grad_norm": 4.941211587059468, "learning_rate": 7.182031375495585e-07, "loss": 0.1934, "step": 12239 }, { "epoch": 1.66, "grad_norm": 5.50968074366646, "learning_rate": 7.176344277928544e-07, "loss": 0.1511, "step": 12240 }, { "epoch": 1.67, "grad_norm": 3.649508917618225, "learning_rate": 7.170659258834733e-07, "loss": 0.137, "step": 12241 }, { "epoch": 1.67, "grad_norm": 3.187892997824659, "learning_rate": 7.164976318490058e-07, "loss": 0.1408, "step": 12242 }, { "epoch": 1.67, "grad_norm": 4.7323242285255676, "learning_rate": 7.15929545717034e-07, "loss": 0.1853, "step": 12243 }, { "epoch": 1.67, "grad_norm": 3.349874782139026, "learning_rate": 7.153616675151315e-07, "loss": 0.1574, "step": 12244 }, { "epoch": 1.67, "grad_norm": 2.2011190568536887, "learning_rate": 7.147939972708601e-07, "loss": 0.1242, "step": 12245 }, { "epoch": 1.67, "grad_norm": 3.615107044857001, "learning_rate": 7.142265350117716e-07, "loss": 0.1692, "step": 12246 }, { "epoch": 1.67, "grad_norm": 3.019793247254122, "learning_rate": 7.136592807654085e-07, "loss": 0.1851, "step": 12247 }, { "epoch": 1.67, "grad_norm": 3.023049138084228, "learning_rate": 7.130922345593034e-07, "loss": 0.1313, "step": 12248 }, { "epoch": 1.67, "grad_norm": 2.6484990818236342, "learning_rate": 7.125253964209771e-07, "loss": 0.1372, "step": 12249 }, { "epoch": 1.67, "grad_norm": 3.031128822212595, "learning_rate": 7.119587663779415e-07, "loss": 0.1667, "step": 12250 }, { "epoch": 1.67, "grad_norm": 3.86873928117654, "learning_rate": 7.113923444577009e-07, "loss": 0.1513, "step": 12251 }, { "epoch": 1.67, "grad_norm": 2.3275323651336084, "learning_rate": 7.108261306877423e-07, "loss": 0.0871, "step": 12252 }, { "epoch": 1.67, "grad_norm": 3.977725132600962, "learning_rate": 7.102601250955526e-07, "loss": 0.1675, "step": 12253 }, { "epoch": 1.67, "grad_norm": 3.630905188324041, "learning_rate": 7.096943277085982e-07, "loss": 0.1851, "step": 12254 }, { "epoch": 1.67, "grad_norm": 3.0308534286456683, "learning_rate": 7.091287385543439e-07, "loss": 0.1153, "step": 12255 }, { "epoch": 1.67, "grad_norm": 3.3067805437771036, "learning_rate": 7.085633576602391e-07, "loss": 0.1746, "step": 12256 }, { "epoch": 1.67, "grad_norm": 3.373725643837014, "learning_rate": 7.079981850537266e-07, "loss": 0.165, "step": 12257 }, { "epoch": 1.67, "grad_norm": 3.1221753781617676, "learning_rate": 7.074332207622359e-07, "loss": 0.1567, "step": 12258 }, { "epoch": 1.67, "grad_norm": 2.427153554308585, "learning_rate": 7.068684648131885e-07, "loss": 0.1075, "step": 12259 }, { "epoch": 1.67, "grad_norm": 3.743083086345931, "learning_rate": 7.063039172339959e-07, "loss": 0.1482, "step": 12260 }, { "epoch": 1.67, "grad_norm": 3.468921317011023, "learning_rate": 7.05739578052056e-07, "loss": 0.1276, "step": 12261 }, { "epoch": 1.67, "grad_norm": 4.34697641053571, "learning_rate": 7.051754472947625e-07, "loss": 0.1796, "step": 12262 }, { "epoch": 1.67, "grad_norm": 3.306058240224288, "learning_rate": 7.046115249894947e-07, "loss": 0.1949, "step": 12263 }, { "epoch": 1.67, "grad_norm": 2.9126806633133473, "learning_rate": 7.040478111636229e-07, "loss": 0.16, "step": 12264 }, { "epoch": 1.67, "grad_norm": 3.323749873972703, "learning_rate": 7.034843058445074e-07, "loss": 0.1382, "step": 12265 }, { "epoch": 1.67, "grad_norm": 4.119517642012277, "learning_rate": 7.029210090594979e-07, "loss": 0.166, "step": 12266 }, { "epoch": 1.67, "grad_norm": 3.356830345822007, "learning_rate": 7.023579208359349e-07, "loss": 0.1125, "step": 12267 }, { "epoch": 1.67, "grad_norm": 3.589322118123474, "learning_rate": 7.017950412011482e-07, "loss": 0.1465, "step": 12268 }, { "epoch": 1.67, "grad_norm": 4.417276767913517, "learning_rate": 7.012323701824558e-07, "loss": 0.1452, "step": 12269 }, { "epoch": 1.67, "grad_norm": 3.8253432314509315, "learning_rate": 7.006699078071711e-07, "loss": 0.1447, "step": 12270 }, { "epoch": 1.67, "grad_norm": 3.126041910378813, "learning_rate": 7.001076541025891e-07, "loss": 0.1314, "step": 12271 }, { "epoch": 1.67, "grad_norm": 3.690324072700113, "learning_rate": 6.995456090960034e-07, "loss": 0.1344, "step": 12272 }, { "epoch": 1.67, "grad_norm": 3.1121094426057923, "learning_rate": 6.989837728146892e-07, "loss": 0.1241, "step": 12273 }, { "epoch": 1.67, "grad_norm": 2.578651812461673, "learning_rate": 6.984221452859185e-07, "loss": 0.1536, "step": 12274 }, { "epoch": 1.67, "grad_norm": 4.13109967912283, "learning_rate": 6.978607265369492e-07, "loss": 0.1707, "step": 12275 }, { "epoch": 1.67, "grad_norm": 3.9596558601643204, "learning_rate": 6.972995165950303e-07, "loss": 0.1354, "step": 12276 }, { "epoch": 1.67, "grad_norm": 5.242009696354776, "learning_rate": 6.967385154874001e-07, "loss": 0.1493, "step": 12277 }, { "epoch": 1.67, "grad_norm": 4.497466256279634, "learning_rate": 6.961777232412875e-07, "loss": 0.1748, "step": 12278 }, { "epoch": 1.67, "grad_norm": 4.858763021256503, "learning_rate": 6.956171398839113e-07, "loss": 0.166, "step": 12279 }, { "epoch": 1.67, "grad_norm": 3.289976477063542, "learning_rate": 6.950567654424784e-07, "loss": 0.117, "step": 12280 }, { "epoch": 1.67, "grad_norm": 4.49116976085878, "learning_rate": 6.944965999441883e-07, "loss": 0.1539, "step": 12281 }, { "epoch": 1.67, "grad_norm": 3.9300208943564576, "learning_rate": 6.939366434162287e-07, "loss": 0.1386, "step": 12282 }, { "epoch": 1.67, "grad_norm": 4.261918169345613, "learning_rate": 6.933768958857756e-07, "loss": 0.1742, "step": 12283 }, { "epoch": 1.67, "grad_norm": 2.9186147601719963, "learning_rate": 6.928173573800007e-07, "loss": 0.1625, "step": 12284 }, { "epoch": 1.67, "grad_norm": 2.9759997678731027, "learning_rate": 6.922580279260571e-07, "loss": 0.1332, "step": 12285 }, { "epoch": 1.67, "grad_norm": 3.0829014573119764, "learning_rate": 6.916989075510966e-07, "loss": 0.1574, "step": 12286 }, { "epoch": 1.67, "grad_norm": 4.190510462301855, "learning_rate": 6.911399962822518e-07, "loss": 0.1181, "step": 12287 }, { "epoch": 1.67, "grad_norm": 2.3506316276726853, "learning_rate": 6.905812941466533e-07, "loss": 0.1255, "step": 12288 }, { "epoch": 1.67, "grad_norm": 3.6931812966017423, "learning_rate": 6.900228011714166e-07, "loss": 0.1418, "step": 12289 }, { "epoch": 1.67, "grad_norm": 3.3872514266038936, "learning_rate": 6.894645173836496e-07, "loss": 0.1781, "step": 12290 }, { "epoch": 1.67, "grad_norm": 4.38751629707223, "learning_rate": 6.889064428104475e-07, "loss": 0.1276, "step": 12291 }, { "epoch": 1.67, "grad_norm": 4.605558248947174, "learning_rate": 6.883485774788973e-07, "loss": 0.1575, "step": 12292 }, { "epoch": 1.67, "grad_norm": 3.3921844842035687, "learning_rate": 6.87790921416076e-07, "loss": 0.125, "step": 12293 }, { "epoch": 1.67, "grad_norm": 4.217770288469826, "learning_rate": 6.872334746490494e-07, "loss": 0.2106, "step": 12294 }, { "epoch": 1.67, "grad_norm": 3.6868062446136523, "learning_rate": 6.866762372048736e-07, "loss": 0.1563, "step": 12295 }, { "epoch": 1.67, "grad_norm": 3.49421348112106, "learning_rate": 6.861192091105939e-07, "loss": 0.1517, "step": 12296 }, { "epoch": 1.67, "grad_norm": 3.3869526806216776, "learning_rate": 6.855623903932457e-07, "loss": 0.1448, "step": 12297 }, { "epoch": 1.67, "grad_norm": 3.6656346106593083, "learning_rate": 6.850057810798571e-07, "loss": 0.1585, "step": 12298 }, { "epoch": 1.67, "grad_norm": 3.7471604135493917, "learning_rate": 6.844493811974407e-07, "loss": 0.1847, "step": 12299 }, { "epoch": 1.67, "grad_norm": 2.8729916497162216, "learning_rate": 6.838931907730029e-07, "loss": 0.1329, "step": 12300 }, { "epoch": 1.67, "grad_norm": 3.309089161663105, "learning_rate": 6.833372098335383e-07, "loss": 0.1465, "step": 12301 }, { "epoch": 1.67, "grad_norm": 3.837048976067695, "learning_rate": 6.82781438406031e-07, "loss": 0.1768, "step": 12302 }, { "epoch": 1.67, "grad_norm": 3.9278952771837363, "learning_rate": 6.822258765174594e-07, "loss": 0.1077, "step": 12303 }, { "epoch": 1.67, "grad_norm": 3.1334025584500003, "learning_rate": 6.816705241947824e-07, "loss": 0.1452, "step": 12304 }, { "epoch": 1.67, "grad_norm": 2.7970271795445014, "learning_rate": 6.811153814649601e-07, "loss": 0.1465, "step": 12305 }, { "epoch": 1.67, "grad_norm": 3.117754232960953, "learning_rate": 6.805604483549322e-07, "loss": 0.1441, "step": 12306 }, { "epoch": 1.67, "grad_norm": 3.0480172778424914, "learning_rate": 6.800057248916347e-07, "loss": 0.1412, "step": 12307 }, { "epoch": 1.67, "grad_norm": 3.561637840391755, "learning_rate": 6.794512111019918e-07, "loss": 0.1804, "step": 12308 }, { "epoch": 1.67, "grad_norm": 2.8477311635583096, "learning_rate": 6.788969070129165e-07, "loss": 0.1631, "step": 12309 }, { "epoch": 1.67, "grad_norm": 2.8730565907434613, "learning_rate": 6.783428126513125e-07, "loss": 0.1548, "step": 12310 }, { "epoch": 1.67, "grad_norm": 3.498892580398703, "learning_rate": 6.77788928044073e-07, "loss": 0.1285, "step": 12311 }, { "epoch": 1.67, "grad_norm": 3.1715087480840554, "learning_rate": 6.772352532180815e-07, "loss": 0.1525, "step": 12312 }, { "epoch": 1.67, "grad_norm": 2.8329300505064587, "learning_rate": 6.766817882002102e-07, "loss": 0.1204, "step": 12313 }, { "epoch": 1.68, "grad_norm": 3.3708232149625923, "learning_rate": 6.761285330173223e-07, "loss": 0.1448, "step": 12314 }, { "epoch": 1.68, "grad_norm": 3.055168033808243, "learning_rate": 6.755754876962711e-07, "loss": 0.1278, "step": 12315 }, { "epoch": 1.68, "grad_norm": 4.3718041335362745, "learning_rate": 6.750226522638964e-07, "loss": 0.1443, "step": 12316 }, { "epoch": 1.68, "grad_norm": 3.833181942546795, "learning_rate": 6.74470026747035e-07, "loss": 0.1683, "step": 12317 }, { "epoch": 1.68, "grad_norm": 3.4807368339497002, "learning_rate": 6.739176111725044e-07, "loss": 0.1606, "step": 12318 }, { "epoch": 1.68, "grad_norm": 2.9825433161929022, "learning_rate": 6.733654055671202e-07, "loss": 0.1425, "step": 12319 }, { "epoch": 1.68, "grad_norm": 3.4147805320217812, "learning_rate": 6.728134099576794e-07, "loss": 0.0927, "step": 12320 }, { "epoch": 1.68, "grad_norm": 3.395535762513639, "learning_rate": 6.722616243709779e-07, "loss": 0.1421, "step": 12321 }, { "epoch": 1.68, "grad_norm": 3.1073626200241407, "learning_rate": 6.717100488337952e-07, "loss": 0.1492, "step": 12322 }, { "epoch": 1.68, "grad_norm": 3.6164332774943406, "learning_rate": 6.711586833729028e-07, "loss": 0.155, "step": 12323 }, { "epoch": 1.68, "grad_norm": 5.455266467956049, "learning_rate": 6.706075280150609e-07, "loss": 0.1804, "step": 12324 }, { "epoch": 1.68, "grad_norm": 4.019520146768063, "learning_rate": 6.700565827870203e-07, "loss": 0.1268, "step": 12325 }, { "epoch": 1.68, "grad_norm": 2.406920102714904, "learning_rate": 6.695058477155219e-07, "loss": 0.1069, "step": 12326 }, { "epoch": 1.68, "grad_norm": 3.7589816909332416, "learning_rate": 6.689553228272955e-07, "loss": 0.1473, "step": 12327 }, { "epoch": 1.68, "grad_norm": 4.660577135580543, "learning_rate": 6.684050081490617e-07, "loss": 0.1368, "step": 12328 }, { "epoch": 1.68, "grad_norm": 2.596718976766292, "learning_rate": 6.678549037075299e-07, "loss": 0.1436, "step": 12329 }, { "epoch": 1.68, "grad_norm": 3.4339126378288025, "learning_rate": 6.673050095293992e-07, "loss": 0.1407, "step": 12330 }, { "epoch": 1.68, "grad_norm": 2.8112020154071367, "learning_rate": 6.667553256413617e-07, "loss": 0.1387, "step": 12331 }, { "epoch": 1.68, "grad_norm": 2.766970470634516, "learning_rate": 6.662058520700926e-07, "loss": 0.1652, "step": 12332 }, { "epoch": 1.68, "grad_norm": 3.289072442191488, "learning_rate": 6.656565888422655e-07, "loss": 0.1572, "step": 12333 }, { "epoch": 1.68, "grad_norm": 4.152169780342039, "learning_rate": 6.651075359845344e-07, "loss": 0.1408, "step": 12334 }, { "epoch": 1.68, "grad_norm": 2.8832267142856653, "learning_rate": 6.645586935235521e-07, "loss": 0.1513, "step": 12335 }, { "epoch": 1.68, "grad_norm": 2.9012607729904225, "learning_rate": 6.640100614859546e-07, "loss": 0.1466, "step": 12336 }, { "epoch": 1.68, "grad_norm": 3.7609806252436817, "learning_rate": 6.634616398983712e-07, "loss": 0.1955, "step": 12337 }, { "epoch": 1.68, "grad_norm": 2.916151072016637, "learning_rate": 6.629134287874189e-07, "loss": 0.1806, "step": 12338 }, { "epoch": 1.68, "grad_norm": 2.7854559277646653, "learning_rate": 6.623654281797065e-07, "loss": 0.143, "step": 12339 }, { "epoch": 1.68, "grad_norm": 3.4270813790955255, "learning_rate": 6.618176381018316e-07, "loss": 0.1818, "step": 12340 }, { "epoch": 1.68, "grad_norm": 3.1840886856117905, "learning_rate": 6.612700585803806e-07, "loss": 0.1471, "step": 12341 }, { "epoch": 1.68, "grad_norm": 3.7059340555234974, "learning_rate": 6.607226896419305e-07, "loss": 0.1737, "step": 12342 }, { "epoch": 1.68, "grad_norm": 3.471580196078821, "learning_rate": 6.601755313130498e-07, "loss": 0.1738, "step": 12343 }, { "epoch": 1.68, "grad_norm": 4.20361406549318, "learning_rate": 6.596285836202937e-07, "loss": 0.1506, "step": 12344 }, { "epoch": 1.68, "grad_norm": 3.4800745628617316, "learning_rate": 6.590818465902088e-07, "loss": 0.1643, "step": 12345 }, { "epoch": 1.68, "grad_norm": 4.190998334229175, "learning_rate": 6.585353202493322e-07, "loss": 0.2042, "step": 12346 }, { "epoch": 1.68, "grad_norm": 2.948537099396501, "learning_rate": 6.579890046241888e-07, "loss": 0.1062, "step": 12347 }, { "epoch": 1.68, "grad_norm": 2.940223209912966, "learning_rate": 6.574428997412952e-07, "loss": 0.1571, "step": 12348 }, { "epoch": 1.68, "grad_norm": 2.978671706553361, "learning_rate": 6.568970056271556e-07, "loss": 0.1742, "step": 12349 }, { "epoch": 1.68, "grad_norm": 2.827111834542289, "learning_rate": 6.563513223082684e-07, "loss": 0.1032, "step": 12350 }, { "epoch": 1.68, "grad_norm": 2.803240760499029, "learning_rate": 6.558058498111147e-07, "loss": 0.123, "step": 12351 }, { "epoch": 1.68, "grad_norm": 4.565404361888007, "learning_rate": 6.552605881621732e-07, "loss": 0.2041, "step": 12352 }, { "epoch": 1.68, "grad_norm": 3.5597727955630263, "learning_rate": 6.54715537387905e-07, "loss": 0.148, "step": 12353 }, { "epoch": 1.68, "grad_norm": 3.3451534343858964, "learning_rate": 6.541706975147666e-07, "loss": 0.1639, "step": 12354 }, { "epoch": 1.68, "grad_norm": 3.5325202966487534, "learning_rate": 6.53626068569202e-07, "loss": 0.1743, "step": 12355 }, { "epoch": 1.68, "grad_norm": 2.720536049453653, "learning_rate": 6.530816505776444e-07, "loss": 0.1559, "step": 12356 }, { "epoch": 1.68, "grad_norm": 3.3505552048411333, "learning_rate": 6.525374435665183e-07, "loss": 0.1422, "step": 12357 }, { "epoch": 1.68, "grad_norm": 3.7948119725509435, "learning_rate": 6.519934475622363e-07, "loss": 0.1561, "step": 12358 }, { "epoch": 1.68, "grad_norm": 3.4318808502063898, "learning_rate": 6.51449662591202e-07, "loss": 0.1818, "step": 12359 }, { "epoch": 1.68, "grad_norm": 4.066709975890936, "learning_rate": 6.509060886798086e-07, "loss": 0.1405, "step": 12360 }, { "epoch": 1.68, "grad_norm": 2.941451808293998, "learning_rate": 6.503627258544382e-07, "loss": 0.1779, "step": 12361 }, { "epoch": 1.68, "grad_norm": 3.7464469086291676, "learning_rate": 6.498195741414637e-07, "loss": 0.1823, "step": 12362 }, { "epoch": 1.68, "grad_norm": 3.657128998260466, "learning_rate": 6.492766335672462e-07, "loss": 0.1562, "step": 12363 }, { "epoch": 1.68, "grad_norm": 5.414937370909808, "learning_rate": 6.487339041581408e-07, "loss": 0.2086, "step": 12364 }, { "epoch": 1.68, "grad_norm": 3.391735874158645, "learning_rate": 6.481913859404848e-07, "loss": 0.1314, "step": 12365 }, { "epoch": 1.68, "grad_norm": 2.9890958833284436, "learning_rate": 6.476490789406142e-07, "loss": 0.1352, "step": 12366 }, { "epoch": 1.68, "grad_norm": 3.671185649099435, "learning_rate": 6.471069831848453e-07, "loss": 0.1581, "step": 12367 }, { "epoch": 1.68, "grad_norm": 2.7311633815815175, "learning_rate": 6.465650986994937e-07, "loss": 0.1699, "step": 12368 }, { "epoch": 1.68, "grad_norm": 3.1067293855619824, "learning_rate": 6.460234255108572e-07, "loss": 0.1584, "step": 12369 }, { "epoch": 1.68, "grad_norm": 4.362095575019027, "learning_rate": 6.454819636452276e-07, "loss": 0.1524, "step": 12370 }, { "epoch": 1.68, "grad_norm": 6.249341847557854, "learning_rate": 6.449407131288848e-07, "loss": 0.1486, "step": 12371 }, { "epoch": 1.68, "grad_norm": 3.0312560213335673, "learning_rate": 6.443996739880981e-07, "loss": 0.1357, "step": 12372 }, { "epoch": 1.68, "grad_norm": 4.19239474740159, "learning_rate": 6.438588462491279e-07, "loss": 0.1495, "step": 12373 }, { "epoch": 1.68, "grad_norm": 3.2461671630307087, "learning_rate": 6.433182299382235e-07, "loss": 0.1778, "step": 12374 }, { "epoch": 1.68, "grad_norm": 3.685284664452195, "learning_rate": 6.427778250816242e-07, "loss": 0.1316, "step": 12375 }, { "epoch": 1.68, "grad_norm": 2.920691216169297, "learning_rate": 6.422376317055579e-07, "loss": 0.1319, "step": 12376 }, { "epoch": 1.68, "grad_norm": 4.400302542040861, "learning_rate": 6.416976498362432e-07, "loss": 0.2078, "step": 12377 }, { "epoch": 1.68, "grad_norm": 4.946280548686105, "learning_rate": 6.411578794998918e-07, "loss": 0.1494, "step": 12378 }, { "epoch": 1.68, "grad_norm": 4.555506274721217, "learning_rate": 6.406183207226973e-07, "loss": 0.164, "step": 12379 }, { "epoch": 1.68, "grad_norm": 2.6947158744037574, "learning_rate": 6.400789735308505e-07, "loss": 0.1254, "step": 12380 }, { "epoch": 1.68, "grad_norm": 3.491999062559405, "learning_rate": 6.395398379505269e-07, "loss": 0.1268, "step": 12381 }, { "epoch": 1.68, "grad_norm": 4.258035687346964, "learning_rate": 6.39000914007894e-07, "loss": 0.1656, "step": 12382 }, { "epoch": 1.68, "grad_norm": 2.6859285597513587, "learning_rate": 6.38462201729112e-07, "loss": 0.1398, "step": 12383 }, { "epoch": 1.68, "grad_norm": 4.19549747105553, "learning_rate": 6.379237011403227e-07, "loss": 0.1474, "step": 12384 }, { "epoch": 1.68, "grad_norm": 4.111514431992162, "learning_rate": 6.373854122676675e-07, "loss": 0.1325, "step": 12385 }, { "epoch": 1.68, "grad_norm": 3.6129564393830083, "learning_rate": 6.368473351372684e-07, "loss": 0.1499, "step": 12386 }, { "epoch": 1.68, "grad_norm": 2.611901342501828, "learning_rate": 6.363094697752436e-07, "loss": 0.1269, "step": 12387 }, { "epoch": 1.69, "grad_norm": 3.758022309525067, "learning_rate": 6.357718162076987e-07, "loss": 0.1778, "step": 12388 }, { "epoch": 1.69, "grad_norm": 2.9485706168203434, "learning_rate": 6.352343744607282e-07, "loss": 0.1424, "step": 12389 }, { "epoch": 1.69, "grad_norm": 2.6984578231120597, "learning_rate": 6.346971445604178e-07, "loss": 0.1344, "step": 12390 }, { "epoch": 1.69, "grad_norm": 3.7250650924134425, "learning_rate": 6.341601265328417e-07, "loss": 0.16, "step": 12391 }, { "epoch": 1.69, "grad_norm": 4.440374929217496, "learning_rate": 6.336233204040654e-07, "loss": 0.1251, "step": 12392 }, { "epoch": 1.69, "grad_norm": 3.1444634634679636, "learning_rate": 6.330867262001422e-07, "loss": 0.1618, "step": 12393 }, { "epoch": 1.69, "grad_norm": 3.1125489263590222, "learning_rate": 6.325503439471159e-07, "loss": 0.1599, "step": 12394 }, { "epoch": 1.69, "grad_norm": 3.4847511033717007, "learning_rate": 6.320141736710211e-07, "loss": 0.1433, "step": 12395 }, { "epoch": 1.69, "grad_norm": 3.097064850449355, "learning_rate": 6.314782153978799e-07, "loss": 0.1691, "step": 12396 }, { "epoch": 1.69, "grad_norm": 2.857463695151511, "learning_rate": 6.309424691537075e-07, "loss": 0.159, "step": 12397 }, { "epoch": 1.69, "grad_norm": 3.20987615641645, "learning_rate": 6.304069349645037e-07, "loss": 0.113, "step": 12398 }, { "epoch": 1.69, "grad_norm": 3.2434032502747785, "learning_rate": 6.298716128562643e-07, "loss": 0.165, "step": 12399 }, { "epoch": 1.69, "grad_norm": 4.687237546686249, "learning_rate": 6.293365028549681e-07, "loss": 0.2129, "step": 12400 }, { "epoch": 1.69, "grad_norm": 2.8606586871009956, "learning_rate": 6.288016049865897e-07, "loss": 0.1326, "step": 12401 }, { "epoch": 1.69, "grad_norm": 3.16013251827372, "learning_rate": 6.282669192770896e-07, "loss": 0.1336, "step": 12402 }, { "epoch": 1.69, "grad_norm": 2.5669218944944396, "learning_rate": 6.277324457524192e-07, "loss": 0.119, "step": 12403 }, { "epoch": 1.69, "grad_norm": 3.3512646647322075, "learning_rate": 6.271981844385194e-07, "loss": 0.1334, "step": 12404 }, { "epoch": 1.69, "grad_norm": 2.6821951630166536, "learning_rate": 6.266641353613212e-07, "loss": 0.1362, "step": 12405 }, { "epoch": 1.69, "grad_norm": 4.375647868338992, "learning_rate": 6.261302985467449e-07, "loss": 0.1761, "step": 12406 }, { "epoch": 1.69, "grad_norm": 4.145115634154091, "learning_rate": 6.255966740207003e-07, "loss": 0.1641, "step": 12407 }, { "epoch": 1.69, "grad_norm": 3.5053608973375234, "learning_rate": 6.250632618090868e-07, "loss": 0.1387, "step": 12408 }, { "epoch": 1.69, "grad_norm": 4.105924089486473, "learning_rate": 6.245300619377964e-07, "loss": 0.1705, "step": 12409 }, { "epoch": 1.69, "grad_norm": 3.481813351805684, "learning_rate": 6.239970744327045e-07, "loss": 0.1662, "step": 12410 }, { "epoch": 1.69, "grad_norm": 4.223959694505581, "learning_rate": 6.234642993196838e-07, "loss": 0.1472, "step": 12411 }, { "epoch": 1.69, "grad_norm": 3.037847937590613, "learning_rate": 6.229317366245891e-07, "loss": 0.1583, "step": 12412 }, { "epoch": 1.69, "grad_norm": 4.029506544401053, "learning_rate": 6.22399386373273e-07, "loss": 0.1698, "step": 12413 }, { "epoch": 1.69, "grad_norm": 3.6900728675122823, "learning_rate": 6.218672485915684e-07, "loss": 0.1779, "step": 12414 }, { "epoch": 1.69, "grad_norm": 4.0506758408078705, "learning_rate": 6.213353233053066e-07, "loss": 0.1878, "step": 12415 }, { "epoch": 1.69, "grad_norm": 3.530547387024542, "learning_rate": 6.208036105403042e-07, "loss": 0.1731, "step": 12416 }, { "epoch": 1.69, "grad_norm": 2.7153575615352907, "learning_rate": 6.20272110322368e-07, "loss": 0.1486, "step": 12417 }, { "epoch": 1.69, "grad_norm": 2.9353755870377727, "learning_rate": 6.197408226772944e-07, "loss": 0.1163, "step": 12418 }, { "epoch": 1.69, "grad_norm": 2.8705892377246145, "learning_rate": 6.192097476308701e-07, "loss": 0.139, "step": 12419 }, { "epoch": 1.69, "grad_norm": 3.0901790656871957, "learning_rate": 6.18678885208871e-07, "loss": 0.1485, "step": 12420 }, { "epoch": 1.69, "grad_norm": 3.1341352982433923, "learning_rate": 6.181482354370633e-07, "loss": 0.1349, "step": 12421 }, { "epoch": 1.69, "grad_norm": 2.615891606607016, "learning_rate": 6.176177983412013e-07, "loss": 0.1601, "step": 12422 }, { "epoch": 1.69, "grad_norm": 2.9637364360227183, "learning_rate": 6.170875739470311e-07, "loss": 0.1453, "step": 12423 }, { "epoch": 1.69, "grad_norm": 3.6648937896324836, "learning_rate": 6.165575622802877e-07, "loss": 0.1981, "step": 12424 }, { "epoch": 1.69, "grad_norm": 3.6700755571101378, "learning_rate": 6.160277633666945e-07, "loss": 0.1705, "step": 12425 }, { "epoch": 1.69, "grad_norm": 2.8820990400060276, "learning_rate": 6.154981772319662e-07, "loss": 0.1868, "step": 12426 }, { "epoch": 1.69, "grad_norm": 2.716436341458492, "learning_rate": 6.14968803901807e-07, "loss": 0.124, "step": 12427 }, { "epoch": 1.69, "grad_norm": 2.9738156736607206, "learning_rate": 6.144396434019101e-07, "loss": 0.0939, "step": 12428 }, { "epoch": 1.69, "grad_norm": 4.368093909432782, "learning_rate": 6.139106957579566e-07, "loss": 0.181, "step": 12429 }, { "epoch": 1.69, "grad_norm": 4.890863613388599, "learning_rate": 6.133819609956243e-07, "loss": 0.114, "step": 12430 }, { "epoch": 1.69, "grad_norm": 2.529923374006871, "learning_rate": 6.1285343914057e-07, "loss": 0.1242, "step": 12431 }, { "epoch": 1.69, "grad_norm": 3.3594879222217275, "learning_rate": 6.123251302184502e-07, "loss": 0.1189, "step": 12432 }, { "epoch": 1.69, "grad_norm": 3.2553311749137586, "learning_rate": 6.117970342549029e-07, "loss": 0.1893, "step": 12433 }, { "epoch": 1.69, "grad_norm": 4.1299879017363175, "learning_rate": 6.112691512755631e-07, "loss": 0.1841, "step": 12434 }, { "epoch": 1.69, "grad_norm": 3.1452551355062424, "learning_rate": 6.107414813060503e-07, "loss": 0.1401, "step": 12435 }, { "epoch": 1.69, "grad_norm": 3.622162882609997, "learning_rate": 6.102140243719751e-07, "loss": 0.1476, "step": 12436 }, { "epoch": 1.69, "grad_norm": 3.9294756064280327, "learning_rate": 6.096867804989387e-07, "loss": 0.163, "step": 12437 }, { "epoch": 1.69, "grad_norm": 2.9069203198237177, "learning_rate": 6.091597497125307e-07, "loss": 0.1556, "step": 12438 }, { "epoch": 1.69, "grad_norm": 3.9317996875660617, "learning_rate": 6.086329320383311e-07, "loss": 0.1521, "step": 12439 }, { "epoch": 1.69, "grad_norm": 2.931968485795329, "learning_rate": 6.081063275019089e-07, "loss": 0.1412, "step": 12440 }, { "epoch": 1.69, "grad_norm": 3.0416260123581096, "learning_rate": 6.075799361288226e-07, "loss": 0.138, "step": 12441 }, { "epoch": 1.69, "grad_norm": 4.574095735477147, "learning_rate": 6.07053757944624e-07, "loss": 0.1662, "step": 12442 }, { "epoch": 1.69, "grad_norm": 4.040637286979368, "learning_rate": 6.06527792974847e-07, "loss": 0.1448, "step": 12443 }, { "epoch": 1.69, "grad_norm": 2.974951833018684, "learning_rate": 6.060020412450241e-07, "loss": 0.1488, "step": 12444 }, { "epoch": 1.69, "grad_norm": 3.312491071451042, "learning_rate": 6.054765027806686e-07, "loss": 0.1257, "step": 12445 }, { "epoch": 1.69, "grad_norm": 4.473646575107777, "learning_rate": 6.049511776072925e-07, "loss": 0.1709, "step": 12446 }, { "epoch": 1.69, "grad_norm": 3.1032468499661676, "learning_rate": 6.044260657503881e-07, "loss": 0.1713, "step": 12447 }, { "epoch": 1.69, "grad_norm": 4.260016837603281, "learning_rate": 6.039011672354456e-07, "loss": 0.1554, "step": 12448 }, { "epoch": 1.69, "grad_norm": 3.6517183814605083, "learning_rate": 6.033764820879401e-07, "loss": 0.1514, "step": 12449 }, { "epoch": 1.69, "grad_norm": 3.901449891744106, "learning_rate": 6.028520103333369e-07, "loss": 0.139, "step": 12450 }, { "epoch": 1.69, "grad_norm": 2.983904003985916, "learning_rate": 6.023277519970927e-07, "loss": 0.1171, "step": 12451 }, { "epoch": 1.69, "grad_norm": 3.0551067519000146, "learning_rate": 6.018037071046518e-07, "loss": 0.1449, "step": 12452 }, { "epoch": 1.69, "grad_norm": 4.5231307127922475, "learning_rate": 6.012798756814491e-07, "loss": 0.166, "step": 12453 }, { "epoch": 1.69, "grad_norm": 4.02021545343481, "learning_rate": 6.007562577529097e-07, "loss": 0.163, "step": 12454 }, { "epoch": 1.69, "grad_norm": 2.785161903169859, "learning_rate": 6.002328533444462e-07, "loss": 0.1256, "step": 12455 }, { "epoch": 1.69, "grad_norm": 4.469511554293251, "learning_rate": 5.997096624814658e-07, "loss": 0.1567, "step": 12456 }, { "epoch": 1.69, "grad_norm": 3.7131842312479226, "learning_rate": 5.991866851893569e-07, "loss": 0.1489, "step": 12457 }, { "epoch": 1.69, "grad_norm": 2.778316870536782, "learning_rate": 5.98663921493508e-07, "loss": 0.1489, "step": 12458 }, { "epoch": 1.69, "grad_norm": 4.085996928924109, "learning_rate": 5.98141371419288e-07, "loss": 0.1453, "step": 12459 }, { "epoch": 1.69, "grad_norm": 3.1139746419656835, "learning_rate": 5.976190349920591e-07, "loss": 0.1434, "step": 12460 }, { "epoch": 1.7, "grad_norm": 4.4118240857292115, "learning_rate": 5.970969122371761e-07, "loss": 0.1287, "step": 12461 }, { "epoch": 1.7, "grad_norm": 2.544642812795601, "learning_rate": 5.965750031799772e-07, "loss": 0.1612, "step": 12462 }, { "epoch": 1.7, "grad_norm": 3.2316075970499796, "learning_rate": 5.960533078457969e-07, "loss": 0.1407, "step": 12463 }, { "epoch": 1.7, "grad_norm": 3.1613263613790856, "learning_rate": 5.955318262599524e-07, "loss": 0.139, "step": 12464 }, { "epoch": 1.7, "grad_norm": 2.87187433556128, "learning_rate": 5.950105584477583e-07, "loss": 0.1579, "step": 12465 }, { "epoch": 1.7, "grad_norm": 3.1422909274451656, "learning_rate": 5.9448950443451e-07, "loss": 0.1566, "step": 12466 }, { "epoch": 1.7, "grad_norm": 2.7479559259946966, "learning_rate": 5.939686642455012e-07, "loss": 0.1767, "step": 12467 }, { "epoch": 1.7, "grad_norm": 4.617280374953936, "learning_rate": 5.934480379060093e-07, "loss": 0.1315, "step": 12468 }, { "epoch": 1.7, "grad_norm": 4.250203840436865, "learning_rate": 5.929276254413035e-07, "loss": 0.1746, "step": 12469 }, { "epoch": 1.7, "grad_norm": 3.9054130420631954, "learning_rate": 5.924074268766422e-07, "loss": 0.1661, "step": 12470 }, { "epoch": 1.7, "grad_norm": 2.6463958603094415, "learning_rate": 5.918874422372739e-07, "loss": 0.1551, "step": 12471 }, { "epoch": 1.7, "grad_norm": 4.822765923071397, "learning_rate": 5.913676715484363e-07, "loss": 0.1389, "step": 12472 }, { "epoch": 1.7, "grad_norm": 4.882408396985145, "learning_rate": 5.908481148353568e-07, "loss": 0.167, "step": 12473 }, { "epoch": 1.7, "grad_norm": 3.080989490437072, "learning_rate": 5.903287721232515e-07, "loss": 0.1617, "step": 12474 }, { "epoch": 1.7, "grad_norm": 3.698445571486332, "learning_rate": 5.898096434373296e-07, "loss": 0.1553, "step": 12475 }, { "epoch": 1.7, "grad_norm": 3.3193650098947387, "learning_rate": 5.892907288027838e-07, "loss": 0.1114, "step": 12476 }, { "epoch": 1.7, "grad_norm": 3.9877616382325862, "learning_rate": 5.887720282448034e-07, "loss": 0.1485, "step": 12477 }, { "epoch": 1.7, "grad_norm": 3.204208093627824, "learning_rate": 5.882535417885604e-07, "loss": 0.0933, "step": 12478 }, { "epoch": 1.7, "grad_norm": 3.8616725855221206, "learning_rate": 5.877352694592231e-07, "loss": 0.1518, "step": 12479 }, { "epoch": 1.7, "grad_norm": 2.9014963692671363, "learning_rate": 5.872172112819447e-07, "loss": 0.1401, "step": 12480 }, { "epoch": 1.7, "grad_norm": 3.1289134444563467, "learning_rate": 5.866993672818688e-07, "loss": 0.1365, "step": 12481 }, { "epoch": 1.7, "grad_norm": 3.506596843837648, "learning_rate": 5.861817374841311e-07, "loss": 0.1269, "step": 12482 }, { "epoch": 1.7, "grad_norm": 3.142088091020532, "learning_rate": 5.856643219138531e-07, "loss": 0.1474, "step": 12483 }, { "epoch": 1.7, "grad_norm": 3.6414125656967964, "learning_rate": 5.851471205961496e-07, "loss": 0.1538, "step": 12484 }, { "epoch": 1.7, "grad_norm": 4.508663126892632, "learning_rate": 5.846301335561221e-07, "loss": 0.1483, "step": 12485 }, { "epoch": 1.7, "grad_norm": 3.6855205023162827, "learning_rate": 5.841133608188632e-07, "loss": 0.149, "step": 12486 }, { "epoch": 1.7, "grad_norm": 4.39931384684515, "learning_rate": 5.835968024094551e-07, "loss": 0.1607, "step": 12487 }, { "epoch": 1.7, "grad_norm": 4.545491819887, "learning_rate": 5.830804583529681e-07, "loss": 0.1211, "step": 12488 }, { "epoch": 1.7, "grad_norm": 3.0130901554185736, "learning_rate": 5.825643286744659e-07, "loss": 0.1553, "step": 12489 }, { "epoch": 1.7, "grad_norm": 4.0113696428580665, "learning_rate": 5.820484133989962e-07, "loss": 0.1641, "step": 12490 }, { "epoch": 1.7, "grad_norm": 3.5144856019082193, "learning_rate": 5.815327125516024e-07, "loss": 0.1427, "step": 12491 }, { "epoch": 1.7, "grad_norm": 3.9942277735385994, "learning_rate": 5.810172261573099e-07, "loss": 0.1915, "step": 12492 }, { "epoch": 1.7, "grad_norm": 2.6117772945156843, "learning_rate": 5.805019542411427e-07, "loss": 0.1565, "step": 12493 }, { "epoch": 1.7, "grad_norm": 3.694192550311636, "learning_rate": 5.799868968281075e-07, "loss": 0.1624, "step": 12494 }, { "epoch": 1.7, "grad_norm": 3.6555263318205364, "learning_rate": 5.794720539432042e-07, "loss": 0.1313, "step": 12495 }, { "epoch": 1.7, "grad_norm": 3.0356615291788596, "learning_rate": 5.789574256114194e-07, "loss": 0.1241, "step": 12496 }, { "epoch": 1.7, "grad_norm": 3.553738017018232, "learning_rate": 5.784430118577322e-07, "loss": 0.1391, "step": 12497 }, { "epoch": 1.7, "grad_norm": 3.633436497281922, "learning_rate": 5.779288127071092e-07, "loss": 0.1853, "step": 12498 }, { "epoch": 1.7, "grad_norm": 3.8059489715533203, "learning_rate": 5.774148281845082e-07, "loss": 0.1375, "step": 12499 }, { "epoch": 1.7, "grad_norm": 4.083436296553965, "learning_rate": 5.769010583148755e-07, "loss": 0.1368, "step": 12500 }, { "epoch": 1.7, "grad_norm": 3.538404351841344, "learning_rate": 5.763875031231464e-07, "loss": 0.1811, "step": 12501 }, { "epoch": 1.7, "grad_norm": 3.2210567984622123, "learning_rate": 5.758741626342479e-07, "loss": 0.1661, "step": 12502 }, { "epoch": 1.7, "grad_norm": 2.9384812426429714, "learning_rate": 5.753610368730944e-07, "loss": 0.1689, "step": 12503 }, { "epoch": 1.7, "grad_norm": 2.900613064533169, "learning_rate": 5.748481258645911e-07, "loss": 0.1487, "step": 12504 }, { "epoch": 1.7, "grad_norm": 3.9740919421483265, "learning_rate": 5.743354296336323e-07, "loss": 0.1575, "step": 12505 }, { "epoch": 1.7, "grad_norm": 3.2317777341489275, "learning_rate": 5.738229482051022e-07, "loss": 0.1273, "step": 12506 }, { "epoch": 1.7, "grad_norm": 2.8143011136329186, "learning_rate": 5.733106816038736e-07, "loss": 0.1423, "step": 12507 }, { "epoch": 1.7, "grad_norm": 3.4295865061584667, "learning_rate": 5.727986298548121e-07, "loss": 0.1254, "step": 12508 }, { "epoch": 1.7, "grad_norm": 3.8467579454573153, "learning_rate": 5.722867929827669e-07, "loss": 0.1949, "step": 12509 }, { "epoch": 1.7, "grad_norm": 2.8390032521085287, "learning_rate": 5.717751710125846e-07, "loss": 0.1439, "step": 12510 }, { "epoch": 1.7, "grad_norm": 3.645185020925562, "learning_rate": 5.712637639690927e-07, "loss": 0.1706, "step": 12511 }, { "epoch": 1.7, "grad_norm": 3.2338368960789423, "learning_rate": 5.707525718771151e-07, "loss": 0.1796, "step": 12512 }, { "epoch": 1.7, "grad_norm": 3.2164262184558314, "learning_rate": 5.702415947614631e-07, "loss": 0.1308, "step": 12513 }, { "epoch": 1.7, "grad_norm": 3.919040525200297, "learning_rate": 5.697308326469364e-07, "loss": 0.1903, "step": 12514 }, { "epoch": 1.7, "grad_norm": 4.488884856602742, "learning_rate": 5.692202855583257e-07, "loss": 0.1492, "step": 12515 }, { "epoch": 1.7, "grad_norm": 4.583901101724941, "learning_rate": 5.687099535204099e-07, "loss": 0.2047, "step": 12516 }, { "epoch": 1.7, "grad_norm": 4.101794245048617, "learning_rate": 5.681998365579594e-07, "loss": 0.1608, "step": 12517 }, { "epoch": 1.7, "grad_norm": 3.076723005076036, "learning_rate": 5.676899346957321e-07, "loss": 0.1796, "step": 12518 }, { "epoch": 1.7, "grad_norm": 3.0621319910401104, "learning_rate": 5.671802479584765e-07, "loss": 0.166, "step": 12519 }, { "epoch": 1.7, "grad_norm": 3.7327729650645978, "learning_rate": 5.666707763709312e-07, "loss": 0.1845, "step": 12520 }, { "epoch": 1.7, "grad_norm": 2.852725148385438, "learning_rate": 5.661615199578225e-07, "loss": 0.1643, "step": 12521 }, { "epoch": 1.7, "grad_norm": 3.263096184918358, "learning_rate": 5.6565247874387e-07, "loss": 0.1575, "step": 12522 }, { "epoch": 1.7, "grad_norm": 4.547024134470933, "learning_rate": 5.651436527537768e-07, "loss": 0.1889, "step": 12523 }, { "epoch": 1.7, "grad_norm": 3.7363864079402105, "learning_rate": 5.646350420122431e-07, "loss": 0.1518, "step": 12524 }, { "epoch": 1.7, "grad_norm": 3.474445747229389, "learning_rate": 5.641266465439504e-07, "loss": 0.1627, "step": 12525 }, { "epoch": 1.7, "grad_norm": 2.5578479968249797, "learning_rate": 5.636184663735772e-07, "loss": 0.1343, "step": 12526 }, { "epoch": 1.7, "grad_norm": 3.5326530603582813, "learning_rate": 5.631105015257871e-07, "loss": 0.1503, "step": 12527 }, { "epoch": 1.7, "grad_norm": 3.345330253964742, "learning_rate": 5.626027520252348e-07, "loss": 0.1173, "step": 12528 }, { "epoch": 1.7, "grad_norm": 3.040396916246078, "learning_rate": 5.620952178965638e-07, "loss": 0.1415, "step": 12529 }, { "epoch": 1.7, "grad_norm": 3.2397755718090413, "learning_rate": 5.615878991644086e-07, "loss": 0.1612, "step": 12530 }, { "epoch": 1.7, "grad_norm": 3.3280653014822437, "learning_rate": 5.610807958533909e-07, "loss": 0.1516, "step": 12531 }, { "epoch": 1.7, "grad_norm": 3.8175627488258876, "learning_rate": 5.60573907988124e-07, "loss": 0.1402, "step": 12532 }, { "epoch": 1.7, "grad_norm": 4.737839543009821, "learning_rate": 5.600672355932102e-07, "loss": 0.167, "step": 12533 }, { "epoch": 1.7, "grad_norm": 3.287796604388368, "learning_rate": 5.595607786932406e-07, "loss": 0.167, "step": 12534 }, { "epoch": 1.71, "grad_norm": 3.829738015020729, "learning_rate": 5.590545373127959e-07, "loss": 0.1201, "step": 12535 }, { "epoch": 1.71, "grad_norm": 5.876628917441026, "learning_rate": 5.585485114764494e-07, "loss": 0.1189, "step": 12536 }, { "epoch": 1.71, "grad_norm": 3.1699274322258786, "learning_rate": 5.58042701208758e-07, "loss": 0.1871, "step": 12537 }, { "epoch": 1.71, "grad_norm": 3.1851481457603024, "learning_rate": 5.575371065342749e-07, "loss": 0.1368, "step": 12538 }, { "epoch": 1.71, "grad_norm": 3.074567240765858, "learning_rate": 5.570317274775367e-07, "loss": 0.1428, "step": 12539 }, { "epoch": 1.71, "grad_norm": 3.2367112251404175, "learning_rate": 5.565265640630724e-07, "loss": 0.1546, "step": 12540 }, { "epoch": 1.71, "grad_norm": 3.35692669023755, "learning_rate": 5.560216163154031e-07, "loss": 0.1388, "step": 12541 }, { "epoch": 1.71, "grad_norm": 3.1860427572512773, "learning_rate": 5.55516884259033e-07, "loss": 0.1544, "step": 12542 }, { "epoch": 1.71, "grad_norm": 3.884536139545198, "learning_rate": 5.550123679184638e-07, "loss": 0.1579, "step": 12543 }, { "epoch": 1.71, "grad_norm": 4.959221498798487, "learning_rate": 5.545080673181785e-07, "loss": 0.1294, "step": 12544 }, { "epoch": 1.71, "grad_norm": 3.905987057886814, "learning_rate": 5.540039824826559e-07, "loss": 0.1086, "step": 12545 }, { "epoch": 1.71, "grad_norm": 2.506508062762234, "learning_rate": 5.535001134363621e-07, "loss": 0.1265, "step": 12546 }, { "epoch": 1.71, "grad_norm": 2.9469231519844987, "learning_rate": 5.529964602037519e-07, "loss": 0.1747, "step": 12547 }, { "epoch": 1.71, "grad_norm": 3.591901912491945, "learning_rate": 5.524930228092712e-07, "loss": 0.1242, "step": 12548 }, { "epoch": 1.71, "grad_norm": 3.467007418110556, "learning_rate": 5.519898012773533e-07, "loss": 0.135, "step": 12549 }, { "epoch": 1.71, "grad_norm": 2.5613536547253206, "learning_rate": 5.514867956324243e-07, "loss": 0.1238, "step": 12550 }, { "epoch": 1.71, "grad_norm": 4.199112761140413, "learning_rate": 5.509840058988963e-07, "loss": 0.1709, "step": 12551 }, { "epoch": 1.71, "grad_norm": 4.192725851441189, "learning_rate": 5.504814321011732e-07, "loss": 0.1608, "step": 12552 }, { "epoch": 1.71, "grad_norm": 4.365908295429091, "learning_rate": 5.499790742636474e-07, "loss": 0.1392, "step": 12553 }, { "epoch": 1.71, "grad_norm": 5.7065305379499325, "learning_rate": 5.494769324107008e-07, "loss": 0.1542, "step": 12554 }, { "epoch": 1.71, "grad_norm": 3.096538192050524, "learning_rate": 5.489750065667076e-07, "loss": 0.1349, "step": 12555 }, { "epoch": 1.71, "grad_norm": 2.5461478457182856, "learning_rate": 5.484732967560252e-07, "loss": 0.1184, "step": 12556 }, { "epoch": 1.71, "grad_norm": 4.871148744332325, "learning_rate": 5.479718030030084e-07, "loss": 0.1981, "step": 12557 }, { "epoch": 1.71, "grad_norm": 2.7304754041306065, "learning_rate": 5.474705253319934e-07, "loss": 0.1252, "step": 12558 }, { "epoch": 1.71, "grad_norm": 3.375362663551157, "learning_rate": 5.469694637673134e-07, "loss": 0.1576, "step": 12559 }, { "epoch": 1.71, "grad_norm": 2.8377159681434363, "learning_rate": 5.464686183332868e-07, "loss": 0.1416, "step": 12560 }, { "epoch": 1.71, "grad_norm": 2.419118755673843, "learning_rate": 5.459679890542218e-07, "loss": 0.1171, "step": 12561 }, { "epoch": 1.71, "grad_norm": 4.064265441174059, "learning_rate": 5.454675759544176e-07, "loss": 0.1554, "step": 12562 }, { "epoch": 1.71, "grad_norm": 3.5944435073673757, "learning_rate": 5.449673790581611e-07, "loss": 0.1507, "step": 12563 }, { "epoch": 1.71, "grad_norm": 4.609175029016242, "learning_rate": 5.444673983897303e-07, "loss": 0.1675, "step": 12564 }, { "epoch": 1.71, "grad_norm": 3.070058518347679, "learning_rate": 5.439676339733918e-07, "loss": 0.148, "step": 12565 }, { "epoch": 1.71, "grad_norm": 3.1627779686335993, "learning_rate": 5.434680858334024e-07, "loss": 0.1622, "step": 12566 }, { "epoch": 1.71, "grad_norm": 3.977577990307445, "learning_rate": 5.429687539940076e-07, "loss": 0.1715, "step": 12567 }, { "epoch": 1.71, "grad_norm": 2.931149991588373, "learning_rate": 5.42469638479442e-07, "loss": 0.1526, "step": 12568 }, { "epoch": 1.71, "grad_norm": 3.0262545978097957, "learning_rate": 5.419707393139334e-07, "loss": 0.1274, "step": 12569 }, { "epoch": 1.71, "grad_norm": 3.318472814313513, "learning_rate": 5.414720565216924e-07, "loss": 0.1736, "step": 12570 }, { "epoch": 1.71, "grad_norm": 2.9840074297385732, "learning_rate": 5.409735901269264e-07, "loss": 0.1219, "step": 12571 }, { "epoch": 1.71, "grad_norm": 2.5544847803050166, "learning_rate": 5.404753401538249e-07, "loss": 0.1478, "step": 12572 }, { "epoch": 1.71, "grad_norm": 4.005739611168214, "learning_rate": 5.39977306626574e-07, "loss": 0.1496, "step": 12573 }, { "epoch": 1.71, "grad_norm": 2.62752017055552, "learning_rate": 5.394794895693451e-07, "loss": 0.154, "step": 12574 }, { "epoch": 1.71, "grad_norm": 3.2927591235749225, "learning_rate": 5.389818890063003e-07, "loss": 0.1582, "step": 12575 }, { "epoch": 1.71, "grad_norm": 3.6040998462625997, "learning_rate": 5.3848450496159e-07, "loss": 0.1828, "step": 12576 }, { "epoch": 1.71, "grad_norm": 3.062177299525632, "learning_rate": 5.379873374593563e-07, "loss": 0.138, "step": 12577 }, { "epoch": 1.71, "grad_norm": 2.767533325621996, "learning_rate": 5.374903865237285e-07, "loss": 0.114, "step": 12578 }, { "epoch": 1.71, "grad_norm": 3.937414452652728, "learning_rate": 5.369936521788266e-07, "loss": 0.1473, "step": 12579 }, { "epoch": 1.71, "grad_norm": 3.508121273690953, "learning_rate": 5.364971344487608e-07, "loss": 0.1456, "step": 12580 }, { "epoch": 1.71, "grad_norm": 4.248174579670686, "learning_rate": 5.360008333576288e-07, "loss": 0.1917, "step": 12581 }, { "epoch": 1.71, "grad_norm": 2.822904427866725, "learning_rate": 5.355047489295195e-07, "loss": 0.1886, "step": 12582 }, { "epoch": 1.71, "grad_norm": 3.6619878367186014, "learning_rate": 5.350088811885107e-07, "loss": 0.1803, "step": 12583 }, { "epoch": 1.71, "grad_norm": 2.7060885239715677, "learning_rate": 5.345132301586692e-07, "loss": 0.1299, "step": 12584 }, { "epoch": 1.71, "grad_norm": 3.0004627743970285, "learning_rate": 5.340177958640519e-07, "loss": 0.1504, "step": 12585 }, { "epoch": 1.71, "grad_norm": 2.6899950925668183, "learning_rate": 5.335225783287051e-07, "loss": 0.1465, "step": 12586 }, { "epoch": 1.71, "grad_norm": 3.472332301295438, "learning_rate": 5.330275775766642e-07, "loss": 0.1539, "step": 12587 }, { "epoch": 1.71, "grad_norm": 2.7913668856459224, "learning_rate": 5.325327936319558e-07, "loss": 0.1325, "step": 12588 }, { "epoch": 1.71, "grad_norm": 2.9042709289062416, "learning_rate": 5.320382265185919e-07, "loss": 0.159, "step": 12589 }, { "epoch": 1.71, "grad_norm": 3.039164344920759, "learning_rate": 5.315438762605801e-07, "loss": 0.1456, "step": 12590 }, { "epoch": 1.71, "grad_norm": 3.80872289244352, "learning_rate": 5.310497428819106e-07, "loss": 0.1533, "step": 12591 }, { "epoch": 1.71, "grad_norm": 4.289812725522589, "learning_rate": 5.30555826406568e-07, "loss": 0.1268, "step": 12592 }, { "epoch": 1.71, "grad_norm": 4.1186583378215635, "learning_rate": 5.300621268585254e-07, "loss": 0.1556, "step": 12593 }, { "epoch": 1.71, "grad_norm": 2.8909603320007604, "learning_rate": 5.295686442617442e-07, "loss": 0.1494, "step": 12594 }, { "epoch": 1.71, "grad_norm": 2.417595468708943, "learning_rate": 5.290753786401758e-07, "loss": 0.1158, "step": 12595 }, { "epoch": 1.71, "grad_norm": 3.508547426451997, "learning_rate": 5.285823300177612e-07, "loss": 0.1485, "step": 12596 }, { "epoch": 1.71, "grad_norm": 3.308175122370146, "learning_rate": 5.28089498418431e-07, "loss": 0.1507, "step": 12597 }, { "epoch": 1.71, "grad_norm": 2.8917828450832244, "learning_rate": 5.27596883866105e-07, "loss": 0.1634, "step": 12598 }, { "epoch": 1.71, "grad_norm": 3.5811018691133674, "learning_rate": 5.271044863846925e-07, "loss": 0.1656, "step": 12599 }, { "epoch": 1.71, "grad_norm": 2.663137502266206, "learning_rate": 5.266123059980926e-07, "loss": 0.1507, "step": 12600 }, { "epoch": 1.71, "grad_norm": 3.2157548533618803, "learning_rate": 5.26120342730192e-07, "loss": 0.1423, "step": 12601 }, { "epoch": 1.71, "grad_norm": 3.469105285399547, "learning_rate": 5.256285966048719e-07, "loss": 0.2053, "step": 12602 }, { "epoch": 1.71, "grad_norm": 3.141614927288135, "learning_rate": 5.251370676459955e-07, "loss": 0.1712, "step": 12603 }, { "epoch": 1.71, "grad_norm": 3.2098801104411976, "learning_rate": 5.246457558774232e-07, "loss": 0.131, "step": 12604 }, { "epoch": 1.71, "grad_norm": 3.5618218324252453, "learning_rate": 5.241546613229976e-07, "loss": 0.1654, "step": 12605 }, { "epoch": 1.71, "grad_norm": 4.284196309840078, "learning_rate": 5.236637840065568e-07, "loss": 0.1658, "step": 12606 }, { "epoch": 1.71, "grad_norm": 3.011908445995858, "learning_rate": 5.23173123951925e-07, "loss": 0.1313, "step": 12607 }, { "epoch": 1.72, "grad_norm": 3.569326344666334, "learning_rate": 5.226826811829166e-07, "loss": 0.1388, "step": 12608 }, { "epoch": 1.72, "grad_norm": 3.5072048160296445, "learning_rate": 5.22192455723336e-07, "loss": 0.1604, "step": 12609 }, { "epoch": 1.72, "grad_norm": 3.693312296662291, "learning_rate": 5.217024475969762e-07, "loss": 0.1469, "step": 12610 }, { "epoch": 1.72, "grad_norm": 2.943175232306087, "learning_rate": 5.212126568276205e-07, "loss": 0.1378, "step": 12611 }, { "epoch": 1.72, "grad_norm": 2.825100320544984, "learning_rate": 5.207230834390403e-07, "loss": 0.1488, "step": 12612 }, { "epoch": 1.72, "grad_norm": 2.541100967599345, "learning_rate": 5.202337274549979e-07, "loss": 0.1222, "step": 12613 }, { "epoch": 1.72, "grad_norm": 4.331727926467584, "learning_rate": 5.197445888992447e-07, "loss": 0.1514, "step": 12614 }, { "epoch": 1.72, "grad_norm": 3.6084038353878265, "learning_rate": 5.192556677955202e-07, "loss": 0.1423, "step": 12615 }, { "epoch": 1.72, "grad_norm": 2.466505018544298, "learning_rate": 5.187669641675574e-07, "loss": 0.1557, "step": 12616 }, { "epoch": 1.72, "grad_norm": 3.516909482741401, "learning_rate": 5.182784780390721e-07, "loss": 0.1679, "step": 12617 }, { "epoch": 1.72, "grad_norm": 3.017391871283778, "learning_rate": 5.177902094337767e-07, "loss": 0.1866, "step": 12618 }, { "epoch": 1.72, "grad_norm": 3.5355047081414246, "learning_rate": 5.173021583753663e-07, "loss": 0.1665, "step": 12619 }, { "epoch": 1.72, "grad_norm": 3.443174053812376, "learning_rate": 5.168143248875312e-07, "loss": 0.103, "step": 12620 }, { "epoch": 1.72, "grad_norm": 3.4330212134231974, "learning_rate": 5.163267089939488e-07, "loss": 0.1203, "step": 12621 }, { "epoch": 1.72, "grad_norm": 3.55590654950614, "learning_rate": 5.158393107182835e-07, "loss": 0.1361, "step": 12622 }, { "epoch": 1.72, "grad_norm": 4.294365417667987, "learning_rate": 5.153521300841946e-07, "loss": 0.1508, "step": 12623 }, { "epoch": 1.72, "grad_norm": 3.26558367091219, "learning_rate": 5.148651671153243e-07, "loss": 0.1347, "step": 12624 }, { "epoch": 1.72, "grad_norm": 3.388386830007467, "learning_rate": 5.143784218353104e-07, "loss": 0.1468, "step": 12625 }, { "epoch": 1.72, "grad_norm": 2.7228918062382084, "learning_rate": 5.138918942677768e-07, "loss": 0.1644, "step": 12626 }, { "epoch": 1.72, "grad_norm": 4.557525755918153, "learning_rate": 5.134055844363367e-07, "loss": 0.1681, "step": 12627 }, { "epoch": 1.72, "grad_norm": 3.527321561834449, "learning_rate": 5.129194923645936e-07, "loss": 0.148, "step": 12628 }, { "epoch": 1.72, "grad_norm": 3.484157429936098, "learning_rate": 5.124336180761413e-07, "loss": 0.1793, "step": 12629 }, { "epoch": 1.72, "grad_norm": 3.522832649495713, "learning_rate": 5.119479615945605e-07, "loss": 0.1666, "step": 12630 }, { "epoch": 1.72, "grad_norm": 2.949078508637893, "learning_rate": 5.114625229434239e-07, "loss": 0.1433, "step": 12631 }, { "epoch": 1.72, "grad_norm": 3.673059028037429, "learning_rate": 5.109773021462921e-07, "loss": 0.136, "step": 12632 }, { "epoch": 1.72, "grad_norm": 4.2204707114711, "learning_rate": 5.104922992267164e-07, "loss": 0.1812, "step": 12633 }, { "epoch": 1.72, "grad_norm": 3.4358769170472634, "learning_rate": 5.100075142082345e-07, "loss": 0.1561, "step": 12634 }, { "epoch": 1.72, "grad_norm": 4.659056516046483, "learning_rate": 5.095229471143792e-07, "loss": 0.1606, "step": 12635 }, { "epoch": 1.72, "grad_norm": 3.4624280928566438, "learning_rate": 5.090385979686658e-07, "loss": 0.1529, "step": 12636 }, { "epoch": 1.72, "grad_norm": 2.8601545916314013, "learning_rate": 5.085544667946057e-07, "loss": 0.1081, "step": 12637 }, { "epoch": 1.72, "grad_norm": 2.996535500232649, "learning_rate": 5.080705536156933e-07, "loss": 0.1255, "step": 12638 }, { "epoch": 1.72, "grad_norm": 3.64315562194958, "learning_rate": 5.075868584554177e-07, "loss": 0.1606, "step": 12639 }, { "epoch": 1.72, "grad_norm": 4.144520819991469, "learning_rate": 5.071033813372556e-07, "loss": 0.137, "step": 12640 }, { "epoch": 1.72, "grad_norm": 2.583969930198406, "learning_rate": 5.066201222846718e-07, "loss": 0.1367, "step": 12641 }, { "epoch": 1.72, "grad_norm": 2.753696990167325, "learning_rate": 5.061370813211219e-07, "loss": 0.1039, "step": 12642 }, { "epoch": 1.72, "grad_norm": 4.258841362713545, "learning_rate": 5.056542584700507e-07, "loss": 0.1516, "step": 12643 }, { "epoch": 1.72, "grad_norm": 3.210499629996837, "learning_rate": 5.051716537548923e-07, "loss": 0.1628, "step": 12644 }, { "epoch": 1.72, "grad_norm": 3.2264286872218233, "learning_rate": 5.046892671990699e-07, "loss": 0.161, "step": 12645 }, { "epoch": 1.72, "grad_norm": 3.2944856140914816, "learning_rate": 5.042070988259956e-07, "loss": 0.1193, "step": 12646 }, { "epoch": 1.72, "grad_norm": 3.346547676432335, "learning_rate": 5.037251486590755e-07, "loss": 0.1246, "step": 12647 }, { "epoch": 1.72, "grad_norm": 6.539340921658134, "learning_rate": 5.032434167216965e-07, "loss": 0.159, "step": 12648 }, { "epoch": 1.72, "grad_norm": 3.950138147261897, "learning_rate": 5.027619030372438e-07, "loss": 0.1706, "step": 12649 }, { "epoch": 1.72, "grad_norm": 2.8762783632873856, "learning_rate": 5.022806076290843e-07, "loss": 0.1185, "step": 12650 }, { "epoch": 1.72, "grad_norm": 3.8606586633578974, "learning_rate": 5.017995305205814e-07, "loss": 0.1509, "step": 12651 }, { "epoch": 1.72, "grad_norm": 4.0640835902494485, "learning_rate": 5.013186717350815e-07, "loss": 0.1581, "step": 12652 }, { "epoch": 1.72, "grad_norm": 3.6916319682939656, "learning_rate": 5.008380312959255e-07, "loss": 0.1766, "step": 12653 }, { "epoch": 1.72, "grad_norm": 2.4021317540524514, "learning_rate": 5.003576092264406e-07, "loss": 0.1141, "step": 12654 }, { "epoch": 1.72, "grad_norm": 4.198627550409467, "learning_rate": 4.99877405549945e-07, "loss": 0.179, "step": 12655 }, { "epoch": 1.72, "grad_norm": 3.774059445119162, "learning_rate": 4.993974202897456e-07, "loss": 0.1297, "step": 12656 }, { "epoch": 1.72, "grad_norm": 4.373733232299156, "learning_rate": 4.989176534691381e-07, "loss": 0.1587, "step": 12657 }, { "epoch": 1.72, "grad_norm": 4.677937054496387, "learning_rate": 4.98438105111409e-07, "loss": 0.2102, "step": 12658 }, { "epoch": 1.72, "grad_norm": 3.186930492193774, "learning_rate": 4.979587752398329e-07, "loss": 0.1609, "step": 12659 }, { "epoch": 1.72, "grad_norm": 3.542316058685809, "learning_rate": 4.974796638776752e-07, "loss": 0.1693, "step": 12660 }, { "epoch": 1.72, "grad_norm": 3.741229941042619, "learning_rate": 4.970007710481889e-07, "loss": 0.1455, "step": 12661 }, { "epoch": 1.72, "grad_norm": 3.8757046683915637, "learning_rate": 4.965220967746181e-07, "loss": 0.157, "step": 12662 }, { "epoch": 1.72, "grad_norm": 3.903554324817257, "learning_rate": 4.96043641080195e-07, "loss": 0.163, "step": 12663 }, { "epoch": 1.72, "grad_norm": 5.029712871902069, "learning_rate": 4.955654039881424e-07, "loss": 0.1454, "step": 12664 }, { "epoch": 1.72, "grad_norm": 3.0664363765176104, "learning_rate": 4.950873855216715e-07, "loss": 0.1467, "step": 12665 }, { "epoch": 1.72, "grad_norm": 3.5785673056836993, "learning_rate": 4.946095857039829e-07, "loss": 0.1416, "step": 12666 }, { "epoch": 1.72, "grad_norm": 3.4299727452270283, "learning_rate": 4.94132004558266e-07, "loss": 0.1795, "step": 12667 }, { "epoch": 1.72, "grad_norm": 3.1466014112590504, "learning_rate": 4.936546421077043e-07, "loss": 0.1763, "step": 12668 }, { "epoch": 1.72, "grad_norm": 4.486500904486997, "learning_rate": 4.931774983754623e-07, "loss": 0.1597, "step": 12669 }, { "epoch": 1.72, "grad_norm": 3.163191516116747, "learning_rate": 4.927005733847017e-07, "loss": 0.1409, "step": 12670 }, { "epoch": 1.72, "grad_norm": 3.9133790657874696, "learning_rate": 4.922238671585683e-07, "loss": 0.1309, "step": 12671 }, { "epoch": 1.72, "grad_norm": 2.9743036592663232, "learning_rate": 4.917473797202005e-07, "loss": 0.1503, "step": 12672 }, { "epoch": 1.72, "grad_norm": 3.9698080616297133, "learning_rate": 4.912711110927249e-07, "loss": 0.1402, "step": 12673 }, { "epoch": 1.72, "grad_norm": 3.2906919499988763, "learning_rate": 4.907950612992569e-07, "loss": 0.1547, "step": 12674 }, { "epoch": 1.72, "grad_norm": 3.660925087321501, "learning_rate": 4.903192303629023e-07, "loss": 0.1455, "step": 12675 }, { "epoch": 1.72, "grad_norm": 3.5117833527852915, "learning_rate": 4.898436183067556e-07, "loss": 0.179, "step": 12676 }, { "epoch": 1.72, "grad_norm": 3.818532728856471, "learning_rate": 4.893682251539012e-07, "loss": 0.168, "step": 12677 }, { "epoch": 1.72, "grad_norm": 3.762260261943262, "learning_rate": 4.888930509274125e-07, "loss": 0.1456, "step": 12678 }, { "epoch": 1.72, "grad_norm": 3.873762601079674, "learning_rate": 4.884180956503515e-07, "loss": 0.1498, "step": 12679 }, { "epoch": 1.72, "grad_norm": 4.447601122985336, "learning_rate": 4.879433593457733e-07, "loss": 0.177, "step": 12680 }, { "epoch": 1.72, "grad_norm": 3.0888966242135862, "learning_rate": 4.874688420367152e-07, "loss": 0.1443, "step": 12681 }, { "epoch": 1.73, "grad_norm": 2.786680874017952, "learning_rate": 4.869945437462126e-07, "loss": 0.1223, "step": 12682 }, { "epoch": 1.73, "grad_norm": 3.3575645568648533, "learning_rate": 4.865204644972821e-07, "loss": 0.1671, "step": 12683 }, { "epoch": 1.73, "grad_norm": 4.304483962048664, "learning_rate": 4.860466043129369e-07, "loss": 0.1309, "step": 12684 }, { "epoch": 1.73, "grad_norm": 3.179013594227199, "learning_rate": 4.855729632161727e-07, "loss": 0.1903, "step": 12685 }, { "epoch": 1.73, "grad_norm": 3.0803559871856323, "learning_rate": 4.850995412299803e-07, "loss": 0.1314, "step": 12686 }, { "epoch": 1.73, "grad_norm": 3.227271445838166, "learning_rate": 4.846263383773364e-07, "loss": 0.147, "step": 12687 }, { "epoch": 1.73, "grad_norm": 3.7890956961313966, "learning_rate": 4.841533546812094e-07, "loss": 0.1662, "step": 12688 }, { "epoch": 1.73, "grad_norm": 3.415096318119207, "learning_rate": 4.836805901645547e-07, "loss": 0.1178, "step": 12689 }, { "epoch": 1.73, "grad_norm": 2.698646726734078, "learning_rate": 4.832080448503191e-07, "loss": 0.1518, "step": 12690 }, { "epoch": 1.73, "grad_norm": 3.5695918128879702, "learning_rate": 4.827357187614368e-07, "loss": 0.1569, "step": 12691 }, { "epoch": 1.73, "grad_norm": 4.654636891233494, "learning_rate": 4.822636119208335e-07, "loss": 0.1827, "step": 12692 }, { "epoch": 1.73, "grad_norm": 4.6143515869361735, "learning_rate": 4.817917243514214e-07, "loss": 0.1713, "step": 12693 }, { "epoch": 1.73, "grad_norm": 3.5736930951885566, "learning_rate": 4.813200560761078e-07, "loss": 0.15, "step": 12694 }, { "epoch": 1.73, "grad_norm": 2.949136135487377, "learning_rate": 4.808486071177804e-07, "loss": 0.1176, "step": 12695 }, { "epoch": 1.73, "grad_norm": 2.582481369963999, "learning_rate": 4.803773774993259e-07, "loss": 0.1117, "step": 12696 }, { "epoch": 1.73, "grad_norm": 3.196233483109473, "learning_rate": 4.799063672436111e-07, "loss": 0.1269, "step": 12697 }, { "epoch": 1.73, "grad_norm": 3.188357303369242, "learning_rate": 4.794355763735009e-07, "loss": 0.1507, "step": 12698 }, { "epoch": 1.73, "grad_norm": 3.171804356868779, "learning_rate": 4.789650049118432e-07, "loss": 0.1583, "step": 12699 }, { "epoch": 1.73, "grad_norm": 3.097382505059315, "learning_rate": 4.784946528814782e-07, "loss": 0.1423, "step": 12700 }, { "epoch": 1.73, "grad_norm": 4.415727916996611, "learning_rate": 4.780245203052359e-07, "loss": 0.1753, "step": 12701 }, { "epoch": 1.73, "grad_norm": 4.234343422152406, "learning_rate": 4.775546072059311e-07, "loss": 0.1611, "step": 12702 }, { "epoch": 1.73, "grad_norm": 3.873752598087158, "learning_rate": 4.770849136063754e-07, "loss": 0.1551, "step": 12703 }, { "epoch": 1.73, "grad_norm": 2.721783586426403, "learning_rate": 4.766154395293615e-07, "loss": 0.141, "step": 12704 }, { "epoch": 1.73, "grad_norm": 3.707913072169541, "learning_rate": 4.761461849976784e-07, "loss": 0.1395, "step": 12705 }, { "epoch": 1.73, "grad_norm": 2.613916078715866, "learning_rate": 4.756771500341018e-07, "loss": 0.1361, "step": 12706 }, { "epoch": 1.73, "grad_norm": 2.8185603769764414, "learning_rate": 4.752083346613956e-07, "loss": 0.1576, "step": 12707 }, { "epoch": 1.73, "grad_norm": 3.928442353783392, "learning_rate": 4.7473973890231395e-07, "loss": 0.1448, "step": 12708 }, { "epoch": 1.73, "grad_norm": 3.5518795960700453, "learning_rate": 4.742713627796014e-07, "loss": 0.16, "step": 12709 }, { "epoch": 1.73, "grad_norm": 3.554128052993514, "learning_rate": 4.738032063159898e-07, "loss": 0.1594, "step": 12710 }, { "epoch": 1.73, "grad_norm": 3.83132902914165, "learning_rate": 4.7333526953420194e-07, "loss": 0.1899, "step": 12711 }, { "epoch": 1.73, "grad_norm": 3.475524230056582, "learning_rate": 4.728675524569487e-07, "loss": 0.1675, "step": 12712 }, { "epoch": 1.73, "grad_norm": 3.6975289115081424, "learning_rate": 4.724000551069335e-07, "loss": 0.1449, "step": 12713 }, { "epoch": 1.73, "grad_norm": 2.52846737707369, "learning_rate": 4.719327775068433e-07, "loss": 0.1043, "step": 12714 }, { "epoch": 1.73, "grad_norm": 3.3901290348139423, "learning_rate": 4.714657196793604e-07, "loss": 0.1113, "step": 12715 }, { "epoch": 1.73, "grad_norm": 3.6853951339363533, "learning_rate": 4.709988816471511e-07, "loss": 0.1355, "step": 12716 }, { "epoch": 1.73, "grad_norm": 3.2327645264633342, "learning_rate": 4.7053226343287626e-07, "loss": 0.1465, "step": 12717 }, { "epoch": 1.73, "grad_norm": 4.192593719449562, "learning_rate": 4.7006586505918273e-07, "loss": 0.1289, "step": 12718 }, { "epoch": 1.73, "grad_norm": 2.7528967142992036, "learning_rate": 4.6959968654870624e-07, "loss": 0.1409, "step": 12719 }, { "epoch": 1.73, "grad_norm": 3.0867623022478083, "learning_rate": 4.691337279240743e-07, "loss": 0.1097, "step": 12720 }, { "epoch": 1.73, "grad_norm": 3.0646831598572386, "learning_rate": 4.686679892079027e-07, "loss": 0.1052, "step": 12721 }, { "epoch": 1.73, "grad_norm": 3.4738086051136223, "learning_rate": 4.68202470422795e-07, "loss": 0.1267, "step": 12722 }, { "epoch": 1.73, "grad_norm": 2.290742183985762, "learning_rate": 4.6773717159134704e-07, "loss": 0.1129, "step": 12723 }, { "epoch": 1.73, "grad_norm": 2.733094995730916, "learning_rate": 4.672720927361413e-07, "loss": 0.1395, "step": 12724 }, { "epoch": 1.73, "grad_norm": 3.4079995687975795, "learning_rate": 4.668072338797508e-07, "loss": 0.2157, "step": 12725 }, { "epoch": 1.73, "grad_norm": 3.9257331606107666, "learning_rate": 4.66342595044737e-07, "loss": 0.1764, "step": 12726 }, { "epoch": 1.73, "grad_norm": 3.4121026665321543, "learning_rate": 4.6587817625365406e-07, "loss": 0.1113, "step": 12727 }, { "epoch": 1.73, "grad_norm": 3.2698490013568384, "learning_rate": 4.6541397752904006e-07, "loss": 0.1544, "step": 12728 }, { "epoch": 1.73, "grad_norm": 4.8896172375056475, "learning_rate": 4.649499988934275e-07, "loss": 0.1602, "step": 12729 }, { "epoch": 1.73, "grad_norm": 3.44342431300764, "learning_rate": 4.6448624036933285e-07, "loss": 0.1834, "step": 12730 }, { "epoch": 1.73, "grad_norm": 3.4638305491167145, "learning_rate": 4.6402270197926813e-07, "loss": 0.1835, "step": 12731 }, { "epoch": 1.73, "grad_norm": 2.880495021458653, "learning_rate": 4.6355938374572975e-07, "loss": 0.1248, "step": 12732 }, { "epoch": 1.73, "grad_norm": 3.80484989015968, "learning_rate": 4.630962856912058e-07, "loss": 0.0906, "step": 12733 }, { "epoch": 1.73, "grad_norm": 3.5235029606112684, "learning_rate": 4.626334078381722e-07, "loss": 0.1683, "step": 12734 }, { "epoch": 1.73, "grad_norm": 3.248599297800011, "learning_rate": 4.621707502090961e-07, "loss": 0.1257, "step": 12735 }, { "epoch": 1.73, "grad_norm": 2.4685102302309216, "learning_rate": 4.6170831282643156e-07, "loss": 0.1288, "step": 12736 }, { "epoch": 1.73, "grad_norm": 2.9906793746376015, "learning_rate": 4.612460957126247e-07, "loss": 0.1772, "step": 12737 }, { "epoch": 1.73, "grad_norm": 4.1458356183992935, "learning_rate": 4.6078409889010856e-07, "loss": 0.1823, "step": 12738 }, { "epoch": 1.73, "grad_norm": 3.260114853700921, "learning_rate": 4.603223223813069e-07, "loss": 0.1473, "step": 12739 }, { "epoch": 1.73, "grad_norm": 3.5937559381103386, "learning_rate": 4.5986076620863185e-07, "loss": 0.1631, "step": 12740 }, { "epoch": 1.73, "grad_norm": 3.177264805912465, "learning_rate": 4.59399430394486e-07, "loss": 0.14, "step": 12741 }, { "epoch": 1.73, "grad_norm": 3.750849685710505, "learning_rate": 4.589383149612603e-07, "loss": 0.1728, "step": 12742 }, { "epoch": 1.73, "grad_norm": 2.911852826743248, "learning_rate": 4.584774199313352e-07, "loss": 0.1338, "step": 12743 }, { "epoch": 1.73, "grad_norm": 3.7357590301294312, "learning_rate": 4.5801674532708004e-07, "loss": 0.1392, "step": 12744 }, { "epoch": 1.73, "grad_norm": 3.40174195509637, "learning_rate": 4.575562911708542e-07, "loss": 0.1346, "step": 12745 }, { "epoch": 1.73, "grad_norm": 3.005448614764629, "learning_rate": 4.570960574850075e-07, "loss": 0.1046, "step": 12746 }, { "epoch": 1.73, "grad_norm": 3.680076418395993, "learning_rate": 4.5663604429187547e-07, "loss": 0.1624, "step": 12747 }, { "epoch": 1.73, "grad_norm": 3.000796171606813, "learning_rate": 4.5617625161378745e-07, "loss": 0.1081, "step": 12748 }, { "epoch": 1.73, "grad_norm": 4.376488314128439, "learning_rate": 4.557166794730572e-07, "loss": 0.1542, "step": 12749 }, { "epoch": 1.73, "grad_norm": 3.274567097889444, "learning_rate": 4.5525732789199193e-07, "loss": 0.1527, "step": 12750 }, { "epoch": 1.73, "grad_norm": 2.7700403672615224, "learning_rate": 4.5479819689288664e-07, "loss": 0.158, "step": 12751 }, { "epoch": 1.73, "grad_norm": 3.849347394613242, "learning_rate": 4.543392864980256e-07, "loss": 0.1874, "step": 12752 }, { "epoch": 1.73, "grad_norm": 2.9518756981110688, "learning_rate": 4.5388059672968164e-07, "loss": 0.166, "step": 12753 }, { "epoch": 1.73, "grad_norm": 3.3072844762106834, "learning_rate": 4.534221276101175e-07, "loss": 0.1283, "step": 12754 }, { "epoch": 1.74, "grad_norm": 3.755178498335298, "learning_rate": 4.529638791615859e-07, "loss": 0.1367, "step": 12755 }, { "epoch": 1.74, "grad_norm": 4.132700395884465, "learning_rate": 4.5250585140632796e-07, "loss": 0.1474, "step": 12756 }, { "epoch": 1.74, "grad_norm": 4.660635426464985, "learning_rate": 4.5204804436657423e-07, "loss": 0.1737, "step": 12757 }, { "epoch": 1.74, "grad_norm": 3.212547274947246, "learning_rate": 4.515904580645447e-07, "loss": 0.1338, "step": 12758 }, { "epoch": 1.74, "grad_norm": 3.2902285395124826, "learning_rate": 4.5113309252244773e-07, "loss": 0.1411, "step": 12759 }, { "epoch": 1.74, "grad_norm": 2.6545186451765153, "learning_rate": 4.50675947762485e-07, "loss": 0.1259, "step": 12760 }, { "epoch": 1.74, "grad_norm": 2.707264334687848, "learning_rate": 4.5021902380683993e-07, "loss": 0.1386, "step": 12761 }, { "epoch": 1.74, "grad_norm": 4.017287909739534, "learning_rate": 4.4976232067769356e-07, "loss": 0.1259, "step": 12762 }, { "epoch": 1.74, "grad_norm": 5.329898221587165, "learning_rate": 4.493058383972088e-07, "loss": 0.1397, "step": 12763 }, { "epoch": 1.74, "grad_norm": 3.597359524801611, "learning_rate": 4.4884957698754396e-07, "loss": 0.1031, "step": 12764 }, { "epoch": 1.74, "grad_norm": 4.201011611304514, "learning_rate": 4.4839353647084303e-07, "loss": 0.167, "step": 12765 }, { "epoch": 1.74, "grad_norm": 3.433115719831096, "learning_rate": 4.4793771686924003e-07, "loss": 0.1633, "step": 12766 }, { "epoch": 1.74, "grad_norm": 2.635823994598984, "learning_rate": 4.474821182048583e-07, "loss": 0.1446, "step": 12767 }, { "epoch": 1.74, "grad_norm": 4.056056851342701, "learning_rate": 4.470267404998113e-07, "loss": 0.1842, "step": 12768 }, { "epoch": 1.74, "grad_norm": 3.1842246383823376, "learning_rate": 4.465715837762008e-07, "loss": 0.1916, "step": 12769 }, { "epoch": 1.74, "grad_norm": 2.7985601541044867, "learning_rate": 4.46116648056118e-07, "loss": 0.1512, "step": 12770 }, { "epoch": 1.74, "grad_norm": 3.4163490017165303, "learning_rate": 4.4566193336164353e-07, "loss": 0.1368, "step": 12771 }, { "epoch": 1.74, "grad_norm": 4.109565756071108, "learning_rate": 4.45207439714847e-07, "loss": 0.1317, "step": 12772 }, { "epoch": 1.74, "grad_norm": 3.832959837964477, "learning_rate": 4.447531671377869e-07, "loss": 0.1311, "step": 12773 }, { "epoch": 1.74, "grad_norm": 3.584048642058596, "learning_rate": 4.4429911565251505e-07, "loss": 0.1189, "step": 12774 }, { "epoch": 1.74, "grad_norm": 3.4722518267335927, "learning_rate": 4.4384528528106373e-07, "loss": 0.1283, "step": 12775 }, { "epoch": 1.74, "grad_norm": 2.5407688609604744, "learning_rate": 4.433916760454654e-07, "loss": 0.1369, "step": 12776 }, { "epoch": 1.74, "grad_norm": 3.338106383122275, "learning_rate": 4.4293828796773133e-07, "loss": 0.1595, "step": 12777 }, { "epoch": 1.74, "grad_norm": 5.190314710376253, "learning_rate": 4.424851210698705e-07, "loss": 0.1267, "step": 12778 }, { "epoch": 1.74, "grad_norm": 3.6301229747670667, "learning_rate": 4.4203217537387653e-07, "loss": 0.1658, "step": 12779 }, { "epoch": 1.74, "grad_norm": 2.8427839746197097, "learning_rate": 4.4157945090173294e-07, "loss": 0.1729, "step": 12780 }, { "epoch": 1.74, "grad_norm": 2.8848163824409476, "learning_rate": 4.411269476754149e-07, "loss": 0.1515, "step": 12781 }, { "epoch": 1.74, "grad_norm": 2.9668391743494165, "learning_rate": 4.406746657168809e-07, "loss": 0.1529, "step": 12782 }, { "epoch": 1.74, "grad_norm": 3.5474381509808244, "learning_rate": 4.402226050480868e-07, "loss": 0.1506, "step": 12783 }, { "epoch": 1.74, "grad_norm": 4.098616073449658, "learning_rate": 4.397707656909717e-07, "loss": 0.1483, "step": 12784 }, { "epoch": 1.74, "grad_norm": 3.758653800030771, "learning_rate": 4.393191476674663e-07, "loss": 0.1133, "step": 12785 }, { "epoch": 1.74, "grad_norm": 4.026165647856139, "learning_rate": 4.388677509994904e-07, "loss": 0.1518, "step": 12786 }, { "epoch": 1.74, "grad_norm": 3.8698875842876084, "learning_rate": 4.384165757089526e-07, "loss": 0.1325, "step": 12787 }, { "epoch": 1.74, "grad_norm": 3.6230567391383826, "learning_rate": 4.379656218177514e-07, "loss": 0.113, "step": 12788 }, { "epoch": 1.74, "grad_norm": 3.1402504280452628, "learning_rate": 4.375148893477732e-07, "loss": 0.1459, "step": 12789 }, { "epoch": 1.74, "grad_norm": 3.61423481074132, "learning_rate": 4.37064378320895e-07, "loss": 0.1559, "step": 12790 }, { "epoch": 1.74, "grad_norm": 3.6579500713733486, "learning_rate": 4.3661408875898313e-07, "loss": 0.1415, "step": 12791 }, { "epoch": 1.74, "grad_norm": 4.310656190611441, "learning_rate": 4.361640206838913e-07, "loss": 0.1681, "step": 12792 }, { "epoch": 1.74, "grad_norm": 3.8004938769216863, "learning_rate": 4.35714174117467e-07, "loss": 0.1466, "step": 12793 }, { "epoch": 1.74, "grad_norm": 3.0264108388677995, "learning_rate": 4.3526454908153934e-07, "loss": 0.1617, "step": 12794 }, { "epoch": 1.74, "grad_norm": 3.2565218254168227, "learning_rate": 4.348151455979355e-07, "loss": 0.1165, "step": 12795 }, { "epoch": 1.74, "grad_norm": 2.9224172472174543, "learning_rate": 4.343659636884634e-07, "loss": 0.1829, "step": 12796 }, { "epoch": 1.74, "grad_norm": 2.8505163476660855, "learning_rate": 4.339170033749279e-07, "loss": 0.1709, "step": 12797 }, { "epoch": 1.74, "grad_norm": 3.968194722809541, "learning_rate": 4.334682646791183e-07, "loss": 0.2115, "step": 12798 }, { "epoch": 1.74, "grad_norm": 2.6482367676977407, "learning_rate": 4.330197476228137e-07, "loss": 0.0907, "step": 12799 }, { "epoch": 1.74, "grad_norm": 4.344457626695032, "learning_rate": 4.325714522277841e-07, "loss": 0.1839, "step": 12800 }, { "epoch": 1.74, "grad_norm": 4.925588780719445, "learning_rate": 4.3212337851578743e-07, "loss": 0.1654, "step": 12801 }, { "epoch": 1.74, "grad_norm": 2.9551587039444134, "learning_rate": 4.316755265085715e-07, "loss": 0.1721, "step": 12802 }, { "epoch": 1.74, "grad_norm": 3.2705592765208986, "learning_rate": 4.3122789622787264e-07, "loss": 0.1699, "step": 12803 }, { "epoch": 1.74, "grad_norm": 3.1742149267070383, "learning_rate": 4.3078048769541745e-07, "loss": 0.1423, "step": 12804 }, { "epoch": 1.74, "grad_norm": 2.973099030258177, "learning_rate": 4.3033330093292026e-07, "loss": 0.1516, "step": 12805 }, { "epoch": 1.74, "grad_norm": 4.143284656677488, "learning_rate": 4.298863359620853e-07, "loss": 0.1574, "step": 12806 }, { "epoch": 1.74, "grad_norm": 3.3443245211670445, "learning_rate": 4.294395928046091e-07, "loss": 0.1635, "step": 12807 }, { "epoch": 1.74, "grad_norm": 2.51368383498774, "learning_rate": 4.2899307148217107e-07, "loss": 0.1644, "step": 12808 }, { "epoch": 1.74, "grad_norm": 2.7056688054381186, "learning_rate": 4.2854677201644657e-07, "loss": 0.1794, "step": 12809 }, { "epoch": 1.74, "grad_norm": 2.788339580341659, "learning_rate": 4.281006944290933e-07, "loss": 0.1421, "step": 12810 }, { "epoch": 1.74, "grad_norm": 3.7111253686801957, "learning_rate": 4.276548387417656e-07, "loss": 0.191, "step": 12811 }, { "epoch": 1.74, "grad_norm": 3.4477090664456225, "learning_rate": 4.272092049761012e-07, "loss": 0.1439, "step": 12812 }, { "epoch": 1.74, "grad_norm": 3.5634627953257554, "learning_rate": 4.2676379315372995e-07, "loss": 0.1679, "step": 12813 }, { "epoch": 1.74, "grad_norm": 2.635398209876367, "learning_rate": 4.263186032962702e-07, "loss": 0.1514, "step": 12814 }, { "epoch": 1.74, "grad_norm": 4.511567307834142, "learning_rate": 4.2587363542532954e-07, "loss": 0.1536, "step": 12815 }, { "epoch": 1.74, "grad_norm": 4.363684552895211, "learning_rate": 4.2542888956250475e-07, "loss": 0.1656, "step": 12816 }, { "epoch": 1.74, "grad_norm": 4.207638083524543, "learning_rate": 4.2498436572938117e-07, "loss": 0.1663, "step": 12817 }, { "epoch": 1.74, "grad_norm": 2.6005700647667775, "learning_rate": 4.2454006394753497e-07, "loss": 0.1669, "step": 12818 }, { "epoch": 1.74, "grad_norm": 4.045345454491429, "learning_rate": 4.2409598423853e-07, "loss": 0.1377, "step": 12819 }, { "epoch": 1.74, "grad_norm": 2.430489040988619, "learning_rate": 4.2365212662392076e-07, "loss": 0.1027, "step": 12820 }, { "epoch": 1.74, "grad_norm": 3.4210600719272746, "learning_rate": 4.2320849112524877e-07, "loss": 0.1582, "step": 12821 }, { "epoch": 1.74, "grad_norm": 2.9535509968797777, "learning_rate": 4.227650777640474e-07, "loss": 0.1146, "step": 12822 }, { "epoch": 1.74, "grad_norm": 3.412603396513941, "learning_rate": 4.223218865618378e-07, "loss": 0.1552, "step": 12823 }, { "epoch": 1.74, "grad_norm": 4.612511992924801, "learning_rate": 4.2187891754013e-07, "loss": 0.1503, "step": 12824 }, { "epoch": 1.74, "grad_norm": 2.983577145192426, "learning_rate": 4.214361707204234e-07, "loss": 0.1782, "step": 12825 }, { "epoch": 1.74, "grad_norm": 3.542732299066318, "learning_rate": 4.2099364612420923e-07, "loss": 0.1357, "step": 12826 }, { "epoch": 1.74, "grad_norm": 3.378256442975458, "learning_rate": 4.2055134377296245e-07, "loss": 0.1515, "step": 12827 }, { "epoch": 1.74, "grad_norm": 3.151304971829911, "learning_rate": 4.201092636881543e-07, "loss": 0.1076, "step": 12828 }, { "epoch": 1.75, "grad_norm": 4.287964432167385, "learning_rate": 4.1966740589123754e-07, "loss": 0.1551, "step": 12829 }, { "epoch": 1.75, "grad_norm": 2.618942811175069, "learning_rate": 4.192257704036601e-07, "loss": 0.1465, "step": 12830 }, { "epoch": 1.75, "grad_norm": 2.9768035059259956, "learning_rate": 4.187843572468575e-07, "loss": 0.1605, "step": 12831 }, { "epoch": 1.75, "grad_norm": 3.0633482899990088, "learning_rate": 4.183431664422527e-07, "loss": 0.1138, "step": 12832 }, { "epoch": 1.75, "grad_norm": 3.2991164285975807, "learning_rate": 4.1790219801125964e-07, "loss": 0.1391, "step": 12833 }, { "epoch": 1.75, "grad_norm": 3.612468818409769, "learning_rate": 4.174614519752812e-07, "loss": 0.1562, "step": 12834 }, { "epoch": 1.75, "grad_norm": 2.4362823309167174, "learning_rate": 4.170209283557092e-07, "loss": 0.153, "step": 12835 }, { "epoch": 1.75, "grad_norm": 3.8039981268328042, "learning_rate": 4.1658062717392476e-07, "loss": 0.1628, "step": 12836 }, { "epoch": 1.75, "grad_norm": 3.121282885960823, "learning_rate": 4.1614054845129814e-07, "loss": 0.1245, "step": 12837 }, { "epoch": 1.75, "grad_norm": 3.5700249499746795, "learning_rate": 4.1570069220918883e-07, "loss": 0.1347, "step": 12838 }, { "epoch": 1.75, "grad_norm": 3.598235867510097, "learning_rate": 4.152610584689443e-07, "loss": 0.17, "step": 12839 }, { "epoch": 1.75, "grad_norm": 3.0588398257784326, "learning_rate": 4.1482164725190564e-07, "loss": 0.1409, "step": 12840 }, { "epoch": 1.75, "grad_norm": 2.659192795467922, "learning_rate": 4.1438245857939595e-07, "loss": 0.1457, "step": 12841 }, { "epoch": 1.75, "grad_norm": 4.314762266038461, "learning_rate": 4.139434924727359e-07, "loss": 0.1759, "step": 12842 }, { "epoch": 1.75, "grad_norm": 3.14769669927848, "learning_rate": 4.135047489532262e-07, "loss": 0.1488, "step": 12843 }, { "epoch": 1.75, "grad_norm": 3.8196906453267285, "learning_rate": 4.130662280421649e-07, "loss": 0.1745, "step": 12844 }, { "epoch": 1.75, "grad_norm": 3.4515500049884666, "learning_rate": 4.1262792976083545e-07, "loss": 0.1247, "step": 12845 }, { "epoch": 1.75, "grad_norm": 6.035299170068039, "learning_rate": 4.121898541305103e-07, "loss": 0.1457, "step": 12846 }, { "epoch": 1.75, "grad_norm": 4.189411410599323, "learning_rate": 4.1175200117245127e-07, "loss": 0.1343, "step": 12847 }, { "epoch": 1.75, "grad_norm": 3.2063445995425814, "learning_rate": 4.113143709079109e-07, "loss": 0.1475, "step": 12848 }, { "epoch": 1.75, "grad_norm": 3.8231618117948525, "learning_rate": 4.108769633581294e-07, "loss": 0.1653, "step": 12849 }, { "epoch": 1.75, "grad_norm": 2.3582315673872114, "learning_rate": 4.1043977854433646e-07, "loss": 0.1427, "step": 12850 }, { "epoch": 1.75, "grad_norm": 2.793348212998718, "learning_rate": 4.1000281648775174e-07, "loss": 0.1339, "step": 12851 }, { "epoch": 1.75, "grad_norm": 3.9432933167927637, "learning_rate": 4.095660772095822e-07, "loss": 0.127, "step": 12852 }, { "epoch": 1.75, "grad_norm": 3.672397312038037, "learning_rate": 4.091295607310258e-07, "loss": 0.1293, "step": 12853 }, { "epoch": 1.75, "grad_norm": 5.339475043543176, "learning_rate": 4.086932670732707e-07, "loss": 0.1508, "step": 12854 }, { "epoch": 1.75, "grad_norm": 2.1753065170612955, "learning_rate": 4.0825719625748994e-07, "loss": 0.0975, "step": 12855 }, { "epoch": 1.75, "grad_norm": 3.347339067090264, "learning_rate": 4.078213483048515e-07, "loss": 0.1532, "step": 12856 }, { "epoch": 1.75, "grad_norm": 4.798864389503527, "learning_rate": 4.0738572323650636e-07, "loss": 0.1182, "step": 12857 }, { "epoch": 1.75, "grad_norm": 4.047973570736115, "learning_rate": 4.069503210736003e-07, "loss": 0.1789, "step": 12858 }, { "epoch": 1.75, "grad_norm": 2.577595520838833, "learning_rate": 4.0651514183726535e-07, "loss": 0.1094, "step": 12859 }, { "epoch": 1.75, "grad_norm": 3.62128618603428, "learning_rate": 4.0608018554862237e-07, "loss": 0.1077, "step": 12860 }, { "epoch": 1.75, "grad_norm": 3.040378905640199, "learning_rate": 4.056454522287839e-07, "loss": 0.1465, "step": 12861 }, { "epoch": 1.75, "grad_norm": 3.678126499330609, "learning_rate": 4.05210941898847e-07, "loss": 0.1499, "step": 12862 }, { "epoch": 1.75, "grad_norm": 3.6261704843656424, "learning_rate": 4.047766545799037e-07, "loss": 0.1493, "step": 12863 }, { "epoch": 1.75, "grad_norm": 2.843008123987075, "learning_rate": 4.043425902930315e-07, "loss": 0.1413, "step": 12864 }, { "epoch": 1.75, "grad_norm": 4.025916978155658, "learning_rate": 4.039087490592974e-07, "loss": 0.1722, "step": 12865 }, { "epoch": 1.75, "grad_norm": 3.4565198219317224, "learning_rate": 4.034751308997592e-07, "loss": 0.1483, "step": 12866 }, { "epoch": 1.75, "grad_norm": 5.839576670964853, "learning_rate": 4.0304173583546214e-07, "loss": 0.1668, "step": 12867 }, { "epoch": 1.75, "grad_norm": 3.9845935092729645, "learning_rate": 4.026085638874416e-07, "loss": 0.1712, "step": 12868 }, { "epoch": 1.75, "grad_norm": 2.799847685016618, "learning_rate": 4.0217561507672143e-07, "loss": 0.1799, "step": 12869 }, { "epoch": 1.75, "grad_norm": 3.61814546128876, "learning_rate": 4.0174288942431585e-07, "loss": 0.1425, "step": 12870 }, { "epoch": 1.75, "grad_norm": 3.071668365226482, "learning_rate": 4.013103869512269e-07, "loss": 0.1471, "step": 12871 }, { "epoch": 1.75, "grad_norm": 4.132145675560085, "learning_rate": 4.008781076784457e-07, "loss": 0.1447, "step": 12872 }, { "epoch": 1.75, "grad_norm": 3.893386670766784, "learning_rate": 4.004460516269554e-07, "loss": 0.1724, "step": 12873 }, { "epoch": 1.75, "grad_norm": 3.4943646664208097, "learning_rate": 4.0001421881772306e-07, "loss": 0.1417, "step": 12874 }, { "epoch": 1.75, "grad_norm": 4.407328399002605, "learning_rate": 3.9958260927171146e-07, "loss": 0.1612, "step": 12875 }, { "epoch": 1.75, "grad_norm": 3.9190442261958376, "learning_rate": 3.991512230098654e-07, "loss": 0.1269, "step": 12876 }, { "epoch": 1.75, "grad_norm": 3.3570016819609942, "learning_rate": 3.9872006005312545e-07, "loss": 0.1312, "step": 12877 }, { "epoch": 1.75, "grad_norm": 3.5779701250907774, "learning_rate": 3.9828912042241653e-07, "loss": 0.1358, "step": 12878 }, { "epoch": 1.75, "grad_norm": 3.4174338305690273, "learning_rate": 3.978584041386557e-07, "loss": 0.1374, "step": 12879 }, { "epoch": 1.75, "grad_norm": 3.8789169946714113, "learning_rate": 3.974279112227475e-07, "loss": 0.1477, "step": 12880 }, { "epoch": 1.75, "grad_norm": 2.6081551718311213, "learning_rate": 3.969976416955862e-07, "loss": 0.12, "step": 12881 }, { "epoch": 1.75, "grad_norm": 2.652416024237806, "learning_rate": 3.965675955780551e-07, "loss": 0.1487, "step": 12882 }, { "epoch": 1.75, "grad_norm": 3.3693451959250873, "learning_rate": 3.961377728910276e-07, "loss": 0.1372, "step": 12883 }, { "epoch": 1.75, "grad_norm": 3.922572911541994, "learning_rate": 3.957081736553631e-07, "loss": 0.1352, "step": 12884 }, { "epoch": 1.75, "grad_norm": 3.338072847385398, "learning_rate": 3.952787978919159e-07, "loss": 0.1684, "step": 12885 }, { "epoch": 1.75, "grad_norm": 2.962177445268759, "learning_rate": 3.948496456215228e-07, "loss": 0.1373, "step": 12886 }, { "epoch": 1.75, "grad_norm": 2.8696801713066034, "learning_rate": 3.9442071686501605e-07, "loss": 0.1087, "step": 12887 }, { "epoch": 1.75, "grad_norm": 4.057885272328229, "learning_rate": 3.9399201164321055e-07, "loss": 0.2144, "step": 12888 }, { "epoch": 1.75, "grad_norm": 2.919929230705662, "learning_rate": 3.935635299769175e-07, "loss": 0.0967, "step": 12889 }, { "epoch": 1.75, "grad_norm": 3.1428188870530067, "learning_rate": 3.9313527188692966e-07, "loss": 0.137, "step": 12890 }, { "epoch": 1.75, "grad_norm": 3.60594501813947, "learning_rate": 3.927072373940355e-07, "loss": 0.1607, "step": 12891 }, { "epoch": 1.75, "grad_norm": 5.00774613232419, "learning_rate": 3.9227942651900943e-07, "loss": 0.1546, "step": 12892 }, { "epoch": 1.75, "grad_norm": 3.221794094856114, "learning_rate": 3.918518392826154e-07, "loss": 0.1848, "step": 12893 }, { "epoch": 1.75, "grad_norm": 3.063863274822255, "learning_rate": 3.9142447570560627e-07, "loss": 0.1511, "step": 12894 }, { "epoch": 1.75, "grad_norm": 3.0368830378469265, "learning_rate": 3.9099733580872436e-07, "loss": 0.1397, "step": 12895 }, { "epoch": 1.75, "grad_norm": 4.588016516863192, "learning_rate": 3.9057041961270194e-07, "loss": 0.1756, "step": 12896 }, { "epoch": 1.75, "grad_norm": 3.059176875180293, "learning_rate": 3.901437271382591e-07, "loss": 0.1203, "step": 12897 }, { "epoch": 1.75, "grad_norm": 4.142440343067998, "learning_rate": 3.897172584061043e-07, "loss": 0.1445, "step": 12898 }, { "epoch": 1.75, "grad_norm": 2.8304865956818053, "learning_rate": 3.8929101343693986e-07, "loss": 0.13, "step": 12899 }, { "epoch": 1.75, "grad_norm": 3.1564372582675566, "learning_rate": 3.888649922514515e-07, "loss": 0.1853, "step": 12900 }, { "epoch": 1.75, "grad_norm": 3.4668511541299964, "learning_rate": 3.8843919487031644e-07, "loss": 0.1538, "step": 12901 }, { "epoch": 1.76, "grad_norm": 2.7693498845393476, "learning_rate": 3.8801362131420105e-07, "loss": 0.132, "step": 12902 }, { "epoch": 1.76, "grad_norm": 2.9529262754243315, "learning_rate": 3.8758827160376156e-07, "loss": 0.1804, "step": 12903 }, { "epoch": 1.76, "grad_norm": 3.0085244229065937, "learning_rate": 3.8716314575964197e-07, "loss": 0.1661, "step": 12904 }, { "epoch": 1.76, "grad_norm": 6.1661952963034405, "learning_rate": 3.8673824380247527e-07, "loss": 0.1639, "step": 12905 }, { "epoch": 1.76, "grad_norm": 3.673131522381376, "learning_rate": 3.863135657528871e-07, "loss": 0.1646, "step": 12906 }, { "epoch": 1.76, "grad_norm": 7.705562485618195, "learning_rate": 3.858891116314861e-07, "loss": 0.1436, "step": 12907 }, { "epoch": 1.76, "grad_norm": 4.679739461933187, "learning_rate": 3.8546488145887627e-07, "loss": 0.1885, "step": 12908 }, { "epoch": 1.76, "grad_norm": 2.809639129140267, "learning_rate": 3.8504087525564504e-07, "loss": 0.1465, "step": 12909 }, { "epoch": 1.76, "grad_norm": 3.4134829770721087, "learning_rate": 3.8461709304237484e-07, "loss": 0.1277, "step": 12910 }, { "epoch": 1.76, "grad_norm": 4.024482601277629, "learning_rate": 3.8419353483963195e-07, "loss": 0.1722, "step": 12911 }, { "epoch": 1.76, "grad_norm": 3.4212419096524305, "learning_rate": 3.8377020066797557e-07, "loss": 0.1592, "step": 12912 }, { "epoch": 1.76, "grad_norm": 2.9957600754525044, "learning_rate": 3.8334709054795136e-07, "loss": 0.1086, "step": 12913 }, { "epoch": 1.76, "grad_norm": 3.298434854964663, "learning_rate": 3.829242045000953e-07, "loss": 0.1418, "step": 12914 }, { "epoch": 1.76, "grad_norm": 4.850704195161271, "learning_rate": 3.8250154254493354e-07, "loss": 0.1477, "step": 12915 }, { "epoch": 1.76, "grad_norm": 3.693051380492536, "learning_rate": 3.820791047029787e-07, "loss": 0.1525, "step": 12916 }, { "epoch": 1.76, "grad_norm": 2.682229050910223, "learning_rate": 3.8165689099473436e-07, "loss": 0.1872, "step": 12917 }, { "epoch": 1.76, "grad_norm": 3.013623039671439, "learning_rate": 3.8123490144069517e-07, "loss": 0.1583, "step": 12918 }, { "epoch": 1.76, "grad_norm": 4.563299115173682, "learning_rate": 3.808131360613393e-07, "loss": 0.1847, "step": 12919 }, { "epoch": 1.76, "grad_norm": 3.682088860792354, "learning_rate": 3.803915948771403e-07, "loss": 0.1587, "step": 12920 }, { "epoch": 1.76, "grad_norm": 2.6455216673431643, "learning_rate": 3.799702779085557e-07, "loss": 0.141, "step": 12921 }, { "epoch": 1.76, "grad_norm": 3.351514264750131, "learning_rate": 3.7954918517603636e-07, "loss": 0.2129, "step": 12922 }, { "epoch": 1.76, "grad_norm": 3.6487636150099663, "learning_rate": 3.791283167000176e-07, "loss": 0.1639, "step": 12923 }, { "epoch": 1.76, "grad_norm": 3.691258031262292, "learning_rate": 3.787076725009292e-07, "loss": 0.1407, "step": 12924 }, { "epoch": 1.76, "grad_norm": 2.5829536453419912, "learning_rate": 3.782872525991865e-07, "loss": 0.1316, "step": 12925 }, { "epoch": 1.76, "grad_norm": 2.8880313485570652, "learning_rate": 3.7786705701519434e-07, "loss": 0.1343, "step": 12926 }, { "epoch": 1.76, "grad_norm": 3.5123464293618825, "learning_rate": 3.7744708576934795e-07, "loss": 0.1782, "step": 12927 }, { "epoch": 1.76, "grad_norm": 4.284023345763442, "learning_rate": 3.7702733888203003e-07, "loss": 0.1526, "step": 12928 }, { "epoch": 1.76, "grad_norm": 3.805335260709425, "learning_rate": 3.766078163736142e-07, "loss": 0.1535, "step": 12929 }, { "epoch": 1.76, "grad_norm": 3.73691841263418, "learning_rate": 3.761885182644615e-07, "loss": 0.1411, "step": 12930 }, { "epoch": 1.76, "grad_norm": 3.7805436336331053, "learning_rate": 3.757694445749227e-07, "loss": 0.1692, "step": 12931 }, { "epoch": 1.76, "grad_norm": 3.1297098970363026, "learning_rate": 3.7535059532533945e-07, "loss": 0.1264, "step": 12932 }, { "epoch": 1.76, "grad_norm": 4.021611585061382, "learning_rate": 3.7493197053603827e-07, "loss": 0.1723, "step": 12933 }, { "epoch": 1.76, "grad_norm": 3.3212491166793634, "learning_rate": 3.7451357022734e-07, "loss": 0.1429, "step": 12934 }, { "epoch": 1.76, "grad_norm": 3.4306349274684256, "learning_rate": 3.740953944195497e-07, "loss": 0.1666, "step": 12935 }, { "epoch": 1.76, "grad_norm": 3.0155324144738347, "learning_rate": 3.7367744313296536e-07, "loss": 0.1512, "step": 12936 }, { "epoch": 1.76, "grad_norm": 2.8694391479158634, "learning_rate": 3.732597163878715e-07, "loss": 0.0985, "step": 12937 }, { "epoch": 1.76, "grad_norm": 4.160326535474042, "learning_rate": 3.7284221420454404e-07, "loss": 0.1449, "step": 12938 }, { "epoch": 1.76, "grad_norm": 4.315430130023734, "learning_rate": 3.7242493660324507e-07, "loss": 0.1696, "step": 12939 }, { "epoch": 1.76, "grad_norm": 5.004466469033155, "learning_rate": 3.72007883604229e-07, "loss": 0.1615, "step": 12940 }, { "epoch": 1.76, "grad_norm": 4.543282972005459, "learning_rate": 3.7159105522773743e-07, "loss": 0.1801, "step": 12941 }, { "epoch": 1.76, "grad_norm": 3.706284699919632, "learning_rate": 3.711744514939991e-07, "loss": 0.1764, "step": 12942 }, { "epoch": 1.76, "grad_norm": 4.140664138416094, "learning_rate": 3.707580724232368e-07, "loss": 0.1886, "step": 12943 }, { "epoch": 1.76, "grad_norm": 3.058354342617318, "learning_rate": 3.7034191803565933e-07, "loss": 0.1815, "step": 12944 }, { "epoch": 1.76, "grad_norm": 2.726441925941312, "learning_rate": 3.6992598835146445e-07, "loss": 0.1264, "step": 12945 }, { "epoch": 1.76, "grad_norm": 2.9106944642318444, "learning_rate": 3.6951028339083985e-07, "loss": 0.1595, "step": 12946 }, { "epoch": 1.76, "grad_norm": 3.2929418458802346, "learning_rate": 3.690948031739622e-07, "loss": 0.166, "step": 12947 }, { "epoch": 1.76, "grad_norm": 3.1056073808439137, "learning_rate": 3.686795477209959e-07, "loss": 0.1213, "step": 12948 }, { "epoch": 1.76, "grad_norm": 3.4388232713223874, "learning_rate": 3.682645170520976e-07, "loss": 0.1436, "step": 12949 }, { "epoch": 1.76, "grad_norm": 3.2084477880957043, "learning_rate": 3.6784971118740854e-07, "loss": 0.1355, "step": 12950 }, { "epoch": 1.76, "grad_norm": 4.016101414333543, "learning_rate": 3.6743513014706464e-07, "loss": 0.1455, "step": 12951 }, { "epoch": 1.76, "grad_norm": 5.076620414712262, "learning_rate": 3.67020773951185e-07, "loss": 0.1511, "step": 12952 }, { "epoch": 1.76, "grad_norm": 2.6220614371485054, "learning_rate": 3.666066426198839e-07, "loss": 0.1174, "step": 12953 }, { "epoch": 1.76, "grad_norm": 4.391890535202303, "learning_rate": 3.66192736173257e-07, "loss": 0.1474, "step": 12954 }, { "epoch": 1.76, "grad_norm": 3.2447878677972755, "learning_rate": 3.657790546313983e-07, "loss": 0.1261, "step": 12955 }, { "epoch": 1.76, "grad_norm": 3.5483308714329467, "learning_rate": 3.653655980143822e-07, "loss": 0.1625, "step": 12956 }, { "epoch": 1.76, "grad_norm": 2.3562279656609575, "learning_rate": 3.649523663422783e-07, "loss": 0.0947, "step": 12957 }, { "epoch": 1.76, "grad_norm": 2.888171759431341, "learning_rate": 3.6453935963514264e-07, "loss": 0.1386, "step": 12958 }, { "epoch": 1.76, "grad_norm": 3.1016725693618596, "learning_rate": 3.641265779130204e-07, "loss": 0.1176, "step": 12959 }, { "epoch": 1.76, "grad_norm": 3.9359994434473244, "learning_rate": 3.6371402119594667e-07, "loss": 0.1442, "step": 12960 }, { "epoch": 1.76, "grad_norm": 4.647147623935033, "learning_rate": 3.633016895039443e-07, "loss": 0.201, "step": 12961 }, { "epoch": 1.76, "grad_norm": 3.681950521682162, "learning_rate": 3.6288958285702726e-07, "loss": 0.1478, "step": 12962 }, { "epoch": 1.76, "grad_norm": 4.037474444440159, "learning_rate": 3.624777012751962e-07, "loss": 0.1543, "step": 12963 }, { "epoch": 1.76, "grad_norm": 3.0953826531173227, "learning_rate": 3.6206604477844245e-07, "loss": 0.1209, "step": 12964 }, { "epoch": 1.76, "grad_norm": 4.334976788427056, "learning_rate": 3.616546133867471e-07, "loss": 0.1217, "step": 12965 }, { "epoch": 1.76, "grad_norm": 3.2926418809002174, "learning_rate": 3.612434071200771e-07, "loss": 0.1412, "step": 12966 }, { "epoch": 1.76, "grad_norm": 3.7124080889824027, "learning_rate": 3.6083242599839365e-07, "loss": 0.1686, "step": 12967 }, { "epoch": 1.76, "grad_norm": 5.385808618808439, "learning_rate": 3.604216700416402e-07, "loss": 0.1711, "step": 12968 }, { "epoch": 1.76, "grad_norm": 3.768732502978412, "learning_rate": 3.600111392697564e-07, "loss": 0.1342, "step": 12969 }, { "epoch": 1.76, "grad_norm": 5.076256395847555, "learning_rate": 3.5960083370266573e-07, "loss": 0.1823, "step": 12970 }, { "epoch": 1.76, "grad_norm": 2.6451326526846164, "learning_rate": 3.591907533602834e-07, "loss": 0.1228, "step": 12971 }, { "epoch": 1.76, "grad_norm": 4.434393345982781, "learning_rate": 3.587808982625124e-07, "loss": 0.1944, "step": 12972 }, { "epoch": 1.76, "grad_norm": 5.642714520421332, "learning_rate": 3.5837126842924564e-07, "loss": 0.168, "step": 12973 }, { "epoch": 1.76, "grad_norm": 3.0348220479354833, "learning_rate": 3.57961863880365e-07, "loss": 0.1113, "step": 12974 }, { "epoch": 1.76, "grad_norm": 2.7770854663391553, "learning_rate": 3.575526846357402e-07, "loss": 0.1427, "step": 12975 }, { "epoch": 1.77, "grad_norm": 2.780488334975641, "learning_rate": 3.5714373071523257e-07, "loss": 0.1142, "step": 12976 }, { "epoch": 1.77, "grad_norm": 4.928969730428315, "learning_rate": 3.567350021386895e-07, "loss": 0.1167, "step": 12977 }, { "epoch": 1.77, "grad_norm": 3.224493539554428, "learning_rate": 3.563264989259485e-07, "loss": 0.1263, "step": 12978 }, { "epoch": 1.77, "grad_norm": 4.7356533694403105, "learning_rate": 3.559182210968398e-07, "loss": 0.1664, "step": 12979 }, { "epoch": 1.77, "grad_norm": 3.3128885740426632, "learning_rate": 3.5551016867117595e-07, "loss": 0.1605, "step": 12980 }, { "epoch": 1.77, "grad_norm": 2.9091696302471566, "learning_rate": 3.5510234166876265e-07, "loss": 0.1457, "step": 12981 }, { "epoch": 1.77, "grad_norm": 2.282862110059961, "learning_rate": 3.546947401093953e-07, "loss": 0.1098, "step": 12982 }, { "epoch": 1.77, "grad_norm": 7.3244885429060105, "learning_rate": 3.5428736401285526e-07, "loss": 0.14, "step": 12983 }, { "epoch": 1.77, "grad_norm": 3.1864499121169825, "learning_rate": 3.5388021339891776e-07, "loss": 0.19, "step": 12984 }, { "epoch": 1.77, "grad_norm": 2.841763838048385, "learning_rate": 3.534732882873404e-07, "loss": 0.1267, "step": 12985 }, { "epoch": 1.77, "grad_norm": 3.1330253209197854, "learning_rate": 3.5306658869787736e-07, "loss": 0.1508, "step": 12986 }, { "epoch": 1.77, "grad_norm": 3.5566214387469928, "learning_rate": 3.5266011465026394e-07, "loss": 0.1623, "step": 12987 }, { "epoch": 1.77, "grad_norm": 5.1299180824031065, "learning_rate": 3.5225386616423216e-07, "loss": 0.175, "step": 12988 }, { "epoch": 1.77, "grad_norm": 4.090629382517436, "learning_rate": 3.5184784325949794e-07, "loss": 0.1402, "step": 12989 }, { "epoch": 1.77, "grad_norm": 3.5856558301741717, "learning_rate": 3.5144204595576824e-07, "loss": 0.1572, "step": 12990 }, { "epoch": 1.77, "grad_norm": 3.0617181632200414, "learning_rate": 3.5103647427273843e-07, "loss": 0.1418, "step": 12991 }, { "epoch": 1.77, "grad_norm": 3.6924522030735396, "learning_rate": 3.506311282300934e-07, "loss": 0.1577, "step": 12992 }, { "epoch": 1.77, "grad_norm": 3.3470742944269785, "learning_rate": 3.502260078475062e-07, "loss": 0.1471, "step": 12993 }, { "epoch": 1.77, "grad_norm": 3.0995984927829516, "learning_rate": 3.4982111314464105e-07, "loss": 0.1472, "step": 12994 }, { "epoch": 1.77, "grad_norm": 3.9440935174020555, "learning_rate": 3.494164441411485e-07, "loss": 0.1801, "step": 12995 }, { "epoch": 1.77, "grad_norm": 3.497954702721681, "learning_rate": 3.490120008566694e-07, "loss": 0.121, "step": 12996 }, { "epoch": 1.77, "grad_norm": 3.4245307494558004, "learning_rate": 3.486077833108342e-07, "loss": 0.1288, "step": 12997 }, { "epoch": 1.77, "grad_norm": 3.368773713838767, "learning_rate": 3.482037915232628e-07, "loss": 0.1796, "step": 12998 }, { "epoch": 1.77, "grad_norm": 3.0224614272239343, "learning_rate": 3.4780002551356054e-07, "loss": 0.1858, "step": 12999 }, { "epoch": 1.77, "grad_norm": 4.480945178196552, "learning_rate": 3.473964853013273e-07, "loss": 0.1435, "step": 13000 }, { "epoch": 1.77, "grad_norm": 2.6840401588187195, "learning_rate": 3.469931709061469e-07, "loss": 0.1157, "step": 13001 }, { "epoch": 1.77, "grad_norm": 4.461247997521023, "learning_rate": 3.4659008234759597e-07, "loss": 0.1374, "step": 13002 }, { "epoch": 1.77, "grad_norm": 3.0129570314355303, "learning_rate": 3.461872196452382e-07, "loss": 0.1492, "step": 13003 }, { "epoch": 1.77, "grad_norm": 4.075780500547376, "learning_rate": 3.457845828186268e-07, "loss": 0.1749, "step": 13004 }, { "epoch": 1.77, "grad_norm": 3.2253364409247225, "learning_rate": 3.4538217188730404e-07, "loss": 0.1262, "step": 13005 }, { "epoch": 1.77, "grad_norm": 3.6334681746709454, "learning_rate": 3.449799868708015e-07, "loss": 0.1753, "step": 13006 }, { "epoch": 1.77, "grad_norm": 3.8687709629802964, "learning_rate": 3.4457802778863846e-07, "loss": 0.1704, "step": 13007 }, { "epoch": 1.77, "grad_norm": 3.1985338943020554, "learning_rate": 3.44176294660325e-07, "loss": 0.1535, "step": 13008 }, { "epoch": 1.77, "grad_norm": 3.5342041494782155, "learning_rate": 3.437747875053593e-07, "loss": 0.1147, "step": 13009 }, { "epoch": 1.77, "grad_norm": 2.9306421066697577, "learning_rate": 3.433735063432292e-07, "loss": 0.1333, "step": 13010 }, { "epoch": 1.77, "grad_norm": 3.1007965602676424, "learning_rate": 3.429724511934096e-07, "loss": 0.1355, "step": 13011 }, { "epoch": 1.77, "grad_norm": 3.883897711071021, "learning_rate": 3.4257162207536887e-07, "loss": 0.1478, "step": 13012 }, { "epoch": 1.77, "grad_norm": 5.154615142373456, "learning_rate": 3.421710190085581e-07, "loss": 0.1676, "step": 13013 }, { "epoch": 1.77, "grad_norm": 3.3753058036494568, "learning_rate": 3.417706420124245e-07, "loss": 0.1604, "step": 13014 }, { "epoch": 1.77, "grad_norm": 3.444397591031855, "learning_rate": 3.413704911063964e-07, "loss": 0.134, "step": 13015 }, { "epoch": 1.77, "grad_norm": 3.659798677628594, "learning_rate": 3.4097056630989887e-07, "loss": 0.1774, "step": 13016 }, { "epoch": 1.77, "grad_norm": 3.553521057834371, "learning_rate": 3.405708676423408e-07, "loss": 0.166, "step": 13017 }, { "epoch": 1.77, "grad_norm": 3.38749482520944, "learning_rate": 3.401713951231228e-07, "loss": 0.1207, "step": 13018 }, { "epoch": 1.77, "grad_norm": 3.135262968861972, "learning_rate": 3.3977214877163266e-07, "loss": 0.1175, "step": 13019 }, { "epoch": 1.77, "grad_norm": 3.0150428969582967, "learning_rate": 3.3937312860724826e-07, "loss": 0.1048, "step": 13020 }, { "epoch": 1.77, "grad_norm": 4.213683584261016, "learning_rate": 3.389743346493368e-07, "loss": 0.181, "step": 13021 }, { "epoch": 1.77, "grad_norm": 2.5763084391929425, "learning_rate": 3.3857576691725346e-07, "loss": 0.1427, "step": 13022 }, { "epoch": 1.77, "grad_norm": 3.0601880656803404, "learning_rate": 3.381774254303427e-07, "loss": 0.1438, "step": 13023 }, { "epoch": 1.77, "grad_norm": 3.210705589659284, "learning_rate": 3.377793102079391e-07, "loss": 0.1674, "step": 13024 }, { "epoch": 1.77, "grad_norm": 3.045964191259625, "learning_rate": 3.373814212693649e-07, "loss": 0.1727, "step": 13025 }, { "epoch": 1.77, "grad_norm": 2.4926877712287667, "learning_rate": 3.3698375863393197e-07, "loss": 0.1236, "step": 13026 }, { "epoch": 1.77, "grad_norm": 4.246335004262558, "learning_rate": 3.365863223209409e-07, "loss": 0.1463, "step": 13027 }, { "epoch": 1.77, "grad_norm": 3.9634415165108443, "learning_rate": 3.361891123496824e-07, "loss": 0.1755, "step": 13028 }, { "epoch": 1.77, "grad_norm": 3.6699842579504347, "learning_rate": 3.3579212873943447e-07, "loss": 0.1248, "step": 13029 }, { "epoch": 1.77, "grad_norm": 4.879304513925246, "learning_rate": 3.353953715094643e-07, "loss": 0.1679, "step": 13030 }, { "epoch": 1.77, "grad_norm": 3.6764367839819307, "learning_rate": 3.349988406790311e-07, "loss": 0.1416, "step": 13031 }, { "epoch": 1.77, "grad_norm": 3.6600359750970437, "learning_rate": 3.3460253626737774e-07, "loss": 0.1583, "step": 13032 }, { "epoch": 1.77, "grad_norm": 2.6218092549280603, "learning_rate": 3.342064582937421e-07, "loss": 0.1292, "step": 13033 }, { "epoch": 1.77, "grad_norm": 3.528014834031398, "learning_rate": 3.3381060677734556e-07, "loss": 0.164, "step": 13034 }, { "epoch": 1.77, "grad_norm": 3.104443316747667, "learning_rate": 3.3341498173740214e-07, "loss": 0.1433, "step": 13035 }, { "epoch": 1.77, "grad_norm": 4.313465809538259, "learning_rate": 3.330195831931143e-07, "loss": 0.1605, "step": 13036 }, { "epoch": 1.77, "grad_norm": 4.2511332999429845, "learning_rate": 3.3262441116367174e-07, "loss": 0.1268, "step": 13037 }, { "epoch": 1.77, "grad_norm": 3.5980594074757066, "learning_rate": 3.3222946566825565e-07, "loss": 0.1299, "step": 13038 }, { "epoch": 1.77, "grad_norm": 3.2944076707060703, "learning_rate": 3.318347467260341e-07, "loss": 0.1646, "step": 13039 }, { "epoch": 1.77, "grad_norm": 3.190347618752408, "learning_rate": 3.314402543561651e-07, "loss": 0.1067, "step": 13040 }, { "epoch": 1.77, "grad_norm": 3.5218689967607832, "learning_rate": 3.310459885777956e-07, "loss": 0.1526, "step": 13041 }, { "epoch": 1.77, "grad_norm": 3.3860288035787796, "learning_rate": 3.306519494100618e-07, "loss": 0.1329, "step": 13042 }, { "epoch": 1.77, "grad_norm": 3.004552115218267, "learning_rate": 3.302581368720881e-07, "loss": 0.1598, "step": 13043 }, { "epoch": 1.77, "grad_norm": 5.247219238537878, "learning_rate": 3.298645509829884e-07, "loss": 0.122, "step": 13044 }, { "epoch": 1.77, "grad_norm": 3.66285237672116, "learning_rate": 3.294711917618676e-07, "loss": 0.1563, "step": 13045 }, { "epoch": 1.77, "grad_norm": 4.127505063497186, "learning_rate": 3.290780592278148e-07, "loss": 0.1482, "step": 13046 }, { "epoch": 1.77, "grad_norm": 3.8542899002917195, "learning_rate": 3.286851533999136e-07, "loss": 0.1515, "step": 13047 }, { "epoch": 1.77, "grad_norm": 4.096171038366063, "learning_rate": 3.2829247429723097e-07, "loss": 0.1708, "step": 13048 }, { "epoch": 1.78, "grad_norm": 3.317097727374147, "learning_rate": 3.2790002193882844e-07, "loss": 0.1272, "step": 13049 }, { "epoch": 1.78, "grad_norm": 3.3951131841122444, "learning_rate": 3.2750779634375286e-07, "loss": 0.1492, "step": 13050 }, { "epoch": 1.78, "grad_norm": 4.723264320555418, "learning_rate": 3.271157975310407e-07, "loss": 0.1837, "step": 13051 }, { "epoch": 1.78, "grad_norm": 3.4319750408982355, "learning_rate": 3.2672402551971903e-07, "loss": 0.1608, "step": 13052 }, { "epoch": 1.78, "grad_norm": 3.874410311271244, "learning_rate": 3.26332480328802e-07, "loss": 0.1508, "step": 13053 }, { "epoch": 1.78, "grad_norm": 3.3833785088045185, "learning_rate": 3.2594116197729343e-07, "loss": 0.1826, "step": 13054 }, { "epoch": 1.78, "grad_norm": 3.2766520579800953, "learning_rate": 3.255500704841863e-07, "loss": 0.1658, "step": 13055 }, { "epoch": 1.78, "grad_norm": 2.822245141169486, "learning_rate": 3.2515920586846273e-07, "loss": 0.1225, "step": 13056 }, { "epoch": 1.78, "grad_norm": 5.360611227042005, "learning_rate": 3.2476856814909364e-07, "loss": 0.1594, "step": 13057 }, { "epoch": 1.78, "grad_norm": 3.7947218361518344, "learning_rate": 3.2437815734503774e-07, "loss": 0.1139, "step": 13058 }, { "epoch": 1.78, "grad_norm": 4.045573936497064, "learning_rate": 3.2398797347524656e-07, "loss": 0.1295, "step": 13059 }, { "epoch": 1.78, "grad_norm": 4.6708786444788535, "learning_rate": 3.23598016558655e-07, "loss": 0.1482, "step": 13060 }, { "epoch": 1.78, "grad_norm": 3.1826131221524636, "learning_rate": 3.2320828661419113e-07, "loss": 0.1311, "step": 13061 }, { "epoch": 1.78, "grad_norm": 3.150655419307079, "learning_rate": 3.2281878366077046e-07, "loss": 0.1573, "step": 13062 }, { "epoch": 1.78, "grad_norm": 3.4521090109597856, "learning_rate": 3.2242950771729785e-07, "loss": 0.1587, "step": 13063 }, { "epoch": 1.78, "grad_norm": 3.9227317071849854, "learning_rate": 3.2204045880266823e-07, "loss": 0.1601, "step": 13064 }, { "epoch": 1.78, "grad_norm": 2.603279059401244, "learning_rate": 3.216516369357614e-07, "loss": 0.1465, "step": 13065 }, { "epoch": 1.78, "grad_norm": 4.485680028520616, "learning_rate": 3.212630421354529e-07, "loss": 0.1271, "step": 13066 }, { "epoch": 1.78, "grad_norm": 3.134360178396366, "learning_rate": 3.208746744205998e-07, "loss": 0.1359, "step": 13067 }, { "epoch": 1.78, "grad_norm": 2.492464009355689, "learning_rate": 3.2048653381005377e-07, "loss": 0.1342, "step": 13068 }, { "epoch": 1.78, "grad_norm": 4.636756488141402, "learning_rate": 3.200986203226536e-07, "loss": 0.1583, "step": 13069 }, { "epoch": 1.78, "grad_norm": 3.608822426813502, "learning_rate": 3.197109339772264e-07, "loss": 0.1466, "step": 13070 }, { "epoch": 1.78, "grad_norm": 3.8124816922978644, "learning_rate": 3.193234747925883e-07, "loss": 0.1778, "step": 13071 }, { "epoch": 1.78, "grad_norm": 4.306735398944874, "learning_rate": 3.1893624278754587e-07, "loss": 0.1593, "step": 13072 }, { "epoch": 1.78, "grad_norm": 7.499597826843081, "learning_rate": 3.18549237980893e-07, "loss": 0.1541, "step": 13073 }, { "epoch": 1.78, "grad_norm": 2.9916701415546862, "learning_rate": 3.181624603914135e-07, "loss": 0.1303, "step": 13074 }, { "epoch": 1.78, "grad_norm": 2.6527125623643855, "learning_rate": 3.1777591003787965e-07, "loss": 0.1221, "step": 13075 }, { "epoch": 1.78, "grad_norm": 3.1731811455975065, "learning_rate": 3.1738958693905306e-07, "loss": 0.1002, "step": 13076 }, { "epoch": 1.78, "grad_norm": 3.0553052257164532, "learning_rate": 3.170034911136832e-07, "loss": 0.1705, "step": 13077 }, { "epoch": 1.78, "grad_norm": 2.752880946560536, "learning_rate": 3.1661762258051174e-07, "loss": 0.1676, "step": 13078 }, { "epoch": 1.78, "grad_norm": 5.767939829185425, "learning_rate": 3.162319813582643e-07, "loss": 0.1573, "step": 13079 }, { "epoch": 1.78, "grad_norm": 2.69952420136485, "learning_rate": 3.1584656746566134e-07, "loss": 0.1359, "step": 13080 }, { "epoch": 1.78, "grad_norm": 3.4770667385925247, "learning_rate": 3.1546138092140575e-07, "loss": 0.1164, "step": 13081 }, { "epoch": 1.78, "grad_norm": 3.5516474900535333, "learning_rate": 3.150764217441954e-07, "loss": 0.1654, "step": 13082 }, { "epoch": 1.78, "grad_norm": 4.108019317544951, "learning_rate": 3.146916899527136e-07, "loss": 0.1556, "step": 13083 }, { "epoch": 1.78, "grad_norm": 3.7522659948685853, "learning_rate": 3.143071855656338e-07, "loss": 0.1576, "step": 13084 }, { "epoch": 1.78, "grad_norm": 3.2321671653255324, "learning_rate": 3.1392290860161765e-07, "loss": 0.1537, "step": 13085 }, { "epoch": 1.78, "grad_norm": 2.878344681743226, "learning_rate": 3.1353885907931646e-07, "loss": 0.126, "step": 13086 }, { "epoch": 1.78, "grad_norm": 4.36853220808556, "learning_rate": 3.131550370173703e-07, "loss": 0.1604, "step": 13087 }, { "epoch": 1.78, "grad_norm": 2.492259713602177, "learning_rate": 3.127714424344086e-07, "loss": 0.158, "step": 13088 }, { "epoch": 1.78, "grad_norm": 2.7448813522685858, "learning_rate": 3.1238807534904937e-07, "loss": 0.1365, "step": 13089 }, { "epoch": 1.78, "grad_norm": 3.441785220660602, "learning_rate": 3.1200493577989875e-07, "loss": 0.1277, "step": 13090 }, { "epoch": 1.78, "grad_norm": 3.66075018231124, "learning_rate": 3.116220237455525e-07, "loss": 0.1573, "step": 13091 }, { "epoch": 1.78, "grad_norm": 3.3149960322158214, "learning_rate": 3.112393392645985e-07, "loss": 0.1654, "step": 13092 }, { "epoch": 1.78, "grad_norm": 6.115103917268281, "learning_rate": 3.1085688235560575e-07, "loss": 0.1301, "step": 13093 }, { "epoch": 1.78, "grad_norm": 3.677419209560944, "learning_rate": 3.1047465303714164e-07, "loss": 0.1368, "step": 13094 }, { "epoch": 1.78, "grad_norm": 3.930265768275167, "learning_rate": 3.1009265132775466e-07, "loss": 0.157, "step": 13095 }, { "epoch": 1.78, "grad_norm": 3.90999090528545, "learning_rate": 3.097108772459867e-07, "loss": 0.1596, "step": 13096 }, { "epoch": 1.78, "grad_norm": 3.5893170265473944, "learning_rate": 3.093293308103679e-07, "loss": 0.15, "step": 13097 }, { "epoch": 1.78, "grad_norm": 3.591769595092795, "learning_rate": 3.0894801203941616e-07, "loss": 0.1568, "step": 13098 }, { "epoch": 1.78, "grad_norm": 3.432251569025286, "learning_rate": 3.085669209516395e-07, "loss": 0.1409, "step": 13099 }, { "epoch": 1.78, "grad_norm": 3.6014364851366363, "learning_rate": 3.081860575655338e-07, "loss": 0.1619, "step": 13100 }, { "epoch": 1.78, "grad_norm": 3.1918535114835844, "learning_rate": 3.0780542189958516e-07, "loss": 0.1412, "step": 13101 }, { "epoch": 1.78, "grad_norm": 3.6332716631956807, "learning_rate": 3.074250139722679e-07, "loss": 0.1765, "step": 13102 }, { "epoch": 1.78, "grad_norm": 3.0358910211060914, "learning_rate": 3.070448338020443e-07, "loss": 0.1579, "step": 13103 }, { "epoch": 1.78, "grad_norm": 4.854298711366788, "learning_rate": 3.0666488140736805e-07, "loss": 0.1574, "step": 13104 }, { "epoch": 1.78, "grad_norm": 3.681691299590374, "learning_rate": 3.062851568066799e-07, "loss": 0.1264, "step": 13105 }, { "epoch": 1.78, "grad_norm": 2.7312356236129762, "learning_rate": 3.059056600184096e-07, "loss": 0.131, "step": 13106 }, { "epoch": 1.78, "grad_norm": 3.3653654454720603, "learning_rate": 3.0552639106097684e-07, "loss": 0.1459, "step": 13107 }, { "epoch": 1.78, "grad_norm": 3.7069031425494017, "learning_rate": 3.0514734995278906e-07, "loss": 0.1738, "step": 13108 }, { "epoch": 1.78, "grad_norm": 2.9896782414106053, "learning_rate": 3.0476853671224436e-07, "loss": 0.1646, "step": 13109 }, { "epoch": 1.78, "grad_norm": 3.3123939244987244, "learning_rate": 3.0438995135772644e-07, "loss": 0.1342, "step": 13110 }, { "epoch": 1.78, "grad_norm": 3.135826776512226, "learning_rate": 3.040115939076138e-07, "loss": 0.1827, "step": 13111 }, { "epoch": 1.78, "grad_norm": 3.228669857628864, "learning_rate": 3.0363346438026633e-07, "loss": 0.1448, "step": 13112 }, { "epoch": 1.78, "grad_norm": 2.769147109443357, "learning_rate": 3.032555627940398e-07, "loss": 0.1189, "step": 13113 }, { "epoch": 1.78, "grad_norm": 3.756855799348985, "learning_rate": 3.0287788916727343e-07, "loss": 0.1571, "step": 13114 }, { "epoch": 1.78, "grad_norm": 3.8326139944670046, "learning_rate": 3.0250044351829987e-07, "loss": 0.1668, "step": 13115 }, { "epoch": 1.78, "grad_norm": 3.1435442514886893, "learning_rate": 3.021232258654383e-07, "loss": 0.155, "step": 13116 }, { "epoch": 1.78, "grad_norm": 3.2654387839430057, "learning_rate": 3.0174623622699685e-07, "loss": 0.1341, "step": 13117 }, { "epoch": 1.78, "grad_norm": 3.205794213676905, "learning_rate": 3.013694746212725e-07, "loss": 0.1253, "step": 13118 }, { "epoch": 1.78, "grad_norm": 2.939167765914327, "learning_rate": 3.009929410665524e-07, "loss": 0.167, "step": 13119 }, { "epoch": 1.78, "grad_norm": 4.829308220461484, "learning_rate": 3.006166355811113e-07, "loss": 0.156, "step": 13120 }, { "epoch": 1.78, "grad_norm": 6.314149076370579, "learning_rate": 3.002405581832135e-07, "loss": 0.136, "step": 13121 }, { "epoch": 1.78, "grad_norm": 2.5976626745766205, "learning_rate": 2.998647088911127e-07, "loss": 0.1219, "step": 13122 }, { "epoch": 1.79, "grad_norm": 2.8619322060016494, "learning_rate": 2.994890877230505e-07, "loss": 0.1499, "step": 13123 }, { "epoch": 1.79, "grad_norm": 3.7491321662172252, "learning_rate": 2.991136946972567e-07, "loss": 0.1747, "step": 13124 }, { "epoch": 1.79, "grad_norm": 2.8822798553318605, "learning_rate": 2.987385298319545e-07, "loss": 0.0925, "step": 13125 }, { "epoch": 1.79, "grad_norm": 3.587292962808706, "learning_rate": 2.9836359314534936e-07, "loss": 0.1796, "step": 13126 }, { "epoch": 1.79, "grad_norm": 4.073653705744827, "learning_rate": 2.9798888465564226e-07, "loss": 0.1332, "step": 13127 }, { "epoch": 1.79, "grad_norm": 3.6790858474530115, "learning_rate": 2.9761440438101587e-07, "loss": 0.1817, "step": 13128 }, { "epoch": 1.79, "grad_norm": 2.7528526415258248, "learning_rate": 2.972401523396495e-07, "loss": 0.1129, "step": 13129 }, { "epoch": 1.79, "grad_norm": 4.065972744486228, "learning_rate": 2.96866128549706e-07, "loss": 0.1509, "step": 13130 }, { "epoch": 1.79, "grad_norm": 3.875603556458085, "learning_rate": 2.9649233302933955e-07, "loss": 0.1476, "step": 13131 }, { "epoch": 1.79, "grad_norm": 2.529640760642678, "learning_rate": 2.961187657966919e-07, "loss": 0.1501, "step": 13132 }, { "epoch": 1.79, "grad_norm": 3.847499277884872, "learning_rate": 2.957454268698945e-07, "loss": 0.1605, "step": 13133 }, { "epoch": 1.79, "grad_norm": 2.4806537484674176, "learning_rate": 2.9537231626706797e-07, "loss": 0.1496, "step": 13134 }, { "epoch": 1.79, "grad_norm": 3.2230251459695096, "learning_rate": 2.949994340063206e-07, "loss": 0.1304, "step": 13135 }, { "epoch": 1.79, "grad_norm": 2.951828403489312, "learning_rate": 2.9462678010575117e-07, "loss": 0.1363, "step": 13136 }, { "epoch": 1.79, "grad_norm": 4.135390410348699, "learning_rate": 2.942543545834475e-07, "loss": 0.1674, "step": 13137 }, { "epoch": 1.79, "grad_norm": 3.9692737840773216, "learning_rate": 2.9388215745748347e-07, "loss": 0.1634, "step": 13138 }, { "epoch": 1.79, "grad_norm": 3.035063871808652, "learning_rate": 2.9351018874592684e-07, "loss": 0.1557, "step": 13139 }, { "epoch": 1.79, "grad_norm": 3.238771609750409, "learning_rate": 2.9313844846682815e-07, "loss": 0.1294, "step": 13140 }, { "epoch": 1.79, "grad_norm": 3.1561807922205816, "learning_rate": 2.9276693663823186e-07, "loss": 0.1675, "step": 13141 }, { "epoch": 1.79, "grad_norm": 2.8391318623056745, "learning_rate": 2.923956532781691e-07, "loss": 0.1411, "step": 13142 }, { "epoch": 1.79, "grad_norm": 3.3032507715499078, "learning_rate": 2.9202459840465934e-07, "loss": 0.1758, "step": 13143 }, { "epoch": 1.79, "grad_norm": 2.953142442167425, "learning_rate": 2.916537720357143e-07, "loss": 0.1767, "step": 13144 }, { "epoch": 1.79, "grad_norm": 3.2453737866552244, "learning_rate": 2.9128317418933015e-07, "loss": 0.1522, "step": 13145 }, { "epoch": 1.79, "grad_norm": 4.249640859569087, "learning_rate": 2.909128048834958e-07, "loss": 0.1473, "step": 13146 }, { "epoch": 1.79, "grad_norm": 7.332106472310891, "learning_rate": 2.9054266413618525e-07, "loss": 0.1581, "step": 13147 }, { "epoch": 1.79, "grad_norm": 3.901606069189625, "learning_rate": 2.901727519653652e-07, "loss": 0.1237, "step": 13148 }, { "epoch": 1.79, "grad_norm": 2.932611709846144, "learning_rate": 2.898030683889891e-07, "loss": 0.108, "step": 13149 }, { "epoch": 1.79, "grad_norm": 3.6959403690385724, "learning_rate": 2.894336134250003e-07, "loss": 0.1212, "step": 13150 }, { "epoch": 1.79, "grad_norm": 4.334975820314282, "learning_rate": 2.890643870913296e-07, "loss": 0.1479, "step": 13151 }, { "epoch": 1.79, "grad_norm": 4.662776133797864, "learning_rate": 2.88695389405898e-07, "loss": 0.1833, "step": 13152 }, { "epoch": 1.79, "grad_norm": 4.153963993544118, "learning_rate": 2.8832662038661475e-07, "loss": 0.1178, "step": 13153 }, { "epoch": 1.79, "grad_norm": 4.016015371682982, "learning_rate": 2.879580800513793e-07, "loss": 0.1508, "step": 13154 }, { "epoch": 1.79, "grad_norm": 4.689834432308683, "learning_rate": 2.875897684180767e-07, "loss": 0.1761, "step": 13155 }, { "epoch": 1.79, "grad_norm": 3.3892830370473748, "learning_rate": 2.872216855045873e-07, "loss": 0.1366, "step": 13156 }, { "epoch": 1.79, "grad_norm": 3.0321062610285567, "learning_rate": 2.8685383132877163e-07, "loss": 0.1584, "step": 13157 }, { "epoch": 1.79, "grad_norm": 3.432359465312993, "learning_rate": 2.864862059084872e-07, "loss": 0.1277, "step": 13158 }, { "epoch": 1.79, "grad_norm": 3.388964370047326, "learning_rate": 2.861188092615741e-07, "loss": 0.1448, "step": 13159 }, { "epoch": 1.79, "grad_norm": 4.755621487983607, "learning_rate": 2.8575164140586695e-07, "loss": 0.1608, "step": 13160 }, { "epoch": 1.79, "grad_norm": 2.837918951580101, "learning_rate": 2.8538470235918436e-07, "loss": 0.1822, "step": 13161 }, { "epoch": 1.79, "grad_norm": 3.046779399214914, "learning_rate": 2.8501799213933646e-07, "loss": 0.1475, "step": 13162 }, { "epoch": 1.79, "grad_norm": 2.7455647825348297, "learning_rate": 2.846515107641229e-07, "loss": 0.1027, "step": 13163 }, { "epoch": 1.79, "grad_norm": 3.390685175750213, "learning_rate": 2.8428525825133004e-07, "loss": 0.1147, "step": 13164 }, { "epoch": 1.79, "grad_norm": 4.425657521106465, "learning_rate": 2.8391923461873363e-07, "loss": 0.1684, "step": 13165 }, { "epoch": 1.79, "grad_norm": 2.8544740698364426, "learning_rate": 2.8355343988410054e-07, "loss": 0.1105, "step": 13166 }, { "epoch": 1.79, "grad_norm": 3.149863438208705, "learning_rate": 2.831878740651833e-07, "loss": 0.175, "step": 13167 }, { "epoch": 1.79, "grad_norm": 2.931358748746952, "learning_rate": 2.828225371797255e-07, "loss": 0.1614, "step": 13168 }, { "epoch": 1.79, "grad_norm": 4.648755723874439, "learning_rate": 2.824574292454585e-07, "loss": 0.1737, "step": 13169 }, { "epoch": 1.79, "grad_norm": 3.9418495669495552, "learning_rate": 2.8209255028010474e-07, "loss": 0.1651, "step": 13170 }, { "epoch": 1.79, "grad_norm": 3.0505074530065683, "learning_rate": 2.8172790030137123e-07, "loss": 0.1427, "step": 13171 }, { "epoch": 1.79, "grad_norm": 5.956998541084342, "learning_rate": 2.8136347932695926e-07, "loss": 0.1194, "step": 13172 }, { "epoch": 1.79, "grad_norm": 2.909768222986883, "learning_rate": 2.809992873745532e-07, "loss": 0.1366, "step": 13173 }, { "epoch": 1.79, "grad_norm": 3.5832817357264677, "learning_rate": 2.8063532446183204e-07, "loss": 0.1263, "step": 13174 }, { "epoch": 1.79, "grad_norm": 3.4854100448118013, "learning_rate": 2.8027159060646014e-07, "loss": 0.146, "step": 13175 }, { "epoch": 1.79, "grad_norm": 2.9773054081922528, "learning_rate": 2.7990808582609055e-07, "loss": 0.1157, "step": 13176 }, { "epoch": 1.79, "grad_norm": 3.412530815588294, "learning_rate": 2.7954481013836744e-07, "loss": 0.1213, "step": 13177 }, { "epoch": 1.79, "grad_norm": 2.8814536159141286, "learning_rate": 2.791817635609223e-07, "loss": 0.1589, "step": 13178 }, { "epoch": 1.79, "grad_norm": 3.8858627121442573, "learning_rate": 2.788189461113755e-07, "loss": 0.165, "step": 13179 }, { "epoch": 1.79, "grad_norm": 3.188636149271774, "learning_rate": 2.7845635780733684e-07, "loss": 0.1189, "step": 13180 }, { "epoch": 1.79, "grad_norm": 3.10533366765998, "learning_rate": 2.780939986664044e-07, "loss": 0.1179, "step": 13181 }, { "epoch": 1.79, "grad_norm": 3.8353533148222816, "learning_rate": 2.7773186870616585e-07, "loss": 0.1821, "step": 13182 }, { "epoch": 1.79, "grad_norm": 4.398576702660362, "learning_rate": 2.7736996794419767e-07, "loss": 0.1881, "step": 13183 }, { "epoch": 1.79, "grad_norm": 3.532718600874238, "learning_rate": 2.770082963980647e-07, "loss": 0.1471, "step": 13184 }, { "epoch": 1.79, "grad_norm": 3.4798603008600093, "learning_rate": 2.766468540853207e-07, "loss": 0.1252, "step": 13185 }, { "epoch": 1.79, "grad_norm": 2.578651101750962, "learning_rate": 2.762856410235082e-07, "loss": 0.1228, "step": 13186 }, { "epoch": 1.79, "grad_norm": 2.849555270349647, "learning_rate": 2.759246572301599e-07, "loss": 0.1135, "step": 13187 }, { "epoch": 1.79, "grad_norm": 3.6002398250240817, "learning_rate": 2.755639027227952e-07, "loss": 0.135, "step": 13188 }, { "epoch": 1.79, "grad_norm": 3.848855553802563, "learning_rate": 2.75203377518925e-07, "loss": 0.1693, "step": 13189 }, { "epoch": 1.79, "grad_norm": 3.3869513729739658, "learning_rate": 2.748430816360459e-07, "loss": 0.1792, "step": 13190 }, { "epoch": 1.79, "grad_norm": 4.331621312007082, "learning_rate": 2.744830150916472e-07, "loss": 0.1457, "step": 13191 }, { "epoch": 1.79, "grad_norm": 4.637780310943272, "learning_rate": 2.741231779032022e-07, "loss": 0.1669, "step": 13192 }, { "epoch": 1.79, "grad_norm": 4.478722558314654, "learning_rate": 2.737635700881791e-07, "loss": 0.187, "step": 13193 }, { "epoch": 1.79, "grad_norm": 3.5456990569703972, "learning_rate": 2.734041916640284e-07, "loss": 0.1645, "step": 13194 }, { "epoch": 1.79, "grad_norm": 3.3721667386538132, "learning_rate": 2.730450426481951e-07, "loss": 0.1667, "step": 13195 }, { "epoch": 1.8, "grad_norm": 4.087460822726553, "learning_rate": 2.726861230581096e-07, "loss": 0.1293, "step": 13196 }, { "epoch": 1.8, "grad_norm": 2.5066547248220403, "learning_rate": 2.72327432911193e-07, "loss": 0.1665, "step": 13197 }, { "epoch": 1.8, "grad_norm": 3.144934314521533, "learning_rate": 2.7196897222485363e-07, "loss": 0.1601, "step": 13198 }, { "epoch": 1.8, "grad_norm": 3.480794405616622, "learning_rate": 2.7161074101649085e-07, "loss": 0.1447, "step": 13199 }, { "epoch": 1.8, "grad_norm": 3.1441402857264147, "learning_rate": 2.712527393034903e-07, "loss": 0.1379, "step": 13200 }, { "epoch": 1.8, "grad_norm": 4.256525729903238, "learning_rate": 2.7089496710322905e-07, "loss": 0.1912, "step": 13201 }, { "epoch": 1.8, "grad_norm": 4.093013425342953, "learning_rate": 2.7053742443307054e-07, "loss": 0.1352, "step": 13202 }, { "epoch": 1.8, "grad_norm": 4.543315480679587, "learning_rate": 2.701801113103697e-07, "loss": 0.1553, "step": 13203 }, { "epoch": 1.8, "grad_norm": 3.5436323596571384, "learning_rate": 2.6982302775246774e-07, "loss": 0.1516, "step": 13204 }, { "epoch": 1.8, "grad_norm": 3.971864977575163, "learning_rate": 2.694661737766974e-07, "loss": 0.1326, "step": 13205 }, { "epoch": 1.8, "grad_norm": 3.075600344428043, "learning_rate": 2.691095494003759e-07, "loss": 0.1613, "step": 13206 }, { "epoch": 1.8, "grad_norm": 3.625253331438959, "learning_rate": 2.6875315464081566e-07, "loss": 0.1655, "step": 13207 }, { "epoch": 1.8, "grad_norm": 4.040112984068949, "learning_rate": 2.683969895153127e-07, "loss": 0.1474, "step": 13208 }, { "epoch": 1.8, "grad_norm": 4.486352770435686, "learning_rate": 2.6804105404115434e-07, "loss": 0.1021, "step": 13209 }, { "epoch": 1.8, "grad_norm": 2.691919942326958, "learning_rate": 2.6768534823561566e-07, "loss": 0.148, "step": 13210 }, { "epoch": 1.8, "grad_norm": 3.26664604318568, "learning_rate": 2.6732987211596173e-07, "loss": 0.1527, "step": 13211 }, { "epoch": 1.8, "grad_norm": 5.1768381811125606, "learning_rate": 2.669746256994449e-07, "loss": 0.1831, "step": 13212 }, { "epoch": 1.8, "grad_norm": 5.945873547464783, "learning_rate": 2.66619609003308e-07, "loss": 0.1382, "step": 13213 }, { "epoch": 1.8, "grad_norm": 3.196071481083685, "learning_rate": 2.662648220447811e-07, "loss": 0.1607, "step": 13214 }, { "epoch": 1.8, "grad_norm": 4.0283435871136835, "learning_rate": 2.659102648410855e-07, "loss": 0.1922, "step": 13215 }, { "epoch": 1.8, "grad_norm": 3.514453534267281, "learning_rate": 2.6555593740942797e-07, "loss": 0.1463, "step": 13216 }, { "epoch": 1.8, "grad_norm": 3.4007383613775772, "learning_rate": 2.652018397670081e-07, "loss": 0.1743, "step": 13217 }, { "epoch": 1.8, "grad_norm": 3.362869193183972, "learning_rate": 2.6484797193100985e-07, "loss": 0.1308, "step": 13218 }, { "epoch": 1.8, "grad_norm": 2.7250864387359823, "learning_rate": 2.644943339186118e-07, "loss": 0.141, "step": 13219 }, { "epoch": 1.8, "grad_norm": 4.13044258559639, "learning_rate": 2.641409257469746e-07, "loss": 0.1566, "step": 13220 }, { "epoch": 1.8, "grad_norm": 2.6259888708531434, "learning_rate": 2.6378774743325173e-07, "loss": 0.1039, "step": 13221 }, { "epoch": 1.8, "grad_norm": 3.5275719395953575, "learning_rate": 2.6343479899458737e-07, "loss": 0.1282, "step": 13222 }, { "epoch": 1.8, "grad_norm": 3.326352034756198, "learning_rate": 2.630820804481088e-07, "loss": 0.1707, "step": 13223 }, { "epoch": 1.8, "grad_norm": 3.669478098875118, "learning_rate": 2.627295918109385e-07, "loss": 0.1239, "step": 13224 }, { "epoch": 1.8, "grad_norm": 3.1229300078414948, "learning_rate": 2.623773331001822e-07, "loss": 0.1206, "step": 13225 }, { "epoch": 1.8, "grad_norm": 3.975124648045258, "learning_rate": 2.620253043329385e-07, "loss": 0.1205, "step": 13226 }, { "epoch": 1.8, "grad_norm": 3.443041081578046, "learning_rate": 2.616735055262931e-07, "loss": 0.0916, "step": 13227 }, { "epoch": 1.8, "grad_norm": 3.380071088830435, "learning_rate": 2.613219366973202e-07, "loss": 0.168, "step": 13228 }, { "epoch": 1.8, "grad_norm": 3.7255024060983977, "learning_rate": 2.609705978630839e-07, "loss": 0.1743, "step": 13229 }, { "epoch": 1.8, "grad_norm": 3.321540810191172, "learning_rate": 2.6061948904063663e-07, "loss": 0.1619, "step": 13230 }, { "epoch": 1.8, "grad_norm": 4.755199350940654, "learning_rate": 2.6026861024701977e-07, "loss": 0.1315, "step": 13231 }, { "epoch": 1.8, "grad_norm": 4.220536666872749, "learning_rate": 2.5991796149926306e-07, "loss": 0.1739, "step": 13232 }, { "epoch": 1.8, "grad_norm": 3.7790575438995795, "learning_rate": 2.5956754281438567e-07, "loss": 0.1511, "step": 13233 }, { "epoch": 1.8, "grad_norm": 3.045712042252651, "learning_rate": 2.5921735420939565e-07, "loss": 0.1386, "step": 13234 }, { "epoch": 1.8, "grad_norm": 2.725174128061434, "learning_rate": 2.588673957012883e-07, "loss": 0.1307, "step": 13235 }, { "epoch": 1.8, "grad_norm": 3.297655551421813, "learning_rate": 2.5851766730705164e-07, "loss": 0.1644, "step": 13236 }, { "epoch": 1.8, "grad_norm": 2.799951860531767, "learning_rate": 2.5816816904365715e-07, "loss": 0.1428, "step": 13237 }, { "epoch": 1.8, "grad_norm": 3.3824102847929844, "learning_rate": 2.578189009280707e-07, "loss": 0.1388, "step": 13238 }, { "epoch": 1.8, "grad_norm": 3.3698842799041877, "learning_rate": 2.574698629772415e-07, "loss": 0.1734, "step": 13239 }, { "epoch": 1.8, "grad_norm": 3.378034677981021, "learning_rate": 2.5712105520811216e-07, "loss": 0.1756, "step": 13240 }, { "epoch": 1.8, "grad_norm": 3.069334538839688, "learning_rate": 2.567724776376118e-07, "loss": 0.1194, "step": 13241 }, { "epoch": 1.8, "grad_norm": 2.9276553729194252, "learning_rate": 2.5642413028265867e-07, "loss": 0.1394, "step": 13242 }, { "epoch": 1.8, "grad_norm": 2.7543797033071122, "learning_rate": 2.5607601316016084e-07, "loss": 0.1472, "step": 13243 }, { "epoch": 1.8, "grad_norm": 4.2296828062876175, "learning_rate": 2.5572812628701315e-07, "loss": 0.1662, "step": 13244 }, { "epoch": 1.8, "grad_norm": 3.4855597122125292, "learning_rate": 2.5538046968010097e-07, "loss": 0.1699, "step": 13245 }, { "epoch": 1.8, "grad_norm": 3.3632402351329547, "learning_rate": 2.5503304335629806e-07, "loss": 0.1192, "step": 13246 }, { "epoch": 1.8, "grad_norm": 3.5880114174783793, "learning_rate": 2.546858473324676e-07, "loss": 0.1701, "step": 13247 }, { "epoch": 1.8, "grad_norm": 2.966506071069425, "learning_rate": 2.5433888162545993e-07, "loss": 0.1763, "step": 13248 }, { "epoch": 1.8, "grad_norm": 3.8003064490577825, "learning_rate": 2.5399214625211556e-07, "loss": 0.1975, "step": 13249 }, { "epoch": 1.8, "grad_norm": 3.1749072390566653, "learning_rate": 2.5364564122926483e-07, "loss": 0.1091, "step": 13250 }, { "epoch": 1.8, "grad_norm": 3.7177984848921906, "learning_rate": 2.532993665737232e-07, "loss": 0.1454, "step": 13251 }, { "epoch": 1.8, "grad_norm": 2.8900085005058767, "learning_rate": 2.529533223022995e-07, "loss": 0.1501, "step": 13252 }, { "epoch": 1.8, "grad_norm": 3.670557764226545, "learning_rate": 2.526075084317875e-07, "loss": 0.1623, "step": 13253 }, { "epoch": 1.8, "grad_norm": 3.79617386471881, "learning_rate": 2.522619249789726e-07, "loss": 0.1876, "step": 13254 }, { "epoch": 1.8, "grad_norm": 2.2728744546747146, "learning_rate": 2.519165719606276e-07, "loss": 0.1176, "step": 13255 }, { "epoch": 1.8, "grad_norm": 2.983791526205209, "learning_rate": 2.515714493935145e-07, "loss": 0.1245, "step": 13256 }, { "epoch": 1.8, "grad_norm": 5.1476858637850516, "learning_rate": 2.5122655729438393e-07, "loss": 0.2071, "step": 13257 }, { "epoch": 1.8, "grad_norm": 2.656814427734121, "learning_rate": 2.508818956799758e-07, "loss": 0.1552, "step": 13258 }, { "epoch": 1.8, "grad_norm": 4.17739743270458, "learning_rate": 2.5053746456701776e-07, "loss": 0.1642, "step": 13259 }, { "epoch": 1.8, "grad_norm": 5.627916333224689, "learning_rate": 2.5019326397222767e-07, "loss": 0.1619, "step": 13260 }, { "epoch": 1.8, "grad_norm": 3.142384588625985, "learning_rate": 2.498492939123109e-07, "loss": 0.1293, "step": 13261 }, { "epoch": 1.8, "grad_norm": 4.286302805730373, "learning_rate": 2.495055544039632e-07, "loss": 0.1696, "step": 13262 }, { "epoch": 1.8, "grad_norm": 2.8312854275327446, "learning_rate": 2.491620454638671e-07, "loss": 0.1055, "step": 13263 }, { "epoch": 1.8, "grad_norm": 3.7721778537742146, "learning_rate": 2.4881876710869556e-07, "loss": 0.1873, "step": 13264 }, { "epoch": 1.8, "grad_norm": 4.796010217711505, "learning_rate": 2.484757193551102e-07, "loss": 0.1851, "step": 13265 }, { "epoch": 1.8, "grad_norm": 2.9067907021564583, "learning_rate": 2.481329022197598e-07, "loss": 0.1583, "step": 13266 }, { "epoch": 1.8, "grad_norm": 3.9930705904790944, "learning_rate": 2.477903157192846e-07, "loss": 0.1665, "step": 13267 }, { "epoch": 1.8, "grad_norm": 3.659253964503815, "learning_rate": 2.474479598703111e-07, "loss": 0.1307, "step": 13268 }, { "epoch": 1.8, "grad_norm": 3.9626587945910843, "learning_rate": 2.4710583468945726e-07, "loss": 0.1652, "step": 13269 }, { "epoch": 1.81, "grad_norm": 3.377004468019754, "learning_rate": 2.467639401933264e-07, "loss": 0.1396, "step": 13270 }, { "epoch": 1.81, "grad_norm": 3.4671599469596597, "learning_rate": 2.4642227639851466e-07, "loss": 0.1382, "step": 13271 }, { "epoch": 1.81, "grad_norm": 3.247354794413992, "learning_rate": 2.4608084332160277e-07, "loss": 0.1521, "step": 13272 }, { "epoch": 1.81, "grad_norm": 4.968069692108519, "learning_rate": 2.4573964097916405e-07, "loss": 0.2212, "step": 13273 }, { "epoch": 1.81, "grad_norm": 3.7194116885686013, "learning_rate": 2.4539866938775856e-07, "loss": 0.1815, "step": 13274 }, { "epoch": 1.81, "grad_norm": 3.0782745403938994, "learning_rate": 2.450579285639354e-07, "loss": 0.1361, "step": 13275 }, { "epoch": 1.81, "grad_norm": 2.814737479667616, "learning_rate": 2.447174185242324e-07, "loss": 0.1433, "step": 13276 }, { "epoch": 1.81, "grad_norm": 3.5501917467860444, "learning_rate": 2.443771392851768e-07, "loss": 0.1537, "step": 13277 }, { "epoch": 1.81, "grad_norm": 4.15918999518931, "learning_rate": 2.440370908632844e-07, "loss": 0.1661, "step": 13278 }, { "epoch": 1.81, "grad_norm": 3.0772712293257936, "learning_rate": 2.4369727327505923e-07, "loss": 0.1511, "step": 13279 }, { "epoch": 1.81, "grad_norm": 2.926286980869659, "learning_rate": 2.4335768653699464e-07, "loss": 0.1274, "step": 13280 }, { "epoch": 1.81, "grad_norm": 5.523845182871645, "learning_rate": 2.4301833066557254e-07, "loss": 0.1869, "step": 13281 }, { "epoch": 1.81, "grad_norm": 4.358442599465757, "learning_rate": 2.4267920567726364e-07, "loss": 0.137, "step": 13282 }, { "epoch": 1.81, "grad_norm": 3.859876465208829, "learning_rate": 2.4234031158852977e-07, "loss": 0.1833, "step": 13283 }, { "epoch": 1.81, "grad_norm": 2.5165184545301247, "learning_rate": 2.4200164841581553e-07, "loss": 0.1406, "step": 13284 }, { "epoch": 1.81, "grad_norm": 3.190548835816069, "learning_rate": 2.416632161755622e-07, "loss": 0.1384, "step": 13285 }, { "epoch": 1.81, "grad_norm": 3.490463542516634, "learning_rate": 2.4132501488419167e-07, "loss": 0.1605, "step": 13286 }, { "epoch": 1.81, "grad_norm": 3.1552047321632077, "learning_rate": 2.409870445581225e-07, "loss": 0.1289, "step": 13287 }, { "epoch": 1.81, "grad_norm": 3.81008527018905, "learning_rate": 2.4064930521375596e-07, "loss": 0.1511, "step": 13288 }, { "epoch": 1.81, "grad_norm": 4.264478865990662, "learning_rate": 2.4031179686748563e-07, "loss": 0.1727, "step": 13289 }, { "epoch": 1.81, "grad_norm": 3.015974993909219, "learning_rate": 2.399745195356917e-07, "loss": 0.15, "step": 13290 }, { "epoch": 1.81, "grad_norm": 2.810949605378914, "learning_rate": 2.3963747323474506e-07, "loss": 0.1247, "step": 13291 }, { "epoch": 1.81, "grad_norm": 2.5230657325549033, "learning_rate": 2.393006579810037e-07, "loss": 0.1084, "step": 13292 }, { "epoch": 1.81, "grad_norm": 2.5321615445946875, "learning_rate": 2.389640737908161e-07, "loss": 0.1114, "step": 13293 }, { "epoch": 1.81, "grad_norm": 3.195024817191981, "learning_rate": 2.3862772068051765e-07, "loss": 0.1725, "step": 13294 }, { "epoch": 1.81, "grad_norm": 4.3976511606034965, "learning_rate": 2.3829159866643358e-07, "loss": 0.1887, "step": 13295 }, { "epoch": 1.81, "grad_norm": 3.193884405267297, "learning_rate": 2.379557077648781e-07, "loss": 0.1928, "step": 13296 }, { "epoch": 1.81, "grad_norm": 2.6220948767299412, "learning_rate": 2.3762004799215422e-07, "loss": 0.1234, "step": 13297 }, { "epoch": 1.81, "grad_norm": 5.861610399488229, "learning_rate": 2.3728461936455228e-07, "loss": 0.1581, "step": 13298 }, { "epoch": 1.81, "grad_norm": 3.3988502200795034, "learning_rate": 2.3694942189835423e-07, "loss": 0.1386, "step": 13299 }, { "epoch": 1.81, "grad_norm": 4.707270030899574, "learning_rate": 2.366144556098271e-07, "loss": 0.1678, "step": 13300 }, { "epoch": 1.81, "grad_norm": 3.398355159330622, "learning_rate": 2.3627972051523062e-07, "loss": 0.1028, "step": 13301 }, { "epoch": 1.81, "grad_norm": 3.148904430814289, "learning_rate": 2.3594521663081072e-07, "loss": 0.1008, "step": 13302 }, { "epoch": 1.81, "grad_norm": 4.258866332462312, "learning_rate": 2.3561094397280104e-07, "loss": 0.1522, "step": 13303 }, { "epoch": 1.81, "grad_norm": 3.9294810458214577, "learning_rate": 2.3527690255742864e-07, "loss": 0.163, "step": 13304 }, { "epoch": 1.81, "grad_norm": 4.355676225028605, "learning_rate": 2.349430924009033e-07, "loss": 0.1534, "step": 13305 }, { "epoch": 1.81, "grad_norm": 3.1062691105977094, "learning_rate": 2.3460951351942928e-07, "loss": 0.2091, "step": 13306 }, { "epoch": 1.81, "grad_norm": 3.0796053110114148, "learning_rate": 2.3427616592919587e-07, "loss": 0.1738, "step": 13307 }, { "epoch": 1.81, "grad_norm": 3.561186379041164, "learning_rate": 2.3394304964638292e-07, "loss": 0.1683, "step": 13308 }, { "epoch": 1.81, "grad_norm": 2.890331785768558, "learning_rate": 2.3361016468715802e-07, "loss": 0.126, "step": 13309 }, { "epoch": 1.81, "grad_norm": 2.448000268674163, "learning_rate": 2.332775110676777e-07, "loss": 0.1472, "step": 13310 }, { "epoch": 1.81, "grad_norm": 3.776254131298495, "learning_rate": 2.3294508880408795e-07, "loss": 0.132, "step": 13311 }, { "epoch": 1.81, "grad_norm": 3.3704035502844336, "learning_rate": 2.3261289791252306e-07, "loss": 0.1601, "step": 13312 }, { "epoch": 1.81, "grad_norm": 2.8208133359043064, "learning_rate": 2.3228093840910627e-07, "loss": 0.1252, "step": 13313 }, { "epoch": 1.81, "grad_norm": 4.658420220457187, "learning_rate": 2.3194921030994855e-07, "loss": 0.144, "step": 13314 }, { "epoch": 1.81, "grad_norm": 3.111369232880609, "learning_rate": 2.3161771363115092e-07, "loss": 0.1476, "step": 13315 }, { "epoch": 1.81, "grad_norm": 2.705791748162034, "learning_rate": 2.3128644838880444e-07, "loss": 0.126, "step": 13316 }, { "epoch": 1.81, "grad_norm": 2.847263674807238, "learning_rate": 2.3095541459898452e-07, "loss": 0.1464, "step": 13317 }, { "epoch": 1.81, "grad_norm": 4.908554390070404, "learning_rate": 2.3062461227776056e-07, "loss": 0.1218, "step": 13318 }, { "epoch": 1.81, "grad_norm": 2.711940985794382, "learning_rate": 2.3029404144118585e-07, "loss": 0.1566, "step": 13319 }, { "epoch": 1.81, "grad_norm": 3.968821614852571, "learning_rate": 2.2996370210530693e-07, "loss": 0.1444, "step": 13320 }, { "epoch": 1.81, "grad_norm": 3.3514709956776705, "learning_rate": 2.2963359428615607e-07, "loss": 0.1037, "step": 13321 }, { "epoch": 1.81, "grad_norm": 2.9212280240862034, "learning_rate": 2.2930371799975593e-07, "loss": 0.1207, "step": 13322 }, { "epoch": 1.81, "grad_norm": 3.5620271051523775, "learning_rate": 2.289740732621165e-07, "loss": 0.1529, "step": 13323 }, { "epoch": 1.81, "grad_norm": 3.483124983497684, "learning_rate": 2.2864466008923725e-07, "loss": 0.2109, "step": 13324 }, { "epoch": 1.81, "grad_norm": 3.3739000065989, "learning_rate": 2.2831547849710755e-07, "loss": 0.1513, "step": 13325 }, { "epoch": 1.81, "grad_norm": 4.120830773876481, "learning_rate": 2.2798652850170299e-07, "loss": 0.165, "step": 13326 }, { "epoch": 1.81, "grad_norm": 2.853844716576904, "learning_rate": 2.2765781011899025e-07, "loss": 0.1466, "step": 13327 }, { "epoch": 1.81, "grad_norm": 3.0591715794480865, "learning_rate": 2.2732932336492375e-07, "loss": 0.1196, "step": 13328 }, { "epoch": 1.81, "grad_norm": 3.5827946886926867, "learning_rate": 2.2700106825544578e-07, "loss": 0.1368, "step": 13329 }, { "epoch": 1.81, "grad_norm": 3.77135292756085, "learning_rate": 2.2667304480649134e-07, "loss": 0.1166, "step": 13330 }, { "epoch": 1.81, "grad_norm": 3.3262580056523157, "learning_rate": 2.2634525303397714e-07, "loss": 0.1825, "step": 13331 }, { "epoch": 1.81, "grad_norm": 3.7093286523200684, "learning_rate": 2.260176929538166e-07, "loss": 0.1557, "step": 13332 }, { "epoch": 1.81, "grad_norm": 3.341479289973558, "learning_rate": 2.256903645819053e-07, "loss": 0.1533, "step": 13333 }, { "epoch": 1.81, "grad_norm": 3.8321500915985713, "learning_rate": 2.2536326793413164e-07, "loss": 0.1447, "step": 13334 }, { "epoch": 1.81, "grad_norm": 3.288534526817507, "learning_rate": 2.250364030263713e-07, "loss": 0.1559, "step": 13335 }, { "epoch": 1.81, "grad_norm": 2.9896634684149457, "learning_rate": 2.247097698744888e-07, "loss": 0.1544, "step": 13336 }, { "epoch": 1.81, "grad_norm": 2.774022086679917, "learning_rate": 2.243833684943375e-07, "loss": 0.1232, "step": 13337 }, { "epoch": 1.81, "grad_norm": 3.070750953386716, "learning_rate": 2.240571989017598e-07, "loss": 0.1353, "step": 13338 }, { "epoch": 1.81, "grad_norm": 3.9611494692761164, "learning_rate": 2.2373126111258582e-07, "loss": 0.178, "step": 13339 }, { "epoch": 1.81, "grad_norm": 2.9920711639309516, "learning_rate": 2.2340555514263562e-07, "loss": 0.1361, "step": 13340 }, { "epoch": 1.81, "grad_norm": 3.4755918540046404, "learning_rate": 2.2308008100771773e-07, "loss": 0.1111, "step": 13341 }, { "epoch": 1.81, "grad_norm": 3.465847497382927, "learning_rate": 2.2275483872362835e-07, "loss": 0.1554, "step": 13342 }, { "epoch": 1.82, "grad_norm": 3.4124032863950555, "learning_rate": 2.2242982830615434e-07, "loss": 0.1222, "step": 13343 }, { "epoch": 1.82, "grad_norm": 3.964425562096498, "learning_rate": 2.2210504977107027e-07, "loss": 0.1179, "step": 13344 }, { "epoch": 1.82, "grad_norm": 2.5465649688955474, "learning_rate": 2.2178050313413856e-07, "loss": 0.1551, "step": 13345 }, { "epoch": 1.82, "grad_norm": 3.300597690481472, "learning_rate": 2.2145618841111217e-07, "loss": 0.1455, "step": 13346 }, { "epoch": 1.82, "grad_norm": 2.482748175792796, "learning_rate": 2.2113210561773124e-07, "loss": 0.1001, "step": 13347 }, { "epoch": 1.82, "grad_norm": 3.023877850324883, "learning_rate": 2.2080825476972544e-07, "loss": 0.1209, "step": 13348 }, { "epoch": 1.82, "grad_norm": 4.073954329184516, "learning_rate": 2.2048463588281443e-07, "loss": 0.1459, "step": 13349 }, { "epoch": 1.82, "grad_norm": 2.9632565212022137, "learning_rate": 2.2016124897270287e-07, "loss": 0.1441, "step": 13350 }, { "epoch": 1.82, "grad_norm": 4.354347916157522, "learning_rate": 2.198380940550887e-07, "loss": 0.1382, "step": 13351 }, { "epoch": 1.82, "grad_norm": 5.2298856114776635, "learning_rate": 2.1951517114565446e-07, "loss": 0.1396, "step": 13352 }, { "epoch": 1.82, "grad_norm": 3.5340065577379365, "learning_rate": 2.1919248026007534e-07, "loss": 0.1661, "step": 13353 }, { "epoch": 1.82, "grad_norm": 3.019861950047235, "learning_rate": 2.188700214140127e-07, "loss": 0.1526, "step": 13354 }, { "epoch": 1.82, "grad_norm": 3.5467862870970372, "learning_rate": 2.1854779462311682e-07, "loss": 0.1923, "step": 13355 }, { "epoch": 1.82, "grad_norm": 3.152902028177325, "learning_rate": 2.1822579990302794e-07, "loss": 0.1493, "step": 13356 }, { "epoch": 1.82, "grad_norm": 3.161560067241424, "learning_rate": 2.179040372693736e-07, "loss": 0.126, "step": 13357 }, { "epoch": 1.82, "grad_norm": 3.51160376837311, "learning_rate": 2.175825067377707e-07, "loss": 0.1686, "step": 13358 }, { "epoch": 1.82, "grad_norm": 3.6475207796484175, "learning_rate": 2.1726120832382513e-07, "loss": 0.134, "step": 13359 }, { "epoch": 1.82, "grad_norm": 3.2475635066110673, "learning_rate": 2.1694014204313163e-07, "loss": 0.1345, "step": 13360 }, { "epoch": 1.82, "grad_norm": 4.110302184004263, "learning_rate": 2.1661930791127327e-07, "loss": 0.1615, "step": 13361 }, { "epoch": 1.82, "grad_norm": 3.2164085828614954, "learning_rate": 2.162987059438204e-07, "loss": 0.1086, "step": 13362 }, { "epoch": 1.82, "grad_norm": 4.516535118942812, "learning_rate": 2.1597833615633722e-07, "loss": 0.0983, "step": 13363 }, { "epoch": 1.82, "grad_norm": 3.3312057780568773, "learning_rate": 2.156581985643691e-07, "loss": 0.1413, "step": 13364 }, { "epoch": 1.82, "grad_norm": 2.5767492156535114, "learning_rate": 2.153382931834569e-07, "loss": 0.1204, "step": 13365 }, { "epoch": 1.82, "grad_norm": 2.643840131379209, "learning_rate": 2.1501862002912544e-07, "loss": 0.1401, "step": 13366 }, { "epoch": 1.82, "grad_norm": 4.965420738681771, "learning_rate": 2.1469917911689232e-07, "loss": 0.1171, "step": 13367 }, { "epoch": 1.82, "grad_norm": 3.055961379326575, "learning_rate": 2.1437997046226012e-07, "loss": 0.1352, "step": 13368 }, { "epoch": 1.82, "grad_norm": 3.0814081766939925, "learning_rate": 2.1406099408072256e-07, "loss": 0.1192, "step": 13369 }, { "epoch": 1.82, "grad_norm": 3.291559347871508, "learning_rate": 2.1374224998776116e-07, "loss": 0.1499, "step": 13370 }, { "epoch": 1.82, "grad_norm": 3.1808005094719065, "learning_rate": 2.1342373819884688e-07, "loss": 0.1149, "step": 13371 }, { "epoch": 1.82, "grad_norm": 2.7415380732443304, "learning_rate": 2.1310545872943788e-07, "loss": 0.102, "step": 13372 }, { "epoch": 1.82, "grad_norm": 4.646504314236283, "learning_rate": 2.1278741159498296e-07, "loss": 0.1935, "step": 13373 }, { "epoch": 1.82, "grad_norm": 2.174365070024812, "learning_rate": 2.124695968109175e-07, "loss": 0.1099, "step": 13374 }, { "epoch": 1.82, "grad_norm": 3.4408063161962934, "learning_rate": 2.121520143926692e-07, "loss": 0.1182, "step": 13375 }, { "epoch": 1.82, "grad_norm": 3.263676595309555, "learning_rate": 2.118346643556496e-07, "loss": 0.1501, "step": 13376 }, { "epoch": 1.82, "grad_norm": 3.612621198251367, "learning_rate": 2.115175467152636e-07, "loss": 0.1073, "step": 13377 }, { "epoch": 1.82, "grad_norm": 3.243403211263072, "learning_rate": 2.1120066148690056e-07, "loss": 0.1365, "step": 13378 }, { "epoch": 1.82, "grad_norm": 2.874820231377407, "learning_rate": 2.1088400868594316e-07, "loss": 0.1664, "step": 13379 }, { "epoch": 1.82, "grad_norm": 3.8574111196630794, "learning_rate": 2.1056758832775746e-07, "loss": 0.1207, "step": 13380 }, { "epoch": 1.82, "grad_norm": 3.9091879242870387, "learning_rate": 2.1025140042770343e-07, "loss": 0.1797, "step": 13381 }, { "epoch": 1.82, "grad_norm": 4.385916089096187, "learning_rate": 2.0993544500112706e-07, "loss": 0.1863, "step": 13382 }, { "epoch": 1.82, "grad_norm": 2.9728139243247083, "learning_rate": 2.0961972206336168e-07, "loss": 0.1161, "step": 13383 }, { "epoch": 1.82, "grad_norm": 4.112384165534354, "learning_rate": 2.0930423162973445e-07, "loss": 0.1286, "step": 13384 }, { "epoch": 1.82, "grad_norm": 2.796796674350816, "learning_rate": 2.0898897371555427e-07, "loss": 0.1191, "step": 13385 }, { "epoch": 1.82, "grad_norm": 3.06260986103866, "learning_rate": 2.086739483361244e-07, "loss": 0.1848, "step": 13386 }, { "epoch": 1.82, "grad_norm": 2.8267881457734196, "learning_rate": 2.0835915550673492e-07, "loss": 0.1692, "step": 13387 }, { "epoch": 1.82, "grad_norm": 2.9756810224916763, "learning_rate": 2.0804459524266353e-07, "loss": 0.1585, "step": 13388 }, { "epoch": 1.82, "grad_norm": 2.985161839399234, "learning_rate": 2.0773026755917858e-07, "loss": 0.1194, "step": 13389 }, { "epoch": 1.82, "grad_norm": 3.8346496666535645, "learning_rate": 2.0741617247153511e-07, "loss": 0.1781, "step": 13390 }, { "epoch": 1.82, "grad_norm": 3.1509638570719596, "learning_rate": 2.0710230999497872e-07, "loss": 0.1156, "step": 13391 }, { "epoch": 1.82, "grad_norm": 3.4893298896558957, "learning_rate": 2.0678868014474328e-07, "loss": 0.1226, "step": 13392 }, { "epoch": 1.82, "grad_norm": 3.180736529840348, "learning_rate": 2.064752829360489e-07, "loss": 0.1513, "step": 13393 }, { "epoch": 1.82, "grad_norm": 4.144907078499116, "learning_rate": 2.0616211838411003e-07, "loss": 0.1858, "step": 13394 }, { "epoch": 1.82, "grad_norm": 2.8717769794129073, "learning_rate": 2.058491865041229e-07, "loss": 0.1093, "step": 13395 }, { "epoch": 1.82, "grad_norm": 3.6400572090216947, "learning_rate": 2.0553648731127806e-07, "loss": 0.1524, "step": 13396 }, { "epoch": 1.82, "grad_norm": 3.1731931949280487, "learning_rate": 2.0522402082075121e-07, "loss": 0.1368, "step": 13397 }, { "epoch": 1.82, "grad_norm": 2.905358406281554, "learning_rate": 2.0491178704770963e-07, "loss": 0.1485, "step": 13398 }, { "epoch": 1.82, "grad_norm": 3.2470207375358187, "learning_rate": 2.0459978600730567e-07, "loss": 0.1474, "step": 13399 }, { "epoch": 1.82, "grad_norm": 3.239367495815973, "learning_rate": 2.0428801771468388e-07, "loss": 0.118, "step": 13400 }, { "epoch": 1.82, "grad_norm": 2.955567659255716, "learning_rate": 2.03976482184976e-07, "loss": 0.1323, "step": 13401 }, { "epoch": 1.82, "grad_norm": 4.079053210866436, "learning_rate": 2.0366517943330278e-07, "loss": 0.1527, "step": 13402 }, { "epoch": 1.82, "grad_norm": 3.6764379340836393, "learning_rate": 2.033541094747732e-07, "loss": 0.1132, "step": 13403 }, { "epoch": 1.82, "grad_norm": 3.711155679572006, "learning_rate": 2.0304327232448573e-07, "loss": 0.1478, "step": 13404 }, { "epoch": 1.82, "grad_norm": 3.5387951851202644, "learning_rate": 2.0273266799752611e-07, "loss": 0.1432, "step": 13405 }, { "epoch": 1.82, "grad_norm": 2.568794077127652, "learning_rate": 2.0242229650897005e-07, "loss": 0.1606, "step": 13406 }, { "epoch": 1.82, "grad_norm": 3.4645179138053774, "learning_rate": 2.0211215787388105e-07, "loss": 0.1598, "step": 13407 }, { "epoch": 1.82, "grad_norm": 4.56542755952249, "learning_rate": 2.018022521073143e-07, "loss": 0.1909, "step": 13408 }, { "epoch": 1.82, "grad_norm": 3.6680299024903005, "learning_rate": 2.0149257922430832e-07, "loss": 0.0975, "step": 13409 }, { "epoch": 1.82, "grad_norm": 4.604391926293821, "learning_rate": 2.0118313923989552e-07, "loss": 0.1541, "step": 13410 }, { "epoch": 1.82, "grad_norm": 3.389203248784187, "learning_rate": 2.0087393216909278e-07, "loss": 0.1549, "step": 13411 }, { "epoch": 1.82, "grad_norm": 3.095220212782377, "learning_rate": 2.0056495802690923e-07, "loss": 0.1855, "step": 13412 }, { "epoch": 1.82, "grad_norm": 4.727640946368157, "learning_rate": 2.0025621682834007e-07, "loss": 0.1874, "step": 13413 }, { "epoch": 1.82, "grad_norm": 2.46099316687919, "learning_rate": 1.999477085883711e-07, "loss": 0.121, "step": 13414 }, { "epoch": 1.82, "grad_norm": 4.002711692242877, "learning_rate": 1.9963943332197587e-07, "loss": 0.1592, "step": 13415 }, { "epoch": 1.82, "grad_norm": 3.930236724707353, "learning_rate": 1.9933139104411635e-07, "loss": 0.1754, "step": 13416 }, { "epoch": 1.83, "grad_norm": 3.6659290742294592, "learning_rate": 1.9902358176974335e-07, "loss": 0.1578, "step": 13417 }, { "epoch": 1.83, "grad_norm": 2.979411223436375, "learning_rate": 1.9871600551379712e-07, "loss": 0.1476, "step": 13418 }, { "epoch": 1.83, "grad_norm": 5.402606882685825, "learning_rate": 1.984086622912057e-07, "loss": 0.1453, "step": 13419 }, { "epoch": 1.83, "grad_norm": 3.9733600417772172, "learning_rate": 1.9810155211688665e-07, "loss": 0.1843, "step": 13420 }, { "epoch": 1.83, "grad_norm": 3.544976263490584, "learning_rate": 1.9779467500574523e-07, "loss": 0.1239, "step": 13421 }, { "epoch": 1.83, "grad_norm": 3.2040791159068718, "learning_rate": 1.974880309726762e-07, "loss": 0.1353, "step": 13422 }, { "epoch": 1.83, "grad_norm": 3.99192412438743, "learning_rate": 1.971816200325627e-07, "loss": 0.1163, "step": 13423 }, { "epoch": 1.83, "grad_norm": 2.636423695926224, "learning_rate": 1.968754422002761e-07, "loss": 0.136, "step": 13424 }, { "epoch": 1.83, "grad_norm": 2.7377525595560828, "learning_rate": 1.965694974906779e-07, "loss": 0.142, "step": 13425 }, { "epoch": 1.83, "grad_norm": 3.05298387591352, "learning_rate": 1.9626378591861616e-07, "loss": 0.184, "step": 13426 }, { "epoch": 1.83, "grad_norm": 3.549380382064684, "learning_rate": 1.959583074989302e-07, "loss": 0.1296, "step": 13427 }, { "epoch": 1.83, "grad_norm": 3.1013578796081633, "learning_rate": 1.9565306224644476e-07, "loss": 0.1663, "step": 13428 }, { "epoch": 1.83, "grad_norm": 3.6260324533446076, "learning_rate": 1.9534805017597745e-07, "loss": 0.1737, "step": 13429 }, { "epoch": 1.83, "grad_norm": 2.719963492715461, "learning_rate": 1.9504327130232981e-07, "loss": 0.1431, "step": 13430 }, { "epoch": 1.83, "grad_norm": 3.273765186804302, "learning_rate": 1.947387256402966e-07, "loss": 0.1608, "step": 13431 }, { "epoch": 1.83, "grad_norm": 3.615444583211379, "learning_rate": 1.9443441320465716e-07, "loss": 0.1695, "step": 13432 }, { "epoch": 1.83, "grad_norm": 2.924058587231205, "learning_rate": 1.94130334010183e-07, "loss": 0.1287, "step": 13433 }, { "epoch": 1.83, "grad_norm": 2.693174473027435, "learning_rate": 1.9382648807163228e-07, "loss": 0.1194, "step": 13434 }, { "epoch": 1.83, "grad_norm": 3.4236452355193467, "learning_rate": 1.935228754037527e-07, "loss": 0.129, "step": 13435 }, { "epoch": 1.83, "grad_norm": 2.876767747568921, "learning_rate": 1.9321949602127964e-07, "loss": 0.1305, "step": 13436 }, { "epoch": 1.83, "grad_norm": 3.2473195042488605, "learning_rate": 1.9291634993893803e-07, "loss": 0.1603, "step": 13437 }, { "epoch": 1.83, "grad_norm": 4.183476964516499, "learning_rate": 1.9261343717144109e-07, "loss": 0.187, "step": 13438 }, { "epoch": 1.83, "grad_norm": 3.5731308544826055, "learning_rate": 1.9231075773349152e-07, "loss": 0.1733, "step": 13439 }, { "epoch": 1.83, "grad_norm": 2.7519496653874826, "learning_rate": 1.9200831163977862e-07, "loss": 0.1267, "step": 13440 }, { "epoch": 1.83, "grad_norm": 3.4099414511345505, "learning_rate": 1.917060989049846e-07, "loss": 0.1396, "step": 13441 }, { "epoch": 1.83, "grad_norm": 2.718498158809535, "learning_rate": 1.9140411954377437e-07, "loss": 0.1296, "step": 13442 }, { "epoch": 1.83, "grad_norm": 2.989402490098003, "learning_rate": 1.911023735708073e-07, "loss": 0.128, "step": 13443 }, { "epoch": 1.83, "grad_norm": 3.468956205076209, "learning_rate": 1.9080086100072613e-07, "loss": 0.1163, "step": 13444 }, { "epoch": 1.83, "grad_norm": 7.809892018444984, "learning_rate": 1.9049958184816753e-07, "loss": 0.1907, "step": 13445 }, { "epoch": 1.83, "grad_norm": 3.345414094356554, "learning_rate": 1.901985361277531e-07, "loss": 0.1969, "step": 13446 }, { "epoch": 1.83, "grad_norm": 3.249767775495393, "learning_rate": 1.8989772385409445e-07, "loss": 0.1298, "step": 13447 }, { "epoch": 1.83, "grad_norm": 2.7523285759547926, "learning_rate": 1.8959714504179104e-07, "loss": 0.1133, "step": 13448 }, { "epoch": 1.83, "grad_norm": 3.950923854705081, "learning_rate": 1.892967997054329e-07, "loss": 0.1606, "step": 13449 }, { "epoch": 1.83, "grad_norm": 3.514310264473379, "learning_rate": 1.8899668785959668e-07, "loss": 0.1648, "step": 13450 }, { "epoch": 1.83, "grad_norm": 3.234495107044928, "learning_rate": 1.886968095188485e-07, "loss": 0.1599, "step": 13451 }, { "epoch": 1.83, "grad_norm": 3.2061120190527594, "learning_rate": 1.883971646977434e-07, "loss": 0.1379, "step": 13452 }, { "epoch": 1.83, "grad_norm": 3.474113071922474, "learning_rate": 1.880977534108247e-07, "loss": 0.1868, "step": 13453 }, { "epoch": 1.83, "grad_norm": 3.8023509165529132, "learning_rate": 1.8779857567262417e-07, "loss": 0.148, "step": 13454 }, { "epoch": 1.83, "grad_norm": 3.2142172221359124, "learning_rate": 1.8749963149766403e-07, "loss": 0.1337, "step": 13455 }, { "epoch": 1.83, "grad_norm": 4.1426229744893215, "learning_rate": 1.8720092090045106e-07, "loss": 0.1371, "step": 13456 }, { "epoch": 1.83, "grad_norm": 3.8264890324243703, "learning_rate": 1.8690244389548694e-07, "loss": 0.1196, "step": 13457 }, { "epoch": 1.83, "grad_norm": 3.5039755981022584, "learning_rate": 1.866042004972546e-07, "loss": 0.1658, "step": 13458 }, { "epoch": 1.83, "grad_norm": 3.3674764027739874, "learning_rate": 1.8630619072023238e-07, "loss": 0.153, "step": 13459 }, { "epoch": 1.83, "grad_norm": 3.498682241380119, "learning_rate": 1.8600841457888264e-07, "loss": 0.1431, "step": 13460 }, { "epoch": 1.83, "grad_norm": 4.413001937607064, "learning_rate": 1.8571087208765936e-07, "loss": 0.1268, "step": 13461 }, { "epoch": 1.83, "grad_norm": 3.993135850237351, "learning_rate": 1.8541356326100436e-07, "loss": 0.1378, "step": 13462 }, { "epoch": 1.83, "grad_norm": 3.4230443544171094, "learning_rate": 1.8511648811334492e-07, "loss": 0.1376, "step": 13463 }, { "epoch": 1.83, "grad_norm": 3.704763969697814, "learning_rate": 1.8481964665910235e-07, "loss": 0.143, "step": 13464 }, { "epoch": 1.83, "grad_norm": 3.3045590007653103, "learning_rate": 1.8452303891268342e-07, "loss": 0.1338, "step": 13465 }, { "epoch": 1.83, "grad_norm": 3.6024428684017464, "learning_rate": 1.842266648884844e-07, "loss": 0.1613, "step": 13466 }, { "epoch": 1.83, "grad_norm": 2.8912166436722293, "learning_rate": 1.8393052460088877e-07, "loss": 0.1635, "step": 13467 }, { "epoch": 1.83, "grad_norm": 2.545692152284542, "learning_rate": 1.8363461806427173e-07, "loss": 0.133, "step": 13468 }, { "epoch": 1.83, "grad_norm": 2.701750998042961, "learning_rate": 1.83338945292994e-07, "loss": 0.1222, "step": 13469 }, { "epoch": 1.83, "grad_norm": 2.429423895200848, "learning_rate": 1.8304350630140631e-07, "loss": 0.1471, "step": 13470 }, { "epoch": 1.83, "grad_norm": 3.7269545653411233, "learning_rate": 1.8274830110384833e-07, "loss": 0.1419, "step": 13471 }, { "epoch": 1.83, "grad_norm": 3.9275940731273185, "learning_rate": 1.8245332971464803e-07, "loss": 0.1618, "step": 13472 }, { "epoch": 1.83, "grad_norm": 3.2795937977027805, "learning_rate": 1.8215859214812114e-07, "loss": 0.1821, "step": 13473 }, { "epoch": 1.83, "grad_norm": 3.913525400808142, "learning_rate": 1.8186408841857517e-07, "loss": 0.1779, "step": 13474 }, { "epoch": 1.83, "grad_norm": 3.7085203196908543, "learning_rate": 1.815698185403014e-07, "loss": 0.1444, "step": 13475 }, { "epoch": 1.83, "grad_norm": 4.236679197295303, "learning_rate": 1.812757825275846e-07, "loss": 0.1837, "step": 13476 }, { "epoch": 1.83, "grad_norm": 3.9609123520603573, "learning_rate": 1.8098198039469438e-07, "loss": 0.1478, "step": 13477 }, { "epoch": 1.83, "grad_norm": 3.8452734530810355, "learning_rate": 1.8068841215589106e-07, "loss": 0.1382, "step": 13478 }, { "epoch": 1.83, "grad_norm": 2.9709095878722525, "learning_rate": 1.8039507782542377e-07, "loss": 0.1165, "step": 13479 }, { "epoch": 1.83, "grad_norm": 3.581530628500995, "learning_rate": 1.8010197741752944e-07, "loss": 0.1616, "step": 13480 }, { "epoch": 1.83, "grad_norm": 3.4113433492442375, "learning_rate": 1.798091109464334e-07, "loss": 0.1505, "step": 13481 }, { "epoch": 1.83, "grad_norm": 3.899797249844771, "learning_rate": 1.7951647842635035e-07, "loss": 0.1417, "step": 13482 }, { "epoch": 1.83, "grad_norm": 3.4821610047901346, "learning_rate": 1.7922407987148394e-07, "loss": 0.1396, "step": 13483 }, { "epoch": 1.83, "grad_norm": 3.4116228879080754, "learning_rate": 1.7893191529602506e-07, "loss": 0.1739, "step": 13484 }, { "epoch": 1.83, "grad_norm": 3.173954516930052, "learning_rate": 1.7863998471415456e-07, "loss": 0.1468, "step": 13485 }, { "epoch": 1.83, "grad_norm": 3.964541435869074, "learning_rate": 1.7834828814004114e-07, "loss": 0.145, "step": 13486 }, { "epoch": 1.83, "grad_norm": 3.427832676603887, "learning_rate": 1.780568255878423e-07, "loss": 0.1681, "step": 13487 }, { "epoch": 1.83, "grad_norm": 3.502915195946693, "learning_rate": 1.777655970717057e-07, "loss": 0.1678, "step": 13488 }, { "epoch": 1.83, "grad_norm": 3.462395028126152, "learning_rate": 1.77474602605765e-07, "loss": 0.1628, "step": 13489 }, { "epoch": 1.83, "grad_norm": 2.911607502609167, "learning_rate": 1.771838422041444e-07, "loss": 0.1158, "step": 13490 }, { "epoch": 1.84, "grad_norm": 2.4188981964722718, "learning_rate": 1.7689331588095548e-07, "loss": 0.1298, "step": 13491 }, { "epoch": 1.84, "grad_norm": 3.870470478298881, "learning_rate": 1.7660302365029969e-07, "loss": 0.1503, "step": 13492 }, { "epoch": 1.84, "grad_norm": 3.304341667448493, "learning_rate": 1.7631296552626687e-07, "loss": 0.1549, "step": 13493 }, { "epoch": 1.84, "grad_norm": 2.1435146320297016, "learning_rate": 1.760231415229341e-07, "loss": 0.113, "step": 13494 }, { "epoch": 1.84, "grad_norm": 2.9087971125635232, "learning_rate": 1.75733551654369e-07, "loss": 0.1422, "step": 13495 }, { "epoch": 1.84, "grad_norm": 4.110135574935015, "learning_rate": 1.7544419593462702e-07, "loss": 0.1527, "step": 13496 }, { "epoch": 1.84, "grad_norm": 4.945870983312663, "learning_rate": 1.7515507437775193e-07, "loss": 0.182, "step": 13497 }, { "epoch": 1.84, "grad_norm": 3.608563134776034, "learning_rate": 1.7486618699777634e-07, "loss": 0.1741, "step": 13498 }, { "epoch": 1.84, "grad_norm": 3.1335587641969096, "learning_rate": 1.7457753380872188e-07, "loss": 0.1445, "step": 13499 }, { "epoch": 1.84, "grad_norm": 4.718398487794374, "learning_rate": 1.7428911482459843e-07, "loss": 0.1356, "step": 13500 }, { "epoch": 1.84, "grad_norm": 3.14907121743844, "learning_rate": 1.7400093005940476e-07, "loss": 0.1323, "step": 13501 }, { "epoch": 1.84, "grad_norm": 3.66809015682203, "learning_rate": 1.7371297952712752e-07, "loss": 0.1268, "step": 13502 }, { "epoch": 1.84, "grad_norm": 3.7075900241359685, "learning_rate": 1.7342526324174268e-07, "loss": 0.1385, "step": 13503 }, { "epoch": 1.84, "grad_norm": 3.802975100164563, "learning_rate": 1.7313778121721524e-07, "loss": 0.1387, "step": 13504 }, { "epoch": 1.84, "grad_norm": 4.657310323773131, "learning_rate": 1.728505334674979e-07, "loss": 0.1547, "step": 13505 }, { "epoch": 1.84, "grad_norm": 2.629414385909141, "learning_rate": 1.725635200065323e-07, "loss": 0.1449, "step": 13506 }, { "epoch": 1.84, "grad_norm": 2.3742888224303496, "learning_rate": 1.722767408482501e-07, "loss": 0.0912, "step": 13507 }, { "epoch": 1.84, "grad_norm": 4.104130917085763, "learning_rate": 1.719901960065684e-07, "loss": 0.1615, "step": 13508 }, { "epoch": 1.84, "grad_norm": 3.618547205628074, "learning_rate": 1.7170388549539619e-07, "loss": 0.1448, "step": 13509 }, { "epoch": 1.84, "grad_norm": 4.010282882439662, "learning_rate": 1.7141780932862896e-07, "loss": 0.1511, "step": 13510 }, { "epoch": 1.84, "grad_norm": 7.790189573593783, "learning_rate": 1.7113196752015172e-07, "loss": 0.1661, "step": 13511 }, { "epoch": 1.84, "grad_norm": 3.1212328034920023, "learning_rate": 1.7084636008383837e-07, "loss": 0.1671, "step": 13512 }, { "epoch": 1.84, "grad_norm": 3.719074202932711, "learning_rate": 1.7056098703355118e-07, "loss": 0.1465, "step": 13513 }, { "epoch": 1.84, "grad_norm": 3.1023223885737603, "learning_rate": 1.7027584838314015e-07, "loss": 0.1734, "step": 13514 }, { "epoch": 1.84, "grad_norm": 2.966789334904392, "learning_rate": 1.6999094414644534e-07, "loss": 0.1619, "step": 13515 }, { "epoch": 1.84, "grad_norm": 2.953428406657333, "learning_rate": 1.697062743372946e-07, "loss": 0.1322, "step": 13516 }, { "epoch": 1.84, "grad_norm": 2.5078596632335968, "learning_rate": 1.6942183896950458e-07, "loss": 0.1131, "step": 13517 }, { "epoch": 1.84, "grad_norm": 2.370172905407194, "learning_rate": 1.6913763805687988e-07, "loss": 0.1496, "step": 13518 }, { "epoch": 1.84, "grad_norm": 3.9343130686410275, "learning_rate": 1.6885367161321552e-07, "loss": 0.1142, "step": 13519 }, { "epoch": 1.84, "grad_norm": 2.6880537250367125, "learning_rate": 1.6856993965229218e-07, "loss": 0.1057, "step": 13520 }, { "epoch": 1.84, "grad_norm": 2.540384638311622, "learning_rate": 1.682864421878838e-07, "loss": 0.179, "step": 13521 }, { "epoch": 1.84, "grad_norm": 3.486582144177051, "learning_rate": 1.680031792337472e-07, "loss": 0.1691, "step": 13522 }, { "epoch": 1.84, "grad_norm": 2.7649541527648056, "learning_rate": 1.6772015080363303e-07, "loss": 0.1107, "step": 13523 }, { "epoch": 1.84, "grad_norm": 2.7642800341017186, "learning_rate": 1.6743735691127639e-07, "loss": 0.1634, "step": 13524 }, { "epoch": 1.84, "grad_norm": 2.897135911339002, "learning_rate": 1.671547975704041e-07, "loss": 0.1227, "step": 13525 }, { "epoch": 1.84, "grad_norm": 3.1841031070641295, "learning_rate": 1.6687247279472962e-07, "loss": 0.1429, "step": 13526 }, { "epoch": 1.84, "grad_norm": 2.708383817440138, "learning_rate": 1.6659038259795644e-07, "loss": 0.1684, "step": 13527 }, { "epoch": 1.84, "grad_norm": 3.977887835716653, "learning_rate": 1.6630852699377586e-07, "loss": 0.1266, "step": 13528 }, { "epoch": 1.84, "grad_norm": 4.954310304552419, "learning_rate": 1.6602690599586747e-07, "loss": 0.1532, "step": 13529 }, { "epoch": 1.84, "grad_norm": 4.09493293996371, "learning_rate": 1.6574551961789974e-07, "loss": 0.144, "step": 13530 }, { "epoch": 1.84, "grad_norm": 3.4624186872120943, "learning_rate": 1.6546436787353072e-07, "loss": 0.1314, "step": 13531 }, { "epoch": 1.84, "grad_norm": 3.505173213089094, "learning_rate": 1.6518345077640606e-07, "loss": 0.1565, "step": 13532 }, { "epoch": 1.84, "grad_norm": 4.240598069761964, "learning_rate": 1.6490276834015994e-07, "loss": 0.1315, "step": 13533 }, { "epoch": 1.84, "grad_norm": 3.825415709063473, "learning_rate": 1.6462232057841532e-07, "loss": 0.1779, "step": 13534 }, { "epoch": 1.84, "grad_norm": 3.0927924768507, "learning_rate": 1.6434210750478518e-07, "loss": 0.1508, "step": 13535 }, { "epoch": 1.84, "grad_norm": 3.931862165050008, "learning_rate": 1.6406212913286757e-07, "loss": 0.1287, "step": 13536 }, { "epoch": 1.84, "grad_norm": 3.098266656776366, "learning_rate": 1.6378238547625436e-07, "loss": 0.1279, "step": 13537 }, { "epoch": 1.84, "grad_norm": 2.6493166538549686, "learning_rate": 1.635028765485197e-07, "loss": 0.1233, "step": 13538 }, { "epoch": 1.84, "grad_norm": 6.710461330959729, "learning_rate": 1.632236023632322e-07, "loss": 0.1518, "step": 13539 }, { "epoch": 1.84, "grad_norm": 3.4646017276404986, "learning_rate": 1.62944562933946e-07, "loss": 0.1745, "step": 13540 }, { "epoch": 1.84, "grad_norm": 4.044369342874201, "learning_rate": 1.626657582742047e-07, "loss": 0.1556, "step": 13541 }, { "epoch": 1.84, "grad_norm": 3.249068666978164, "learning_rate": 1.6238718839753975e-07, "loss": 0.1537, "step": 13542 }, { "epoch": 1.84, "grad_norm": 4.346138390107313, "learning_rate": 1.621088533174714e-07, "loss": 0.1673, "step": 13543 }, { "epoch": 1.84, "grad_norm": 3.521750307177005, "learning_rate": 1.6183075304750938e-07, "loss": 0.1429, "step": 13544 }, { "epoch": 1.84, "grad_norm": 2.19553872002986, "learning_rate": 1.6155288760115183e-07, "loss": 0.105, "step": 13545 }, { "epoch": 1.84, "grad_norm": 3.4492908834913902, "learning_rate": 1.6127525699188407e-07, "loss": 0.1098, "step": 13546 }, { "epoch": 1.84, "grad_norm": 3.3221253094910614, "learning_rate": 1.609978612331825e-07, "loss": 0.1558, "step": 13547 }, { "epoch": 1.84, "grad_norm": 3.039277559008478, "learning_rate": 1.6072070033850917e-07, "loss": 0.1512, "step": 13548 }, { "epoch": 1.84, "grad_norm": 4.062332523641109, "learning_rate": 1.604437743213172e-07, "loss": 0.1264, "step": 13549 }, { "epoch": 1.84, "grad_norm": 2.6868818720263756, "learning_rate": 1.6016708319504693e-07, "loss": 0.177, "step": 13550 }, { "epoch": 1.84, "grad_norm": 2.660262809910094, "learning_rate": 1.5989062697312818e-07, "loss": 0.1236, "step": 13551 }, { "epoch": 1.84, "grad_norm": 3.4235960331530824, "learning_rate": 1.5961440566897913e-07, "loss": 0.1304, "step": 13552 }, { "epoch": 1.84, "grad_norm": 3.3647551010183534, "learning_rate": 1.593384192960046e-07, "loss": 0.1289, "step": 13553 }, { "epoch": 1.84, "grad_norm": 2.9442433917174187, "learning_rate": 1.5906266786760273e-07, "loss": 0.1614, "step": 13554 }, { "epoch": 1.84, "grad_norm": 3.0436000277883606, "learning_rate": 1.5878715139715395e-07, "loss": 0.1383, "step": 13555 }, { "epoch": 1.84, "grad_norm": 3.3407555495335965, "learning_rate": 1.5851186989803369e-07, "loss": 0.1475, "step": 13556 }, { "epoch": 1.84, "grad_norm": 2.9762504179139952, "learning_rate": 1.582368233836007e-07, "loss": 0.1553, "step": 13557 }, { "epoch": 1.84, "grad_norm": 3.9439512127279865, "learning_rate": 1.5796201186720595e-07, "loss": 0.1403, "step": 13558 }, { "epoch": 1.84, "grad_norm": 4.059686483511197, "learning_rate": 1.5768743536218657e-07, "loss": 0.1545, "step": 13559 }, { "epoch": 1.84, "grad_norm": 3.46694341425255, "learning_rate": 1.5741309388186964e-07, "loss": 0.1403, "step": 13560 }, { "epoch": 1.84, "grad_norm": 2.944824818915871, "learning_rate": 1.5713898743957124e-07, "loss": 0.0991, "step": 13561 }, { "epoch": 1.84, "grad_norm": 3.6401290434741203, "learning_rate": 1.5686511604859456e-07, "loss": 0.1718, "step": 13562 }, { "epoch": 1.84, "grad_norm": 3.905624816631028, "learning_rate": 1.5659147972223177e-07, "loss": 0.1708, "step": 13563 }, { "epoch": 1.85, "grad_norm": 3.0575676472580233, "learning_rate": 1.5631807847376446e-07, "loss": 0.1406, "step": 13564 }, { "epoch": 1.85, "grad_norm": 3.266206761061434, "learning_rate": 1.5604491231646256e-07, "loss": 0.1591, "step": 13565 }, { "epoch": 1.85, "grad_norm": 4.185756717825795, "learning_rate": 1.5577198126358438e-07, "loss": 0.1702, "step": 13566 }, { "epoch": 1.85, "grad_norm": 2.7980016418137224, "learning_rate": 1.5549928532837544e-07, "loss": 0.1466, "step": 13567 }, { "epoch": 1.85, "grad_norm": 3.236630761536605, "learning_rate": 1.5522682452407402e-07, "loss": 0.1332, "step": 13568 }, { "epoch": 1.85, "grad_norm": 5.61676259512781, "learning_rate": 1.5495459886390063e-07, "loss": 0.152, "step": 13569 }, { "epoch": 1.85, "grad_norm": 3.6503720317659933, "learning_rate": 1.546826083610714e-07, "loss": 0.1094, "step": 13570 }, { "epoch": 1.85, "grad_norm": 3.7840764736550763, "learning_rate": 1.5441085302878466e-07, "loss": 0.1382, "step": 13571 }, { "epoch": 1.85, "grad_norm": 2.97088363688359, "learning_rate": 1.5413933288023207e-07, "loss": 0.1391, "step": 13572 }, { "epoch": 1.85, "grad_norm": 2.6779508465283777, "learning_rate": 1.5386804792859144e-07, "loss": 0.1234, "step": 13573 }, { "epoch": 1.85, "grad_norm": 3.6476472575934134, "learning_rate": 1.5359699818702944e-07, "loss": 0.1778, "step": 13574 }, { "epoch": 1.85, "grad_norm": 3.1154849203805868, "learning_rate": 1.533261836687022e-07, "loss": 0.1021, "step": 13575 }, { "epoch": 1.85, "grad_norm": 3.915089599520134, "learning_rate": 1.5305560438675372e-07, "loss": 0.1663, "step": 13576 }, { "epoch": 1.85, "grad_norm": 3.131646546736224, "learning_rate": 1.5278526035431673e-07, "loss": 0.125, "step": 13577 }, { "epoch": 1.85, "grad_norm": 2.9991251324897594, "learning_rate": 1.525151515845119e-07, "loss": 0.1458, "step": 13578 }, { "epoch": 1.85, "grad_norm": 3.274454228065918, "learning_rate": 1.5224527809044986e-07, "loss": 0.1369, "step": 13579 }, { "epoch": 1.85, "grad_norm": 4.275923536299921, "learning_rate": 1.5197563988522955e-07, "loss": 0.1624, "step": 13580 }, { "epoch": 1.85, "grad_norm": 3.970917010108656, "learning_rate": 1.5170623698193722e-07, "loss": 0.1239, "step": 13581 }, { "epoch": 1.85, "grad_norm": 3.8006305533813682, "learning_rate": 1.5143706939364844e-07, "loss": 0.1821, "step": 13582 }, { "epoch": 1.85, "grad_norm": 4.925238322400765, "learning_rate": 1.5116813713342782e-07, "loss": 0.1513, "step": 13583 }, { "epoch": 1.85, "grad_norm": 4.110576590323459, "learning_rate": 1.5089944021432767e-07, "loss": 0.175, "step": 13584 }, { "epoch": 1.85, "grad_norm": 3.664920462556057, "learning_rate": 1.506309786493898e-07, "loss": 0.1815, "step": 13585 }, { "epoch": 1.85, "grad_norm": 2.6357232097122796, "learning_rate": 1.5036275245164377e-07, "loss": 0.0883, "step": 13586 }, { "epoch": 1.85, "grad_norm": 5.553463209505059, "learning_rate": 1.5009476163410975e-07, "loss": 0.1363, "step": 13587 }, { "epoch": 1.85, "grad_norm": 2.8632612084406666, "learning_rate": 1.4982700620979175e-07, "loss": 0.1448, "step": 13588 }, { "epoch": 1.85, "grad_norm": 4.215264641965827, "learning_rate": 1.495594861916888e-07, "loss": 0.1539, "step": 13589 }, { "epoch": 1.85, "grad_norm": 5.987013121760618, "learning_rate": 1.4929220159278224e-07, "loss": 0.0945, "step": 13590 }, { "epoch": 1.85, "grad_norm": 3.4428318389093375, "learning_rate": 1.4902515242604666e-07, "loss": 0.1469, "step": 13591 }, { "epoch": 1.85, "grad_norm": 2.6087782439543177, "learning_rate": 1.4875833870444334e-07, "loss": 0.1439, "step": 13592 }, { "epoch": 1.85, "grad_norm": 3.041108242885539, "learning_rate": 1.4849176044092195e-07, "loss": 0.1412, "step": 13593 }, { "epoch": 1.85, "grad_norm": 3.396780267350313, "learning_rate": 1.4822541764842047e-07, "loss": 0.1601, "step": 13594 }, { "epoch": 1.85, "grad_norm": 2.57719257985481, "learning_rate": 1.4795931033986744e-07, "loss": 0.1097, "step": 13595 }, { "epoch": 1.85, "grad_norm": 3.6636780730901686, "learning_rate": 1.4769343852817697e-07, "loss": 0.1478, "step": 13596 }, { "epoch": 1.85, "grad_norm": 2.7743451290485757, "learning_rate": 1.474278022262543e-07, "loss": 0.1474, "step": 13597 }, { "epoch": 1.85, "grad_norm": 2.914402550820858, "learning_rate": 1.471624014469919e-07, "loss": 0.1257, "step": 13598 }, { "epoch": 1.85, "grad_norm": 3.6304302453141197, "learning_rate": 1.4689723620327113e-07, "loss": 0.114, "step": 13599 }, { "epoch": 1.85, "grad_norm": 4.55222468837999, "learning_rate": 1.4663230650796167e-07, "loss": 0.173, "step": 13600 }, { "epoch": 1.85, "grad_norm": 3.411336400756495, "learning_rate": 1.4636761237392328e-07, "loss": 0.1068, "step": 13601 }, { "epoch": 1.85, "grad_norm": 3.262283996371542, "learning_rate": 1.4610315381400175e-07, "loss": 0.1386, "step": 13602 }, { "epoch": 1.85, "grad_norm": 4.03331558049215, "learning_rate": 1.4583893084103352e-07, "loss": 0.1863, "step": 13603 }, { "epoch": 1.85, "grad_norm": 4.61736027470223, "learning_rate": 1.455749434678416e-07, "loss": 0.1363, "step": 13604 }, { "epoch": 1.85, "grad_norm": 3.7376594876312383, "learning_rate": 1.4531119170724083e-07, "loss": 0.1754, "step": 13605 }, { "epoch": 1.85, "grad_norm": 3.505100580654623, "learning_rate": 1.4504767557203092e-07, "loss": 0.1324, "step": 13606 }, { "epoch": 1.85, "grad_norm": 5.403205397023527, "learning_rate": 1.4478439507500218e-07, "loss": 0.1573, "step": 13607 }, { "epoch": 1.85, "grad_norm": 3.853741434608973, "learning_rate": 1.4452135022893387e-07, "loss": 0.1555, "step": 13608 }, { "epoch": 1.85, "grad_norm": 3.0416737992649434, "learning_rate": 1.4425854104659187e-07, "loss": 0.1187, "step": 13609 }, { "epoch": 1.85, "grad_norm": 3.368633215343349, "learning_rate": 1.439959675407321e-07, "loss": 0.1715, "step": 13610 }, { "epoch": 1.85, "grad_norm": 3.529586947779578, "learning_rate": 1.437336297240993e-07, "loss": 0.1708, "step": 13611 }, { "epoch": 1.85, "grad_norm": 3.597308194801457, "learning_rate": 1.4347152760942507e-07, "loss": 0.1566, "step": 13612 }, { "epoch": 1.85, "grad_norm": 3.786518524320496, "learning_rate": 1.4320966120943304e-07, "loss": 0.1827, "step": 13613 }, { "epoch": 1.85, "grad_norm": 4.236718022208604, "learning_rate": 1.429480305368297e-07, "loss": 0.1708, "step": 13614 }, { "epoch": 1.85, "grad_norm": 2.712418206008547, "learning_rate": 1.4268663560431718e-07, "loss": 0.1517, "step": 13615 }, { "epoch": 1.85, "grad_norm": 3.4525830382841027, "learning_rate": 1.424254764245786e-07, "loss": 0.1757, "step": 13616 }, { "epoch": 1.85, "grad_norm": 3.111896477883797, "learning_rate": 1.4216455301029274e-07, "loss": 0.1534, "step": 13617 }, { "epoch": 1.85, "grad_norm": 2.9482616620994175, "learning_rate": 1.419038653741217e-07, "loss": 0.1505, "step": 13618 }, { "epoch": 1.85, "grad_norm": 3.871996725753897, "learning_rate": 1.4164341352871925e-07, "loss": 0.1422, "step": 13619 }, { "epoch": 1.85, "grad_norm": 3.7580911462331, "learning_rate": 1.413831974867258e-07, "loss": 0.1795, "step": 13620 }, { "epoch": 1.85, "grad_norm": 4.062006241144133, "learning_rate": 1.4112321726077127e-07, "loss": 0.1331, "step": 13621 }, { "epoch": 1.85, "grad_norm": 3.6426157409653346, "learning_rate": 1.4086347286347502e-07, "loss": 0.1415, "step": 13622 }, { "epoch": 1.85, "grad_norm": 3.3756336625590775, "learning_rate": 1.406039643074414e-07, "loss": 0.1371, "step": 13623 }, { "epoch": 1.85, "grad_norm": 2.8009655498552264, "learning_rate": 1.4034469160526864e-07, "loss": 0.1353, "step": 13624 }, { "epoch": 1.85, "grad_norm": 3.64855148524458, "learning_rate": 1.4008565476953895e-07, "loss": 0.1413, "step": 13625 }, { "epoch": 1.85, "grad_norm": 3.092678547993879, "learning_rate": 1.3982685381282557e-07, "loss": 0.1322, "step": 13626 }, { "epoch": 1.85, "grad_norm": 3.117133787341269, "learning_rate": 1.3956828874768901e-07, "loss": 0.185, "step": 13627 }, { "epoch": 1.85, "grad_norm": 2.778513623985916, "learning_rate": 1.393099595866798e-07, "loss": 0.1294, "step": 13628 }, { "epoch": 1.85, "grad_norm": 3.1383825625599155, "learning_rate": 1.3905186634233515e-07, "loss": 0.1417, "step": 13629 }, { "epoch": 1.85, "grad_norm": 3.8863856924688833, "learning_rate": 1.387940090271822e-07, "loss": 0.144, "step": 13630 }, { "epoch": 1.85, "grad_norm": 3.568712959238118, "learning_rate": 1.38536387653736e-07, "loss": 0.1616, "step": 13631 }, { "epoch": 1.85, "grad_norm": 3.80718617989656, "learning_rate": 1.3827900223450152e-07, "loss": 0.1171, "step": 13632 }, { "epoch": 1.85, "grad_norm": 3.6925533469858416, "learning_rate": 1.3802185278196934e-07, "loss": 0.1013, "step": 13633 }, { "epoch": 1.85, "grad_norm": 3.5794878629148714, "learning_rate": 1.3776493930862168e-07, "loss": 0.1186, "step": 13634 }, { "epoch": 1.85, "grad_norm": 3.3770493953388097, "learning_rate": 1.3750826182692688e-07, "loss": 0.1835, "step": 13635 }, { "epoch": 1.85, "grad_norm": 3.284764116167596, "learning_rate": 1.3725182034934447e-07, "loss": 0.1554, "step": 13636 }, { "epoch": 1.85, "grad_norm": 2.5742486739696546, "learning_rate": 1.3699561488831892e-07, "loss": 0.129, "step": 13637 }, { "epoch": 1.86, "grad_norm": 3.620601224432504, "learning_rate": 1.367396454562875e-07, "loss": 0.1705, "step": 13638 }, { "epoch": 1.86, "grad_norm": 2.6267200985650287, "learning_rate": 1.3648391206567247e-07, "loss": 0.1099, "step": 13639 }, { "epoch": 1.86, "grad_norm": 3.2896922902757284, "learning_rate": 1.3622841472888671e-07, "loss": 0.1684, "step": 13640 }, { "epoch": 1.86, "grad_norm": 3.883862917154777, "learning_rate": 1.359731534583303e-07, "loss": 0.1515, "step": 13641 }, { "epoch": 1.86, "grad_norm": 3.0130594725768916, "learning_rate": 1.357181282663933e-07, "loss": 0.1458, "step": 13642 }, { "epoch": 1.86, "grad_norm": 3.9371006577740006, "learning_rate": 1.3546333916545308e-07, "loss": 0.1806, "step": 13643 }, { "epoch": 1.86, "grad_norm": 3.5618978808590445, "learning_rate": 1.3520878616787525e-07, "loss": 0.2196, "step": 13644 }, { "epoch": 1.86, "grad_norm": 4.22061455730204, "learning_rate": 1.349544692860155e-07, "loss": 0.1497, "step": 13645 }, { "epoch": 1.86, "grad_norm": 3.4111429956552706, "learning_rate": 1.3470038853221845e-07, "loss": 0.1105, "step": 13646 }, { "epoch": 1.86, "grad_norm": 3.6782598287927444, "learning_rate": 1.3444654391881306e-07, "loss": 0.1517, "step": 13647 }, { "epoch": 1.86, "grad_norm": 2.854070281039254, "learning_rate": 1.341929354581234e-07, "loss": 0.134, "step": 13648 }, { "epoch": 1.86, "grad_norm": 3.6946943925262716, "learning_rate": 1.3393956316245515e-07, "loss": 0.1677, "step": 13649 }, { "epoch": 1.86, "grad_norm": 2.936695261889526, "learning_rate": 1.3368642704410795e-07, "loss": 0.1528, "step": 13650 }, { "epoch": 1.86, "grad_norm": 3.6694204480006567, "learning_rate": 1.334335271153675e-07, "loss": 0.1676, "step": 13651 }, { "epoch": 1.86, "grad_norm": 3.898592708482722, "learning_rate": 1.3318086338850843e-07, "loss": 0.1455, "step": 13652 }, { "epoch": 1.86, "grad_norm": 3.034559096152888, "learning_rate": 1.329284358757943e-07, "loss": 0.1662, "step": 13653 }, { "epoch": 1.86, "grad_norm": 2.6819392252438066, "learning_rate": 1.326762445894758e-07, "loss": 0.1364, "step": 13654 }, { "epoch": 1.86, "grad_norm": 2.945456402074687, "learning_rate": 1.324242895417943e-07, "loss": 0.1195, "step": 13655 }, { "epoch": 1.86, "grad_norm": 3.23811987004272, "learning_rate": 1.321725707449778e-07, "loss": 0.118, "step": 13656 }, { "epoch": 1.86, "grad_norm": 2.681051984760622, "learning_rate": 1.3192108821124428e-07, "loss": 0.1123, "step": 13657 }, { "epoch": 1.86, "grad_norm": 4.064608894834438, "learning_rate": 1.3166984195279898e-07, "loss": 0.173, "step": 13658 }, { "epoch": 1.86, "grad_norm": 3.0105994302187127, "learning_rate": 1.3141883198183603e-07, "loss": 0.1615, "step": 13659 }, { "epoch": 1.86, "grad_norm": 2.3444777005086634, "learning_rate": 1.3116805831054014e-07, "loss": 0.1349, "step": 13660 }, { "epoch": 1.86, "grad_norm": 3.9182250775517327, "learning_rate": 1.30917520951081e-07, "loss": 0.1809, "step": 13661 }, { "epoch": 1.86, "grad_norm": 3.4103362024665875, "learning_rate": 1.3066721991561891e-07, "loss": 0.16, "step": 13662 }, { "epoch": 1.86, "grad_norm": 3.491880587703576, "learning_rate": 1.304171552163025e-07, "loss": 0.1699, "step": 13663 }, { "epoch": 1.86, "grad_norm": 3.0514725814155006, "learning_rate": 1.3016732686526866e-07, "loss": 0.1128, "step": 13664 }, { "epoch": 1.86, "grad_norm": 3.7686474192661814, "learning_rate": 1.2991773487464442e-07, "loss": 0.1572, "step": 13665 }, { "epoch": 1.86, "grad_norm": 2.833413765117678, "learning_rate": 1.2966837925654173e-07, "loss": 0.1422, "step": 13666 }, { "epoch": 1.86, "grad_norm": 3.225563655268161, "learning_rate": 1.2941926002306536e-07, "loss": 0.0977, "step": 13667 }, { "epoch": 1.86, "grad_norm": 2.92486673696146, "learning_rate": 1.2917037718630397e-07, "loss": 0.1329, "step": 13668 }, { "epoch": 1.86, "grad_norm": 3.4037593816117075, "learning_rate": 1.289217307583396e-07, "loss": 0.1613, "step": 13669 }, { "epoch": 1.86, "grad_norm": 3.4365069584078913, "learning_rate": 1.2867332075123863e-07, "loss": 0.1488, "step": 13670 }, { "epoch": 1.86, "grad_norm": 4.697230886126015, "learning_rate": 1.2842514717705868e-07, "loss": 0.1673, "step": 13671 }, { "epoch": 1.86, "grad_norm": 3.862524145595444, "learning_rate": 1.2817721004784568e-07, "loss": 0.1335, "step": 13672 }, { "epoch": 1.86, "grad_norm": 5.2140951192606115, "learning_rate": 1.279295093756322e-07, "loss": 0.1617, "step": 13673 }, { "epoch": 1.86, "grad_norm": 3.880498815497015, "learning_rate": 1.2768204517244087e-07, "loss": 0.1477, "step": 13674 }, { "epoch": 1.86, "grad_norm": 4.023851380296186, "learning_rate": 1.274348174502832e-07, "loss": 0.1791, "step": 13675 }, { "epoch": 1.86, "grad_norm": 4.692035023949516, "learning_rate": 1.2718782622115789e-07, "loss": 0.1346, "step": 13676 }, { "epoch": 1.86, "grad_norm": 3.1000629190530713, "learning_rate": 1.2694107149705258e-07, "loss": 0.1496, "step": 13677 }, { "epoch": 1.86, "grad_norm": 3.2550363895165475, "learning_rate": 1.2669455328994385e-07, "loss": 0.1636, "step": 13678 }, { "epoch": 1.86, "grad_norm": 3.708684518916629, "learning_rate": 1.2644827161179763e-07, "loss": 0.1909, "step": 13679 }, { "epoch": 1.86, "grad_norm": 4.216945727953139, "learning_rate": 1.2620222647456604e-07, "loss": 0.14, "step": 13680 }, { "epoch": 1.86, "grad_norm": 3.1284833917460118, "learning_rate": 1.2595641789019176e-07, "loss": 0.1584, "step": 13681 }, { "epoch": 1.86, "grad_norm": 2.955240571346667, "learning_rate": 1.2571084587060466e-07, "loss": 0.1538, "step": 13682 }, { "epoch": 1.86, "grad_norm": 2.6136954294080357, "learning_rate": 1.2546551042772415e-07, "loss": 0.1056, "step": 13683 }, { "epoch": 1.86, "grad_norm": 3.931560017405195, "learning_rate": 1.252204115734579e-07, "loss": 0.1664, "step": 13684 }, { "epoch": 1.86, "grad_norm": 3.828078171109783, "learning_rate": 1.2497554931970192e-07, "loss": 0.1441, "step": 13685 }, { "epoch": 1.86, "grad_norm": 3.823149256942461, "learning_rate": 1.247309236783406e-07, "loss": 0.133, "step": 13686 }, { "epoch": 1.86, "grad_norm": 3.8820088094271874, "learning_rate": 1.2448653466124672e-07, "loss": 0.1295, "step": 13687 }, { "epoch": 1.86, "grad_norm": 2.5544414165647447, "learning_rate": 1.242423822802824e-07, "loss": 0.1255, "step": 13688 }, { "epoch": 1.86, "grad_norm": 4.213349773620547, "learning_rate": 1.2399846654729707e-07, "loss": 0.1868, "step": 13689 }, { "epoch": 1.86, "grad_norm": 2.546752356801435, "learning_rate": 1.2375478747413017e-07, "loss": 0.1199, "step": 13690 }, { "epoch": 1.86, "grad_norm": 3.1924607993506586, "learning_rate": 1.2351134507260775e-07, "loss": 0.1435, "step": 13691 }, { "epoch": 1.86, "grad_norm": 2.8783502697473327, "learning_rate": 1.2326813935454596e-07, "loss": 0.1426, "step": 13692 }, { "epoch": 1.86, "grad_norm": 3.043592255880213, "learning_rate": 1.2302517033174976e-07, "loss": 0.1252, "step": 13693 }, { "epoch": 1.86, "grad_norm": 2.9876136113707625, "learning_rate": 1.2278243801601032e-07, "loss": 0.1381, "step": 13694 }, { "epoch": 1.86, "grad_norm": 3.4395039070927886, "learning_rate": 1.2253994241911038e-07, "loss": 0.1373, "step": 13695 }, { "epoch": 1.86, "grad_norm": 2.898807806874658, "learning_rate": 1.222976835528178e-07, "loss": 0.1644, "step": 13696 }, { "epoch": 1.86, "grad_norm": 3.494944045368664, "learning_rate": 1.2205566142889257e-07, "loss": 0.1725, "step": 13697 }, { "epoch": 1.86, "grad_norm": 4.146403050946557, "learning_rate": 1.2181387605908034e-07, "loss": 0.1179, "step": 13698 }, { "epoch": 1.86, "grad_norm": 4.136790520084567, "learning_rate": 1.2157232745511673e-07, "loss": 0.1818, "step": 13699 }, { "epoch": 1.86, "grad_norm": 3.316135238995914, "learning_rate": 1.2133101562872507e-07, "loss": 0.1682, "step": 13700 }, { "epoch": 1.86, "grad_norm": 2.9216961768918037, "learning_rate": 1.210899405916177e-07, "loss": 0.1573, "step": 13701 }, { "epoch": 1.86, "grad_norm": 4.797743105365075, "learning_rate": 1.2084910235549586e-07, "loss": 0.0751, "step": 13702 }, { "epoch": 1.86, "grad_norm": 3.103720519513556, "learning_rate": 1.206085009320479e-07, "loss": 0.1368, "step": 13703 }, { "epoch": 1.86, "grad_norm": 3.33289783441064, "learning_rate": 1.203681363329523e-07, "loss": 0.1335, "step": 13704 }, { "epoch": 1.86, "grad_norm": 2.3815634452760426, "learning_rate": 1.2012800856987472e-07, "loss": 0.109, "step": 13705 }, { "epoch": 1.86, "grad_norm": 3.5218984535782, "learning_rate": 1.1988811765447083e-07, "loss": 0.1561, "step": 13706 }, { "epoch": 1.86, "grad_norm": 3.735719303725023, "learning_rate": 1.19648463598383e-07, "loss": 0.1453, "step": 13707 }, { "epoch": 1.86, "grad_norm": 3.1754015870988055, "learning_rate": 1.19409046413243e-07, "loss": 0.1325, "step": 13708 }, { "epoch": 1.86, "grad_norm": 3.224497687634843, "learning_rate": 1.1916986611067161e-07, "loss": 0.1455, "step": 13709 }, { "epoch": 1.86, "grad_norm": 3.2466169818746944, "learning_rate": 1.1893092270227724e-07, "loss": 0.1243, "step": 13710 }, { "epoch": 1.87, "grad_norm": 3.4870140843631767, "learning_rate": 1.1869221619965732e-07, "loss": 0.1598, "step": 13711 }, { "epoch": 1.87, "grad_norm": 2.881319642347931, "learning_rate": 1.1845374661439813e-07, "loss": 0.1348, "step": 13712 }, { "epoch": 1.87, "grad_norm": 3.008252487172666, "learning_rate": 1.1821551395807207e-07, "loss": 0.1562, "step": 13713 }, { "epoch": 1.87, "grad_norm": 2.855667192019793, "learning_rate": 1.1797751824224491e-07, "loss": 0.1187, "step": 13714 }, { "epoch": 1.87, "grad_norm": 3.201223977423667, "learning_rate": 1.177397594784646e-07, "loss": 0.1408, "step": 13715 }, { "epoch": 1.87, "grad_norm": 2.6715012923735455, "learning_rate": 1.1750223767827363e-07, "loss": 0.1208, "step": 13716 }, { "epoch": 1.87, "grad_norm": 6.316831732246301, "learning_rate": 1.1726495285319883e-07, "loss": 0.1541, "step": 13717 }, { "epoch": 1.87, "grad_norm": 2.819905927433771, "learning_rate": 1.1702790501475769e-07, "loss": 0.1413, "step": 13718 }, { "epoch": 1.87, "grad_norm": 3.4126578360699567, "learning_rate": 1.1679109417445489e-07, "loss": 0.1604, "step": 13719 }, { "epoch": 1.87, "grad_norm": 3.352883041527934, "learning_rate": 1.1655452034378401e-07, "loss": 0.1506, "step": 13720 }, { "epoch": 1.87, "grad_norm": 2.837500412648355, "learning_rate": 1.1631818353422809e-07, "loss": 0.1412, "step": 13721 }, { "epoch": 1.87, "grad_norm": 2.8881156154226852, "learning_rate": 1.1608208375725794e-07, "loss": 0.1314, "step": 13722 }, { "epoch": 1.87, "grad_norm": 3.5113016598549267, "learning_rate": 1.1584622102433218e-07, "loss": 0.1491, "step": 13723 }, { "epoch": 1.87, "grad_norm": 3.157581262499099, "learning_rate": 1.1561059534689834e-07, "loss": 0.1101, "step": 13724 }, { "epoch": 1.87, "grad_norm": 3.4345324439353697, "learning_rate": 1.153752067363928e-07, "loss": 0.1316, "step": 13725 }, { "epoch": 1.87, "grad_norm": 3.5740072725322634, "learning_rate": 1.1514005520424198e-07, "loss": 0.174, "step": 13726 }, { "epoch": 1.87, "grad_norm": 3.0269686883510407, "learning_rate": 1.1490514076185621e-07, "loss": 0.1408, "step": 13727 }, { "epoch": 1.87, "grad_norm": 3.0505752937114807, "learning_rate": 1.1467046342064026e-07, "loss": 0.1254, "step": 13728 }, { "epoch": 1.87, "grad_norm": 3.171889159391038, "learning_rate": 1.1443602319198166e-07, "loss": 0.1368, "step": 13729 }, { "epoch": 1.87, "grad_norm": 2.708948417991542, "learning_rate": 1.1420182008726022e-07, "loss": 0.1337, "step": 13730 }, { "epoch": 1.87, "grad_norm": 3.807665372777192, "learning_rate": 1.139678541178435e-07, "loss": 0.1652, "step": 13731 }, { "epoch": 1.87, "grad_norm": 3.1484141742332747, "learning_rate": 1.1373412529508687e-07, "loss": 0.1368, "step": 13732 }, { "epoch": 1.87, "grad_norm": 3.40204327807049, "learning_rate": 1.1350063363033459e-07, "loss": 0.1583, "step": 13733 }, { "epoch": 1.87, "grad_norm": 4.83673584122786, "learning_rate": 1.1326737913491925e-07, "loss": 0.1672, "step": 13734 }, { "epoch": 1.87, "grad_norm": 3.1374089987161353, "learning_rate": 1.1303436182016181e-07, "loss": 0.1241, "step": 13735 }, { "epoch": 1.87, "grad_norm": 4.1276806711550815, "learning_rate": 1.1280158169737265e-07, "loss": 0.1007, "step": 13736 }, { "epoch": 1.87, "grad_norm": 3.528788997497871, "learning_rate": 1.1256903877784886e-07, "loss": 0.129, "step": 13737 }, { "epoch": 1.87, "grad_norm": 3.7102850829958594, "learning_rate": 1.1233673307287752e-07, "loss": 0.2134, "step": 13738 }, { "epoch": 1.87, "grad_norm": 3.4673150341136916, "learning_rate": 1.1210466459373348e-07, "loss": 0.1626, "step": 13739 }, { "epoch": 1.87, "grad_norm": 3.7367508660990083, "learning_rate": 1.1187283335168109e-07, "loss": 0.1401, "step": 13740 }, { "epoch": 1.87, "grad_norm": 3.607470046656225, "learning_rate": 1.1164123935797189e-07, "loss": 0.1283, "step": 13741 }, { "epoch": 1.87, "grad_norm": 4.057798601119893, "learning_rate": 1.1140988262384633e-07, "loss": 0.1751, "step": 13742 }, { "epoch": 1.87, "grad_norm": 2.8082157379117416, "learning_rate": 1.1117876316053378e-07, "loss": 0.1216, "step": 13743 }, { "epoch": 1.87, "grad_norm": 4.681213563303381, "learning_rate": 1.109478809792508e-07, "loss": 0.1581, "step": 13744 }, { "epoch": 1.87, "grad_norm": 4.78273127998252, "learning_rate": 1.1071723609120566e-07, "loss": 0.1353, "step": 13745 }, { "epoch": 1.87, "grad_norm": 3.7524845514081924, "learning_rate": 1.1048682850758996e-07, "loss": 0.1724, "step": 13746 }, { "epoch": 1.87, "grad_norm": 3.014986351196649, "learning_rate": 1.1025665823958975e-07, "loss": 0.125, "step": 13747 }, { "epoch": 1.87, "grad_norm": 3.4903998071947906, "learning_rate": 1.1002672529837333e-07, "loss": 0.1645, "step": 13748 }, { "epoch": 1.87, "grad_norm": 4.075755937509131, "learning_rate": 1.0979702969510286e-07, "loss": 0.1996, "step": 13749 }, { "epoch": 1.87, "grad_norm": 3.3976313154833044, "learning_rate": 1.0956757144092555e-07, "loss": 0.1227, "step": 13750 }, { "epoch": 1.87, "grad_norm": 3.0204116332123454, "learning_rate": 1.0933835054697973e-07, "loss": 0.1317, "step": 13751 }, { "epoch": 1.87, "grad_norm": 4.771000346957373, "learning_rate": 1.0910936702438924e-07, "loss": 0.1503, "step": 13752 }, { "epoch": 1.87, "grad_norm": 4.658591502474296, "learning_rate": 1.0888062088426853e-07, "loss": 0.1553, "step": 13753 }, { "epoch": 1.87, "grad_norm": 2.7855615529514135, "learning_rate": 1.0865211213772042e-07, "loss": 0.1358, "step": 13754 }, { "epoch": 1.87, "grad_norm": 3.1464828891881713, "learning_rate": 1.084238407958349e-07, "loss": 0.1466, "step": 13755 }, { "epoch": 1.87, "grad_norm": 3.1718225515353593, "learning_rate": 1.08195806869692e-07, "loss": 0.1488, "step": 13756 }, { "epoch": 1.87, "grad_norm": 3.6053582006077023, "learning_rate": 1.0796801037035898e-07, "loss": 0.1752, "step": 13757 }, { "epoch": 1.87, "grad_norm": 4.335444506121868, "learning_rate": 1.07740451308892e-07, "loss": 0.1746, "step": 13758 }, { "epoch": 1.87, "grad_norm": 2.45149093623015, "learning_rate": 1.0751312969633665e-07, "loss": 0.1249, "step": 13759 }, { "epoch": 1.87, "grad_norm": 3.4281562447747147, "learning_rate": 1.0728604554372468e-07, "loss": 0.0912, "step": 13760 }, { "epoch": 1.87, "grad_norm": 4.078058637541729, "learning_rate": 1.0705919886207949e-07, "loss": 0.1549, "step": 13761 }, { "epoch": 1.87, "grad_norm": 3.094393180190398, "learning_rate": 1.068325896624095e-07, "loss": 0.1144, "step": 13762 }, { "epoch": 1.87, "grad_norm": 3.637334213878056, "learning_rate": 1.0660621795571424e-07, "loss": 0.1543, "step": 13763 }, { "epoch": 1.87, "grad_norm": 3.728233229777894, "learning_rate": 1.0638008375298159e-07, "loss": 0.1681, "step": 13764 }, { "epoch": 1.87, "grad_norm": 3.0078529407146064, "learning_rate": 1.0615418706518554e-07, "loss": 0.1246, "step": 13765 }, { "epoch": 1.87, "grad_norm": 3.450479931381706, "learning_rate": 1.0592852790329122e-07, "loss": 0.1626, "step": 13766 }, { "epoch": 1.87, "grad_norm": 2.8182124985120125, "learning_rate": 1.0570310627825042e-07, "loss": 0.1332, "step": 13767 }, { "epoch": 1.87, "grad_norm": 4.067722511905333, "learning_rate": 1.054779222010044e-07, "loss": 0.1638, "step": 13768 }, { "epoch": 1.87, "grad_norm": 3.432795058068547, "learning_rate": 1.0525297568248327e-07, "loss": 0.1182, "step": 13769 }, { "epoch": 1.87, "grad_norm": 3.3142016911667747, "learning_rate": 1.0502826673360388e-07, "loss": 0.1642, "step": 13770 }, { "epoch": 1.87, "grad_norm": 3.42696578342177, "learning_rate": 1.0480379536527307e-07, "loss": 0.1375, "step": 13771 }, { "epoch": 1.87, "grad_norm": 3.8300107535913934, "learning_rate": 1.0457956158838545e-07, "loss": 0.1324, "step": 13772 }, { "epoch": 1.87, "grad_norm": 5.0780774221348635, "learning_rate": 1.0435556541382563e-07, "loss": 0.1398, "step": 13773 }, { "epoch": 1.87, "grad_norm": 3.678618249954216, "learning_rate": 1.0413180685246327e-07, "loss": 0.1552, "step": 13774 }, { "epoch": 1.87, "grad_norm": 3.247288737021477, "learning_rate": 1.0390828591516078e-07, "loss": 0.1144, "step": 13775 }, { "epoch": 1.87, "grad_norm": 3.5540102548464163, "learning_rate": 1.0368500261276503e-07, "loss": 0.1091, "step": 13776 }, { "epoch": 1.87, "grad_norm": 3.4896650042743653, "learning_rate": 1.0346195695611461e-07, "loss": 0.1581, "step": 13777 }, { "epoch": 1.87, "grad_norm": 3.5343408174561355, "learning_rate": 1.0323914895603415e-07, "loss": 0.1442, "step": 13778 }, { "epoch": 1.87, "grad_norm": 3.4001380874773166, "learning_rate": 1.0301657862333891e-07, "loss": 0.1593, "step": 13779 }, { "epoch": 1.87, "grad_norm": 3.7069469575022658, "learning_rate": 1.0279424596883025e-07, "loss": 0.1968, "step": 13780 }, { "epoch": 1.87, "grad_norm": 2.895065848324601, "learning_rate": 1.0257215100330009e-07, "loss": 0.144, "step": 13781 }, { "epoch": 1.87, "grad_norm": 2.7444470425908563, "learning_rate": 1.0235029373752758e-07, "loss": 0.1032, "step": 13782 }, { "epoch": 1.87, "grad_norm": 3.2205950806983203, "learning_rate": 1.0212867418228078e-07, "loss": 0.1646, "step": 13783 }, { "epoch": 1.87, "grad_norm": 3.260186176112959, "learning_rate": 1.019072923483161e-07, "loss": 0.1623, "step": 13784 }, { "epoch": 1.88, "grad_norm": 2.9913880660901424, "learning_rate": 1.0168614824637879e-07, "loss": 0.1168, "step": 13785 }, { "epoch": 1.88, "grad_norm": 4.606741321835379, "learning_rate": 1.0146524188720197e-07, "loss": 0.141, "step": 13786 }, { "epoch": 1.88, "grad_norm": 3.888806241423139, "learning_rate": 1.0124457328150705e-07, "loss": 0.1369, "step": 13787 }, { "epoch": 1.88, "grad_norm": 3.134365955334161, "learning_rate": 1.0102414244000491e-07, "loss": 0.1301, "step": 13788 }, { "epoch": 1.88, "grad_norm": 4.934862103436506, "learning_rate": 1.0080394937339422e-07, "loss": 0.1898, "step": 13789 }, { "epoch": 1.88, "grad_norm": 3.475912384242351, "learning_rate": 1.0058399409236141e-07, "loss": 0.1567, "step": 13790 }, { "epoch": 1.88, "grad_norm": 3.3262149158968644, "learning_rate": 1.0036427660758297e-07, "loss": 0.1322, "step": 13791 }, { "epoch": 1.88, "grad_norm": 3.757629837005505, "learning_rate": 1.0014479692972368e-07, "loss": 0.1499, "step": 13792 }, { "epoch": 1.88, "grad_norm": 2.527654371304884, "learning_rate": 9.992555506943447e-08, "loss": 0.1201, "step": 13793 }, { "epoch": 1.88, "grad_norm": 4.74958129785537, "learning_rate": 9.970655103735793e-08, "loss": 0.1686, "step": 13794 }, { "epoch": 1.88, "grad_norm": 3.2919955099687015, "learning_rate": 9.948778484412169e-08, "loss": 0.1455, "step": 13795 }, { "epoch": 1.88, "grad_norm": 2.9836466793718786, "learning_rate": 9.926925650034558e-08, "loss": 0.1387, "step": 13796 }, { "epoch": 1.88, "grad_norm": 2.9884788636842377, "learning_rate": 9.905096601663556e-08, "loss": 0.1162, "step": 13797 }, { "epoch": 1.88, "grad_norm": 3.032042291754833, "learning_rate": 9.883291340358592e-08, "loss": 0.1582, "step": 13798 }, { "epoch": 1.88, "grad_norm": 2.962473693789692, "learning_rate": 9.861509867177987e-08, "loss": 0.1537, "step": 13799 }, { "epoch": 1.88, "grad_norm": 3.21691547750586, "learning_rate": 9.839752183179007e-08, "loss": 0.1527, "step": 13800 }, { "epoch": 1.88, "grad_norm": 4.189747335712495, "learning_rate": 9.818018289417641e-08, "loss": 0.1633, "step": 13801 }, { "epoch": 1.88, "grad_norm": 4.155194915329955, "learning_rate": 9.796308186948711e-08, "loss": 0.1603, "step": 13802 }, { "epoch": 1.88, "grad_norm": 3.127458217702593, "learning_rate": 9.774621876825985e-08, "loss": 0.1265, "step": 13803 }, { "epoch": 1.88, "grad_norm": 3.514739950968557, "learning_rate": 9.75295936010201e-08, "loss": 0.1018, "step": 13804 }, { "epoch": 1.88, "grad_norm": 4.423480354348581, "learning_rate": 9.731320637828113e-08, "loss": 0.2013, "step": 13805 }, { "epoch": 1.88, "grad_norm": 3.803641753801413, "learning_rate": 9.709705711054673e-08, "loss": 0.1735, "step": 13806 }, { "epoch": 1.88, "grad_norm": 4.0196116984633194, "learning_rate": 9.688114580830688e-08, "loss": 0.1528, "step": 13807 }, { "epoch": 1.88, "grad_norm": 3.780526314350199, "learning_rate": 9.666547248204206e-08, "loss": 0.168, "step": 13808 }, { "epoch": 1.88, "grad_norm": 3.2017805222344546, "learning_rate": 9.645003714221835e-08, "loss": 0.1199, "step": 13809 }, { "epoch": 1.88, "grad_norm": 2.3402665597019623, "learning_rate": 9.62348397992935e-08, "loss": 0.1101, "step": 13810 }, { "epoch": 1.88, "grad_norm": 3.019336263285116, "learning_rate": 9.601988046371136e-08, "loss": 0.18, "step": 13811 }, { "epoch": 1.88, "grad_norm": 6.655665988158323, "learning_rate": 9.580515914590637e-08, "loss": 0.1487, "step": 13812 }, { "epoch": 1.88, "grad_norm": 3.145939519493252, "learning_rate": 9.559067585629856e-08, "loss": 0.1237, "step": 13813 }, { "epoch": 1.88, "grad_norm": 3.830710647743197, "learning_rate": 9.537643060529956e-08, "loss": 0.1416, "step": 13814 }, { "epoch": 1.88, "grad_norm": 2.6430612575739496, "learning_rate": 9.516242340330661e-08, "loss": 0.0977, "step": 13815 }, { "epoch": 1.88, "grad_norm": 3.171831233224266, "learning_rate": 9.494865426070754e-08, "loss": 0.1454, "step": 13816 }, { "epoch": 1.88, "grad_norm": 4.293066633971112, "learning_rate": 9.473512318787681e-08, "loss": 0.1171, "step": 13817 }, { "epoch": 1.88, "grad_norm": 4.079558016442551, "learning_rate": 9.452183019518002e-08, "loss": 0.1679, "step": 13818 }, { "epoch": 1.88, "grad_norm": 3.457862810480329, "learning_rate": 9.430877529296722e-08, "loss": 0.1297, "step": 13819 }, { "epoch": 1.88, "grad_norm": 2.8865625639896093, "learning_rate": 9.409595849158127e-08, "loss": 0.1249, "step": 13820 }, { "epoch": 1.88, "grad_norm": 3.584039625108501, "learning_rate": 9.388337980134998e-08, "loss": 0.1483, "step": 13821 }, { "epoch": 1.88, "grad_norm": 2.7447612713453275, "learning_rate": 9.367103923259124e-08, "loss": 0.1472, "step": 13822 }, { "epoch": 1.88, "grad_norm": 3.173232986975358, "learning_rate": 9.34589367956118e-08, "loss": 0.1632, "step": 13823 }, { "epoch": 1.88, "grad_norm": 6.030535677487871, "learning_rate": 9.324707250070509e-08, "loss": 0.1647, "step": 13824 }, { "epoch": 1.88, "grad_norm": 3.953020027490121, "learning_rate": 9.303544635815564e-08, "loss": 0.134, "step": 13825 }, { "epoch": 1.88, "grad_norm": 3.0358631671396297, "learning_rate": 9.28240583782325e-08, "loss": 0.1334, "step": 13826 }, { "epoch": 1.88, "grad_norm": 3.829368481743928, "learning_rate": 9.261290857119853e-08, "loss": 0.1338, "step": 13827 }, { "epoch": 1.88, "grad_norm": 2.3773392044438864, "learning_rate": 9.240199694729946e-08, "loss": 0.1242, "step": 13828 }, { "epoch": 1.88, "grad_norm": 3.1580242310091315, "learning_rate": 9.219132351677318e-08, "loss": 0.1455, "step": 13829 }, { "epoch": 1.88, "grad_norm": 3.2395993924692417, "learning_rate": 9.198088828984487e-08, "loss": 0.157, "step": 13830 }, { "epoch": 1.88, "grad_norm": 2.4855086597208147, "learning_rate": 9.177069127672744e-08, "loss": 0.135, "step": 13831 }, { "epoch": 1.88, "grad_norm": 3.8069522084949, "learning_rate": 9.156073248762387e-08, "loss": 0.1855, "step": 13832 }, { "epoch": 1.88, "grad_norm": 3.233332351618516, "learning_rate": 9.135101193272433e-08, "loss": 0.156, "step": 13833 }, { "epoch": 1.88, "grad_norm": 3.0152402708853696, "learning_rate": 9.114152962220734e-08, "loss": 0.1375, "step": 13834 }, { "epoch": 1.88, "grad_norm": 3.133746858563931, "learning_rate": 9.093228556624035e-08, "loss": 0.1653, "step": 13835 }, { "epoch": 1.88, "grad_norm": 2.64856523071625, "learning_rate": 9.072327977497963e-08, "loss": 0.1214, "step": 13836 }, { "epoch": 1.88, "grad_norm": 5.805974196605542, "learning_rate": 9.051451225856877e-08, "loss": 0.1278, "step": 13837 }, { "epoch": 1.88, "grad_norm": 3.7960177651130356, "learning_rate": 9.030598302714078e-08, "loss": 0.1182, "step": 13838 }, { "epoch": 1.88, "grad_norm": 2.527183858810067, "learning_rate": 9.009769209081753e-08, "loss": 0.1492, "step": 13839 }, { "epoch": 1.88, "grad_norm": 3.5372391141325643, "learning_rate": 8.988963945970707e-08, "loss": 0.1485, "step": 13840 }, { "epoch": 1.88, "grad_norm": 3.385956421385087, "learning_rate": 8.968182514390855e-08, "loss": 0.1607, "step": 13841 }, { "epoch": 1.88, "grad_norm": 2.3894737357849305, "learning_rate": 8.947424915350723e-08, "loss": 0.1363, "step": 13842 }, { "epoch": 1.88, "grad_norm": 4.13269193175634, "learning_rate": 8.926691149857891e-08, "loss": 0.1479, "step": 13843 }, { "epoch": 1.88, "grad_norm": 3.169425116914033, "learning_rate": 8.905981218918669e-08, "loss": 0.1313, "step": 13844 }, { "epoch": 1.88, "grad_norm": 4.451634907565255, "learning_rate": 8.885295123538196e-08, "loss": 0.1205, "step": 13845 }, { "epoch": 1.88, "grad_norm": 3.0951908026829646, "learning_rate": 8.864632864720502e-08, "loss": 0.1316, "step": 13846 }, { "epoch": 1.88, "grad_norm": 4.218195674770983, "learning_rate": 8.843994443468451e-08, "loss": 0.1542, "step": 13847 }, { "epoch": 1.88, "grad_norm": 3.007486235641659, "learning_rate": 8.823379860783687e-08, "loss": 0.1409, "step": 13848 }, { "epoch": 1.88, "grad_norm": 5.264999997822558, "learning_rate": 8.802789117666855e-08, "loss": 0.1603, "step": 13849 }, { "epoch": 1.88, "grad_norm": 3.511261702712903, "learning_rate": 8.782222215117209e-08, "loss": 0.1586, "step": 13850 }, { "epoch": 1.88, "grad_norm": 3.2357109176025314, "learning_rate": 8.761679154133173e-08, "loss": 0.1705, "step": 13851 }, { "epoch": 1.88, "grad_norm": 4.592762048279087, "learning_rate": 8.741159935711563e-08, "loss": 0.1543, "step": 13852 }, { "epoch": 1.88, "grad_norm": 3.565819526855943, "learning_rate": 8.720664560848525e-08, "loss": 0.1282, "step": 13853 }, { "epoch": 1.88, "grad_norm": 3.981101093024718, "learning_rate": 8.70019303053865e-08, "loss": 0.1581, "step": 13854 }, { "epoch": 1.88, "grad_norm": 2.893641522022633, "learning_rate": 8.679745345775648e-08, "loss": 0.1348, "step": 13855 }, { "epoch": 1.88, "grad_norm": 6.01417787117664, "learning_rate": 8.659321507551888e-08, "loss": 0.1657, "step": 13856 }, { "epoch": 1.88, "grad_norm": 3.277587402871174, "learning_rate": 8.638921516858634e-08, "loss": 0.1544, "step": 13857 }, { "epoch": 1.89, "grad_norm": 3.7118786757260986, "learning_rate": 8.618545374686094e-08, "loss": 0.1648, "step": 13858 }, { "epoch": 1.89, "grad_norm": 3.5392131613016944, "learning_rate": 8.598193082023254e-08, "loss": 0.1518, "step": 13859 }, { "epoch": 1.89, "grad_norm": 3.786246130737447, "learning_rate": 8.577864639857825e-08, "loss": 0.2009, "step": 13860 }, { "epoch": 1.89, "grad_norm": 2.60750667987059, "learning_rate": 8.557560049176461e-08, "loss": 0.1344, "step": 13861 }, { "epoch": 1.89, "grad_norm": 3.290520095954842, "learning_rate": 8.537279310964763e-08, "loss": 0.146, "step": 13862 }, { "epoch": 1.89, "grad_norm": 3.476438192519933, "learning_rate": 8.517022426207e-08, "loss": 0.1621, "step": 13863 }, { "epoch": 1.89, "grad_norm": 2.7690570729601505, "learning_rate": 8.496789395886329e-08, "loss": 0.1352, "step": 13864 }, { "epoch": 1.89, "grad_norm": 2.8361209335152964, "learning_rate": 8.476580220984854e-08, "loss": 0.1391, "step": 13865 }, { "epoch": 1.89, "grad_norm": 2.970585691428216, "learning_rate": 8.4563949024834e-08, "loss": 0.1382, "step": 13866 }, { "epoch": 1.89, "grad_norm": 4.065754599311985, "learning_rate": 8.436233441361629e-08, "loss": 0.1367, "step": 13867 }, { "epoch": 1.89, "grad_norm": 2.5952042121092203, "learning_rate": 8.416095838598148e-08, "loss": 0.1504, "step": 13868 }, { "epoch": 1.89, "grad_norm": 3.4612899828622443, "learning_rate": 8.395982095170285e-08, "loss": 0.1384, "step": 13869 }, { "epoch": 1.89, "grad_norm": 2.76194008134239, "learning_rate": 8.37589221205437e-08, "loss": 0.1364, "step": 13870 }, { "epoch": 1.89, "grad_norm": 3.536206392183293, "learning_rate": 8.355826190225347e-08, "loss": 0.1333, "step": 13871 }, { "epoch": 1.89, "grad_norm": 3.043048702962397, "learning_rate": 8.335784030657324e-08, "loss": 0.1495, "step": 13872 }, { "epoch": 1.89, "grad_norm": 2.590411237405859, "learning_rate": 8.315765734322856e-08, "loss": 0.1048, "step": 13873 }, { "epoch": 1.89, "grad_norm": 3.2649202065381204, "learning_rate": 8.295771302193723e-08, "loss": 0.1697, "step": 13874 }, { "epoch": 1.89, "grad_norm": 3.3103361448507167, "learning_rate": 8.275800735240203e-08, "loss": 0.1262, "step": 13875 }, { "epoch": 1.89, "grad_norm": 4.668737612121642, "learning_rate": 8.255854034431687e-08, "loss": 0.1462, "step": 13876 }, { "epoch": 1.89, "grad_norm": 2.8135580386888823, "learning_rate": 8.235931200736235e-08, "loss": 0.1016, "step": 13877 }, { "epoch": 1.89, "grad_norm": 3.1328410931918764, "learning_rate": 8.216032235120908e-08, "loss": 0.1581, "step": 13878 }, { "epoch": 1.89, "grad_norm": 2.3466829312470217, "learning_rate": 8.196157138551431e-08, "loss": 0.1139, "step": 13879 }, { "epoch": 1.89, "grad_norm": 3.2992141456549624, "learning_rate": 8.176305911992477e-08, "loss": 0.1611, "step": 13880 }, { "epoch": 1.89, "grad_norm": 4.932789962651725, "learning_rate": 8.156478556407499e-08, "loss": 0.1403, "step": 13881 }, { "epoch": 1.89, "grad_norm": 2.773872559359049, "learning_rate": 8.136675072758948e-08, "loss": 0.1684, "step": 13882 }, { "epoch": 1.89, "grad_norm": 2.9987011334521556, "learning_rate": 8.116895462007834e-08, "loss": 0.1573, "step": 13883 }, { "epoch": 1.89, "grad_norm": 3.79074602619558, "learning_rate": 8.097139725114334e-08, "loss": 0.1308, "step": 13884 }, { "epoch": 1.89, "grad_norm": 2.7874215658189874, "learning_rate": 8.077407863037179e-08, "loss": 0.1531, "step": 13885 }, { "epoch": 1.89, "grad_norm": 3.0647643788699086, "learning_rate": 8.057699876734216e-08, "loss": 0.1559, "step": 13886 }, { "epoch": 1.89, "grad_norm": 3.2486307538948935, "learning_rate": 8.038015767161789e-08, "loss": 0.1782, "step": 13887 }, { "epoch": 1.89, "grad_norm": 2.4924617354593446, "learning_rate": 8.018355535275468e-08, "loss": 0.1389, "step": 13888 }, { "epoch": 1.89, "grad_norm": 3.5822512390532695, "learning_rate": 7.998719182029269e-08, "loss": 0.1254, "step": 13889 }, { "epoch": 1.89, "grad_norm": 5.035736385497175, "learning_rate": 7.979106708376427e-08, "loss": 0.1628, "step": 13890 }, { "epoch": 1.89, "grad_norm": 3.1255557715102986, "learning_rate": 7.959518115268793e-08, "loss": 0.1658, "step": 13891 }, { "epoch": 1.89, "grad_norm": 4.561200906639016, "learning_rate": 7.939953403657164e-08, "loss": 0.1844, "step": 13892 }, { "epoch": 1.89, "grad_norm": 2.397847576238762, "learning_rate": 7.920412574490999e-08, "loss": 0.1511, "step": 13893 }, { "epoch": 1.89, "grad_norm": 3.4162893902344975, "learning_rate": 7.900895628718819e-08, "loss": 0.1428, "step": 13894 }, { "epoch": 1.89, "grad_norm": 2.458611661139794, "learning_rate": 7.881402567287866e-08, "loss": 0.1147, "step": 13895 }, { "epoch": 1.89, "grad_norm": 3.5317453646075236, "learning_rate": 7.861933391144272e-08, "loss": 0.1491, "step": 13896 }, { "epoch": 1.89, "grad_norm": 2.306071143192388, "learning_rate": 7.842488101232893e-08, "loss": 0.1099, "step": 13897 }, { "epoch": 1.89, "grad_norm": 3.803198582356793, "learning_rate": 7.823066698497695e-08, "loss": 0.1499, "step": 13898 }, { "epoch": 1.89, "grad_norm": 5.46333653155855, "learning_rate": 7.803669183881091e-08, "loss": 0.1584, "step": 13899 }, { "epoch": 1.89, "grad_norm": 4.481855814062581, "learning_rate": 7.784295558324772e-08, "loss": 0.1953, "step": 13900 }, { "epoch": 1.89, "grad_norm": 3.6655001655296577, "learning_rate": 7.764945822768933e-08, "loss": 0.1543, "step": 13901 }, { "epoch": 1.89, "grad_norm": 2.9143079522297155, "learning_rate": 7.745619978152653e-08, "loss": 0.1359, "step": 13902 }, { "epoch": 1.89, "grad_norm": 2.7755255117242457, "learning_rate": 7.726318025414071e-08, "loss": 0.1697, "step": 13903 }, { "epoch": 1.89, "grad_norm": 3.0947695674917846, "learning_rate": 7.707039965489882e-08, "loss": 0.1779, "step": 13904 }, { "epoch": 1.89, "grad_norm": 3.0929963820114397, "learning_rate": 7.687785799315895e-08, "loss": 0.1426, "step": 13905 }, { "epoch": 1.89, "grad_norm": 3.467141887565849, "learning_rate": 7.668555527826527e-08, "loss": 0.119, "step": 13906 }, { "epoch": 1.89, "grad_norm": 2.826886328468483, "learning_rate": 7.649349151955199e-08, "loss": 0.116, "step": 13907 }, { "epoch": 1.89, "grad_norm": 3.7791285047178214, "learning_rate": 7.630166672633999e-08, "loss": 0.1431, "step": 13908 }, { "epoch": 1.89, "grad_norm": 3.364136651498178, "learning_rate": 7.61100809079407e-08, "loss": 0.1798, "step": 13909 }, { "epoch": 1.89, "grad_norm": 3.0462075914532174, "learning_rate": 7.59187340736528e-08, "loss": 0.114, "step": 13910 }, { "epoch": 1.89, "grad_norm": 3.0819800999704587, "learning_rate": 7.572762623276276e-08, "loss": 0.1428, "step": 13911 }, { "epoch": 1.89, "grad_norm": 3.6673642268691333, "learning_rate": 7.553675739454647e-08, "loss": 0.1363, "step": 13912 }, { "epoch": 1.89, "grad_norm": 2.8094599495892303, "learning_rate": 7.534612756826764e-08, "loss": 0.1063, "step": 13913 }, { "epoch": 1.89, "grad_norm": 3.758448909992079, "learning_rate": 7.515573676317945e-08, "loss": 0.1866, "step": 13914 }, { "epoch": 1.89, "grad_norm": 4.433889528408943, "learning_rate": 7.496558498852113e-08, "loss": 0.1594, "step": 13915 }, { "epoch": 1.89, "grad_norm": 3.5054505591041485, "learning_rate": 7.477567225352311e-08, "loss": 0.1475, "step": 13916 }, { "epoch": 1.89, "grad_norm": 2.5949635570786636, "learning_rate": 7.4585998567403e-08, "loss": 0.0976, "step": 13917 }, { "epoch": 1.89, "grad_norm": 3.026352141301138, "learning_rate": 7.439656393936511e-08, "loss": 0.1568, "step": 13918 }, { "epoch": 1.89, "grad_norm": 2.3631924011163186, "learning_rate": 7.420736837860599e-08, "loss": 0.1253, "step": 13919 }, { "epoch": 1.89, "grad_norm": 3.5034432371316093, "learning_rate": 7.401841189430659e-08, "loss": 0.1373, "step": 13920 }, { "epoch": 1.89, "grad_norm": 3.3452929900247095, "learning_rate": 7.382969449563903e-08, "loss": 0.1354, "step": 13921 }, { "epoch": 1.89, "grad_norm": 3.028537349255392, "learning_rate": 7.364121619176213e-08, "loss": 0.1528, "step": 13922 }, { "epoch": 1.89, "grad_norm": 2.9295475300772162, "learning_rate": 7.345297699182464e-08, "loss": 0.1503, "step": 13923 }, { "epoch": 1.89, "grad_norm": 3.0263590160040077, "learning_rate": 7.326497690496204e-08, "loss": 0.1589, "step": 13924 }, { "epoch": 1.89, "grad_norm": 3.065190425341146, "learning_rate": 7.307721594029982e-08, "loss": 0.121, "step": 13925 }, { "epoch": 1.89, "grad_norm": 2.909880393506383, "learning_rate": 7.288969410695068e-08, "loss": 0.1449, "step": 13926 }, { "epoch": 1.89, "grad_norm": 2.435689366752346, "learning_rate": 7.270241141401568e-08, "loss": 0.145, "step": 13927 }, { "epoch": 1.89, "grad_norm": 2.5984911287672876, "learning_rate": 7.251536787058533e-08, "loss": 0.1299, "step": 13928 }, { "epoch": 1.89, "grad_norm": 2.2964016156364506, "learning_rate": 7.232856348573791e-08, "loss": 0.1106, "step": 13929 }, { "epoch": 1.89, "grad_norm": 4.517213090093056, "learning_rate": 7.214199826853952e-08, "loss": 0.1339, "step": 13930 }, { "epoch": 1.89, "grad_norm": 3.7297371797218384, "learning_rate": 7.195567222804568e-08, "loss": 0.1491, "step": 13931 }, { "epoch": 1.9, "grad_norm": 2.8968497171821603, "learning_rate": 7.17695853732997e-08, "loss": 0.1132, "step": 13932 }, { "epoch": 1.9, "grad_norm": 2.4452457282919227, "learning_rate": 7.158373771333383e-08, "loss": 0.1316, "step": 13933 }, { "epoch": 1.9, "grad_norm": 3.1753788152023374, "learning_rate": 7.13981292571675e-08, "loss": 0.1606, "step": 13934 }, { "epoch": 1.9, "grad_norm": 3.0990921415189834, "learning_rate": 7.121276001380961e-08, "loss": 0.1779, "step": 13935 }, { "epoch": 1.9, "grad_norm": 2.4607293037360294, "learning_rate": 7.102762999225799e-08, "loss": 0.1476, "step": 13936 }, { "epoch": 1.9, "grad_norm": 4.2629193761437625, "learning_rate": 7.084273920149654e-08, "loss": 0.1494, "step": 13937 }, { "epoch": 1.9, "grad_norm": 3.3694203021869367, "learning_rate": 7.065808765050086e-08, "loss": 0.1234, "step": 13938 }, { "epoch": 1.9, "grad_norm": 3.3004490188124374, "learning_rate": 7.047367534823158e-08, "loss": 0.1642, "step": 13939 }, { "epoch": 1.9, "grad_norm": 2.8756689896456904, "learning_rate": 7.028950230363984e-08, "loss": 0.1473, "step": 13940 }, { "epoch": 1.9, "grad_norm": 4.334090363872909, "learning_rate": 7.010556852566464e-08, "loss": 0.134, "step": 13941 }, { "epoch": 1.9, "grad_norm": 2.7392785854448185, "learning_rate": 6.99218740232338e-08, "loss": 0.1587, "step": 13942 }, { "epoch": 1.9, "grad_norm": 3.439032148070462, "learning_rate": 6.973841880526188e-08, "loss": 0.1505, "step": 13943 }, { "epoch": 1.9, "grad_norm": 3.1861779139929847, "learning_rate": 6.955520288065398e-08, "loss": 0.1703, "step": 13944 }, { "epoch": 1.9, "grad_norm": 3.486673170988262, "learning_rate": 6.937222625830243e-08, "loss": 0.1509, "step": 13945 }, { "epoch": 1.9, "grad_norm": 3.5221212707106186, "learning_rate": 6.918948894708732e-08, "loss": 0.1429, "step": 13946 }, { "epoch": 1.9, "grad_norm": 3.662741740164336, "learning_rate": 6.900699095587937e-08, "loss": 0.1671, "step": 13947 }, { "epoch": 1.9, "grad_norm": 4.131376303269883, "learning_rate": 6.882473229353481e-08, "loss": 0.1527, "step": 13948 }, { "epoch": 1.9, "grad_norm": 3.672139463496064, "learning_rate": 6.864271296890046e-08, "loss": 0.1459, "step": 13949 }, { "epoch": 1.9, "grad_norm": 2.9983699294678314, "learning_rate": 6.846093299081091e-08, "loss": 0.1081, "step": 13950 }, { "epoch": 1.9, "grad_norm": 3.3437818402154162, "learning_rate": 6.8279392368088e-08, "loss": 0.1027, "step": 13951 }, { "epoch": 1.9, "grad_norm": 3.419426622503934, "learning_rate": 6.809809110954413e-08, "loss": 0.141, "step": 13952 }, { "epoch": 1.9, "grad_norm": 3.559952960189187, "learning_rate": 6.791702922397781e-08, "loss": 0.146, "step": 13953 }, { "epoch": 1.9, "grad_norm": 5.176342741229525, "learning_rate": 6.773620672017811e-08, "loss": 0.1538, "step": 13954 }, { "epoch": 1.9, "grad_norm": 3.62751659794988, "learning_rate": 6.755562360692025e-08, "loss": 0.1545, "step": 13955 }, { "epoch": 1.9, "grad_norm": 2.924198778799465, "learning_rate": 6.737527989296944e-08, "loss": 0.1105, "step": 13956 }, { "epoch": 1.9, "grad_norm": 5.418453955145722, "learning_rate": 6.719517558707922e-08, "loss": 0.1358, "step": 13957 }, { "epoch": 1.9, "grad_norm": 3.3104691243913824, "learning_rate": 6.701531069799039e-08, "loss": 0.1215, "step": 13958 }, { "epoch": 1.9, "grad_norm": 3.390214500612926, "learning_rate": 6.683568523443262e-08, "loss": 0.1409, "step": 13959 }, { "epoch": 1.9, "grad_norm": 3.2009760052973313, "learning_rate": 6.665629920512507e-08, "loss": 0.1605, "step": 13960 }, { "epoch": 1.9, "grad_norm": 3.594211876395141, "learning_rate": 6.647715261877352e-08, "loss": 0.1347, "step": 13961 }, { "epoch": 1.9, "grad_norm": 3.1316309554260573, "learning_rate": 6.629824548407381e-08, "loss": 0.1334, "step": 13962 }, { "epoch": 1.9, "grad_norm": 3.132566838573646, "learning_rate": 6.61195778097079e-08, "loss": 0.1364, "step": 13963 }, { "epoch": 1.9, "grad_norm": 4.229589966123948, "learning_rate": 6.594114960434938e-08, "loss": 0.1245, "step": 13964 }, { "epoch": 1.9, "grad_norm": 3.590619336078854, "learning_rate": 6.576296087665634e-08, "loss": 0.1525, "step": 13965 }, { "epoch": 1.9, "grad_norm": 3.698315953450175, "learning_rate": 6.558501163527964e-08, "loss": 0.1423, "step": 13966 }, { "epoch": 1.9, "grad_norm": 3.0626225508031513, "learning_rate": 6.540730188885347e-08, "loss": 0.1518, "step": 13967 }, { "epoch": 1.9, "grad_norm": 2.547421993439999, "learning_rate": 6.522983164600538e-08, "loss": 0.1087, "step": 13968 }, { "epoch": 1.9, "grad_norm": 3.5042683866237843, "learning_rate": 6.505260091534793e-08, "loss": 0.1522, "step": 13969 }, { "epoch": 1.9, "grad_norm": 2.975905635661113, "learning_rate": 6.487560970548367e-08, "loss": 0.1081, "step": 13970 }, { "epoch": 1.9, "grad_norm": 3.586536347559922, "learning_rate": 6.469885802500186e-08, "loss": 0.1164, "step": 13971 }, { "epoch": 1.9, "grad_norm": 3.3784210535954298, "learning_rate": 6.452234588248285e-08, "loss": 0.1388, "step": 13972 }, { "epoch": 1.9, "grad_norm": 3.74064629739623, "learning_rate": 6.434607328649256e-08, "loss": 0.1371, "step": 13973 }, { "epoch": 1.9, "grad_norm": 3.249040900426772, "learning_rate": 6.417004024558637e-08, "loss": 0.1158, "step": 13974 }, { "epoch": 1.9, "grad_norm": 2.8937944296343834, "learning_rate": 6.399424676830968e-08, "loss": 0.122, "step": 13975 }, { "epoch": 1.9, "grad_norm": 2.7525076758347247, "learning_rate": 6.381869286319287e-08, "loss": 0.1287, "step": 13976 }, { "epoch": 1.9, "grad_norm": 3.5591464565713498, "learning_rate": 6.364337853875745e-08, "loss": 0.1585, "step": 13977 }, { "epoch": 1.9, "grad_norm": 4.275937969124263, "learning_rate": 6.346830380351276e-08, "loss": 0.186, "step": 13978 }, { "epoch": 1.9, "grad_norm": 3.5010596867434014, "learning_rate": 6.329346866595532e-08, "loss": 0.1919, "step": 13979 }, { "epoch": 1.9, "grad_norm": 3.598212115375908, "learning_rate": 6.311887313457221e-08, "loss": 0.2079, "step": 13980 }, { "epoch": 1.9, "grad_norm": 3.4423834063701304, "learning_rate": 6.294451721783556e-08, "loss": 0.1358, "step": 13981 }, { "epoch": 1.9, "grad_norm": 3.3630701395775455, "learning_rate": 6.277040092420916e-08, "loss": 0.154, "step": 13982 }, { "epoch": 1.9, "grad_norm": 2.9029216965538893, "learning_rate": 6.259652426214403e-08, "loss": 0.1111, "step": 13983 }, { "epoch": 1.9, "grad_norm": 3.1388865854568726, "learning_rate": 6.242288724007839e-08, "loss": 0.1317, "step": 13984 }, { "epoch": 1.9, "grad_norm": 2.4229706661834425, "learning_rate": 6.224948986644108e-08, "loss": 0.1565, "step": 13985 }, { "epoch": 1.9, "grad_norm": 2.6890321464029765, "learning_rate": 6.207633214964704e-08, "loss": 0.1472, "step": 13986 }, { "epoch": 1.9, "grad_norm": 3.2855614821060763, "learning_rate": 6.190341409810063e-08, "loss": 0.13, "step": 13987 }, { "epoch": 1.9, "grad_norm": 3.6883638370662597, "learning_rate": 6.173073572019517e-08, "loss": 0.1228, "step": 13988 }, { "epoch": 1.9, "grad_norm": 4.344672002207572, "learning_rate": 6.15582970243117e-08, "loss": 0.162, "step": 13989 }, { "epoch": 1.9, "grad_norm": 3.08896041772781, "learning_rate": 6.138609801881856e-08, "loss": 0.1126, "step": 13990 }, { "epoch": 1.9, "grad_norm": 3.437004427377066, "learning_rate": 6.12141387120746e-08, "loss": 0.1472, "step": 13991 }, { "epoch": 1.9, "grad_norm": 3.1929527720660937, "learning_rate": 6.104241911242592e-08, "loss": 0.1357, "step": 13992 }, { "epoch": 1.9, "grad_norm": 2.9204988149975883, "learning_rate": 6.087093922820642e-08, "loss": 0.1538, "step": 13993 }, { "epoch": 1.9, "grad_norm": 2.7657017458351096, "learning_rate": 6.069969906773942e-08, "loss": 0.1434, "step": 13994 }, { "epoch": 1.9, "grad_norm": 3.859942653498728, "learning_rate": 6.052869863933663e-08, "loss": 0.1416, "step": 13995 }, { "epoch": 1.9, "grad_norm": 3.468892314404777, "learning_rate": 6.035793795129585e-08, "loss": 0.1866, "step": 13996 }, { "epoch": 1.9, "grad_norm": 2.8036377003409743, "learning_rate": 6.018741701190767e-08, "loss": 0.127, "step": 13997 }, { "epoch": 1.9, "grad_norm": 3.7378763364359044, "learning_rate": 6.001713582944602e-08, "loss": 0.1769, "step": 13998 }, { "epoch": 1.9, "grad_norm": 2.6205205535779683, "learning_rate": 5.984709441217762e-08, "loss": 0.0952, "step": 13999 }, { "epoch": 1.9, "grad_norm": 3.119108093900181, "learning_rate": 5.967729276835421e-08, "loss": 0.1846, "step": 14000 }, { "epoch": 1.9, "grad_norm": 3.4452831645931767, "learning_rate": 5.9507730906217534e-08, "loss": 0.1743, "step": 14001 }, { "epoch": 1.9, "grad_norm": 3.1378831516699606, "learning_rate": 5.933840883399766e-08, "loss": 0.1427, "step": 14002 }, { "epoch": 1.9, "grad_norm": 2.7372257721058064, "learning_rate": 5.916932655991248e-08, "loss": 0.1508, "step": 14003 }, { "epoch": 1.9, "grad_norm": 4.9295242320689185, "learning_rate": 5.9000484092168184e-08, "loss": 0.118, "step": 14004 }, { "epoch": 1.91, "grad_norm": 4.253397064236511, "learning_rate": 5.883188143896046e-08, "loss": 0.1879, "step": 14005 }, { "epoch": 1.91, "grad_norm": 2.953523187371378, "learning_rate": 5.866351860847219e-08, "loss": 0.1441, "step": 14006 }, { "epoch": 1.91, "grad_norm": 5.658280681872923, "learning_rate": 5.8495395608874625e-08, "loss": 0.1916, "step": 14007 }, { "epoch": 1.91, "grad_norm": 4.698950277127028, "learning_rate": 5.832751244832846e-08, "loss": 0.1723, "step": 14008 }, { "epoch": 1.91, "grad_norm": 3.5970098068106577, "learning_rate": 5.815986913498106e-08, "loss": 0.171, "step": 14009 }, { "epoch": 1.91, "grad_norm": 3.1637747664395723, "learning_rate": 5.799246567696926e-08, "loss": 0.1331, "step": 14010 }, { "epoch": 1.91, "grad_norm": 3.0388385962544877, "learning_rate": 5.7825302082419324e-08, "loss": 0.1404, "step": 14011 }, { "epoch": 1.91, "grad_norm": 2.878080123081491, "learning_rate": 5.7658378359443104e-08, "loss": 0.1557, "step": 14012 }, { "epoch": 1.91, "grad_norm": 4.21900707887688, "learning_rate": 5.749169451614356e-08, "loss": 0.1072, "step": 14013 }, { "epoch": 1.91, "grad_norm": 3.195528709179182, "learning_rate": 5.7325250560609225e-08, "loss": 0.1768, "step": 14014 }, { "epoch": 1.91, "grad_norm": 2.879573856445181, "learning_rate": 5.71590465009203e-08, "loss": 0.1132, "step": 14015 }, { "epoch": 1.91, "grad_norm": 2.7178825496847256, "learning_rate": 5.699308234514256e-08, "loss": 0.1242, "step": 14016 }, { "epoch": 1.91, "grad_norm": 3.619234992432821, "learning_rate": 5.6827358101331774e-08, "loss": 0.1713, "step": 14017 }, { "epoch": 1.91, "grad_norm": 3.7386934799146103, "learning_rate": 5.666187377753096e-08, "loss": 0.1743, "step": 14018 }, { "epoch": 1.91, "grad_norm": 2.612393834400543, "learning_rate": 5.649662938177258e-08, "loss": 0.1633, "step": 14019 }, { "epoch": 1.91, "grad_norm": 3.5798575561567993, "learning_rate": 5.633162492207633e-08, "loss": 0.1393, "step": 14020 }, { "epoch": 1.91, "grad_norm": 2.8977355458843705, "learning_rate": 5.616686040645081e-08, "loss": 0.1102, "step": 14021 }, { "epoch": 1.91, "grad_norm": 3.5111670566690054, "learning_rate": 5.600233584289294e-08, "loss": 0.1734, "step": 14022 }, { "epoch": 1.91, "grad_norm": 3.362462302967452, "learning_rate": 5.583805123938857e-08, "loss": 0.1314, "step": 14023 }, { "epoch": 1.91, "grad_norm": 3.3219907466549876, "learning_rate": 5.567400660391131e-08, "loss": 0.1386, "step": 14024 }, { "epoch": 1.91, "grad_norm": 2.961363176835381, "learning_rate": 5.551020194442258e-08, "loss": 0.121, "step": 14025 }, { "epoch": 1.91, "grad_norm": 4.611542234050505, "learning_rate": 5.534663726887324e-08, "loss": 0.1673, "step": 14026 }, { "epoch": 1.91, "grad_norm": 2.5473697300824583, "learning_rate": 5.518331258520138e-08, "loss": 0.1282, "step": 14027 }, { "epoch": 1.91, "grad_norm": 3.8172058566035973, "learning_rate": 5.50202279013351e-08, "loss": 0.1528, "step": 14028 }, { "epoch": 1.91, "grad_norm": 4.613301489163009, "learning_rate": 5.4857383225188076e-08, "loss": 0.1544, "step": 14029 }, { "epoch": 1.91, "grad_norm": 3.9977118421617464, "learning_rate": 5.469477856466676e-08, "loss": 0.1517, "step": 14030 }, { "epoch": 1.91, "grad_norm": 2.6693195476730267, "learning_rate": 5.4532413927660396e-08, "loss": 0.1468, "step": 14031 }, { "epoch": 1.91, "grad_norm": 3.640148324847056, "learning_rate": 5.437028932205213e-08, "loss": 0.1505, "step": 14032 }, { "epoch": 1.91, "grad_norm": 4.54861466716392, "learning_rate": 5.420840475570843e-08, "loss": 0.1822, "step": 14033 }, { "epoch": 1.91, "grad_norm": 3.2717132277439895, "learning_rate": 5.4046760236487473e-08, "loss": 0.1299, "step": 14034 }, { "epoch": 1.91, "grad_norm": 3.799127044935421, "learning_rate": 5.388535577223519e-08, "loss": 0.1457, "step": 14035 }, { "epoch": 1.91, "grad_norm": 2.729941098222647, "learning_rate": 5.372419137078477e-08, "loss": 0.0962, "step": 14036 }, { "epoch": 1.91, "grad_norm": 3.372452202630421, "learning_rate": 5.356326703995884e-08, "loss": 0.1203, "step": 14037 }, { "epoch": 1.91, "grad_norm": 3.415622943656344, "learning_rate": 5.3402582787567824e-08, "loss": 0.1409, "step": 14038 }, { "epoch": 1.91, "grad_norm": 2.81258844271074, "learning_rate": 5.324213862141103e-08, "loss": 0.1357, "step": 14039 }, { "epoch": 1.91, "grad_norm": 3.4616450209125946, "learning_rate": 5.308193454927446e-08, "loss": 0.1554, "step": 14040 }, { "epoch": 1.91, "grad_norm": 4.06108846980877, "learning_rate": 5.292197057893522e-08, "loss": 0.1574, "step": 14041 }, { "epoch": 1.91, "grad_norm": 4.173503906900696, "learning_rate": 5.276224671815655e-08, "loss": 0.1484, "step": 14042 }, { "epoch": 1.91, "grad_norm": 2.6495809147047136, "learning_rate": 5.260276297469058e-08, "loss": 0.149, "step": 14043 }, { "epoch": 1.91, "grad_norm": 3.9326379936341316, "learning_rate": 5.244351935627834e-08, "loss": 0.1515, "step": 14044 }, { "epoch": 1.91, "grad_norm": 3.1882396677098748, "learning_rate": 5.2284515870648645e-08, "loss": 0.1295, "step": 14045 }, { "epoch": 1.91, "grad_norm": 2.5833783206582317, "learning_rate": 5.212575252551921e-08, "loss": 0.1112, "step": 14046 }, { "epoch": 1.91, "grad_norm": 4.01944212675052, "learning_rate": 5.196722932859499e-08, "loss": 0.1756, "step": 14047 }, { "epoch": 1.91, "grad_norm": 3.2206328926437156, "learning_rate": 5.1808946287570386e-08, "loss": 0.1414, "step": 14048 }, { "epoch": 1.91, "grad_norm": 2.581823786478871, "learning_rate": 5.165090341012757e-08, "loss": 0.1158, "step": 14049 }, { "epoch": 1.91, "grad_norm": 2.7053831500852565, "learning_rate": 5.149310070393765e-08, "loss": 0.1449, "step": 14050 }, { "epoch": 1.91, "grad_norm": 2.8491440170982454, "learning_rate": 5.133553817665948e-08, "loss": 0.1219, "step": 14051 }, { "epoch": 1.91, "grad_norm": 3.844281089242734, "learning_rate": 5.117821583594085e-08, "loss": 0.1427, "step": 14052 }, { "epoch": 1.91, "grad_norm": 3.381554563959507, "learning_rate": 5.102113368941675e-08, "loss": 0.1748, "step": 14053 }, { "epoch": 1.91, "grad_norm": 2.9805357267499586, "learning_rate": 5.0864291744711633e-08, "loss": 0.1457, "step": 14054 }, { "epoch": 1.91, "grad_norm": 3.815756359286453, "learning_rate": 5.070769000943776e-08, "loss": 0.1249, "step": 14055 }, { "epoch": 1.91, "grad_norm": 3.9815191204551796, "learning_rate": 5.055132849119626e-08, "loss": 0.1911, "step": 14056 }, { "epoch": 1.91, "grad_norm": 2.472142344926403, "learning_rate": 5.0395207197575516e-08, "loss": 0.1657, "step": 14057 }, { "epoch": 1.91, "grad_norm": 2.8184788527380373, "learning_rate": 5.0239326136154454e-08, "loss": 0.1369, "step": 14058 }, { "epoch": 1.91, "grad_norm": 3.5890488138421834, "learning_rate": 5.008368531449703e-08, "loss": 0.1766, "step": 14059 }, { "epoch": 1.91, "grad_norm": 3.730259618475841, "learning_rate": 4.992828474015887e-08, "loss": 0.1378, "step": 14060 }, { "epoch": 1.91, "grad_norm": 3.26956986553389, "learning_rate": 4.977312442068116e-08, "loss": 0.166, "step": 14061 }, { "epoch": 1.91, "grad_norm": 3.6267397998862974, "learning_rate": 4.9618204363595656e-08, "loss": 0.1645, "step": 14062 }, { "epoch": 1.91, "grad_norm": 2.5888290714008257, "learning_rate": 4.946352457642134e-08, "loss": 0.1541, "step": 14063 }, { "epoch": 1.91, "grad_norm": 4.781475745024865, "learning_rate": 4.9309085066664986e-08, "loss": 0.1453, "step": 14064 }, { "epoch": 1.91, "grad_norm": 3.2083313113571963, "learning_rate": 4.915488584182393e-08, "loss": 0.1394, "step": 14065 }, { "epoch": 1.91, "grad_norm": 4.108643192924233, "learning_rate": 4.900092690938052e-08, "loss": 0.1532, "step": 14066 }, { "epoch": 1.91, "grad_norm": 3.427748605358579, "learning_rate": 4.8847208276808224e-08, "loss": 0.1047, "step": 14067 }, { "epoch": 1.91, "grad_norm": 3.1582982497815975, "learning_rate": 4.8693729951567734e-08, "loss": 0.16, "step": 14068 }, { "epoch": 1.91, "grad_norm": 2.4179554430670995, "learning_rate": 4.854049194110866e-08, "loss": 0.1096, "step": 14069 }, { "epoch": 1.91, "grad_norm": 3.6473024932813436, "learning_rate": 4.838749425286726e-08, "loss": 0.1833, "step": 14070 }, { "epoch": 1.91, "grad_norm": 3.369612781894617, "learning_rate": 4.823473689427094e-08, "loss": 0.1358, "step": 14071 }, { "epoch": 1.91, "grad_norm": 4.518120798921422, "learning_rate": 4.808221987273265e-08, "loss": 0.163, "step": 14072 }, { "epoch": 1.91, "grad_norm": 3.1633293720458293, "learning_rate": 4.7929943195655914e-08, "loss": 0.1177, "step": 14073 }, { "epoch": 1.91, "grad_norm": 4.668778318002002, "learning_rate": 4.777790687043038e-08, "loss": 0.1484, "step": 14074 }, { "epoch": 1.91, "grad_norm": 3.8224618022405425, "learning_rate": 4.762611090443625e-08, "loss": 0.1684, "step": 14075 }, { "epoch": 1.91, "grad_norm": 3.585930526594361, "learning_rate": 4.7474555305040414e-08, "loss": 0.1353, "step": 14076 }, { "epoch": 1.91, "grad_norm": 2.466087815296356, "learning_rate": 4.732324007959921e-08, "loss": 0.1036, "step": 14077 }, { "epoch": 1.91, "grad_norm": 2.950341453344614, "learning_rate": 4.7172165235456204e-08, "loss": 0.141, "step": 14078 }, { "epoch": 1.92, "grad_norm": 3.56637594754129, "learning_rate": 4.7021330779944974e-08, "loss": 0.1336, "step": 14079 }, { "epoch": 1.92, "grad_norm": 2.903790496811367, "learning_rate": 4.6870736720384666e-08, "loss": 0.1254, "step": 14080 }, { "epoch": 1.92, "grad_norm": 4.087463337748784, "learning_rate": 4.6720383064086106e-08, "loss": 0.1832, "step": 14081 }, { "epoch": 1.92, "grad_norm": 2.1722253978650454, "learning_rate": 4.657026981834623e-08, "loss": 0.1475, "step": 14082 }, { "epoch": 1.92, "grad_norm": 3.1481978050074413, "learning_rate": 4.642039699045087e-08, "loss": 0.1673, "step": 14083 }, { "epoch": 1.92, "grad_norm": 2.865329994827911, "learning_rate": 4.627076458767366e-08, "loss": 0.1735, "step": 14084 }, { "epoch": 1.92, "grad_norm": 2.6014474566661248, "learning_rate": 4.6121372617278224e-08, "loss": 0.1193, "step": 14085 }, { "epoch": 1.92, "grad_norm": 3.7295044590173005, "learning_rate": 4.5972221086514335e-08, "loss": 0.1557, "step": 14086 }, { "epoch": 1.92, "grad_norm": 3.414003183759958, "learning_rate": 4.5823310002621745e-08, "loss": 0.1216, "step": 14087 }, { "epoch": 1.92, "grad_norm": 2.850451355359241, "learning_rate": 4.567463937282746e-08, "loss": 0.1451, "step": 14088 }, { "epoch": 1.92, "grad_norm": 4.312821842008665, "learning_rate": 4.5526209204347913e-08, "loss": 0.1191, "step": 14089 }, { "epoch": 1.92, "grad_norm": 2.8838678270233364, "learning_rate": 4.5378019504386806e-08, "loss": 0.1664, "step": 14090 }, { "epoch": 1.92, "grad_norm": 3.2626238369222618, "learning_rate": 4.523007028013726e-08, "loss": 0.1376, "step": 14091 }, { "epoch": 1.92, "grad_norm": 3.451297603847429, "learning_rate": 4.5082361538779095e-08, "loss": 0.1211, "step": 14092 }, { "epoch": 1.92, "grad_norm": 4.477118667746173, "learning_rate": 4.493489328748213e-08, "loss": 0.1774, "step": 14093 }, { "epoch": 1.92, "grad_norm": 3.1096574469102425, "learning_rate": 4.4787665533403414e-08, "loss": 0.1279, "step": 14094 }, { "epoch": 1.92, "grad_norm": 3.0910144792318173, "learning_rate": 4.4640678283688896e-08, "loss": 0.1511, "step": 14095 }, { "epoch": 1.92, "grad_norm": 3.195343744763195, "learning_rate": 4.449393154547343e-08, "loss": 0.1324, "step": 14096 }, { "epoch": 1.92, "grad_norm": 3.4745995463018544, "learning_rate": 4.434742532587855e-08, "loss": 0.1541, "step": 14097 }, { "epoch": 1.92, "grad_norm": 3.775371044016971, "learning_rate": 4.420115963201521e-08, "loss": 0.1551, "step": 14098 }, { "epoch": 1.92, "grad_norm": 3.5524850021433925, "learning_rate": 4.40551344709822e-08, "loss": 0.1365, "step": 14099 }, { "epoch": 1.92, "grad_norm": 2.9561309010904675, "learning_rate": 4.3909349849867723e-08, "loss": 0.1301, "step": 14100 }, { "epoch": 1.92, "grad_norm": 3.231174982820296, "learning_rate": 4.376380577574724e-08, "loss": 0.1662, "step": 14101 }, { "epoch": 1.92, "grad_norm": 3.0657763623664276, "learning_rate": 4.3618502255684533e-08, "loss": 0.2104, "step": 14102 }, { "epoch": 1.92, "grad_norm": 2.5783181670448627, "learning_rate": 4.347343929673231e-08, "loss": 0.1423, "step": 14103 }, { "epoch": 1.92, "grad_norm": 3.2673921175492695, "learning_rate": 4.3328616905931595e-08, "loss": 0.157, "step": 14104 }, { "epoch": 1.92, "grad_norm": 2.989439814663121, "learning_rate": 4.31840350903101e-08, "loss": 0.106, "step": 14105 }, { "epoch": 1.92, "grad_norm": 2.854595527446049, "learning_rate": 4.3039693856886665e-08, "loss": 0.1474, "step": 14106 }, { "epoch": 1.92, "grad_norm": 2.586222452925023, "learning_rate": 4.289559321266623e-08, "loss": 0.1329, "step": 14107 }, { "epoch": 1.92, "grad_norm": 4.8497096143105916, "learning_rate": 4.27517331646432e-08, "loss": 0.1734, "step": 14108 }, { "epoch": 1.92, "grad_norm": 2.698069316889723, "learning_rate": 4.260811371979923e-08, "loss": 0.1361, "step": 14109 }, { "epoch": 1.92, "grad_norm": 4.989757384907413, "learning_rate": 4.2464734885106494e-08, "loss": 0.1254, "step": 14110 }, { "epoch": 1.92, "grad_norm": 3.1805248512397, "learning_rate": 4.2321596667522225e-08, "loss": 0.1695, "step": 14111 }, { "epoch": 1.92, "grad_norm": 2.6034762195786545, "learning_rate": 4.2178699073994744e-08, "loss": 0.1384, "step": 14112 }, { "epoch": 1.92, "grad_norm": 3.4774014696891835, "learning_rate": 4.203604211145851e-08, "loss": 0.1564, "step": 14113 }, { "epoch": 1.92, "grad_norm": 2.469959608540756, "learning_rate": 4.189362578683909e-08, "loss": 0.1512, "step": 14114 }, { "epoch": 1.92, "grad_norm": 2.965363858620668, "learning_rate": 4.1751450107048175e-08, "loss": 0.1241, "step": 14115 }, { "epoch": 1.92, "grad_norm": 3.2496993264987872, "learning_rate": 4.160951507898581e-08, "loss": 0.1549, "step": 14116 }, { "epoch": 1.92, "grad_norm": 2.689761677857253, "learning_rate": 4.1467820709541474e-08, "loss": 0.1352, "step": 14117 }, { "epoch": 1.92, "grad_norm": 3.864953928659685, "learning_rate": 4.1326367005591897e-08, "loss": 0.1871, "step": 14118 }, { "epoch": 1.92, "grad_norm": 2.94266702895023, "learning_rate": 4.1185153974003246e-08, "loss": 0.1439, "step": 14119 }, { "epoch": 1.92, "grad_norm": 3.5315127668472375, "learning_rate": 4.104418162162893e-08, "loss": 0.198, "step": 14120 }, { "epoch": 1.92, "grad_norm": 3.3454645748790974, "learning_rate": 4.090344995531126e-08, "loss": 0.1357, "step": 14121 }, { "epoch": 1.92, "grad_norm": 2.758366221474289, "learning_rate": 4.0762958981880876e-08, "loss": 0.154, "step": 14122 }, { "epoch": 1.92, "grad_norm": 4.644985976943946, "learning_rate": 4.062270870815621e-08, "loss": 0.1442, "step": 14123 }, { "epoch": 1.92, "grad_norm": 2.92565873287427, "learning_rate": 4.048269914094516e-08, "loss": 0.1427, "step": 14124 }, { "epoch": 1.92, "grad_norm": 3.3386061622130243, "learning_rate": 4.034293028704173e-08, "loss": 0.1458, "step": 14125 }, { "epoch": 1.92, "grad_norm": 3.3669343272403887, "learning_rate": 4.02034021532316e-08, "loss": 0.1508, "step": 14126 }, { "epoch": 1.92, "grad_norm": 3.7055389351254906, "learning_rate": 4.006411474628491e-08, "loss": 0.1649, "step": 14127 }, { "epoch": 1.92, "grad_norm": 3.16274145469514, "learning_rate": 3.992506807296348e-08, "loss": 0.1609, "step": 14128 }, { "epoch": 1.92, "grad_norm": 3.748553126202449, "learning_rate": 3.97862621400158e-08, "loss": 0.1656, "step": 14129 }, { "epoch": 1.92, "grad_norm": 2.683514687250982, "learning_rate": 3.964769695417869e-08, "loss": 0.1435, "step": 14130 }, { "epoch": 1.92, "grad_norm": 3.7891063769658158, "learning_rate": 3.950937252217735e-08, "loss": 0.1665, "step": 14131 }, { "epoch": 1.92, "grad_norm": 4.164685127479322, "learning_rate": 3.937128885072528e-08, "loss": 0.1554, "step": 14132 }, { "epoch": 1.92, "grad_norm": 4.016790371685934, "learning_rate": 3.923344594652489e-08, "loss": 0.1173, "step": 14133 }, { "epoch": 1.92, "grad_norm": 3.081743687985529, "learning_rate": 3.9095843816266965e-08, "loss": 0.1295, "step": 14134 }, { "epoch": 1.92, "grad_norm": 4.158850397919779, "learning_rate": 3.8958482466628365e-08, "loss": 0.1254, "step": 14135 }, { "epoch": 1.92, "grad_norm": 4.0581869220493765, "learning_rate": 3.8821361904278206e-08, "loss": 0.1934, "step": 14136 }, { "epoch": 1.92, "grad_norm": 3.1873249342239465, "learning_rate": 3.868448213587006e-08, "loss": 0.1664, "step": 14137 }, { "epoch": 1.92, "grad_norm": 3.6236669849310705, "learning_rate": 3.8547843168048605e-08, "loss": 0.1968, "step": 14138 }, { "epoch": 1.92, "grad_norm": 3.280754617931021, "learning_rate": 3.8411445007444095e-08, "loss": 0.1619, "step": 14139 }, { "epoch": 1.92, "grad_norm": 4.579187145764714, "learning_rate": 3.827528766067845e-08, "loss": 0.1277, "step": 14140 }, { "epoch": 1.92, "grad_norm": 2.7612054547680907, "learning_rate": 3.8139371134359725e-08, "loss": 0.1336, "step": 14141 }, { "epoch": 1.92, "grad_norm": 6.338452698815258, "learning_rate": 3.800369543508431e-08, "loss": 0.1525, "step": 14142 }, { "epoch": 1.92, "grad_norm": 2.8937629680480508, "learning_rate": 3.786826056943749e-08, "loss": 0.1445, "step": 14143 }, { "epoch": 1.92, "grad_norm": 3.14820867923602, "learning_rate": 3.773306654399234e-08, "loss": 0.1553, "step": 14144 }, { "epoch": 1.92, "grad_norm": 4.484168678711918, "learning_rate": 3.7598113365311384e-08, "loss": 0.1693, "step": 14145 }, { "epoch": 1.92, "grad_norm": 3.5048026019279916, "learning_rate": 3.746340103994384e-08, "loss": 0.1226, "step": 14146 }, { "epoch": 1.92, "grad_norm": 2.8376978026337962, "learning_rate": 3.7328929574428354e-08, "loss": 0.0997, "step": 14147 }, { "epoch": 1.92, "grad_norm": 4.182157055689716, "learning_rate": 3.719469897529138e-08, "loss": 0.1357, "step": 14148 }, { "epoch": 1.92, "grad_norm": 4.352161431825068, "learning_rate": 3.7060709249048254e-08, "loss": 0.1441, "step": 14149 }, { "epoch": 1.92, "grad_norm": 3.2733477244649922, "learning_rate": 3.6926960402202674e-08, "loss": 0.141, "step": 14150 }, { "epoch": 1.92, "grad_norm": 3.090692564148693, "learning_rate": 3.6793452441244995e-08, "loss": 0.1282, "step": 14151 }, { "epoch": 1.93, "grad_norm": 2.9985810080062447, "learning_rate": 3.6660185372656144e-08, "loss": 0.1342, "step": 14152 }, { "epoch": 1.93, "grad_norm": 3.4448159659525373, "learning_rate": 3.652715920290428e-08, "loss": 0.1661, "step": 14153 }, { "epoch": 1.93, "grad_norm": 2.9835920877912625, "learning_rate": 3.639437393844481e-08, "loss": 0.1346, "step": 14154 }, { "epoch": 1.93, "grad_norm": 3.101000745015522, "learning_rate": 3.6261829585724215e-08, "loss": 0.1286, "step": 14155 }, { "epoch": 1.93, "grad_norm": 4.659354095602436, "learning_rate": 3.612952615117404e-08, "loss": 0.1293, "step": 14156 }, { "epoch": 1.93, "grad_norm": 3.233462166496531, "learning_rate": 3.5997463641216925e-08, "loss": 0.1958, "step": 14157 }, { "epoch": 1.93, "grad_norm": 3.214767860548768, "learning_rate": 3.586564206226162e-08, "loss": 0.1494, "step": 14158 }, { "epoch": 1.93, "grad_norm": 3.8649566987131947, "learning_rate": 3.5734061420706346e-08, "loss": 0.1773, "step": 14159 }, { "epoch": 1.93, "grad_norm": 3.7143469855636284, "learning_rate": 3.5602721722938774e-08, "loss": 0.1427, "step": 14160 }, { "epoch": 1.93, "grad_norm": 3.1178892734078727, "learning_rate": 3.547162297533158e-08, "loss": 0.1239, "step": 14161 }, { "epoch": 1.93, "grad_norm": 3.2809680868800197, "learning_rate": 3.534076518424967e-08, "loss": 0.1259, "step": 14162 }, { "epoch": 1.93, "grad_norm": 3.0828560635083178, "learning_rate": 3.5210148356042415e-08, "loss": 0.1015, "step": 14163 }, { "epoch": 1.93, "grad_norm": 5.224314413271133, "learning_rate": 3.507977249705086e-08, "loss": 0.1852, "step": 14164 }, { "epoch": 1.93, "grad_norm": 2.4701339085790854, "learning_rate": 3.494963761360215e-08, "loss": 0.1355, "step": 14165 }, { "epoch": 1.93, "grad_norm": 2.448688280148494, "learning_rate": 3.481974371201291e-08, "loss": 0.1419, "step": 14166 }, { "epoch": 1.93, "grad_norm": 3.6710262822782127, "learning_rate": 3.469009079858698e-08, "loss": 0.1732, "step": 14167 }, { "epoch": 1.93, "grad_norm": 2.982014886877094, "learning_rate": 3.456067887961712e-08, "loss": 0.1094, "step": 14168 }, { "epoch": 1.93, "grad_norm": 4.446112988781476, "learning_rate": 3.4431507961385504e-08, "loss": 0.1295, "step": 14169 }, { "epoch": 1.93, "grad_norm": 3.1175642537149324, "learning_rate": 3.430257805016046e-08, "loss": 0.1538, "step": 14170 }, { "epoch": 1.93, "grad_norm": 3.246931916648682, "learning_rate": 3.4173889152200326e-08, "loss": 0.1978, "step": 14171 }, { "epoch": 1.93, "grad_norm": 3.8407751103776544, "learning_rate": 3.404544127375064e-08, "loss": 0.1569, "step": 14172 }, { "epoch": 1.93, "grad_norm": 3.309339556963195, "learning_rate": 3.3917234421045884e-08, "loss": 0.1572, "step": 14173 }, { "epoch": 1.93, "grad_norm": 3.3766993835673604, "learning_rate": 3.3789268600308845e-08, "loss": 0.1638, "step": 14174 }, { "epoch": 1.93, "grad_norm": 2.9002194812557383, "learning_rate": 3.366154381775011e-08, "loss": 0.151, "step": 14175 }, { "epoch": 1.93, "grad_norm": 3.6105695650164136, "learning_rate": 3.3534060079569164e-08, "loss": 0.1553, "step": 14176 }, { "epoch": 1.93, "grad_norm": 3.9076837110300904, "learning_rate": 3.340681739195328e-08, "loss": 0.1694, "step": 14177 }, { "epoch": 1.93, "grad_norm": 2.865437192082293, "learning_rate": 3.3279815761078083e-08, "loss": 0.1415, "step": 14178 }, { "epoch": 1.93, "grad_norm": 3.439256662961902, "learning_rate": 3.3153055193108074e-08, "loss": 0.151, "step": 14179 }, { "epoch": 1.93, "grad_norm": 3.648170771285131, "learning_rate": 3.302653569419556e-08, "loss": 0.1856, "step": 14180 }, { "epoch": 1.93, "grad_norm": 3.306082485981458, "learning_rate": 3.290025727048118e-08, "loss": 0.1637, "step": 14181 }, { "epoch": 1.93, "grad_norm": 4.3236171701812625, "learning_rate": 3.277421992809448e-08, "loss": 0.1821, "step": 14182 }, { "epoch": 1.93, "grad_norm": 3.819314576532098, "learning_rate": 3.264842367315168e-08, "loss": 0.1445, "step": 14183 }, { "epoch": 1.93, "grad_norm": 3.654058235033446, "learning_rate": 3.2522868511759007e-08, "loss": 0.1615, "step": 14184 }, { "epoch": 1.93, "grad_norm": 2.459715906072998, "learning_rate": 3.239755445001047e-08, "loss": 0.1015, "step": 14185 }, { "epoch": 1.93, "grad_norm": 3.1724760627581556, "learning_rate": 3.227248149398787e-08, "loss": 0.1343, "step": 14186 }, { "epoch": 1.93, "grad_norm": 4.331392448743921, "learning_rate": 3.2147649649761914e-08, "loss": 0.1444, "step": 14187 }, { "epoch": 1.93, "grad_norm": 3.0031687360375834, "learning_rate": 3.202305892339164e-08, "loss": 0.1449, "step": 14188 }, { "epoch": 1.93, "grad_norm": 3.5719516765085233, "learning_rate": 3.189870932092332e-08, "loss": 0.1395, "step": 14189 }, { "epoch": 1.93, "grad_norm": 3.19586847319581, "learning_rate": 3.177460084839379e-08, "loss": 0.1419, "step": 14190 }, { "epoch": 1.93, "grad_norm": 3.035607404517107, "learning_rate": 3.165073351182491e-08, "loss": 0.1445, "step": 14191 }, { "epoch": 1.93, "grad_norm": 4.490478739347651, "learning_rate": 3.152710731723019e-08, "loss": 0.1219, "step": 14192 }, { "epoch": 1.93, "grad_norm": 3.2639114385912564, "learning_rate": 3.1403722270609284e-08, "loss": 0.1358, "step": 14193 }, { "epoch": 1.93, "grad_norm": 3.140705399915824, "learning_rate": 3.128057837795129e-08, "loss": 0.1636, "step": 14194 }, { "epoch": 1.93, "grad_norm": 3.62326013643554, "learning_rate": 3.115767564523198e-08, "loss": 0.1747, "step": 14195 }, { "epoch": 1.93, "grad_norm": 4.135690836105003, "learning_rate": 3.1035014078417136e-08, "loss": 0.1709, "step": 14196 }, { "epoch": 1.93, "grad_norm": 3.8643589980457773, "learning_rate": 3.0912593683460336e-08, "loss": 0.1627, "step": 14197 }, { "epoch": 1.93, "grad_norm": 3.607647998157187, "learning_rate": 3.079041446630349e-08, "loss": 0.1287, "step": 14198 }, { "epoch": 1.93, "grad_norm": 2.2849531713787545, "learning_rate": 3.0668476432876293e-08, "loss": 0.131, "step": 14199 }, { "epoch": 1.93, "grad_norm": 3.025935984442156, "learning_rate": 3.05467795890968e-08, "loss": 0.1027, "step": 14200 }, { "epoch": 1.93, "grad_norm": 3.042184881968576, "learning_rate": 3.0425323940871945e-08, "loss": 0.1491, "step": 14201 }, { "epoch": 1.93, "grad_norm": 3.720738781320716, "learning_rate": 3.030410949409701e-08, "loss": 0.1298, "step": 14202 }, { "epoch": 1.93, "grad_norm": 2.859849253923661, "learning_rate": 3.018313625465452e-08, "loss": 0.1205, "step": 14203 }, { "epoch": 1.93, "grad_norm": 4.746530341199026, "learning_rate": 3.0062404228417e-08, "loss": 0.1471, "step": 14204 }, { "epoch": 1.93, "grad_norm": 3.3688816143359674, "learning_rate": 2.994191342124308e-08, "loss": 0.1533, "step": 14205 }, { "epoch": 1.93, "grad_norm": 3.9032210240149454, "learning_rate": 2.9821663838981994e-08, "loss": 0.1362, "step": 14206 }, { "epoch": 1.93, "grad_norm": 5.065679482862626, "learning_rate": 2.9701655487469062e-08, "loss": 0.1316, "step": 14207 }, { "epoch": 1.93, "grad_norm": 4.869177506210316, "learning_rate": 2.958188837252962e-08, "loss": 0.1553, "step": 14208 }, { "epoch": 1.93, "grad_norm": 2.7916986282819742, "learning_rate": 2.946236249997625e-08, "loss": 0.1473, "step": 14209 }, { "epoch": 1.93, "grad_norm": 2.3967294836353528, "learning_rate": 2.934307787561097e-08, "loss": 0.1036, "step": 14210 }, { "epoch": 1.93, "grad_norm": 3.639126008243351, "learning_rate": 2.9224034505222488e-08, "loss": 0.1712, "step": 14211 }, { "epoch": 1.93, "grad_norm": 2.843355405658241, "learning_rate": 2.9105232394588955e-08, "loss": 0.1077, "step": 14212 }, { "epoch": 1.93, "grad_norm": 3.8269116294180776, "learning_rate": 2.898667154947632e-08, "loss": 0.1539, "step": 14213 }, { "epoch": 1.93, "grad_norm": 4.161685671178621, "learning_rate": 2.886835197563942e-08, "loss": 0.1944, "step": 14214 }, { "epoch": 1.93, "grad_norm": 3.638478994031023, "learning_rate": 2.8750273678820327e-08, "loss": 0.1427, "step": 14215 }, { "epoch": 1.93, "grad_norm": 3.07483657827802, "learning_rate": 2.8632436664751127e-08, "loss": 0.1602, "step": 14216 }, { "epoch": 1.93, "grad_norm": 3.068463646499981, "learning_rate": 2.8514840939150023e-08, "loss": 0.1339, "step": 14217 }, { "epoch": 1.93, "grad_norm": 3.2554036621490434, "learning_rate": 2.839748650772578e-08, "loss": 0.1368, "step": 14218 }, { "epoch": 1.93, "grad_norm": 2.8768291450008823, "learning_rate": 2.828037337617273e-08, "loss": 0.1174, "step": 14219 }, { "epoch": 1.93, "grad_norm": 4.132385088766264, "learning_rate": 2.8163501550175775e-08, "loss": 0.1449, "step": 14220 }, { "epoch": 1.93, "grad_norm": 2.8928498920107746, "learning_rate": 2.8046871035408153e-08, "loss": 0.1172, "step": 14221 }, { "epoch": 1.93, "grad_norm": 4.0136492036895595, "learning_rate": 2.793048183752922e-08, "loss": 0.1516, "step": 14222 }, { "epoch": 1.93, "grad_norm": 3.4182407956175744, "learning_rate": 2.781433396218891e-08, "loss": 0.1126, "step": 14223 }, { "epoch": 1.93, "grad_norm": 2.7715280492530496, "learning_rate": 2.7698427415024377e-08, "loss": 0.133, "step": 14224 }, { "epoch": 1.93, "grad_norm": 2.898084001246223, "learning_rate": 2.758276220166056e-08, "loss": 0.1414, "step": 14225 }, { "epoch": 1.94, "grad_norm": 2.796877299343537, "learning_rate": 2.7467338327712424e-08, "loss": 0.1465, "step": 14226 }, { "epoch": 1.94, "grad_norm": 2.616704329086332, "learning_rate": 2.735215579878159e-08, "loss": 0.137, "step": 14227 }, { "epoch": 1.94, "grad_norm": 5.012683551517539, "learning_rate": 2.723721462045803e-08, "loss": 0.167, "step": 14228 }, { "epoch": 1.94, "grad_norm": 4.763134106466115, "learning_rate": 2.712251479832173e-08, "loss": 0.1793, "step": 14229 }, { "epoch": 1.94, "grad_norm": 3.9372342063150287, "learning_rate": 2.7008056337938238e-08, "loss": 0.1562, "step": 14230 }, { "epoch": 1.94, "grad_norm": 3.7281887837318854, "learning_rate": 2.6893839244864216e-08, "loss": 0.1076, "step": 14231 }, { "epoch": 1.94, "grad_norm": 2.6010078744497576, "learning_rate": 2.6779863524642458e-08, "loss": 0.152, "step": 14232 }, { "epoch": 1.94, "grad_norm": 2.86780209231865, "learning_rate": 2.6666129182804646e-08, "loss": 0.1615, "step": 14233 }, { "epoch": 1.94, "grad_norm": 3.7233690213401154, "learning_rate": 2.655263622487192e-08, "loss": 0.1368, "step": 14234 }, { "epoch": 1.94, "grad_norm": 2.881848612192916, "learning_rate": 2.6439384656352096e-08, "loss": 0.1167, "step": 14235 }, { "epoch": 1.94, "grad_norm": 4.533594016662777, "learning_rate": 2.6326374482741335e-08, "loss": 0.1667, "step": 14236 }, { "epoch": 1.94, "grad_norm": 2.939852595516167, "learning_rate": 2.6213605709525803e-08, "loss": 0.1932, "step": 14237 }, { "epoch": 1.94, "grad_norm": 4.4613866727655065, "learning_rate": 2.6101078342178343e-08, "loss": 0.1871, "step": 14238 }, { "epoch": 1.94, "grad_norm": 2.8717702322763565, "learning_rate": 2.5988792386160145e-08, "loss": 0.1329, "step": 14239 }, { "epoch": 1.94, "grad_norm": 4.0335946446224655, "learning_rate": 2.5876747846921846e-08, "loss": 0.1802, "step": 14240 }, { "epoch": 1.94, "grad_norm": 3.539161579277384, "learning_rate": 2.576494472990132e-08, "loss": 0.1245, "step": 14241 }, { "epoch": 1.94, "grad_norm": 3.5005148630113174, "learning_rate": 2.5653383040524228e-08, "loss": 0.1341, "step": 14242 }, { "epoch": 1.94, "grad_norm": 3.1282307462671217, "learning_rate": 2.5542062784206235e-08, "loss": 0.1782, "step": 14243 }, { "epoch": 1.94, "grad_norm": 3.099503222124737, "learning_rate": 2.5430983966350242e-08, "loss": 0.1039, "step": 14244 }, { "epoch": 1.94, "grad_norm": 3.8140380910994236, "learning_rate": 2.532014659234694e-08, "loss": 0.1486, "step": 14245 }, { "epoch": 1.94, "grad_norm": 3.270406985479662, "learning_rate": 2.5209550667576465e-08, "loss": 0.1738, "step": 14246 }, { "epoch": 1.94, "grad_norm": 3.392992974126443, "learning_rate": 2.509919619740675e-08, "loss": 0.1401, "step": 14247 }, { "epoch": 1.94, "grad_norm": 3.1078675708373886, "learning_rate": 2.4989083187192954e-08, "loss": 0.1539, "step": 14248 }, { "epoch": 1.94, "grad_norm": 3.9447205507686776, "learning_rate": 2.48792116422808e-08, "loss": 0.1601, "step": 14249 }, { "epoch": 1.94, "grad_norm": 3.3620903075312225, "learning_rate": 2.476958156800158e-08, "loss": 0.173, "step": 14250 }, { "epoch": 1.94, "grad_norm": 3.6555832547761544, "learning_rate": 2.46601929696777e-08, "loss": 0.195, "step": 14251 }, { "epoch": 1.94, "grad_norm": 3.070094380878646, "learning_rate": 2.4551045852617694e-08, "loss": 0.1563, "step": 14252 }, { "epoch": 1.94, "grad_norm": 3.657756999131551, "learning_rate": 2.444214022211844e-08, "loss": 0.1366, "step": 14253 }, { "epoch": 1.94, "grad_norm": 2.6054031615781117, "learning_rate": 2.4333476083467368e-08, "loss": 0.0864, "step": 14254 }, { "epoch": 1.94, "grad_norm": 3.504255467932299, "learning_rate": 2.422505344193693e-08, "loss": 0.1474, "step": 14255 }, { "epoch": 1.94, "grad_norm": 3.7363168039740615, "learning_rate": 2.411687230279014e-08, "loss": 0.144, "step": 14256 }, { "epoch": 1.94, "grad_norm": 4.275051011590752, "learning_rate": 2.4008932671277795e-08, "loss": 0.1672, "step": 14257 }, { "epoch": 1.94, "grad_norm": 4.252124299456613, "learning_rate": 2.390123455263904e-08, "loss": 0.1527, "step": 14258 }, { "epoch": 1.94, "grad_norm": 4.697465043241915, "learning_rate": 2.379377795210025e-08, "loss": 0.1658, "step": 14259 }, { "epoch": 1.94, "grad_norm": 3.4700714552195615, "learning_rate": 2.368656287487725e-08, "loss": 0.1395, "step": 14260 }, { "epoch": 1.94, "grad_norm": 3.8906964752048214, "learning_rate": 2.357958932617421e-08, "loss": 0.126, "step": 14261 }, { "epoch": 1.94, "grad_norm": 3.3441201027049967, "learning_rate": 2.3472857311183095e-08, "loss": 0.1697, "step": 14262 }, { "epoch": 1.94, "grad_norm": 3.999528807011441, "learning_rate": 2.336636683508364e-08, "loss": 0.1636, "step": 14263 }, { "epoch": 1.94, "grad_norm": 3.5853955969129947, "learning_rate": 2.32601179030445e-08, "loss": 0.1652, "step": 14264 }, { "epoch": 1.94, "grad_norm": 2.7289884732404226, "learning_rate": 2.3154110520223207e-08, "loss": 0.1501, "step": 14265 }, { "epoch": 1.94, "grad_norm": 4.0245113062645395, "learning_rate": 2.3048344691764535e-08, "loss": 0.1419, "step": 14266 }, { "epoch": 1.94, "grad_norm": 2.744017314331698, "learning_rate": 2.294282042280105e-08, "loss": 0.1186, "step": 14267 }, { "epoch": 1.94, "grad_norm": 3.292092874294045, "learning_rate": 2.283753771845587e-08, "loss": 0.1482, "step": 14268 }, { "epoch": 1.94, "grad_norm": 2.7737261365412937, "learning_rate": 2.2732496583838248e-08, "loss": 0.1447, "step": 14269 }, { "epoch": 1.94, "grad_norm": 4.127740893591594, "learning_rate": 2.262769702404688e-08, "loss": 0.1858, "step": 14270 }, { "epoch": 1.94, "grad_norm": 3.839992746310677, "learning_rate": 2.252313904416714e-08, "loss": 0.1229, "step": 14271 }, { "epoch": 1.94, "grad_norm": 3.246229755247038, "learning_rate": 2.2418822649274974e-08, "loss": 0.144, "step": 14272 }, { "epoch": 1.94, "grad_norm": 2.5061365540677323, "learning_rate": 2.2314747844432437e-08, "loss": 0.1185, "step": 14273 }, { "epoch": 1.94, "grad_norm": 3.156495148381555, "learning_rate": 2.221091463469216e-08, "loss": 0.134, "step": 14274 }, { "epoch": 1.94, "grad_norm": 5.083453410551129, "learning_rate": 2.2107323025092886e-08, "loss": 0.188, "step": 14275 }, { "epoch": 1.94, "grad_norm": 3.803211881192395, "learning_rate": 2.2003973020662817e-08, "loss": 0.1568, "step": 14276 }, { "epoch": 1.94, "grad_norm": 3.053639758948676, "learning_rate": 2.1900864626417385e-08, "loss": 0.1583, "step": 14277 }, { "epoch": 1.94, "grad_norm": 3.7214872963025045, "learning_rate": 2.179799784736203e-08, "loss": 0.1953, "step": 14278 }, { "epoch": 1.94, "grad_norm": 2.7906320417690607, "learning_rate": 2.169537268848887e-08, "loss": 0.1335, "step": 14279 }, { "epoch": 1.94, "grad_norm": 3.205060701973699, "learning_rate": 2.1592989154779475e-08, "loss": 0.1555, "step": 14280 }, { "epoch": 1.94, "grad_norm": 2.9737520817019405, "learning_rate": 2.1490847251202097e-08, "loss": 0.1094, "step": 14281 }, { "epoch": 1.94, "grad_norm": 4.895914168386472, "learning_rate": 2.1388946982714986e-08, "loss": 0.1709, "step": 14282 }, { "epoch": 1.94, "grad_norm": 2.4193742313819797, "learning_rate": 2.1287288354263635e-08, "loss": 0.1227, "step": 14283 }, { "epoch": 1.94, "grad_norm": 4.644015065240355, "learning_rate": 2.118587137078243e-08, "loss": 0.1333, "step": 14284 }, { "epoch": 1.94, "grad_norm": 4.184423869891992, "learning_rate": 2.1084696037193543e-08, "loss": 0.1576, "step": 14285 }, { "epoch": 1.94, "grad_norm": 2.854773564015893, "learning_rate": 2.0983762358407488e-08, "loss": 0.1701, "step": 14286 }, { "epoch": 1.94, "grad_norm": 3.982711477507982, "learning_rate": 2.088307033932313e-08, "loss": 0.1381, "step": 14287 }, { "epoch": 1.94, "grad_norm": 3.045514923927698, "learning_rate": 2.0782619984827667e-08, "loss": 0.1655, "step": 14288 }, { "epoch": 1.94, "grad_norm": 2.887670242764729, "learning_rate": 2.0682411299796646e-08, "loss": 0.15, "step": 14289 }, { "epoch": 1.94, "grad_norm": 4.064630062289594, "learning_rate": 2.0582444289093395e-08, "loss": 0.168, "step": 14290 }, { "epoch": 1.94, "grad_norm": 2.9215867050877113, "learning_rate": 2.0482718957570702e-08, "loss": 0.1525, "step": 14291 }, { "epoch": 1.94, "grad_norm": 3.809485807215277, "learning_rate": 2.0383235310068027e-08, "loss": 0.1563, "step": 14292 }, { "epoch": 1.94, "grad_norm": 2.8571838065836923, "learning_rate": 2.0283993351413733e-08, "loss": 0.1579, "step": 14293 }, { "epoch": 1.94, "grad_norm": 2.6317752436391855, "learning_rate": 2.018499308642563e-08, "loss": 0.1088, "step": 14294 }, { "epoch": 1.94, "grad_norm": 3.7035957933452286, "learning_rate": 2.00862345199071e-08, "loss": 0.156, "step": 14295 }, { "epoch": 1.94, "grad_norm": 2.686883489598219, "learning_rate": 1.9987717656653193e-08, "loss": 0.1415, "step": 14296 }, { "epoch": 1.94, "grad_norm": 3.4536777692520424, "learning_rate": 1.9889442501444533e-08, "loss": 0.1765, "step": 14297 }, { "epoch": 1.94, "grad_norm": 3.406649020601835, "learning_rate": 1.9791409059051193e-08, "loss": 0.1714, "step": 14298 }, { "epoch": 1.95, "grad_norm": 3.0673412989795183, "learning_rate": 1.969361733423103e-08, "loss": 0.1496, "step": 14299 }, { "epoch": 1.95, "grad_norm": 3.252955056034579, "learning_rate": 1.9596067331731362e-08, "loss": 0.1226, "step": 14300 }, { "epoch": 1.95, "grad_norm": 3.578718586271023, "learning_rate": 1.949875905628562e-08, "loss": 0.1394, "step": 14301 }, { "epoch": 1.95, "grad_norm": 2.9543145595602396, "learning_rate": 1.9401692512617254e-08, "loss": 0.1484, "step": 14302 }, { "epoch": 1.95, "grad_norm": 2.905843031278512, "learning_rate": 1.930486770543749e-08, "loss": 0.1498, "step": 14303 }, { "epoch": 1.95, "grad_norm": 3.2638558542466116, "learning_rate": 1.9208284639445352e-08, "loss": 0.1749, "step": 14304 }, { "epoch": 1.95, "grad_norm": 4.6762119657147245, "learning_rate": 1.9111943319329308e-08, "loss": 0.1587, "step": 14305 }, { "epoch": 1.95, "grad_norm": 2.793771609926487, "learning_rate": 1.9015843749764507e-08, "loss": 0.131, "step": 14306 }, { "epoch": 1.95, "grad_norm": 3.9863927757664035, "learning_rate": 1.891998593541611e-08, "loss": 0.1795, "step": 14307 }, { "epoch": 1.95, "grad_norm": 3.6050205118617704, "learning_rate": 1.88243698809365e-08, "loss": 0.1624, "step": 14308 }, { "epoch": 1.95, "grad_norm": 5.263815162034296, "learning_rate": 1.8728995590965305e-08, "loss": 0.1592, "step": 14309 }, { "epoch": 1.95, "grad_norm": 2.847828466964528, "learning_rate": 1.8633863070133262e-08, "loss": 0.1057, "step": 14310 }, { "epoch": 1.95, "grad_norm": 3.080205791105365, "learning_rate": 1.8538972323056125e-08, "loss": 0.1047, "step": 14311 }, { "epoch": 1.95, "grad_norm": 3.426055101447011, "learning_rate": 1.8444323354340765e-08, "loss": 0.1409, "step": 14312 }, { "epoch": 1.95, "grad_norm": 2.83640343264845, "learning_rate": 1.8349916168580172e-08, "loss": 0.1495, "step": 14313 }, { "epoch": 1.95, "grad_norm": 3.5638827860815403, "learning_rate": 1.825575077035624e-08, "loss": 0.2047, "step": 14314 }, { "epoch": 1.95, "grad_norm": 3.9559155773841073, "learning_rate": 1.8161827164240865e-08, "loss": 0.199, "step": 14315 }, { "epoch": 1.95, "grad_norm": 3.783715913181034, "learning_rate": 1.8068145354790957e-08, "loss": 0.1671, "step": 14316 }, { "epoch": 1.95, "grad_norm": 3.447636846295325, "learning_rate": 1.7974705346554543e-08, "loss": 0.1701, "step": 14317 }, { "epoch": 1.95, "grad_norm": 5.217106624430448, "learning_rate": 1.7881507144066334e-08, "loss": 0.1519, "step": 14318 }, { "epoch": 1.95, "grad_norm": 3.4726211458205176, "learning_rate": 1.7788550751849377e-08, "loss": 0.1879, "step": 14319 }, { "epoch": 1.95, "grad_norm": 3.3422810454808065, "learning_rate": 1.769583617441617e-08, "loss": 0.1605, "step": 14320 }, { "epoch": 1.95, "grad_norm": 3.1860874054793413, "learning_rate": 1.760336341626645e-08, "loss": 0.1248, "step": 14321 }, { "epoch": 1.95, "grad_norm": 3.090783343594507, "learning_rate": 1.7511132481888293e-08, "loss": 0.1396, "step": 14322 }, { "epoch": 1.95, "grad_norm": 3.495916015947252, "learning_rate": 1.7419143375758117e-08, "loss": 0.1389, "step": 14323 }, { "epoch": 1.95, "grad_norm": 3.882403717084352, "learning_rate": 1.732739610234069e-08, "loss": 0.1655, "step": 14324 }, { "epoch": 1.95, "grad_norm": 2.86895254383319, "learning_rate": 1.7235890666089105e-08, "loss": 0.114, "step": 14325 }, { "epoch": 1.95, "grad_norm": 2.925281984614148, "learning_rate": 1.7144627071444264e-08, "loss": 0.1479, "step": 14326 }, { "epoch": 1.95, "grad_norm": 3.140384066105454, "learning_rate": 1.7053605322837064e-08, "loss": 0.1345, "step": 14327 }, { "epoch": 1.95, "grad_norm": 2.638977016411202, "learning_rate": 1.6962825424683415e-08, "loss": 0.13, "step": 14328 }, { "epoch": 1.95, "grad_norm": 3.2239015119337227, "learning_rate": 1.687228738139035e-08, "loss": 0.1642, "step": 14329 }, { "epoch": 1.95, "grad_norm": 3.2002647961897623, "learning_rate": 1.6781991197352133e-08, "loss": 0.1499, "step": 14330 }, { "epoch": 1.95, "grad_norm": 3.3319528374176124, "learning_rate": 1.669193687695192e-08, "loss": 0.1668, "step": 14331 }, { "epoch": 1.95, "grad_norm": 3.595132318375327, "learning_rate": 1.6602124424558998e-08, "loss": 0.2035, "step": 14332 }, { "epoch": 1.95, "grad_norm": 2.7643113067573553, "learning_rate": 1.651255384453432e-08, "loss": 0.1451, "step": 14333 }, { "epoch": 1.95, "grad_norm": 3.4745645465234696, "learning_rate": 1.6423225141223854e-08, "loss": 0.146, "step": 14334 }, { "epoch": 1.95, "grad_norm": 2.9237680783973357, "learning_rate": 1.6334138318963577e-08, "loss": 0.1335, "step": 14335 }, { "epoch": 1.95, "grad_norm": 3.3435504516729218, "learning_rate": 1.624529338207781e-08, "loss": 0.1234, "step": 14336 }, { "epoch": 1.95, "grad_norm": 3.087328981135023, "learning_rate": 1.6156690334878655e-08, "loss": 0.1583, "step": 14337 }, { "epoch": 1.95, "grad_norm": 2.7796882927209547, "learning_rate": 1.6068329181665453e-08, "loss": 0.1304, "step": 14338 }, { "epoch": 1.95, "grad_norm": 6.1293774651780115, "learning_rate": 1.598020992672866e-08, "loss": 0.1516, "step": 14339 }, { "epoch": 1.95, "grad_norm": 3.3283093842068117, "learning_rate": 1.5892332574343194e-08, "loss": 0.1588, "step": 14340 }, { "epoch": 1.95, "grad_norm": 2.4231168587707597, "learning_rate": 1.5804697128776193e-08, "loss": 0.134, "step": 14341 }, { "epoch": 1.95, "grad_norm": 4.219355207558855, "learning_rate": 1.571730359427981e-08, "loss": 0.1488, "step": 14342 }, { "epoch": 1.95, "grad_norm": 2.898266216000375, "learning_rate": 1.563015197509621e-08, "loss": 0.1155, "step": 14343 }, { "epoch": 1.95, "grad_norm": 3.5963370635737117, "learning_rate": 1.554324227545534e-08, "loss": 0.1555, "step": 14344 }, { "epoch": 1.95, "grad_norm": 3.5587401492719186, "learning_rate": 1.5456574499574937e-08, "loss": 0.1495, "step": 14345 }, { "epoch": 1.95, "grad_norm": 3.463131086426314, "learning_rate": 1.53701486516622e-08, "loss": 0.1645, "step": 14346 }, { "epoch": 1.95, "grad_norm": 3.099861400289469, "learning_rate": 1.5283964735911537e-08, "loss": 0.1315, "step": 14347 }, { "epoch": 1.95, "grad_norm": 3.1544297567954276, "learning_rate": 1.5198022756505727e-08, "loss": 0.1364, "step": 14348 }, { "epoch": 1.95, "grad_norm": 3.879093153833894, "learning_rate": 1.5112322717616424e-08, "loss": 0.1769, "step": 14349 }, { "epoch": 1.95, "grad_norm": 3.025199552866405, "learning_rate": 1.5026864623402527e-08, "loss": 0.1513, "step": 14350 }, { "epoch": 1.95, "grad_norm": 6.4413393603980715, "learning_rate": 1.4941648478012382e-08, "loss": 0.1274, "step": 14351 }, { "epoch": 1.95, "grad_norm": 3.433427929977146, "learning_rate": 1.4856674285582128e-08, "loss": 0.1668, "step": 14352 }, { "epoch": 1.95, "grad_norm": 2.697907786350997, "learning_rate": 1.4771942050235687e-08, "loss": 0.1695, "step": 14353 }, { "epoch": 1.95, "grad_norm": 4.237453671849035, "learning_rate": 1.4687451776085326e-08, "loss": 0.1445, "step": 14354 }, { "epoch": 1.95, "grad_norm": 2.6517993735796432, "learning_rate": 1.4603203467232763e-08, "loss": 0.1403, "step": 14355 }, { "epoch": 1.95, "grad_norm": 3.221986100806882, "learning_rate": 1.4519197127765838e-08, "loss": 0.1638, "step": 14356 }, { "epoch": 1.95, "grad_norm": 3.6817552055711684, "learning_rate": 1.4435432761762958e-08, "loss": 0.1573, "step": 14357 }, { "epoch": 1.95, "grad_norm": 3.0441784147252187, "learning_rate": 1.4351910373288647e-08, "loss": 0.1347, "step": 14358 }, { "epoch": 1.95, "grad_norm": 3.585861183039208, "learning_rate": 1.426862996639744e-08, "loss": 0.1608, "step": 14359 }, { "epoch": 1.95, "grad_norm": 3.0014810568948755, "learning_rate": 1.418559154513166e-08, "loss": 0.1193, "step": 14360 }, { "epoch": 1.95, "grad_norm": 3.5381094491635565, "learning_rate": 1.4102795113520307e-08, "loss": 0.1658, "step": 14361 }, { "epoch": 1.95, "grad_norm": 3.2166384348661383, "learning_rate": 1.4020240675583496e-08, "loss": 0.1747, "step": 14362 }, { "epoch": 1.95, "grad_norm": 2.563402657149303, "learning_rate": 1.3937928235326914e-08, "loss": 0.1342, "step": 14363 }, { "epoch": 1.95, "grad_norm": 2.934607022208722, "learning_rate": 1.3855857796746807e-08, "loss": 0.1667, "step": 14364 }, { "epoch": 1.95, "grad_norm": 3.7043675001684346, "learning_rate": 1.3774029363824992e-08, "loss": 0.1661, "step": 14365 }, { "epoch": 1.95, "grad_norm": 3.385863009726537, "learning_rate": 1.3692442940534401e-08, "loss": 0.1581, "step": 14366 }, { "epoch": 1.95, "grad_norm": 2.9817250830670634, "learning_rate": 1.3611098530834643e-08, "loss": 0.1121, "step": 14367 }, { "epoch": 1.95, "grad_norm": 3.3095693335865493, "learning_rate": 1.3529996138673119e-08, "loss": 0.1323, "step": 14368 }, { "epoch": 1.95, "grad_norm": 3.8930995993068485, "learning_rate": 1.3449135767986676e-08, "loss": 0.1563, "step": 14369 }, { "epoch": 1.95, "grad_norm": 4.331084260250401, "learning_rate": 1.3368517422699956e-08, "loss": 0.162, "step": 14370 }, { "epoch": 1.95, "grad_norm": 3.2789376538217283, "learning_rate": 1.3288141106725938e-08, "loss": 0.1465, "step": 14371 }, { "epoch": 1.95, "grad_norm": 3.788192957639324, "learning_rate": 1.3208006823965391e-08, "loss": 0.1459, "step": 14372 }, { "epoch": 1.96, "grad_norm": 3.5401920289284172, "learning_rate": 1.3128114578307427e-08, "loss": 0.1571, "step": 14373 }, { "epoch": 1.96, "grad_norm": 3.800034158879136, "learning_rate": 1.304846437363061e-08, "loss": 0.1342, "step": 14374 }, { "epoch": 1.96, "grad_norm": 2.4328581055640823, "learning_rate": 1.296905621380018e-08, "loss": 0.132, "step": 14375 }, { "epoch": 1.96, "grad_norm": 4.699390799567378, "learning_rate": 1.2889890102670278e-08, "loss": 0.1808, "step": 14376 }, { "epoch": 1.96, "grad_norm": 4.038321530348146, "learning_rate": 1.2810966044083384e-08, "loss": 0.1526, "step": 14377 }, { "epoch": 1.96, "grad_norm": 4.839751823823959, "learning_rate": 1.2732284041870325e-08, "loss": 0.153, "step": 14378 }, { "epoch": 1.96, "grad_norm": 6.002008381032364, "learning_rate": 1.2653844099849156e-08, "loss": 0.1878, "step": 14379 }, { "epoch": 1.96, "grad_norm": 3.135386855328285, "learning_rate": 1.2575646221828497e-08, "loss": 0.1574, "step": 14380 }, { "epoch": 1.96, "grad_norm": 3.9610799199787468, "learning_rate": 1.2497690411601982e-08, "loss": 0.1934, "step": 14381 }, { "epoch": 1.96, "grad_norm": 3.451512612931562, "learning_rate": 1.241997667295436e-08, "loss": 0.1862, "step": 14382 }, { "epoch": 1.96, "grad_norm": 2.610763788601839, "learning_rate": 1.2342505009657613e-08, "loss": 0.147, "step": 14383 }, { "epoch": 1.96, "grad_norm": 3.3794637251474313, "learning_rate": 1.2265275425471513e-08, "loss": 0.1627, "step": 14384 }, { "epoch": 1.96, "grad_norm": 3.260015578639146, "learning_rate": 1.2188287924144172e-08, "loss": 0.1534, "step": 14385 }, { "epoch": 1.96, "grad_norm": 3.3744767309742434, "learning_rate": 1.21115425094126e-08, "loss": 0.1467, "step": 14386 }, { "epoch": 1.96, "grad_norm": 2.632979168782777, "learning_rate": 1.2035039185001595e-08, "loss": 0.1434, "step": 14387 }, { "epoch": 1.96, "grad_norm": 3.704642967106605, "learning_rate": 1.1958777954624301e-08, "loss": 0.1547, "step": 14388 }, { "epoch": 1.96, "grad_norm": 3.667329157477734, "learning_rate": 1.18827588219822e-08, "loss": 0.134, "step": 14389 }, { "epoch": 1.96, "grad_norm": 3.902604541227342, "learning_rate": 1.1806981790765114e-08, "loss": 0.1481, "step": 14390 }, { "epoch": 1.96, "grad_norm": 2.992777588706737, "learning_rate": 1.1731446864650108e-08, "loss": 0.1364, "step": 14391 }, { "epoch": 1.96, "grad_norm": 3.304248904562372, "learning_rate": 1.1656154047303691e-08, "loss": 0.1577, "step": 14392 }, { "epoch": 1.96, "grad_norm": 3.4136807311859307, "learning_rate": 1.1581103342381273e-08, "loss": 0.1444, "step": 14393 }, { "epoch": 1.96, "grad_norm": 4.333927375502566, "learning_rate": 1.1506294753523829e-08, "loss": 0.1459, "step": 14394 }, { "epoch": 1.96, "grad_norm": 4.762543261788601, "learning_rate": 1.1431728284364007e-08, "loss": 0.1199, "step": 14395 }, { "epoch": 1.96, "grad_norm": 2.8116692310575413, "learning_rate": 1.1357403938518918e-08, "loss": 0.1215, "step": 14396 }, { "epoch": 1.96, "grad_norm": 3.3154285153882417, "learning_rate": 1.128332171959734e-08, "loss": 0.1438, "step": 14397 }, { "epoch": 1.96, "grad_norm": 4.065280740525749, "learning_rate": 1.1209481631194729e-08, "loss": 0.1541, "step": 14398 }, { "epoch": 1.96, "grad_norm": 3.3792832324963737, "learning_rate": 1.1135883676894887e-08, "loss": 0.1509, "step": 14399 }, { "epoch": 1.96, "grad_norm": 3.0558990358574363, "learning_rate": 1.1062527860269956e-08, "loss": 0.1069, "step": 14400 }, { "epoch": 1.96, "grad_norm": 3.3673565909645893, "learning_rate": 1.0989414184879866e-08, "loss": 0.1611, "step": 14401 }, { "epoch": 1.96, "grad_norm": 2.765954004138858, "learning_rate": 1.0916542654273443e-08, "loss": 0.1252, "step": 14402 }, { "epoch": 1.96, "grad_norm": 2.790016017376184, "learning_rate": 1.0843913271987861e-08, "loss": 0.1554, "step": 14403 }, { "epoch": 1.96, "grad_norm": 3.506400901413209, "learning_rate": 1.0771526041548074e-08, "loss": 0.159, "step": 14404 }, { "epoch": 1.96, "grad_norm": 3.2606663757839, "learning_rate": 1.0699380966467388e-08, "loss": 0.1353, "step": 14405 }, { "epoch": 1.96, "grad_norm": 4.038555596721692, "learning_rate": 1.0627478050247442e-08, "loss": 0.161, "step": 14406 }, { "epoch": 1.96, "grad_norm": 4.067894490797818, "learning_rate": 1.0555817296378223e-08, "loss": 0.1583, "step": 14407 }, { "epoch": 1.96, "grad_norm": 2.9311962302019894, "learning_rate": 1.0484398708337507e-08, "loss": 0.1507, "step": 14408 }, { "epoch": 1.96, "grad_norm": 3.9662202625770826, "learning_rate": 1.0413222289591962e-08, "loss": 0.1452, "step": 14409 }, { "epoch": 1.96, "grad_norm": 3.9805474996243966, "learning_rate": 1.0342288043595494e-08, "loss": 0.143, "step": 14410 }, { "epoch": 1.96, "grad_norm": 3.849427054154618, "learning_rate": 1.0271595973792014e-08, "loss": 0.1691, "step": 14411 }, { "epoch": 1.96, "grad_norm": 4.794025520738991, "learning_rate": 1.0201146083612113e-08, "loss": 0.1779, "step": 14412 }, { "epoch": 1.96, "grad_norm": 3.096027759341132, "learning_rate": 1.0130938376474719e-08, "loss": 0.1754, "step": 14413 }, { "epoch": 1.96, "grad_norm": 3.623999151477795, "learning_rate": 1.006097285578822e-08, "loss": 0.1748, "step": 14414 }, { "epoch": 1.96, "grad_norm": 2.8040132986771797, "learning_rate": 9.991249524947676e-09, "loss": 0.1172, "step": 14415 }, { "epoch": 1.96, "grad_norm": 2.9716209235120514, "learning_rate": 9.921768387337605e-09, "loss": 0.1145, "step": 14416 }, { "epoch": 1.96, "grad_norm": 2.7707854138946457, "learning_rate": 9.852529446330306e-09, "loss": 0.1088, "step": 14417 }, { "epoch": 1.96, "grad_norm": 3.100345553491352, "learning_rate": 9.783532705286425e-09, "loss": 0.1422, "step": 14418 }, { "epoch": 1.96, "grad_norm": 3.1634952064334017, "learning_rate": 9.71477816755384e-09, "loss": 0.1631, "step": 14419 }, { "epoch": 1.96, "grad_norm": 3.857215161553794, "learning_rate": 9.646265836470992e-09, "loss": 0.1695, "step": 14420 }, { "epoch": 1.96, "grad_norm": 3.1563994218073215, "learning_rate": 9.577995715361887e-09, "loss": 0.1379, "step": 14421 }, { "epoch": 1.96, "grad_norm": 3.4384022777523877, "learning_rate": 9.509967807541098e-09, "loss": 0.1476, "step": 14422 }, { "epoch": 1.96, "grad_norm": 3.6181810982310214, "learning_rate": 9.442182116309872e-09, "loss": 0.1411, "step": 14423 }, { "epoch": 1.96, "grad_norm": 4.364185135878329, "learning_rate": 9.374638644958357e-09, "loss": 0.1575, "step": 14424 }, { "epoch": 1.96, "grad_norm": 3.8524608981909005, "learning_rate": 9.307337396764482e-09, "loss": 0.1471, "step": 14425 }, { "epoch": 1.96, "grad_norm": 2.7467203105290534, "learning_rate": 9.240278374995637e-09, "loss": 0.0991, "step": 14426 }, { "epoch": 1.96, "grad_norm": 3.187668502413109, "learning_rate": 9.17346158290533e-09, "loss": 0.1237, "step": 14427 }, { "epoch": 1.96, "grad_norm": 2.68033996315591, "learning_rate": 9.106887023737632e-09, "loss": 0.1328, "step": 14428 }, { "epoch": 1.96, "grad_norm": 3.5584725958421797, "learning_rate": 9.040554700723292e-09, "loss": 0.1202, "step": 14429 }, { "epoch": 1.96, "grad_norm": 3.8559110428345376, "learning_rate": 8.974464617081957e-09, "loss": 0.1027, "step": 14430 }, { "epoch": 1.96, "grad_norm": 3.5850244262905244, "learning_rate": 8.908616776021062e-09, "loss": 0.1426, "step": 14431 }, { "epoch": 1.96, "grad_norm": 3.2336123150034144, "learning_rate": 8.843011180736383e-09, "loss": 0.1327, "step": 14432 }, { "epoch": 1.96, "grad_norm": 4.056637872258021, "learning_rate": 8.77764783441315e-09, "loss": 0.193, "step": 14433 }, { "epoch": 1.96, "grad_norm": 3.228958596254558, "learning_rate": 8.71252674022327e-09, "loss": 0.13, "step": 14434 }, { "epoch": 1.96, "grad_norm": 3.8177847277252126, "learning_rate": 8.647647901326994e-09, "loss": 0.1274, "step": 14435 }, { "epoch": 1.96, "grad_norm": 2.6955449533952343, "learning_rate": 8.58301132087347e-09, "loss": 0.1646, "step": 14436 }, { "epoch": 1.96, "grad_norm": 3.790840063225992, "learning_rate": 8.518617002000184e-09, "loss": 0.1738, "step": 14437 }, { "epoch": 1.96, "grad_norm": 3.0173093353203715, "learning_rate": 8.454464947832975e-09, "loss": 0.1681, "step": 14438 }, { "epoch": 1.96, "grad_norm": 4.342829089370126, "learning_rate": 8.390555161483793e-09, "loss": 0.1939, "step": 14439 }, { "epoch": 1.96, "grad_norm": 3.4654714118731986, "learning_rate": 8.326887646056826e-09, "loss": 0.1366, "step": 14440 }, { "epoch": 1.96, "grad_norm": 2.4586023815914753, "learning_rate": 8.263462404640155e-09, "loss": 0.1236, "step": 14441 }, { "epoch": 1.96, "grad_norm": 4.794908277232816, "learning_rate": 8.200279440313541e-09, "loss": 0.1596, "step": 14442 }, { "epoch": 1.96, "grad_norm": 3.495384782397564, "learning_rate": 8.137338756142865e-09, "loss": 0.1443, "step": 14443 }, { "epoch": 1.96, "grad_norm": 5.4162709970167064, "learning_rate": 8.074640355183461e-09, "loss": 0.17, "step": 14444 }, { "epoch": 1.96, "grad_norm": 4.277446826628244, "learning_rate": 8.012184240477893e-09, "loss": 0.1369, "step": 14445 }, { "epoch": 1.97, "grad_norm": 3.84205354552003, "learning_rate": 7.949970415058183e-09, "loss": 0.124, "step": 14446 }, { "epoch": 1.97, "grad_norm": 3.6360736017443602, "learning_rate": 7.88799888194358e-09, "loss": 0.1647, "step": 14447 }, { "epoch": 1.97, "grad_norm": 4.628653912270013, "learning_rate": 7.826269644142237e-09, "loss": 0.1429, "step": 14448 }, { "epoch": 1.97, "grad_norm": 4.529607604020148, "learning_rate": 7.764782704649532e-09, "loss": 0.1783, "step": 14449 }, { "epoch": 1.97, "grad_norm": 3.402940575239597, "learning_rate": 7.703538066449745e-09, "loss": 0.166, "step": 14450 }, { "epoch": 1.97, "grad_norm": 3.73369701503505, "learning_rate": 7.642535732516609e-09, "loss": 0.1128, "step": 14451 }, { "epoch": 1.97, "grad_norm": 3.249258221344137, "learning_rate": 7.581775705809424e-09, "loss": 0.1332, "step": 14452 }, { "epoch": 1.97, "grad_norm": 2.7203218492732053, "learning_rate": 7.521257989278053e-09, "loss": 0.1271, "step": 14453 }, { "epoch": 1.97, "grad_norm": 2.802009398588606, "learning_rate": 7.460982585860144e-09, "loss": 0.107, "step": 14454 }, { "epoch": 1.97, "grad_norm": 2.6443130426240855, "learning_rate": 7.400949498480026e-09, "loss": 0.1276, "step": 14455 }, { "epoch": 1.97, "grad_norm": 3.040748276850913, "learning_rate": 7.341158730052589e-09, "loss": 0.1689, "step": 14456 }, { "epoch": 1.97, "grad_norm": 3.5220279898410287, "learning_rate": 7.281610283479401e-09, "loss": 0.1372, "step": 14457 }, { "epoch": 1.97, "grad_norm": 3.0426028844142348, "learning_rate": 7.2223041616503735e-09, "loss": 0.145, "step": 14458 }, { "epoch": 1.97, "grad_norm": 3.637150717375302, "learning_rate": 7.1632403674437575e-09, "loss": 0.1321, "step": 14459 }, { "epoch": 1.97, "grad_norm": 4.263057705343519, "learning_rate": 7.104418903727262e-09, "loss": 0.1664, "step": 14460 }, { "epoch": 1.97, "grad_norm": 3.5650391942019017, "learning_rate": 7.045839773355268e-09, "loss": 0.1746, "step": 14461 }, { "epoch": 1.97, "grad_norm": 3.325509680990715, "learning_rate": 6.987502979170502e-09, "loss": 0.1363, "step": 14462 }, { "epoch": 1.97, "grad_norm": 3.1251779299198144, "learning_rate": 6.9294085240051436e-09, "loss": 0.1296, "step": 14463 }, { "epoch": 1.97, "grad_norm": 2.9163134438725704, "learning_rate": 6.871556410678603e-09, "loss": 0.1457, "step": 14464 }, { "epoch": 1.97, "grad_norm": 4.244962808621726, "learning_rate": 6.813946641998637e-09, "loss": 0.1391, "step": 14465 }, { "epoch": 1.97, "grad_norm": 3.638426362900443, "learning_rate": 6.756579220761339e-09, "loss": 0.1741, "step": 14466 }, { "epoch": 1.97, "grad_norm": 3.76306019122633, "learning_rate": 6.69945414975115e-09, "loss": 0.0982, "step": 14467 }, { "epoch": 1.97, "grad_norm": 4.5610680621848845, "learning_rate": 6.642571431740297e-09, "loss": 0.1402, "step": 14468 }, { "epoch": 1.97, "grad_norm": 3.232392445899805, "learning_rate": 6.585931069490459e-09, "loss": 0.1775, "step": 14469 }, { "epoch": 1.97, "grad_norm": 2.914793136234766, "learning_rate": 6.529533065750548e-09, "loss": 0.1541, "step": 14470 }, { "epoch": 1.97, "grad_norm": 4.214084448128165, "learning_rate": 6.4733774232567105e-09, "loss": 0.1434, "step": 14471 }, { "epoch": 1.97, "grad_norm": 4.551090943068356, "learning_rate": 6.417464144736208e-09, "loss": 0.1218, "step": 14472 }, { "epoch": 1.97, "grad_norm": 2.9714066584544105, "learning_rate": 6.3617932329013144e-09, "loss": 0.1362, "step": 14473 }, { "epoch": 1.97, "grad_norm": 2.5396537568370476, "learning_rate": 6.306364690454869e-09, "loss": 0.1003, "step": 14474 }, { "epoch": 1.97, "grad_norm": 2.9077624202926904, "learning_rate": 6.251178520087498e-09, "loss": 0.1115, "step": 14475 }, { "epoch": 1.97, "grad_norm": 3.7052880497760468, "learning_rate": 6.196234724476502e-09, "loss": 0.179, "step": 14476 }, { "epoch": 1.97, "grad_norm": 3.8098265588850944, "learning_rate": 6.141533306289749e-09, "loss": 0.1596, "step": 14477 }, { "epoch": 1.97, "grad_norm": 4.46028271770615, "learning_rate": 6.087074268181225e-09, "loss": 0.1642, "step": 14478 }, { "epoch": 1.97, "grad_norm": 2.741138558865164, "learning_rate": 6.032857612794929e-09, "loss": 0.1356, "step": 14479 }, { "epoch": 1.97, "grad_norm": 3.840538396042472, "learning_rate": 5.978883342762642e-09, "loss": 0.1763, "step": 14480 }, { "epoch": 1.97, "grad_norm": 3.4677187162317527, "learning_rate": 5.9251514607028275e-09, "loss": 0.117, "step": 14481 }, { "epoch": 1.97, "grad_norm": 4.7135227990499, "learning_rate": 5.871661969223951e-09, "loss": 0.1558, "step": 14482 }, { "epoch": 1.97, "grad_norm": 2.6294379789612576, "learning_rate": 5.818414870922273e-09, "loss": 0.14, "step": 14483 }, { "epoch": 1.97, "grad_norm": 4.161504723455294, "learning_rate": 5.765410168381835e-09, "loss": 0.1538, "step": 14484 }, { "epoch": 1.97, "grad_norm": 2.4628168618132142, "learning_rate": 5.712647864176135e-09, "loss": 0.1461, "step": 14485 }, { "epoch": 1.97, "grad_norm": 3.490364168109877, "learning_rate": 5.660127960864792e-09, "loss": 0.1135, "step": 14486 }, { "epoch": 1.97, "grad_norm": 3.663621137672347, "learning_rate": 5.6078504609979874e-09, "loss": 0.1546, "step": 14487 }, { "epoch": 1.97, "grad_norm": 3.4030251884996465, "learning_rate": 5.555815367112583e-09, "loss": 0.12, "step": 14488 }, { "epoch": 1.97, "grad_norm": 2.948246318197081, "learning_rate": 5.504022681733778e-09, "loss": 0.1287, "step": 14489 }, { "epoch": 1.97, "grad_norm": 3.129931296849511, "learning_rate": 5.452472407375675e-09, "loss": 0.141, "step": 14490 }, { "epoch": 1.97, "grad_norm": 3.8809407403981577, "learning_rate": 5.401164546540716e-09, "loss": 0.1681, "step": 14491 }, { "epoch": 1.97, "grad_norm": 2.7820614979739817, "learning_rate": 5.350099101718575e-09, "loss": 0.1412, "step": 14492 }, { "epoch": 1.97, "grad_norm": 3.810303872050765, "learning_rate": 5.299276075387827e-09, "loss": 0.1295, "step": 14493 }, { "epoch": 1.97, "grad_norm": 4.729069613031806, "learning_rate": 5.248695470015386e-09, "loss": 0.1515, "step": 14494 }, { "epoch": 1.97, "grad_norm": 4.284312489001034, "learning_rate": 5.1983572880565106e-09, "loss": 0.1784, "step": 14495 }, { "epoch": 1.97, "grad_norm": 2.7946092673618077, "learning_rate": 5.148261531953691e-09, "loss": 0.1778, "step": 14496 }, { "epoch": 1.97, "grad_norm": 5.138084259005446, "learning_rate": 5.098408204138872e-09, "loss": 0.1861, "step": 14497 }, { "epoch": 1.97, "grad_norm": 3.405700459647428, "learning_rate": 5.048797307031228e-09, "loss": 0.1303, "step": 14498 }, { "epoch": 1.97, "grad_norm": 2.8466387508793867, "learning_rate": 4.9994288430388334e-09, "loss": 0.139, "step": 14499 }, { "epoch": 1.97, "grad_norm": 2.783643870754094, "learning_rate": 4.950302814558105e-09, "loss": 0.1448, "step": 14500 }, { "epoch": 1.97, "grad_norm": 3.1032358739399397, "learning_rate": 4.901419223973803e-09, "loss": 0.1668, "step": 14501 }, { "epoch": 1.97, "grad_norm": 2.9898810807467497, "learning_rate": 4.852778073657361e-09, "loss": 0.1314, "step": 14502 }, { "epoch": 1.97, "grad_norm": 2.475752750419503, "learning_rate": 4.8043793659707794e-09, "loss": 0.1331, "step": 14503 }, { "epoch": 1.97, "grad_norm": 3.666249947116254, "learning_rate": 4.756223103262181e-09, "loss": 0.1408, "step": 14504 }, { "epoch": 1.97, "grad_norm": 3.1305919240850235, "learning_rate": 4.708309287869139e-09, "loss": 0.1706, "step": 14505 }, { "epoch": 1.97, "grad_norm": 3.908687090268322, "learning_rate": 4.660637922117572e-09, "loss": 0.1301, "step": 14506 }, { "epoch": 1.97, "grad_norm": 2.7646318955363323, "learning_rate": 4.613209008320629e-09, "loss": 0.1448, "step": 14507 }, { "epoch": 1.97, "grad_norm": 3.572471719516573, "learning_rate": 4.566022548780913e-09, "loss": 0.122, "step": 14508 }, { "epoch": 1.97, "grad_norm": 3.130882626818302, "learning_rate": 4.519078545788258e-09, "loss": 0.1018, "step": 14509 }, { "epoch": 1.97, "grad_norm": 3.8815667841508414, "learning_rate": 4.472377001620843e-09, "loss": 0.1604, "step": 14510 }, { "epoch": 1.97, "grad_norm": 3.933227340400228, "learning_rate": 4.4259179185462964e-09, "loss": 0.137, "step": 14511 }, { "epoch": 1.97, "grad_norm": 3.553097599080517, "learning_rate": 4.379701298818928e-09, "loss": 0.1365, "step": 14512 }, { "epoch": 1.97, "grad_norm": 3.372922863525578, "learning_rate": 4.3337271446813875e-09, "loss": 0.1605, "step": 14513 }, { "epoch": 1.97, "grad_norm": 4.767772864729194, "learning_rate": 4.287995458366334e-09, "loss": 0.1684, "step": 14514 }, { "epoch": 1.97, "grad_norm": 3.3093446541217526, "learning_rate": 4.242506242092548e-09, "loss": 0.1789, "step": 14515 }, { "epoch": 1.97, "grad_norm": 5.412336847798854, "learning_rate": 4.197259498067707e-09, "loss": 0.1283, "step": 14516 }, { "epoch": 1.97, "grad_norm": 2.613851503589191, "learning_rate": 4.152255228487834e-09, "loss": 0.1398, "step": 14517 }, { "epoch": 1.97, "grad_norm": 4.204393664910199, "learning_rate": 4.1074934355384015e-09, "loss": 0.156, "step": 14518 }, { "epoch": 1.97, "grad_norm": 3.303152393361639, "learning_rate": 4.06297412139045e-09, "loss": 0.1635, "step": 14519 }, { "epoch": 1.98, "grad_norm": 2.599231708998189, "learning_rate": 4.018697288206141e-09, "loss": 0.1214, "step": 14520 }, { "epoch": 1.98, "grad_norm": 4.496875867166405, "learning_rate": 3.9746629381332e-09, "loss": 0.1491, "step": 14521 }, { "epoch": 1.98, "grad_norm": 3.410027150426615, "learning_rate": 3.9308710733093616e-09, "loss": 0.1782, "step": 14522 }, { "epoch": 1.98, "grad_norm": 3.8137139803717144, "learning_rate": 3.887321695860702e-09, "loss": 0.1914, "step": 14523 }, { "epoch": 1.98, "grad_norm": 2.964815815669776, "learning_rate": 3.844014807899976e-09, "loss": 0.1617, "step": 14524 }, { "epoch": 1.98, "grad_norm": 3.885220236637562, "learning_rate": 3.800950411529392e-09, "loss": 0.1754, "step": 14525 }, { "epoch": 1.98, "grad_norm": 4.187046118944492, "learning_rate": 3.758128508839498e-09, "loss": 0.1476, "step": 14526 }, { "epoch": 1.98, "grad_norm": 3.815667903746263, "learning_rate": 3.715549101908633e-09, "loss": 0.1213, "step": 14527 }, { "epoch": 1.98, "grad_norm": 3.8307180093231756, "learning_rate": 3.6732121928029218e-09, "loss": 0.142, "step": 14528 }, { "epoch": 1.98, "grad_norm": 3.5186840397906978, "learning_rate": 3.6311177835773868e-09, "loss": 0.1398, "step": 14529 }, { "epoch": 1.98, "grad_norm": 2.8410227895295033, "learning_rate": 3.589265876275394e-09, "loss": 0.1256, "step": 14530 }, { "epoch": 1.98, "grad_norm": 3.7301326671958392, "learning_rate": 3.547656472928096e-09, "loss": 0.115, "step": 14531 }, { "epoch": 1.98, "grad_norm": 4.974638891008808, "learning_rate": 3.5062895755544337e-09, "loss": 0.1493, "step": 14532 }, { "epoch": 1.98, "grad_norm": 3.4730102688366142, "learning_rate": 3.465165186163355e-09, "loss": 0.1602, "step": 14533 }, { "epoch": 1.98, "grad_norm": 3.6438440034603325, "learning_rate": 3.4242833067499314e-09, "loss": 0.1377, "step": 14534 }, { "epoch": 1.98, "grad_norm": 2.727915939011322, "learning_rate": 3.3836439392992414e-09, "loss": 0.1569, "step": 14535 }, { "epoch": 1.98, "grad_norm": 3.688358186655662, "learning_rate": 3.343247085782486e-09, "loss": 0.2049, "step": 14536 }, { "epoch": 1.98, "grad_norm": 2.9577145072337934, "learning_rate": 3.3030927481614294e-09, "loss": 0.1262, "step": 14537 }, { "epoch": 1.98, "grad_norm": 4.803232837338824, "learning_rate": 3.263180928385068e-09, "loss": 0.1552, "step": 14538 }, { "epoch": 1.98, "grad_norm": 3.255422779851963, "learning_rate": 3.2235116283896305e-09, "loss": 0.1341, "step": 14539 }, { "epoch": 1.98, "grad_norm": 2.961357114726015, "learning_rate": 3.1840848501007993e-09, "loss": 0.121, "step": 14540 }, { "epoch": 1.98, "grad_norm": 3.4982845812996244, "learning_rate": 3.1449005954325985e-09, "loss": 0.1322, "step": 14541 }, { "epoch": 1.98, "grad_norm": 3.9791258238479696, "learning_rate": 3.10595886628684e-09, "loss": 0.1498, "step": 14542 }, { "epoch": 1.98, "grad_norm": 3.8567189522186776, "learning_rate": 3.0672596645525686e-09, "loss": 0.18, "step": 14543 }, { "epoch": 1.98, "grad_norm": 2.698866329775613, "learning_rate": 3.028802992109392e-09, "loss": 0.1817, "step": 14544 }, { "epoch": 1.98, "grad_norm": 2.529185117152515, "learning_rate": 2.9905888508235945e-09, "loss": 0.1637, "step": 14545 }, { "epoch": 1.98, "grad_norm": 2.4819527616433916, "learning_rate": 2.9526172425492493e-09, "loss": 0.1248, "step": 14546 }, { "epoch": 1.98, "grad_norm": 3.7998960530446, "learning_rate": 2.9148881691298812e-09, "loss": 0.1786, "step": 14547 }, { "epoch": 1.98, "grad_norm": 3.6806759517631633, "learning_rate": 2.8774016323962484e-09, "loss": 0.1081, "step": 14548 }, { "epoch": 1.98, "grad_norm": 3.1644427699494844, "learning_rate": 2.8401576341685612e-09, "loss": 0.1378, "step": 14549 }, { "epoch": 1.98, "grad_norm": 4.132484042441659, "learning_rate": 2.8031561762537073e-09, "loss": 0.1476, "step": 14550 }, { "epoch": 1.98, "grad_norm": 5.134400555588281, "learning_rate": 2.766397260447473e-09, "loss": 0.1664, "step": 14551 }, { "epoch": 1.98, "grad_norm": 3.211465914695835, "learning_rate": 2.7298808885350968e-09, "loss": 0.1343, "step": 14552 }, { "epoch": 1.98, "grad_norm": 3.8023559131472386, "learning_rate": 2.6936070622879397e-09, "loss": 0.1442, "step": 14553 }, { "epoch": 1.98, "grad_norm": 3.0337782087962637, "learning_rate": 2.6575757834662595e-09, "loss": 0.113, "step": 14554 }, { "epoch": 1.98, "grad_norm": 3.2767215141197124, "learning_rate": 2.621787053819769e-09, "loss": 0.1322, "step": 14555 }, { "epoch": 1.98, "grad_norm": 3.4394723350595933, "learning_rate": 2.5862408750854108e-09, "loss": 0.1356, "step": 14556 }, { "epoch": 1.98, "grad_norm": 3.569466958411025, "learning_rate": 2.550937248987917e-09, "loss": 0.1221, "step": 14557 }, { "epoch": 1.98, "grad_norm": 3.598330493610758, "learning_rate": 2.515876177240917e-09, "loss": 0.1235, "step": 14558 }, { "epoch": 1.98, "grad_norm": 4.114882242689799, "learning_rate": 2.4810576615463823e-09, "loss": 0.1707, "step": 14559 }, { "epoch": 1.98, "grad_norm": 2.765774622578876, "learning_rate": 2.4464817035940726e-09, "loss": 0.1293, "step": 14560 }, { "epoch": 1.98, "grad_norm": 3.5707246261233734, "learning_rate": 2.41214830506209e-09, "loss": 0.1404, "step": 14561 }, { "epoch": 1.98, "grad_norm": 2.995301995458888, "learning_rate": 2.378057467617434e-09, "loss": 0.1688, "step": 14562 }, { "epoch": 1.98, "grad_norm": 3.3761334928359616, "learning_rate": 2.344209192913782e-09, "loss": 0.1792, "step": 14563 }, { "epoch": 1.98, "grad_norm": 2.9116743091955963, "learning_rate": 2.3106034825942647e-09, "loss": 0.1346, "step": 14564 }, { "epoch": 1.98, "grad_norm": 4.649097799260775, "learning_rate": 2.277240338290354e-09, "loss": 0.2102, "step": 14565 }, { "epoch": 1.98, "grad_norm": 2.5514129572542834, "learning_rate": 2.2441197616207554e-09, "loss": 0.1241, "step": 14566 }, { "epoch": 1.98, "grad_norm": 2.7134129865680303, "learning_rate": 2.211241754193627e-09, "loss": 0.1378, "step": 14567 }, { "epoch": 1.98, "grad_norm": 3.6366018925457344, "learning_rate": 2.178606317604359e-09, "loss": 0.1171, "step": 14568 }, { "epoch": 1.98, "grad_norm": 3.0312734237794188, "learning_rate": 2.1462134534372403e-09, "loss": 0.1493, "step": 14569 }, { "epoch": 1.98, "grad_norm": 3.6995403120534536, "learning_rate": 2.114063163263791e-09, "loss": 0.1438, "step": 14570 }, { "epoch": 1.98, "grad_norm": 3.3406014098055223, "learning_rate": 2.0821554486455397e-09, "loss": 0.137, "step": 14571 }, { "epoch": 1.98, "grad_norm": 3.0391890940802218, "learning_rate": 2.050490311130138e-09, "loss": 0.1403, "step": 14572 }, { "epoch": 1.98, "grad_norm": 4.23038177128481, "learning_rate": 2.0190677522546886e-09, "loss": 0.1553, "step": 14573 }, { "epoch": 1.98, "grad_norm": 4.025144981772045, "learning_rate": 1.987887773544639e-09, "loss": 0.1526, "step": 14574 }, { "epoch": 1.98, "grad_norm": 3.3502015213244145, "learning_rate": 1.9569503765132224e-09, "loss": 0.1209, "step": 14575 }, { "epoch": 1.98, "grad_norm": 2.6575159066736598, "learning_rate": 1.926255562662016e-09, "loss": 0.1606, "step": 14576 }, { "epoch": 1.98, "grad_norm": 2.3559373549446465, "learning_rate": 1.8958033334803837e-09, "loss": 0.1282, "step": 14577 }, { "epoch": 1.98, "grad_norm": 2.696004674936664, "learning_rate": 1.865593690446588e-09, "loss": 0.1229, "step": 14578 }, { "epoch": 1.98, "grad_norm": 3.0599427818295113, "learning_rate": 1.8356266350277873e-09, "loss": 0.1505, "step": 14579 }, { "epoch": 1.98, "grad_norm": 2.8601103977026647, "learning_rate": 1.8059021686767098e-09, "loss": 0.1632, "step": 14580 }, { "epoch": 1.98, "grad_norm": 4.798905270766104, "learning_rate": 1.776420292837755e-09, "loss": 0.1269, "step": 14581 }, { "epoch": 1.98, "grad_norm": 3.029651250796882, "learning_rate": 1.7471810089403352e-09, "loss": 0.1342, "step": 14582 }, { "epoch": 1.98, "grad_norm": 3.2399520842794765, "learning_rate": 1.7181843184049807e-09, "loss": 0.1228, "step": 14583 }, { "epoch": 1.98, "grad_norm": 3.7377191199581103, "learning_rate": 1.689430222638344e-09, "loss": 0.1694, "step": 14584 }, { "epoch": 1.98, "grad_norm": 3.0624818500244024, "learning_rate": 1.6609187230354207e-09, "loss": 0.1648, "step": 14585 }, { "epoch": 1.98, "grad_norm": 3.5946583103012957, "learning_rate": 1.6326498209812136e-09, "loss": 0.1151, "step": 14586 }, { "epoch": 1.98, "grad_norm": 3.508571071533131, "learning_rate": 1.6046235178474034e-09, "loss": 0.1849, "step": 14587 }, { "epoch": 1.98, "grad_norm": 3.959624104943235, "learning_rate": 1.5768398149940135e-09, "loss": 0.1082, "step": 14588 }, { "epoch": 1.98, "grad_norm": 3.9701781299168637, "learning_rate": 1.5492987137694094e-09, "loss": 0.1297, "step": 14589 }, { "epoch": 1.98, "grad_norm": 2.873470924345312, "learning_rate": 1.522000215510855e-09, "loss": 0.1606, "step": 14590 }, { "epoch": 1.98, "grad_norm": 3.7236383614425907, "learning_rate": 1.4949443215428461e-09, "loss": 0.1128, "step": 14591 }, { "epoch": 1.98, "grad_norm": 2.702599670615106, "learning_rate": 1.4681310331787767e-09, "loss": 0.1107, "step": 14592 }, { "epoch": 1.99, "grad_norm": 3.2363658882004116, "learning_rate": 1.4415603517203835e-09, "loss": 0.1584, "step": 14593 }, { "epoch": 1.99, "grad_norm": 2.851752902192359, "learning_rate": 1.4152322784566351e-09, "loss": 0.1276, "step": 14594 }, { "epoch": 1.99, "grad_norm": 4.289763708435722, "learning_rate": 1.3891468146653985e-09, "loss": 0.1614, "step": 14595 }, { "epoch": 1.99, "grad_norm": 3.893223669423351, "learning_rate": 1.363303961613438e-09, "loss": 0.1079, "step": 14596 }, { "epoch": 1.99, "grad_norm": 3.27282796263142, "learning_rate": 1.3377037205541954e-09, "loss": 0.1294, "step": 14597 }, { "epoch": 1.99, "grad_norm": 3.439892086693478, "learning_rate": 1.3123460927305653e-09, "loss": 0.1614, "step": 14598 }, { "epoch": 1.99, "grad_norm": 3.0426759505961587, "learning_rate": 1.2872310793737852e-09, "loss": 0.1518, "step": 14599 }, { "epoch": 1.99, "grad_norm": 2.9585970893069726, "learning_rate": 1.2623586817017697e-09, "loss": 0.1339, "step": 14600 }, { "epoch": 1.99, "grad_norm": 4.82333768631016, "learning_rate": 1.2377289009229965e-09, "loss": 0.1118, "step": 14601 }, { "epoch": 1.99, "grad_norm": 3.9382462092772004, "learning_rate": 1.2133417382320656e-09, "loss": 0.1205, "step": 14602 }, { "epoch": 1.99, "grad_norm": 3.3640978740302727, "learning_rate": 1.1891971948124749e-09, "loss": 0.1203, "step": 14603 }, { "epoch": 1.99, "grad_norm": 3.0984892994214515, "learning_rate": 1.1652952718366194e-09, "loss": 0.1342, "step": 14604 }, { "epoch": 1.99, "grad_norm": 2.593668795013529, "learning_rate": 1.1416359704641278e-09, "loss": 0.0939, "step": 14605 }, { "epoch": 1.99, "grad_norm": 3.006273857840782, "learning_rate": 1.1182192918435253e-09, "loss": 0.1097, "step": 14606 }, { "epoch": 1.99, "grad_norm": 3.2403320039433674, "learning_rate": 1.0950452371116805e-09, "loss": 0.1344, "step": 14607 }, { "epoch": 1.99, "grad_norm": 2.7477701982432006, "learning_rate": 1.072113807393249e-09, "loss": 0.1412, "step": 14608 }, { "epoch": 1.99, "grad_norm": 2.6977581750600566, "learning_rate": 1.0494250038006747e-09, "loss": 0.1161, "step": 14609 }, { "epoch": 1.99, "grad_norm": 3.5410388783879494, "learning_rate": 1.026978827435854e-09, "loss": 0.1654, "step": 14610 }, { "epoch": 1.99, "grad_norm": 3.285921093085991, "learning_rate": 1.004775279387915e-09, "loss": 0.1566, "step": 14611 }, { "epoch": 1.99, "grad_norm": 3.1676263701792298, "learning_rate": 9.828143607343298e-10, "loss": 0.1469, "step": 14612 }, { "epoch": 1.99, "grad_norm": 2.921326561978914, "learning_rate": 9.610960725409125e-10, "loss": 0.1391, "step": 14613 }, { "epoch": 1.99, "grad_norm": 3.2947230864153867, "learning_rate": 9.396204158623745e-10, "loss": 0.1318, "step": 14614 }, { "epoch": 1.99, "grad_norm": 4.043452376793645, "learning_rate": 9.183873917406605e-10, "loss": 0.1502, "step": 14615 }, { "epoch": 1.99, "grad_norm": 4.073929237606257, "learning_rate": 8.973970012066124e-10, "loss": 0.1481, "step": 14616 }, { "epoch": 1.99, "grad_norm": 3.7797226300010065, "learning_rate": 8.766492452783048e-10, "loss": 0.1679, "step": 14617 }, { "epoch": 1.99, "grad_norm": 3.8566862576186853, "learning_rate": 8.561441249638202e-10, "loss": 0.1352, "step": 14618 }, { "epoch": 1.99, "grad_norm": 4.236727169943015, "learning_rate": 8.358816412573634e-10, "loss": 0.1159, "step": 14619 }, { "epoch": 1.99, "grad_norm": 3.8975177937525847, "learning_rate": 8.158617951431469e-10, "loss": 0.1728, "step": 14620 }, { "epoch": 1.99, "grad_norm": 3.2873537527959247, "learning_rate": 7.960845875920609e-10, "loss": 0.1361, "step": 14621 }, { "epoch": 1.99, "grad_norm": 3.010059383442071, "learning_rate": 7.765500195650034e-10, "loss": 0.1572, "step": 14622 }, { "epoch": 1.99, "grad_norm": 3.3020102354424767, "learning_rate": 7.572580920089944e-10, "loss": 0.1601, "step": 14623 }, { "epoch": 1.99, "grad_norm": 2.67787932189335, "learning_rate": 7.382088058616177e-10, "loss": 0.1529, "step": 14624 }, { "epoch": 1.99, "grad_norm": 4.215865694730319, "learning_rate": 7.194021620460234e-10, "loss": 0.1835, "step": 14625 }, { "epoch": 1.99, "grad_norm": 4.561281256769123, "learning_rate": 7.008381614764803e-10, "loss": 0.1497, "step": 14626 }, { "epoch": 1.99, "grad_norm": 3.2915565901063766, "learning_rate": 6.825168050528241e-10, "loss": 0.1406, "step": 14627 }, { "epoch": 1.99, "grad_norm": 3.821029336645117, "learning_rate": 6.644380936648987e-10, "loss": 0.1285, "step": 14628 }, { "epoch": 1.99, "grad_norm": 4.987440980544944, "learning_rate": 6.4660202818978e-10, "loss": 0.1807, "step": 14629 }, { "epoch": 1.99, "grad_norm": 3.0167935300269697, "learning_rate": 6.290086094934422e-10, "loss": 0.1656, "step": 14630 }, { "epoch": 1.99, "grad_norm": 3.562483179222686, "learning_rate": 6.116578384296468e-10, "loss": 0.1484, "step": 14631 }, { "epoch": 1.99, "grad_norm": 3.486703678969492, "learning_rate": 5.945497158404979e-10, "loss": 0.1609, "step": 14632 }, { "epoch": 1.99, "grad_norm": 4.070825214242333, "learning_rate": 5.776842425569973e-10, "loss": 0.1783, "step": 14633 }, { "epoch": 1.99, "grad_norm": 3.2544879614952302, "learning_rate": 5.610614193968245e-10, "loss": 0.1459, "step": 14634 }, { "epoch": 1.99, "grad_norm": 4.099331267971455, "learning_rate": 5.446812471671115e-10, "loss": 0.1389, "step": 14635 }, { "epoch": 1.99, "grad_norm": 3.3280408370865056, "learning_rate": 5.285437266633331e-10, "loss": 0.1663, "step": 14636 }, { "epoch": 1.99, "grad_norm": 3.2758798204224706, "learning_rate": 5.126488586676414e-10, "loss": 0.1393, "step": 14637 }, { "epoch": 1.99, "grad_norm": 3.6542052244667222, "learning_rate": 4.969966439527518e-10, "loss": 0.1154, "step": 14638 }, { "epoch": 1.99, "grad_norm": 3.2816664472870896, "learning_rate": 4.815870832775016e-10, "loss": 0.1697, "step": 14639 }, { "epoch": 1.99, "grad_norm": 3.7346546198444113, "learning_rate": 4.664201773896259e-10, "loss": 0.1567, "step": 14640 }, { "epoch": 1.99, "grad_norm": 4.525227678852003, "learning_rate": 4.5149592702631305e-10, "loss": 0.1359, "step": 14641 }, { "epoch": 1.99, "grad_norm": 2.6610068924024435, "learning_rate": 4.368143329114283e-10, "loss": 0.1292, "step": 14642 }, { "epoch": 1.99, "grad_norm": 3.517979005348926, "learning_rate": 4.223753957571797e-10, "loss": 0.1755, "step": 14643 }, { "epoch": 1.99, "grad_norm": 2.659420294657463, "learning_rate": 4.081791162646731e-10, "loss": 0.1236, "step": 14644 }, { "epoch": 1.99, "grad_norm": 3.17431823929678, "learning_rate": 3.942254951228019e-10, "loss": 0.0946, "step": 14645 }, { "epoch": 1.99, "grad_norm": 3.660117996981063, "learning_rate": 3.8051453300880225e-10, "loss": 0.1447, "step": 14646 }, { "epoch": 1.99, "grad_norm": 3.288746001928911, "learning_rate": 3.6704623058825275e-10, "loss": 0.1312, "step": 14647 }, { "epoch": 1.99, "grad_norm": 3.1142084737087696, "learning_rate": 3.5382058851507475e-10, "loss": 0.1336, "step": 14648 }, { "epoch": 1.99, "grad_norm": 3.190738450376309, "learning_rate": 3.408376074309772e-10, "loss": 0.1348, "step": 14649 }, { "epoch": 1.99, "grad_norm": 3.200214435670862, "learning_rate": 3.280972879654565e-10, "loss": 0.1404, "step": 14650 }, { "epoch": 1.99, "grad_norm": 3.509077771240364, "learning_rate": 3.1559963073801714e-10, "loss": 0.1405, "step": 14651 }, { "epoch": 1.99, "grad_norm": 2.9074842311185027, "learning_rate": 3.033446363548409e-10, "loss": 0.1355, "step": 14652 }, { "epoch": 1.99, "grad_norm": 3.32740838636431, "learning_rate": 2.913323054104522e-10, "loss": 0.175, "step": 14653 }, { "epoch": 1.99, "grad_norm": 3.899925673763249, "learning_rate": 2.7956263848771815e-10, "loss": 0.165, "step": 14654 }, { "epoch": 1.99, "grad_norm": 3.652665502652116, "learning_rate": 2.6803563615840356e-10, "loss": 0.1636, "step": 14655 }, { "epoch": 1.99, "grad_norm": 3.4628791524566984, "learning_rate": 2.5675129898206086e-10, "loss": 0.184, "step": 14656 }, { "epoch": 1.99, "grad_norm": 3.4634457809755945, "learning_rate": 2.4570962750547487e-10, "loss": 0.1753, "step": 14657 }, { "epoch": 1.99, "grad_norm": 3.6089386389758493, "learning_rate": 2.3491062226543846e-10, "loss": 0.1307, "step": 14658 }, { "epoch": 1.99, "grad_norm": 3.454716126356579, "learning_rate": 2.243542837859769e-10, "loss": 0.1633, "step": 14659 }, { "epoch": 1.99, "grad_norm": 4.757189925919512, "learning_rate": 2.1404061257945807e-10, "loss": 0.1711, "step": 14660 }, { "epoch": 1.99, "grad_norm": 4.669007749452094, "learning_rate": 2.0396960914603747e-10, "loss": 0.1492, "step": 14661 }, { "epoch": 1.99, "grad_norm": 3.539372123619363, "learning_rate": 1.9414127397476834e-10, "loss": 0.1611, "step": 14662 }, { "epoch": 1.99, "grad_norm": 3.619483736104374, "learning_rate": 1.8455560754249147e-10, "loss": 0.1658, "step": 14663 }, { "epoch": 1.99, "grad_norm": 2.8442786187759217, "learning_rate": 1.752126103149454e-10, "loss": 0.1243, "step": 14664 }, { "epoch": 1.99, "grad_norm": 2.461815487149628, "learning_rate": 1.6611228274510116e-10, "loss": 0.1058, "step": 14665 }, { "epoch": 1.99, "grad_norm": 3.5368288262772203, "learning_rate": 1.5725462527482748e-10, "loss": 0.1707, "step": 14666 }, { "epoch": 2.0, "grad_norm": 2.8773570392743455, "learning_rate": 1.486396383343358e-10, "loss": 0.1331, "step": 14667 }, { "epoch": 2.0, "grad_norm": 3.473007622311862, "learning_rate": 1.402673223416251e-10, "loss": 0.1692, "step": 14668 }, { "epoch": 2.0, "grad_norm": 3.6164878965143368, "learning_rate": 1.3213767770248186e-10, "loss": 0.1625, "step": 14669 }, { "epoch": 2.0, "grad_norm": 4.019085508058464, "learning_rate": 1.2425070481214552e-10, "loss": 0.1355, "step": 14670 }, { "epoch": 2.0, "grad_norm": 3.0219528864281515, "learning_rate": 1.1660640405308787e-10, "loss": 0.1434, "step": 14671 }, { "epoch": 2.0, "grad_norm": 2.720931066455072, "learning_rate": 1.0920477579612342e-10, "loss": 0.1354, "step": 14672 }, { "epoch": 2.0, "grad_norm": 3.625704042202617, "learning_rate": 1.0204582040096444e-10, "loss": 0.1401, "step": 14673 }, { "epoch": 2.0, "grad_norm": 3.589300200288635, "learning_rate": 9.512953821511073e-11, "loss": 0.169, "step": 14674 }, { "epoch": 2.0, "grad_norm": 2.5365506958175605, "learning_rate": 8.845592957384963e-11, "loss": 0.1053, "step": 14675 }, { "epoch": 2.0, "grad_norm": 4.375385843041661, "learning_rate": 8.202499480136628e-11, "loss": 0.1751, "step": 14676 }, { "epoch": 2.0, "grad_norm": 3.1615496579189646, "learning_rate": 7.583673420963333e-11, "loss": 0.1603, "step": 14677 }, { "epoch": 2.0, "grad_norm": 3.1161596642487828, "learning_rate": 6.989114809896613e-11, "loss": 0.1547, "step": 14678 }, { "epoch": 2.0, "grad_norm": 3.9054286201798343, "learning_rate": 6.418823675857777e-11, "loss": 0.1312, "step": 14679 }, { "epoch": 2.0, "grad_norm": 2.5438301221701924, "learning_rate": 5.872800046435867e-11, "loss": 0.1325, "step": 14680 }, { "epoch": 2.0, "grad_norm": 3.2500843144047002, "learning_rate": 5.3510439481652135e-11, "loss": 0.1714, "step": 14681 }, { "epoch": 2.0, "grad_norm": 3.1932471916657525, "learning_rate": 4.8535554063589006e-11, "loss": 0.1478, "step": 14682 }, { "epoch": 2.0, "grad_norm": 3.629982691649399, "learning_rate": 4.380334445219791e-11, "loss": 0.1839, "step": 14683 }, { "epoch": 2.0, "grad_norm": 4.118597536363416, "learning_rate": 3.93138108761848e-11, "loss": 0.1558, "step": 14684 }, { "epoch": 2.0, "grad_norm": 2.2274570857819658, "learning_rate": 3.5066953554263595e-11, "loss": 0.1235, "step": 14685 }, { "epoch": 2.0, "grad_norm": 2.821904867782135, "learning_rate": 3.106277269238067e-11, "loss": 0.1284, "step": 14686 }, { "epoch": 2.0, "grad_norm": 3.681458198038987, "learning_rate": 2.7301268484825062e-11, "loss": 0.1463, "step": 14687 }, { "epoch": 2.0, "grad_norm": 4.719512106307552, "learning_rate": 2.3782441114228448e-11, "loss": 0.1701, "step": 14688 }, { "epoch": 2.0, "grad_norm": 4.510090478471152, "learning_rate": 2.0506290751010073e-11, "loss": 0.1562, "step": 14689 }, { "epoch": 2.0, "grad_norm": 3.842074989354751, "learning_rate": 1.7472817554486933e-11, "loss": 0.1444, "step": 14690 }, { "epoch": 2.0, "grad_norm": 3.633578790828354, "learning_rate": 1.468202167231869e-11, "loss": 0.1664, "step": 14691 }, { "epoch": 2.0, "grad_norm": 2.9181288564449943, "learning_rate": 1.2133903238842337e-11, "loss": 0.1154, "step": 14692 }, { "epoch": 2.0, "grad_norm": 2.9869131642541342, "learning_rate": 9.828462378957959e-12, "loss": 0.1125, "step": 14693 }, { "epoch": 2.0, "grad_norm": 3.5021135053479906, "learning_rate": 7.76569920368786e-12, "loss": 0.182, "step": 14694 }, { "epoch": 2.0, "grad_norm": 3.5851125049746813, "learning_rate": 5.945613813507223e-12, "loss": 0.1498, "step": 14695 }, { "epoch": 2.0, "grad_norm": 3.263335547969631, "learning_rate": 4.368206296678779e-12, "loss": 0.137, "step": 14696 }, { "epoch": 2.0, "grad_norm": 3.0480294455670425, "learning_rate": 3.033476729807916e-12, "loss": 0.1283, "step": 14697 }, { "epoch": 2.0, "grad_norm": 3.162913495329946, "learning_rate": 1.941425177842682e-12, "loss": 0.1467, "step": 14698 }, { "epoch": 2.0, "grad_norm": 3.5723827187384294, "learning_rate": 1.0920516935186699e-12, "loss": 0.1715, "step": 14699 }, { "epoch": 2.0, "grad_norm": 3.7437783896517196, "learning_rate": 4.853563179141318e-13, "loss": 0.1236, "step": 14700 }, { "epoch": 2.0, "grad_norm": 3.2747231158078844, "learning_rate": 1.2133908100508963e-13, "loss": 0.1528, "step": 14701 }, { "epoch": 2.0, "grad_norm": 3.943386548222809, "learning_rate": 0.0, "loss": 0.1873, "step": 14702 }, { "epoch": 2.0, "eval_EK100_accuracy": 0.6539810231023102, "step": 14702 }, { "epoch": 2.0, "step": 14702, "total_flos": 552718546624512.0, "train_loss": 0.23860495874067583, "train_runtime": 42722.2096, "train_samples_per_second": 22.027, "train_steps_per_second": 0.344 } ], "logging_steps": 1.0, "max_steps": 14702, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 5000, "total_flos": 552718546624512.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }