{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 2.0, "eval_steps": 500, "global_step": 7704, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00025960539979231567, "grad_norm": 36.000831604003906, "learning_rate": 0.0, "loss": 0.4004536271095276, "step": 1 }, { "epoch": 0.0005192107995846313, "grad_norm": 48.744239807128906, "learning_rate": 2.1551724137931035e-08, "loss": 0.4256002902984619, "step": 2 }, { "epoch": 0.000778816199376947, "grad_norm": 49.4882926940918, "learning_rate": 4.310344827586207e-08, "loss": 0.428159236907959, "step": 3 }, { "epoch": 0.0010384215991692627, "grad_norm": 33.126190185546875, "learning_rate": 6.465517241379311e-08, "loss": 0.3928731083869934, "step": 4 }, { "epoch": 0.0012980269989615785, "grad_norm": 44.44942092895508, "learning_rate": 8.620689655172414e-08, "loss": 0.40993767976760864, "step": 5 }, { "epoch": 0.001557632398753894, "grad_norm": 35.903099060058594, "learning_rate": 1.0775862068965518e-07, "loss": 0.3903833031654358, "step": 6 }, { "epoch": 0.0018172377985462098, "grad_norm": 36.59717559814453, "learning_rate": 1.2931034482758622e-07, "loss": 0.3978159427642822, "step": 7 }, { "epoch": 0.0020768431983385254, "grad_norm": 22.785905838012695, "learning_rate": 1.5086206896551727e-07, "loss": 0.36123597621917725, "step": 8 }, { "epoch": 0.002336448598130841, "grad_norm": 22.143814086914062, "learning_rate": 1.7241379310344828e-07, "loss": 0.3424040675163269, "step": 9 }, { "epoch": 0.002596053997923157, "grad_norm": 17.124189376831055, "learning_rate": 1.9396551724137932e-07, "loss": 0.3505767285823822, "step": 10 }, { "epoch": 0.0028556593977154725, "grad_norm": 17.779603958129883, "learning_rate": 2.1551724137931036e-07, "loss": 0.34036123752593994, "step": 11 }, { "epoch": 0.003115264797507788, "grad_norm": 26.164113998413086, "learning_rate": 2.3706896551724138e-07, "loss": 0.35741353034973145, "step": 12 }, { "epoch": 0.0033748701973001036, "grad_norm": 17.916975021362305, "learning_rate": 2.5862068965517245e-07, "loss": 0.3332182765007019, "step": 13 }, { "epoch": 0.0036344755970924196, "grad_norm": 19.147138595581055, "learning_rate": 2.801724137931035e-07, "loss": 0.3152492642402649, "step": 14 }, { "epoch": 0.003894080996884735, "grad_norm": 26.42192840576172, "learning_rate": 3.0172413793103453e-07, "loss": 0.3421565890312195, "step": 15 }, { "epoch": 0.004153686396677051, "grad_norm": 19.459726333618164, "learning_rate": 3.2327586206896557e-07, "loss": 0.30212169885635376, "step": 16 }, { "epoch": 0.004413291796469367, "grad_norm": 25.59136962890625, "learning_rate": 3.4482758620689656e-07, "loss": 0.34346455335617065, "step": 17 }, { "epoch": 0.004672897196261682, "grad_norm": 21.618221282958984, "learning_rate": 3.663793103448276e-07, "loss": 0.31766217947006226, "step": 18 }, { "epoch": 0.004932502596053998, "grad_norm": 21.63040542602539, "learning_rate": 3.8793103448275865e-07, "loss": 0.3175572156906128, "step": 19 }, { "epoch": 0.005192107995846314, "grad_norm": 11.867350578308105, "learning_rate": 4.094827586206897e-07, "loss": 0.27640751004219055, "step": 20 }, { "epoch": 0.005451713395638629, "grad_norm": 12.711272239685059, "learning_rate": 4.3103448275862073e-07, "loss": 0.2671198546886444, "step": 21 }, { "epoch": 0.005711318795430945, "grad_norm": 13.229530334472656, "learning_rate": 4.5258620689655177e-07, "loss": 0.2750229239463806, "step": 22 }, { "epoch": 0.005970924195223261, "grad_norm": 9.093338012695312, "learning_rate": 4.7413793103448276e-07, "loss": 0.2673614025115967, "step": 23 }, { "epoch": 0.006230529595015576, "grad_norm": 15.587196350097656, "learning_rate": 4.956896551724138e-07, "loss": 0.2640160918235779, "step": 24 }, { "epoch": 0.006490134994807892, "grad_norm": 13.262322425842285, "learning_rate": 5.172413793103449e-07, "loss": 0.28813672065734863, "step": 25 }, { "epoch": 0.006749740394600207, "grad_norm": 7.8371100425720215, "learning_rate": 5.387931034482759e-07, "loss": 0.25962844491004944, "step": 26 }, { "epoch": 0.007009345794392523, "grad_norm": 13.588038444519043, "learning_rate": 5.60344827586207e-07, "loss": 0.24782322347164154, "step": 27 }, { "epoch": 0.007268951194184839, "grad_norm": 8.397758483886719, "learning_rate": 5.81896551724138e-07, "loss": 0.24025985598564148, "step": 28 }, { "epoch": 0.007528556593977154, "grad_norm": 11.745874404907227, "learning_rate": 6.034482758620691e-07, "loss": 0.2768011689186096, "step": 29 }, { "epoch": 0.00778816199376947, "grad_norm": 9.621773719787598, "learning_rate": 6.25e-07, "loss": 0.2754887640476227, "step": 30 }, { "epoch": 0.008047767393561786, "grad_norm": 7.397965431213379, "learning_rate": 6.465517241379311e-07, "loss": 0.26962393522262573, "step": 31 }, { "epoch": 0.008307372793354102, "grad_norm": 8.559306144714355, "learning_rate": 6.68103448275862e-07, "loss": 0.24626237154006958, "step": 32 }, { "epoch": 0.008566978193146417, "grad_norm": 6.801912784576416, "learning_rate": 6.896551724137931e-07, "loss": 0.2180132269859314, "step": 33 }, { "epoch": 0.008826583592938734, "grad_norm": 7.743856906890869, "learning_rate": 7.112068965517242e-07, "loss": 0.21620649099349976, "step": 34 }, { "epoch": 0.009086188992731049, "grad_norm": 6.976649284362793, "learning_rate": 7.327586206896552e-07, "loss": 0.25069954991340637, "step": 35 }, { "epoch": 0.009345794392523364, "grad_norm": 10.544355392456055, "learning_rate": 7.543103448275863e-07, "loss": 0.28238558769226074, "step": 36 }, { "epoch": 0.00960539979231568, "grad_norm": 6.595256805419922, "learning_rate": 7.758620689655173e-07, "loss": 0.2653074562549591, "step": 37 }, { "epoch": 0.009865005192107996, "grad_norm": 5.759922027587891, "learning_rate": 7.974137931034484e-07, "loss": 0.2490157186985016, "step": 38 }, { "epoch": 0.010124610591900311, "grad_norm": 6.058347225189209, "learning_rate": 8.189655172413794e-07, "loss": 0.2738566994667053, "step": 39 }, { "epoch": 0.010384215991692628, "grad_norm": 7.90873908996582, "learning_rate": 8.405172413793105e-07, "loss": 0.20831885933876038, "step": 40 }, { "epoch": 0.010643821391484943, "grad_norm": 7.461945533752441, "learning_rate": 8.620689655172415e-07, "loss": 0.2334851622581482, "step": 41 }, { "epoch": 0.010903426791277258, "grad_norm": 5.529984474182129, "learning_rate": 8.836206896551724e-07, "loss": 0.23753845691680908, "step": 42 }, { "epoch": 0.011163032191069575, "grad_norm": 12.030415534973145, "learning_rate": 9.051724137931035e-07, "loss": 0.2739925980567932, "step": 43 }, { "epoch": 0.01142263759086189, "grad_norm": 6.446074962615967, "learning_rate": 9.267241379310346e-07, "loss": 0.21448713541030884, "step": 44 }, { "epoch": 0.011682242990654205, "grad_norm": 6.225979804992676, "learning_rate": 9.482758620689655e-07, "loss": 0.20683956146240234, "step": 45 }, { "epoch": 0.011941848390446522, "grad_norm": 5.783356189727783, "learning_rate": 9.698275862068966e-07, "loss": 0.22018496692180634, "step": 46 }, { "epoch": 0.012201453790238837, "grad_norm": 8.743072509765625, "learning_rate": 9.913793103448276e-07, "loss": 0.24101513624191284, "step": 47 }, { "epoch": 0.012461059190031152, "grad_norm": 5.1674065589904785, "learning_rate": 1.0129310344827586e-06, "loss": 0.223082035779953, "step": 48 }, { "epoch": 0.01272066458982347, "grad_norm": 7.7235493659973145, "learning_rate": 1.0344827586206898e-06, "loss": 0.24540969729423523, "step": 49 }, { "epoch": 0.012980269989615784, "grad_norm": 5.428606033325195, "learning_rate": 1.0560344827586208e-06, "loss": 0.18970027565956116, "step": 50 }, { "epoch": 0.0132398753894081, "grad_norm": 6.427073001861572, "learning_rate": 1.0775862068965518e-06, "loss": 0.19460377097129822, "step": 51 }, { "epoch": 0.013499480789200415, "grad_norm": 3.960780382156372, "learning_rate": 1.0991379310344828e-06, "loss": 0.18084293603897095, "step": 52 }, { "epoch": 0.013759086188992731, "grad_norm": 4.838387966156006, "learning_rate": 1.120689655172414e-06, "loss": 0.1886947900056839, "step": 53 }, { "epoch": 0.014018691588785047, "grad_norm": 8.527689933776855, "learning_rate": 1.142241379310345e-06, "loss": 0.24218779802322388, "step": 54 }, { "epoch": 0.014278296988577362, "grad_norm": 7.650318622589111, "learning_rate": 1.163793103448276e-06, "loss": 0.23584404587745667, "step": 55 }, { "epoch": 0.014537902388369679, "grad_norm": 5.936899185180664, "learning_rate": 1.185344827586207e-06, "loss": 0.21349135041236877, "step": 56 }, { "epoch": 0.014797507788161994, "grad_norm": 6.6475419998168945, "learning_rate": 1.2068965517241381e-06, "loss": 0.24574065208435059, "step": 57 }, { "epoch": 0.015057113187954309, "grad_norm": 10.809253692626953, "learning_rate": 1.2284482758620691e-06, "loss": 0.2632869482040405, "step": 58 }, { "epoch": 0.015316718587746626, "grad_norm": 6.916379928588867, "learning_rate": 1.25e-06, "loss": 0.18405872583389282, "step": 59 }, { "epoch": 0.01557632398753894, "grad_norm": 7.406436443328857, "learning_rate": 1.271551724137931e-06, "loss": 0.21621964871883392, "step": 60 }, { "epoch": 0.015835929387331256, "grad_norm": 8.445347785949707, "learning_rate": 1.2931034482758623e-06, "loss": 0.2276899516582489, "step": 61 }, { "epoch": 0.016095534787123573, "grad_norm": 3.903897762298584, "learning_rate": 1.3146551724137933e-06, "loss": 0.1785777062177658, "step": 62 }, { "epoch": 0.016355140186915886, "grad_norm": 4.464757442474365, "learning_rate": 1.336206896551724e-06, "loss": 0.20330852270126343, "step": 63 }, { "epoch": 0.016614745586708203, "grad_norm": 8.190912246704102, "learning_rate": 1.3577586206896553e-06, "loss": 0.2484232485294342, "step": 64 }, { "epoch": 0.01687435098650052, "grad_norm": 6.79339599609375, "learning_rate": 1.3793103448275862e-06, "loss": 0.22851403057575226, "step": 65 }, { "epoch": 0.017133956386292833, "grad_norm": 4.627327919006348, "learning_rate": 1.4008620689655172e-06, "loss": 0.17506536841392517, "step": 66 }, { "epoch": 0.01739356178608515, "grad_norm": 3.154578447341919, "learning_rate": 1.4224137931034484e-06, "loss": 0.19490265846252441, "step": 67 }, { "epoch": 0.017653167185877467, "grad_norm": 5.269001007080078, "learning_rate": 1.4439655172413794e-06, "loss": 0.25001391768455505, "step": 68 }, { "epoch": 0.01791277258566978, "grad_norm": 6.785883903503418, "learning_rate": 1.4655172413793104e-06, "loss": 0.18239900469779968, "step": 69 }, { "epoch": 0.018172377985462097, "grad_norm": 4.631606101989746, "learning_rate": 1.4870689655172416e-06, "loss": 0.18996429443359375, "step": 70 }, { "epoch": 0.018431983385254414, "grad_norm": 7.309718608856201, "learning_rate": 1.5086206896551726e-06, "loss": 0.17979270219802856, "step": 71 }, { "epoch": 0.018691588785046728, "grad_norm": 3.7146410942077637, "learning_rate": 1.5301724137931034e-06, "loss": 0.18085616827011108, "step": 72 }, { "epoch": 0.018951194184839044, "grad_norm": 6.651581764221191, "learning_rate": 1.5517241379310346e-06, "loss": 0.2062850445508957, "step": 73 }, { "epoch": 0.01921079958463136, "grad_norm": 4.280433177947998, "learning_rate": 1.5732758620689656e-06, "loss": 0.21607807278633118, "step": 74 }, { "epoch": 0.019470404984423675, "grad_norm": 2.8871498107910156, "learning_rate": 1.5948275862068968e-06, "loss": 0.16675621271133423, "step": 75 }, { "epoch": 0.01973001038421599, "grad_norm": 3.4952914714813232, "learning_rate": 1.6163793103448278e-06, "loss": 0.17863905429840088, "step": 76 }, { "epoch": 0.01998961578400831, "grad_norm": 7.858782768249512, "learning_rate": 1.6379310344827587e-06, "loss": 0.23102828860282898, "step": 77 }, { "epoch": 0.020249221183800622, "grad_norm": 3.9728784561157227, "learning_rate": 1.65948275862069e-06, "loss": 0.1772451102733612, "step": 78 }, { "epoch": 0.02050882658359294, "grad_norm": 3.9052302837371826, "learning_rate": 1.681034482758621e-06, "loss": 0.21675768494606018, "step": 79 }, { "epoch": 0.020768431983385256, "grad_norm": 4.7596611976623535, "learning_rate": 1.7025862068965517e-06, "loss": 0.2201749086380005, "step": 80 }, { "epoch": 0.02102803738317757, "grad_norm": 6.447850227355957, "learning_rate": 1.724137931034483e-06, "loss": 0.1725550889968872, "step": 81 }, { "epoch": 0.021287642782969886, "grad_norm": 4.408200263977051, "learning_rate": 1.745689655172414e-06, "loss": 0.17029929161071777, "step": 82 }, { "epoch": 0.021547248182762203, "grad_norm": 17.693592071533203, "learning_rate": 1.7672413793103449e-06, "loss": 0.255332350730896, "step": 83 }, { "epoch": 0.021806853582554516, "grad_norm": 5.197042465209961, "learning_rate": 1.788793103448276e-06, "loss": 0.2044193148612976, "step": 84 }, { "epoch": 0.022066458982346833, "grad_norm": 15.99860668182373, "learning_rate": 1.810344827586207e-06, "loss": 0.22098957002162933, "step": 85 }, { "epoch": 0.02232606438213915, "grad_norm": 6.470719337463379, "learning_rate": 1.831896551724138e-06, "loss": 0.23281870782375336, "step": 86 }, { "epoch": 0.022585669781931463, "grad_norm": 8.8930025100708, "learning_rate": 1.8534482758620693e-06, "loss": 0.19912584125995636, "step": 87 }, { "epoch": 0.02284527518172378, "grad_norm": 6.249830722808838, "learning_rate": 1.8750000000000003e-06, "loss": 0.22309300303459167, "step": 88 }, { "epoch": 0.023104880581516097, "grad_norm": 4.471930980682373, "learning_rate": 1.896551724137931e-06, "loss": 0.17745891213417053, "step": 89 }, { "epoch": 0.02336448598130841, "grad_norm": 8.00529956817627, "learning_rate": 1.9181034482758622e-06, "loss": 0.19427117705345154, "step": 90 }, { "epoch": 0.023624091381100727, "grad_norm": 8.376458168029785, "learning_rate": 1.9396551724137932e-06, "loss": 0.21259278059005737, "step": 91 }, { "epoch": 0.023883696780893044, "grad_norm": 5.320324897766113, "learning_rate": 1.9612068965517242e-06, "loss": 0.20793083310127258, "step": 92 }, { "epoch": 0.024143302180685357, "grad_norm": 5.512922763824463, "learning_rate": 1.982758620689655e-06, "loss": 0.16936475038528442, "step": 93 }, { "epoch": 0.024402907580477674, "grad_norm": 10.094862937927246, "learning_rate": 2.004310344827586e-06, "loss": 0.19157373905181885, "step": 94 }, { "epoch": 0.02466251298026999, "grad_norm": 3.151434898376465, "learning_rate": 2.025862068965517e-06, "loss": 0.17058934271335602, "step": 95 }, { "epoch": 0.024922118380062305, "grad_norm": 6.678121566772461, "learning_rate": 2.0474137931034486e-06, "loss": 0.17326629161834717, "step": 96 }, { "epoch": 0.02518172377985462, "grad_norm": 5.289290428161621, "learning_rate": 2.0689655172413796e-06, "loss": 0.18311059474945068, "step": 97 }, { "epoch": 0.02544132917964694, "grad_norm": 6.219593524932861, "learning_rate": 2.0905172413793106e-06, "loss": 0.20244450867176056, "step": 98 }, { "epoch": 0.02570093457943925, "grad_norm": 7.852299213409424, "learning_rate": 2.1120689655172416e-06, "loss": 0.17139403522014618, "step": 99 }, { "epoch": 0.02596053997923157, "grad_norm": 2.846259355545044, "learning_rate": 2.1336206896551725e-06, "loss": 0.16972136497497559, "step": 100 }, { "epoch": 0.026220145379023882, "grad_norm": 6.342812538146973, "learning_rate": 2.1551724137931035e-06, "loss": 0.16823768615722656, "step": 101 }, { "epoch": 0.0264797507788162, "grad_norm": 9.17609691619873, "learning_rate": 2.1767241379310345e-06, "loss": 0.2014111578464508, "step": 102 }, { "epoch": 0.026739356178608516, "grad_norm": 3.1354265213012695, "learning_rate": 2.1982758620689655e-06, "loss": 0.1895349621772766, "step": 103 }, { "epoch": 0.02699896157840083, "grad_norm": 9.057998657226562, "learning_rate": 2.219827586206897e-06, "loss": 0.17370013892650604, "step": 104 }, { "epoch": 0.027258566978193146, "grad_norm": 9.617159843444824, "learning_rate": 2.241379310344828e-06, "loss": 0.21284937858581543, "step": 105 }, { "epoch": 0.027518172377985463, "grad_norm": 8.939170837402344, "learning_rate": 2.262931034482759e-06, "loss": 0.19268397986888885, "step": 106 }, { "epoch": 0.027777777777777776, "grad_norm": 5.360718727111816, "learning_rate": 2.28448275862069e-06, "loss": 0.16941624879837036, "step": 107 }, { "epoch": 0.028037383177570093, "grad_norm": 6.901419639587402, "learning_rate": 2.306034482758621e-06, "loss": 0.1592540442943573, "step": 108 }, { "epoch": 0.02829698857736241, "grad_norm": 3.783787250518799, "learning_rate": 2.327586206896552e-06, "loss": 0.16396544873714447, "step": 109 }, { "epoch": 0.028556593977154723, "grad_norm": 5.210651874542236, "learning_rate": 2.349137931034483e-06, "loss": 0.22442901134490967, "step": 110 }, { "epoch": 0.02881619937694704, "grad_norm": 14.704705238342285, "learning_rate": 2.370689655172414e-06, "loss": 0.24847298860549927, "step": 111 }, { "epoch": 0.029075804776739357, "grad_norm": 5.741897106170654, "learning_rate": 2.392241379310345e-06, "loss": 0.20774176716804504, "step": 112 }, { "epoch": 0.02933541017653167, "grad_norm": 10.968568801879883, "learning_rate": 2.4137931034482762e-06, "loss": 0.22665145993232727, "step": 113 }, { "epoch": 0.029595015576323987, "grad_norm": 4.635924339294434, "learning_rate": 2.4353448275862072e-06, "loss": 0.1802576184272766, "step": 114 }, { "epoch": 0.029854620976116304, "grad_norm": 3.654743194580078, "learning_rate": 2.4568965517241382e-06, "loss": 0.19145122170448303, "step": 115 }, { "epoch": 0.030114226375908618, "grad_norm": 4.433925628662109, "learning_rate": 2.4784482758620692e-06, "loss": 0.20531761646270752, "step": 116 }, { "epoch": 0.030373831775700934, "grad_norm": 3.749790668487549, "learning_rate": 2.5e-06, "loss": 0.1658068299293518, "step": 117 }, { "epoch": 0.03063343717549325, "grad_norm": 4.692869186401367, "learning_rate": 2.521551724137931e-06, "loss": 0.2136266827583313, "step": 118 }, { "epoch": 0.030893042575285565, "grad_norm": 3.97434401512146, "learning_rate": 2.543103448275862e-06, "loss": 0.20519131422042847, "step": 119 }, { "epoch": 0.03115264797507788, "grad_norm": 4.488476753234863, "learning_rate": 2.5646551724137936e-06, "loss": 0.16122350096702576, "step": 120 }, { "epoch": 0.0314122533748702, "grad_norm": 7.700930118560791, "learning_rate": 2.5862068965517246e-06, "loss": 0.2318609654903412, "step": 121 }, { "epoch": 0.03167185877466251, "grad_norm": 6.7368268966674805, "learning_rate": 2.6077586206896556e-06, "loss": 0.1798655390739441, "step": 122 }, { "epoch": 0.031931464174454825, "grad_norm": 7.472613334655762, "learning_rate": 2.6293103448275866e-06, "loss": 0.16895198822021484, "step": 123 }, { "epoch": 0.032191069574247146, "grad_norm": 2.9014663696289062, "learning_rate": 2.6508620689655175e-06, "loss": 0.1727498471736908, "step": 124 }, { "epoch": 0.03245067497403946, "grad_norm": 5.434816837310791, "learning_rate": 2.672413793103448e-06, "loss": 0.22744199633598328, "step": 125 }, { "epoch": 0.03271028037383177, "grad_norm": 11.148801803588867, "learning_rate": 2.69396551724138e-06, "loss": 0.17238348722457886, "step": 126 }, { "epoch": 0.03296988577362409, "grad_norm": 3.2822225093841553, "learning_rate": 2.7155172413793105e-06, "loss": 0.19733279943466187, "step": 127 }, { "epoch": 0.033229491173416406, "grad_norm": 7.458563327789307, "learning_rate": 2.7370689655172415e-06, "loss": 0.18073299527168274, "step": 128 }, { "epoch": 0.03348909657320872, "grad_norm": 7.500495910644531, "learning_rate": 2.7586206896551725e-06, "loss": 0.19854670763015747, "step": 129 }, { "epoch": 0.03374870197300104, "grad_norm": 2.255357265472412, "learning_rate": 2.7801724137931035e-06, "loss": 0.17393359541893005, "step": 130 }, { "epoch": 0.03400830737279335, "grad_norm": 5.850879669189453, "learning_rate": 2.8017241379310345e-06, "loss": 0.1624828279018402, "step": 131 }, { "epoch": 0.03426791277258567, "grad_norm": 5.5082783699035645, "learning_rate": 2.823275862068966e-06, "loss": 0.19870604574680328, "step": 132 }, { "epoch": 0.03452751817237799, "grad_norm": 3.8505401611328125, "learning_rate": 2.844827586206897e-06, "loss": 0.18426847457885742, "step": 133 }, { "epoch": 0.0347871235721703, "grad_norm": 5.397397041320801, "learning_rate": 2.866379310344828e-06, "loss": 0.16499431431293488, "step": 134 }, { "epoch": 0.035046728971962614, "grad_norm": 5.120184898376465, "learning_rate": 2.887931034482759e-06, "loss": 0.17745718359947205, "step": 135 }, { "epoch": 0.035306334371754934, "grad_norm": 3.9481606483459473, "learning_rate": 2.90948275862069e-06, "loss": 0.20657910406589508, "step": 136 }, { "epoch": 0.03556593977154725, "grad_norm": 7.855442523956299, "learning_rate": 2.931034482758621e-06, "loss": 0.21156448125839233, "step": 137 }, { "epoch": 0.03582554517133956, "grad_norm": 2.3497579097747803, "learning_rate": 2.9525862068965522e-06, "loss": 0.16913172602653503, "step": 138 }, { "epoch": 0.03608515057113188, "grad_norm": 2.1113414764404297, "learning_rate": 2.9741379310344832e-06, "loss": 0.1703111082315445, "step": 139 }, { "epoch": 0.036344755970924195, "grad_norm": 8.959603309631348, "learning_rate": 2.9956896551724142e-06, "loss": 0.20626625418663025, "step": 140 }, { "epoch": 0.03660436137071651, "grad_norm": 7.4593825340271, "learning_rate": 3.017241379310345e-06, "loss": 0.23583710193634033, "step": 141 }, { "epoch": 0.03686396677050883, "grad_norm": 8.23781681060791, "learning_rate": 3.0387931034482758e-06, "loss": 0.18471497297286987, "step": 142 }, { "epoch": 0.03712357217030114, "grad_norm": 9.567289352416992, "learning_rate": 3.0603448275862068e-06, "loss": 0.17259125411510468, "step": 143 }, { "epoch": 0.037383177570093455, "grad_norm": 4.7825212478637695, "learning_rate": 3.081896551724138e-06, "loss": 0.18629753589630127, "step": 144 }, { "epoch": 0.037642782969885775, "grad_norm": 9.622005462646484, "learning_rate": 3.103448275862069e-06, "loss": 0.1763424575328827, "step": 145 }, { "epoch": 0.03790238836967809, "grad_norm": 5.008886814117432, "learning_rate": 3.125e-06, "loss": 0.16419553756713867, "step": 146 }, { "epoch": 0.0381619937694704, "grad_norm": 6.055441379547119, "learning_rate": 3.146551724137931e-06, "loss": 0.21428991854190826, "step": 147 }, { "epoch": 0.03842159916926272, "grad_norm": 4.462464332580566, "learning_rate": 3.168103448275862e-06, "loss": 0.21428415179252625, "step": 148 }, { "epoch": 0.038681204569055036, "grad_norm": 3.2144875526428223, "learning_rate": 3.1896551724137935e-06, "loss": 0.15340356528759003, "step": 149 }, { "epoch": 0.03894080996884735, "grad_norm": 3.9997572898864746, "learning_rate": 3.2112068965517245e-06, "loss": 0.17094668745994568, "step": 150 }, { "epoch": 0.03920041536863967, "grad_norm": 6.647733211517334, "learning_rate": 3.2327586206896555e-06, "loss": 0.1739555299282074, "step": 151 }, { "epoch": 0.03946002076843198, "grad_norm": 6.610648155212402, "learning_rate": 3.2543103448275865e-06, "loss": 0.21073564887046814, "step": 152 }, { "epoch": 0.0397196261682243, "grad_norm": 10.679266929626465, "learning_rate": 3.2758620689655175e-06, "loss": 0.1749483346939087, "step": 153 }, { "epoch": 0.03997923156801662, "grad_norm": 7.580510139465332, "learning_rate": 3.2974137931034485e-06, "loss": 0.16683390736579895, "step": 154 }, { "epoch": 0.04023883696780893, "grad_norm": 3.499598264694214, "learning_rate": 3.31896551724138e-06, "loss": 0.1660749465227127, "step": 155 }, { "epoch": 0.040498442367601244, "grad_norm": 7.117327690124512, "learning_rate": 3.340517241379311e-06, "loss": 0.17351442575454712, "step": 156 }, { "epoch": 0.040758047767393564, "grad_norm": 5.276373386383057, "learning_rate": 3.362068965517242e-06, "loss": 0.18035456538200378, "step": 157 }, { "epoch": 0.04101765316718588, "grad_norm": 5.422152519226074, "learning_rate": 3.383620689655173e-06, "loss": 0.1912727653980255, "step": 158 }, { "epoch": 0.04127725856697819, "grad_norm": 2.6480019092559814, "learning_rate": 3.4051724137931034e-06, "loss": 0.15835857391357422, "step": 159 }, { "epoch": 0.04153686396677051, "grad_norm": 3.398489236831665, "learning_rate": 3.4267241379310344e-06, "loss": 0.16798140108585358, "step": 160 }, { "epoch": 0.041796469366562825, "grad_norm": 1.8599700927734375, "learning_rate": 3.448275862068966e-06, "loss": 0.14680664241313934, "step": 161 }, { "epoch": 0.04205607476635514, "grad_norm": 5.639477729797363, "learning_rate": 3.469827586206897e-06, "loss": 0.1919446885585785, "step": 162 }, { "epoch": 0.04231568016614746, "grad_norm": 3.952070951461792, "learning_rate": 3.491379310344828e-06, "loss": 0.1626761108636856, "step": 163 }, { "epoch": 0.04257528556593977, "grad_norm": 3.0708119869232178, "learning_rate": 3.512931034482759e-06, "loss": 0.16932132840156555, "step": 164 }, { "epoch": 0.042834890965732085, "grad_norm": 6.2499165534973145, "learning_rate": 3.5344827586206898e-06, "loss": 0.21028771996498108, "step": 165 }, { "epoch": 0.043094496365524405, "grad_norm": 5.09121561050415, "learning_rate": 3.5560344827586208e-06, "loss": 0.18764573335647583, "step": 166 }, { "epoch": 0.04335410176531672, "grad_norm": 6.381021499633789, "learning_rate": 3.577586206896552e-06, "loss": 0.19519859552383423, "step": 167 }, { "epoch": 0.04361370716510903, "grad_norm": 5.673404693603516, "learning_rate": 3.599137931034483e-06, "loss": 0.20483997464179993, "step": 168 }, { "epoch": 0.04387331256490135, "grad_norm": 3.9901838302612305, "learning_rate": 3.620689655172414e-06, "loss": 0.21053537726402283, "step": 169 }, { "epoch": 0.044132917964693666, "grad_norm": 4.73471212387085, "learning_rate": 3.642241379310345e-06, "loss": 0.22167184948921204, "step": 170 }, { "epoch": 0.04439252336448598, "grad_norm": 1.7850035429000854, "learning_rate": 3.663793103448276e-06, "loss": 0.1671627014875412, "step": 171 }, { "epoch": 0.0446521287642783, "grad_norm": 2.994032621383667, "learning_rate": 3.685344827586207e-06, "loss": 0.23269197344779968, "step": 172 }, { "epoch": 0.04491173416407061, "grad_norm": 5.152878761291504, "learning_rate": 3.7068965517241385e-06, "loss": 0.1774715632200241, "step": 173 }, { "epoch": 0.045171339563862926, "grad_norm": 6.434433460235596, "learning_rate": 3.7284482758620695e-06, "loss": 0.16643106937408447, "step": 174 }, { "epoch": 0.04543094496365525, "grad_norm": 6.448436737060547, "learning_rate": 3.7500000000000005e-06, "loss": 0.19647520780563354, "step": 175 }, { "epoch": 0.04569055036344756, "grad_norm": 8.85893440246582, "learning_rate": 3.771551724137931e-06, "loss": 0.16392821073532104, "step": 176 }, { "epoch": 0.045950155763239874, "grad_norm": 7.348093032836914, "learning_rate": 3.793103448275862e-06, "loss": 0.22272033989429474, "step": 177 }, { "epoch": 0.046209761163032194, "grad_norm": 7.285104751586914, "learning_rate": 3.8146551724137935e-06, "loss": 0.17370232939720154, "step": 178 }, { "epoch": 0.04646936656282451, "grad_norm": 7.860143661499023, "learning_rate": 3.8362068965517245e-06, "loss": 0.18630366027355194, "step": 179 }, { "epoch": 0.04672897196261682, "grad_norm": 4.910839080810547, "learning_rate": 3.857758620689656e-06, "loss": 0.19706512987613678, "step": 180 }, { "epoch": 0.04698857736240914, "grad_norm": 6.993145942687988, "learning_rate": 3.8793103448275865e-06, "loss": 0.24388304352760315, "step": 181 }, { "epoch": 0.047248182762201454, "grad_norm": 3.4418346881866455, "learning_rate": 3.900862068965518e-06, "loss": 0.16107794642448425, "step": 182 }, { "epoch": 0.04750778816199377, "grad_norm": 10.054652214050293, "learning_rate": 3.9224137931034484e-06, "loss": 0.19845399260520935, "step": 183 }, { "epoch": 0.04776739356178609, "grad_norm": 1.8626421689987183, "learning_rate": 3.94396551724138e-06, "loss": 0.1581079661846161, "step": 184 }, { "epoch": 0.0480269989615784, "grad_norm": 3.5025370121002197, "learning_rate": 3.96551724137931e-06, "loss": 0.1907721906900406, "step": 185 }, { "epoch": 0.048286604361370715, "grad_norm": 2.2709851264953613, "learning_rate": 3.987068965517242e-06, "loss": 0.1559697389602661, "step": 186 }, { "epoch": 0.048546209761163035, "grad_norm": 1.6010488271713257, "learning_rate": 4.008620689655172e-06, "loss": 0.1475631594657898, "step": 187 }, { "epoch": 0.04880581516095535, "grad_norm": 2.0182247161865234, "learning_rate": 4.030172413793104e-06, "loss": 0.15417340397834778, "step": 188 }, { "epoch": 0.04906542056074766, "grad_norm": 3.4394755363464355, "learning_rate": 4.051724137931034e-06, "loss": 0.1557568907737732, "step": 189 }, { "epoch": 0.04932502596053998, "grad_norm": 2.7772510051727295, "learning_rate": 4.073275862068966e-06, "loss": 0.16429413855075836, "step": 190 }, { "epoch": 0.049584631360332296, "grad_norm": 3.8975777626037598, "learning_rate": 4.094827586206897e-06, "loss": 0.14402753114700317, "step": 191 }, { "epoch": 0.04984423676012461, "grad_norm": 3.2871761322021484, "learning_rate": 4.116379310344828e-06, "loss": 0.18763604760169983, "step": 192 }, { "epoch": 0.05010384215991693, "grad_norm": 3.7244489192962646, "learning_rate": 4.137931034482759e-06, "loss": 0.17728674411773682, "step": 193 }, { "epoch": 0.05036344755970924, "grad_norm": 10.317301750183105, "learning_rate": 4.15948275862069e-06, "loss": 0.19559919834136963, "step": 194 }, { "epoch": 0.050623052959501556, "grad_norm": 1.9446686506271362, "learning_rate": 4.181034482758621e-06, "loss": 0.15680357813835144, "step": 195 }, { "epoch": 0.05088265835929388, "grad_norm": 10.471190452575684, "learning_rate": 4.2025862068965526e-06, "loss": 0.20365022122859955, "step": 196 }, { "epoch": 0.05114226375908619, "grad_norm": 2.1556951999664307, "learning_rate": 4.224137931034483e-06, "loss": 0.1726238876581192, "step": 197 }, { "epoch": 0.0514018691588785, "grad_norm": 4.372365951538086, "learning_rate": 4.2456896551724145e-06, "loss": 0.16192562878131866, "step": 198 }, { "epoch": 0.051661474558670824, "grad_norm": 6.39937686920166, "learning_rate": 4.267241379310345e-06, "loss": 0.20015473663806915, "step": 199 }, { "epoch": 0.05192107995846314, "grad_norm": 2.3559136390686035, "learning_rate": 4.288793103448276e-06, "loss": 0.20785890519618988, "step": 200 }, { "epoch": 0.05218068535825545, "grad_norm": 5.447387218475342, "learning_rate": 4.310344827586207e-06, "loss": 0.19601623713970184, "step": 201 }, { "epoch": 0.052440290758047764, "grad_norm": 3.238369941711426, "learning_rate": 4.3318965517241385e-06, "loss": 0.18420542776584625, "step": 202 }, { "epoch": 0.052699896157840084, "grad_norm": 6.415649890899658, "learning_rate": 4.353448275862069e-06, "loss": 0.18723170459270477, "step": 203 }, { "epoch": 0.0529595015576324, "grad_norm": 2.089495897293091, "learning_rate": 4.3750000000000005e-06, "loss": 0.1711127758026123, "step": 204 }, { "epoch": 0.05321910695742471, "grad_norm": 2.7710301876068115, "learning_rate": 4.396551724137931e-06, "loss": 0.20039042830467224, "step": 205 }, { "epoch": 0.05347871235721703, "grad_norm": 2.847064256668091, "learning_rate": 4.4181034482758624e-06, "loss": 0.1497441679239273, "step": 206 }, { "epoch": 0.053738317757009345, "grad_norm": 3.455547332763672, "learning_rate": 4.439655172413794e-06, "loss": 0.2059808373451233, "step": 207 }, { "epoch": 0.05399792315680166, "grad_norm": 3.745551109313965, "learning_rate": 4.461206896551724e-06, "loss": 0.1790323257446289, "step": 208 }, { "epoch": 0.05425752855659398, "grad_norm": 3.88889741897583, "learning_rate": 4.482758620689656e-06, "loss": 0.19250234961509705, "step": 209 }, { "epoch": 0.05451713395638629, "grad_norm": 2.3280720710754395, "learning_rate": 4.504310344827586e-06, "loss": 0.1587367057800293, "step": 210 }, { "epoch": 0.054776739356178605, "grad_norm": 4.1559157371521, "learning_rate": 4.525862068965518e-06, "loss": 0.20711275935173035, "step": 211 }, { "epoch": 0.055036344755970926, "grad_norm": 2.774156093597412, "learning_rate": 4.547413793103448e-06, "loss": 0.18325337767601013, "step": 212 }, { "epoch": 0.05529595015576324, "grad_norm": 2.3748111724853516, "learning_rate": 4.56896551724138e-06, "loss": 0.1779455989599228, "step": 213 }, { "epoch": 0.05555555555555555, "grad_norm": 4.905636310577393, "learning_rate": 4.590517241379311e-06, "loss": 0.15088707208633423, "step": 214 }, { "epoch": 0.05581516095534787, "grad_norm": 4.384755611419678, "learning_rate": 4.612068965517242e-06, "loss": 0.16596707701683044, "step": 215 }, { "epoch": 0.056074766355140186, "grad_norm": 2.8599038124084473, "learning_rate": 4.633620689655173e-06, "loss": 0.18111968040466309, "step": 216 }, { "epoch": 0.0563343717549325, "grad_norm": 1.627022385597229, "learning_rate": 4.655172413793104e-06, "loss": 0.14877940714359283, "step": 217 }, { "epoch": 0.05659397715472482, "grad_norm": 3.6703732013702393, "learning_rate": 4.676724137931034e-06, "loss": 0.1529468446969986, "step": 218 }, { "epoch": 0.05685358255451713, "grad_norm": 3.804706335067749, "learning_rate": 4.698275862068966e-06, "loss": 0.24142590165138245, "step": 219 }, { "epoch": 0.05711318795430945, "grad_norm": 5.754454135894775, "learning_rate": 4.719827586206897e-06, "loss": 0.17395402491092682, "step": 220 }, { "epoch": 0.05737279335410177, "grad_norm": 3.184447765350342, "learning_rate": 4.741379310344828e-06, "loss": 0.15367475152015686, "step": 221 }, { "epoch": 0.05763239875389408, "grad_norm": 2.6205673217773438, "learning_rate": 4.762931034482759e-06, "loss": 0.15646353363990784, "step": 222 }, { "epoch": 0.057892004153686394, "grad_norm": 1.5787967443466187, "learning_rate": 4.78448275862069e-06, "loss": 0.1745673418045044, "step": 223 }, { "epoch": 0.058151609553478714, "grad_norm": 3.925577402114868, "learning_rate": 4.806034482758621e-06, "loss": 0.16583478450775146, "step": 224 }, { "epoch": 0.05841121495327103, "grad_norm": 2.2370805740356445, "learning_rate": 4.8275862068965525e-06, "loss": 0.16914653778076172, "step": 225 }, { "epoch": 0.05867082035306334, "grad_norm": 3.418182373046875, "learning_rate": 4.849137931034483e-06, "loss": 0.2114856243133545, "step": 226 }, { "epoch": 0.05893042575285566, "grad_norm": 3.5659217834472656, "learning_rate": 4.8706896551724145e-06, "loss": 0.1822638064622879, "step": 227 }, { "epoch": 0.059190031152647975, "grad_norm": 1.5967700481414795, "learning_rate": 4.892241379310345e-06, "loss": 0.17291058599948883, "step": 228 }, { "epoch": 0.05944963655244029, "grad_norm": 5.470901012420654, "learning_rate": 4.9137931034482765e-06, "loss": 0.17980509996414185, "step": 229 }, { "epoch": 0.05970924195223261, "grad_norm": 3.0544042587280273, "learning_rate": 4.935344827586207e-06, "loss": 0.15065129101276398, "step": 230 }, { "epoch": 0.05996884735202492, "grad_norm": 2.0084054470062256, "learning_rate": 4.9568965517241384e-06, "loss": 0.14492827653884888, "step": 231 }, { "epoch": 0.060228452751817235, "grad_norm": 7.3156819343566895, "learning_rate": 4.97844827586207e-06, "loss": 0.19345435500144958, "step": 232 }, { "epoch": 0.060488058151609556, "grad_norm": 8.135090827941895, "learning_rate": 5e-06, "loss": 0.19252316653728485, "step": 233 }, { "epoch": 0.06074766355140187, "grad_norm": 4.05657958984375, "learning_rate": 4.99999980112576e-06, "loss": 0.2053917646408081, "step": 234 }, { "epoch": 0.06100726895119418, "grad_norm": 7.174722194671631, "learning_rate": 4.999999204503072e-06, "loss": 0.19374054670333862, "step": 235 }, { "epoch": 0.0612668743509865, "grad_norm": 3.1311659812927246, "learning_rate": 4.999998210132044e-06, "loss": 0.17093433439731598, "step": 236 }, { "epoch": 0.061526479750778816, "grad_norm": 3.683199882507324, "learning_rate": 4.999996818012849e-06, "loss": 0.19066283106803894, "step": 237 }, { "epoch": 0.06178608515057113, "grad_norm": 5.104578018188477, "learning_rate": 4.999995028145735e-06, "loss": 0.17664282023906708, "step": 238 }, { "epoch": 0.06204569055036345, "grad_norm": 7.900356292724609, "learning_rate": 4.999992840531019e-06, "loss": 0.24222689867019653, "step": 239 }, { "epoch": 0.06230529595015576, "grad_norm": 3.11885404586792, "learning_rate": 4.999990255169086e-06, "loss": 0.19081714749336243, "step": 240 }, { "epoch": 0.06256490134994808, "grad_norm": 3.2523820400238037, "learning_rate": 4.999987272060394e-06, "loss": 0.18701736629009247, "step": 241 }, { "epoch": 0.0628245067497404, "grad_norm": 1.7797150611877441, "learning_rate": 4.999983891205471e-06, "loss": 0.15969061851501465, "step": 242 }, { "epoch": 0.0630841121495327, "grad_norm": 5.873961448669434, "learning_rate": 4.999980112604912e-06, "loss": 0.20316731929779053, "step": 243 }, { "epoch": 0.06334371754932502, "grad_norm": 4.622780799865723, "learning_rate": 4.999975936259389e-06, "loss": 0.1950325220823288, "step": 244 }, { "epoch": 0.06360332294911734, "grad_norm": 6.385434150695801, "learning_rate": 4.9999713621696375e-06, "loss": 0.20971205830574036, "step": 245 }, { "epoch": 0.06386292834890965, "grad_norm": 1.822005033493042, "learning_rate": 4.999966390336466e-06, "loss": 0.16378480195999146, "step": 246 }, { "epoch": 0.06412253374870197, "grad_norm": 4.92262077331543, "learning_rate": 4.999961020760754e-06, "loss": 0.18115060031414032, "step": 247 }, { "epoch": 0.06438213914849429, "grad_norm": 3.627244710922241, "learning_rate": 4.999955253443451e-06, "loss": 0.1817604899406433, "step": 248 }, { "epoch": 0.0646417445482866, "grad_norm": 5.565239906311035, "learning_rate": 4.999949088385577e-06, "loss": 0.22555764019489288, "step": 249 }, { "epoch": 0.06490134994807892, "grad_norm": 2.5154871940612793, "learning_rate": 4.9999425255882205e-06, "loss": 0.19154538214206696, "step": 250 }, { "epoch": 0.06516095534787124, "grad_norm": 2.6930599212646484, "learning_rate": 4.999935565052543e-06, "loss": 0.17557202279567719, "step": 251 }, { "epoch": 0.06542056074766354, "grad_norm": 6.452319145202637, "learning_rate": 4.999928206779774e-06, "loss": 0.1957147717475891, "step": 252 }, { "epoch": 0.06568016614745587, "grad_norm": 2.3283395767211914, "learning_rate": 4.999920450771214e-06, "loss": 0.18969787657260895, "step": 253 }, { "epoch": 0.06593977154724819, "grad_norm": 4.299205303192139, "learning_rate": 4.999912297028235e-06, "loss": 0.21542949974536896, "step": 254 }, { "epoch": 0.06619937694704049, "grad_norm": 5.899773597717285, "learning_rate": 4.999903745552278e-06, "loss": 0.18611325323581696, "step": 255 }, { "epoch": 0.06645898234683281, "grad_norm": 4.758547306060791, "learning_rate": 4.9998947963448554e-06, "loss": 0.21656712889671326, "step": 256 }, { "epoch": 0.06671858774662513, "grad_norm": 2.475476026535034, "learning_rate": 4.999885449407547e-06, "loss": 0.18134574592113495, "step": 257 }, { "epoch": 0.06697819314641744, "grad_norm": 2.665928363800049, "learning_rate": 4.999875704742009e-06, "loss": 0.1845458447933197, "step": 258 }, { "epoch": 0.06723779854620976, "grad_norm": 3.703798294067383, "learning_rate": 4.99986556234996e-06, "loss": 0.18163293600082397, "step": 259 }, { "epoch": 0.06749740394600208, "grad_norm": 7.7519330978393555, "learning_rate": 4.999855022233195e-06, "loss": 0.1785305142402649, "step": 260 }, { "epoch": 0.06775700934579439, "grad_norm": 5.232489109039307, "learning_rate": 4.999844084393577e-06, "loss": 0.2073153704404831, "step": 261 }, { "epoch": 0.0680166147455867, "grad_norm": 14.755343437194824, "learning_rate": 4.99983274883304e-06, "loss": 0.2292483150959015, "step": 262 }, { "epoch": 0.06827622014537903, "grad_norm": 6.160196304321289, "learning_rate": 4.999821015553586e-06, "loss": 0.16573238372802734, "step": 263 }, { "epoch": 0.06853582554517133, "grad_norm": 8.46606731414795, "learning_rate": 4.999808884557291e-06, "loss": 0.17723099887371063, "step": 264 }, { "epoch": 0.06879543094496365, "grad_norm": 6.082627773284912, "learning_rate": 4.9997963558463e-06, "loss": 0.18407520651817322, "step": 265 }, { "epoch": 0.06905503634475597, "grad_norm": 2.5086913108825684, "learning_rate": 4.999783429422826e-06, "loss": 0.15643930435180664, "step": 266 }, { "epoch": 0.06931464174454828, "grad_norm": 8.443078994750977, "learning_rate": 4.999770105289155e-06, "loss": 0.19754689931869507, "step": 267 }, { "epoch": 0.0695742471443406, "grad_norm": 4.663389205932617, "learning_rate": 4.999756383447642e-06, "loss": 0.19095052778720856, "step": 268 }, { "epoch": 0.06983385254413292, "grad_norm": 3.6825318336486816, "learning_rate": 4.999742263900714e-06, "loss": 0.2099687159061432, "step": 269 }, { "epoch": 0.07009345794392523, "grad_norm": 1.8569071292877197, "learning_rate": 4.9997277466508654e-06, "loss": 0.14638930559158325, "step": 270 }, { "epoch": 0.07035306334371755, "grad_norm": 4.36627197265625, "learning_rate": 4.9997128317006625e-06, "loss": 0.19741317629814148, "step": 271 }, { "epoch": 0.07061266874350987, "grad_norm": 2.0591232776641846, "learning_rate": 4.999697519052743e-06, "loss": 0.15693722665309906, "step": 272 }, { "epoch": 0.07087227414330217, "grad_norm": 6.639446258544922, "learning_rate": 4.999681808709814e-06, "loss": 0.2000424712896347, "step": 273 }, { "epoch": 0.0711318795430945, "grad_norm": 2.5978899002075195, "learning_rate": 4.999665700674651e-06, "loss": 0.20204974710941315, "step": 274 }, { "epoch": 0.07139148494288682, "grad_norm": 4.345844268798828, "learning_rate": 4.999649194950103e-06, "loss": 0.19953733682632446, "step": 275 }, { "epoch": 0.07165109034267912, "grad_norm": 2.9226467609405518, "learning_rate": 4.999632291539088e-06, "loss": 0.16936147212982178, "step": 276 }, { "epoch": 0.07191069574247144, "grad_norm": 3.0732791423797607, "learning_rate": 4.999614990444594e-06, "loss": 0.1638348549604416, "step": 277 }, { "epoch": 0.07217030114226376, "grad_norm": 3.1201748847961426, "learning_rate": 4.999597291669679e-06, "loss": 0.20115284621715546, "step": 278 }, { "epoch": 0.07242990654205607, "grad_norm": 4.638462066650391, "learning_rate": 4.99957919521747e-06, "loss": 0.16076868772506714, "step": 279 }, { "epoch": 0.07268951194184839, "grad_norm": 3.924083709716797, "learning_rate": 4.9995607010911696e-06, "loss": 0.20291592180728912, "step": 280 }, { "epoch": 0.07294911734164071, "grad_norm": 2.268939971923828, "learning_rate": 4.999541809294045e-06, "loss": 0.15374481678009033, "step": 281 }, { "epoch": 0.07320872274143302, "grad_norm": 2.498560667037964, "learning_rate": 4.999522519829435e-06, "loss": 0.1528492122888565, "step": 282 }, { "epoch": 0.07346832814122534, "grad_norm": 1.4783649444580078, "learning_rate": 4.999502832700752e-06, "loss": 0.15590590238571167, "step": 283 }, { "epoch": 0.07372793354101766, "grad_norm": 3.6385931968688965, "learning_rate": 4.999482747911475e-06, "loss": 0.16364845633506775, "step": 284 }, { "epoch": 0.07398753894080996, "grad_norm": 4.789358615875244, "learning_rate": 4.999462265465154e-06, "loss": 0.1467256247997284, "step": 285 }, { "epoch": 0.07424714434060228, "grad_norm": 2.1566290855407715, "learning_rate": 4.99944138536541e-06, "loss": 0.15319114923477173, "step": 286 }, { "epoch": 0.0745067497403946, "grad_norm": 1.7142056226730347, "learning_rate": 4.999420107615935e-06, "loss": 0.15492624044418335, "step": 287 }, { "epoch": 0.07476635514018691, "grad_norm": 2.042846441268921, "learning_rate": 4.9993984322204905e-06, "loss": 0.15964587032794952, "step": 288 }, { "epoch": 0.07502596053997923, "grad_norm": 1.5610624551773071, "learning_rate": 4.999376359182906e-06, "loss": 0.16266444325447083, "step": 289 }, { "epoch": 0.07528556593977155, "grad_norm": 3.150059938430786, "learning_rate": 4.999353888507086e-06, "loss": 0.1508605033159256, "step": 290 }, { "epoch": 0.07554517133956386, "grad_norm": 3.2419323921203613, "learning_rate": 4.999331020197002e-06, "loss": 0.1662689745426178, "step": 291 }, { "epoch": 0.07580477673935618, "grad_norm": 2.800225019454956, "learning_rate": 4.999307754256696e-06, "loss": 0.19623447954654694, "step": 292 }, { "epoch": 0.0760643821391485, "grad_norm": 1.024777889251709, "learning_rate": 4.999284090690282e-06, "loss": 0.14326360821723938, "step": 293 }, { "epoch": 0.0763239875389408, "grad_norm": 1.9561597108840942, "learning_rate": 4.999260029501942e-06, "loss": 0.20734238624572754, "step": 294 }, { "epoch": 0.07658359293873312, "grad_norm": 2.155933141708374, "learning_rate": 4.999235570695931e-06, "loss": 0.15113498270511627, "step": 295 }, { "epoch": 0.07684319833852545, "grad_norm": 2.122751235961914, "learning_rate": 4.9992107142765715e-06, "loss": 0.14312312006950378, "step": 296 }, { "epoch": 0.07710280373831775, "grad_norm": 3.6757278442382812, "learning_rate": 4.999185460248257e-06, "loss": 0.1667429357767105, "step": 297 }, { "epoch": 0.07736240913811007, "grad_norm": 3.356642484664917, "learning_rate": 4.999159808615453e-06, "loss": 0.14521875977516174, "step": 298 }, { "epoch": 0.07762201453790239, "grad_norm": 6.259903430938721, "learning_rate": 4.999133759382693e-06, "loss": 0.20474949479103088, "step": 299 }, { "epoch": 0.0778816199376947, "grad_norm": 2.219115734100342, "learning_rate": 4.999107312554583e-06, "loss": 0.14519649744033813, "step": 300 }, { "epoch": 0.07814122533748702, "grad_norm": 8.034956932067871, "learning_rate": 4.9990804681358e-06, "loss": 0.1713053584098816, "step": 301 }, { "epoch": 0.07840083073727934, "grad_norm": 2.8608381748199463, "learning_rate": 4.999053226131084e-06, "loss": 0.2060289829969406, "step": 302 }, { "epoch": 0.07866043613707165, "grad_norm": 6.435917854309082, "learning_rate": 4.9990255865452565e-06, "loss": 0.16334189474582672, "step": 303 }, { "epoch": 0.07892004153686397, "grad_norm": 5.870814323425293, "learning_rate": 4.9989975493832005e-06, "loss": 0.1542157530784607, "step": 304 }, { "epoch": 0.07917964693665629, "grad_norm": 7.957790851593018, "learning_rate": 4.998969114649873e-06, "loss": 0.20730283856391907, "step": 305 }, { "epoch": 0.0794392523364486, "grad_norm": 1.5185803174972534, "learning_rate": 4.9989402823503006e-06, "loss": 0.18742559850215912, "step": 306 }, { "epoch": 0.07969885773624091, "grad_norm": 3.37115478515625, "learning_rate": 4.9989110524895795e-06, "loss": 0.17474070191383362, "step": 307 }, { "epoch": 0.07995846313603323, "grad_norm": 1.703151822090149, "learning_rate": 4.998881425072879e-06, "loss": 0.17469820380210876, "step": 308 }, { "epoch": 0.08021806853582554, "grad_norm": 3.3629422187805176, "learning_rate": 4.998851400105433e-06, "loss": 0.14530789852142334, "step": 309 }, { "epoch": 0.08047767393561786, "grad_norm": 2.3281378746032715, "learning_rate": 4.998820977592553e-06, "loss": 0.19442838430404663, "step": 310 }, { "epoch": 0.08073727933541018, "grad_norm": 2.733093500137329, "learning_rate": 4.998790157539615e-06, "loss": 0.17125913500785828, "step": 311 }, { "epoch": 0.08099688473520249, "grad_norm": 4.878777980804443, "learning_rate": 4.998758939952068e-06, "loss": 0.2168044149875641, "step": 312 }, { "epoch": 0.08125649013499481, "grad_norm": 2.075904130935669, "learning_rate": 4.998727324835429e-06, "loss": 0.17377503216266632, "step": 313 }, { "epoch": 0.08151609553478713, "grad_norm": 6.275523662567139, "learning_rate": 4.998695312195289e-06, "loss": 0.16824454069137573, "step": 314 }, { "epoch": 0.08177570093457943, "grad_norm": 2.546541690826416, "learning_rate": 4.998662902037306e-06, "loss": 0.15631595253944397, "step": 315 }, { "epoch": 0.08203530633437175, "grad_norm": 1.9625295400619507, "learning_rate": 4.998630094367209e-06, "loss": 0.1560731828212738, "step": 316 }, { "epoch": 0.08229491173416408, "grad_norm": 1.0108023881912231, "learning_rate": 4.998596889190799e-06, "loss": 0.1465441733598709, "step": 317 }, { "epoch": 0.08255451713395638, "grad_norm": 1.2732905149459839, "learning_rate": 4.998563286513944e-06, "loss": 0.15628008544445038, "step": 318 }, { "epoch": 0.0828141225337487, "grad_norm": 2.70456862449646, "learning_rate": 4.998529286342585e-06, "loss": 0.18371176719665527, "step": 319 }, { "epoch": 0.08307372793354102, "grad_norm": 4.028971195220947, "learning_rate": 4.998494888682734e-06, "loss": 0.2226772904396057, "step": 320 }, { "epoch": 0.08333333333333333, "grad_norm": 3.7143261432647705, "learning_rate": 4.99846009354047e-06, "loss": 0.1636454164981842, "step": 321 }, { "epoch": 0.08359293873312565, "grad_norm": 1.6964855194091797, "learning_rate": 4.998424900921944e-06, "loss": 0.18511547148227692, "step": 322 }, { "epoch": 0.08385254413291797, "grad_norm": 3.260631799697876, "learning_rate": 4.998389310833378e-06, "loss": 0.18463361263275146, "step": 323 }, { "epoch": 0.08411214953271028, "grad_norm": 3.1382322311401367, "learning_rate": 4.998353323281063e-06, "loss": 0.1605904996395111, "step": 324 }, { "epoch": 0.0843717549325026, "grad_norm": 1.6565895080566406, "learning_rate": 4.998316938271361e-06, "loss": 0.16560789942741394, "step": 325 }, { "epoch": 0.08463136033229492, "grad_norm": 1.9214900732040405, "learning_rate": 4.998280155810704e-06, "loss": 0.173313170671463, "step": 326 }, { "epoch": 0.08489096573208722, "grad_norm": 2.3755321502685547, "learning_rate": 4.998242975905593e-06, "loss": 0.194078728556633, "step": 327 }, { "epoch": 0.08515057113187954, "grad_norm": 2.2251052856445312, "learning_rate": 4.998205398562603e-06, "loss": 0.18700304627418518, "step": 328 }, { "epoch": 0.08541017653167186, "grad_norm": 1.8253350257873535, "learning_rate": 4.998167423788375e-06, "loss": 0.15261022746562958, "step": 329 }, { "epoch": 0.08566978193146417, "grad_norm": 2.457953929901123, "learning_rate": 4.998129051589624e-06, "loss": 0.2068975418806076, "step": 330 }, { "epoch": 0.08592938733125649, "grad_norm": 3.188765048980713, "learning_rate": 4.998090281973131e-06, "loss": 0.15901055932044983, "step": 331 }, { "epoch": 0.08618899273104881, "grad_norm": 4.4347076416015625, "learning_rate": 4.99805111494575e-06, "loss": 0.1706700325012207, "step": 332 }, { "epoch": 0.08644859813084112, "grad_norm": 3.7042298316955566, "learning_rate": 4.9980115505144065e-06, "loss": 0.1742784082889557, "step": 333 }, { "epoch": 0.08670820353063344, "grad_norm": 2.976675510406494, "learning_rate": 4.997971588686093e-06, "loss": 0.16418549418449402, "step": 334 }, { "epoch": 0.08696780893042576, "grad_norm": 5.933595657348633, "learning_rate": 4.9979312294678745e-06, "loss": 0.20097136497497559, "step": 335 }, { "epoch": 0.08722741433021806, "grad_norm": 1.8812589645385742, "learning_rate": 4.997890472866885e-06, "loss": 0.1777919977903366, "step": 336 }, { "epoch": 0.08748701973001038, "grad_norm": 3.8818359375, "learning_rate": 4.99784931889033e-06, "loss": 0.1894613653421402, "step": 337 }, { "epoch": 0.0877466251298027, "grad_norm": 3.8873653411865234, "learning_rate": 4.997807767545485e-06, "loss": 0.16034212708473206, "step": 338 }, { "epoch": 0.08800623052959501, "grad_norm": 2.252020835876465, "learning_rate": 4.9977658188396935e-06, "loss": 0.15837976336479187, "step": 339 }, { "epoch": 0.08826583592938733, "grad_norm": 5.227093696594238, "learning_rate": 4.997723472780373e-06, "loss": 0.15623529255390167, "step": 340 }, { "epoch": 0.08852544132917965, "grad_norm": 2.178772449493408, "learning_rate": 4.997680729375008e-06, "loss": 0.15140143036842346, "step": 341 }, { "epoch": 0.08878504672897196, "grad_norm": 4.463717937469482, "learning_rate": 4.997637588631156e-06, "loss": 0.1893811821937561, "step": 342 }, { "epoch": 0.08904465212876428, "grad_norm": 5.480949878692627, "learning_rate": 4.997594050556443e-06, "loss": 0.1501477062702179, "step": 343 }, { "epoch": 0.0893042575285566, "grad_norm": 3.144810914993286, "learning_rate": 4.9975501151585635e-06, "loss": 0.17385409772396088, "step": 344 }, { "epoch": 0.0895638629283489, "grad_norm": 1.9951248168945312, "learning_rate": 4.997505782445286e-06, "loss": 0.16279616951942444, "step": 345 }, { "epoch": 0.08982346832814123, "grad_norm": 5.599791526794434, "learning_rate": 4.997461052424447e-06, "loss": 0.1883581578731537, "step": 346 }, { "epoch": 0.09008307372793355, "grad_norm": 4.150071620941162, "learning_rate": 4.997415925103955e-06, "loss": 0.19577229022979736, "step": 347 }, { "epoch": 0.09034267912772585, "grad_norm": 6.093507766723633, "learning_rate": 4.997370400491786e-06, "loss": 0.1665656417608261, "step": 348 }, { "epoch": 0.09060228452751817, "grad_norm": 3.6321566104888916, "learning_rate": 4.997324478595987e-06, "loss": 0.18312698602676392, "step": 349 }, { "epoch": 0.0908618899273105, "grad_norm": 4.979744911193848, "learning_rate": 4.997278159424678e-06, "loss": 0.17565785348415375, "step": 350 }, { "epoch": 0.0911214953271028, "grad_norm": 3.1443018913269043, "learning_rate": 4.997231442986047e-06, "loss": 0.1926029771566391, "step": 351 }, { "epoch": 0.09138110072689512, "grad_norm": 1.976229190826416, "learning_rate": 4.997184329288351e-06, "loss": 0.19054117798805237, "step": 352 }, { "epoch": 0.09164070612668744, "grad_norm": 4.936254024505615, "learning_rate": 4.99713681833992e-06, "loss": 0.190906822681427, "step": 353 }, { "epoch": 0.09190031152647975, "grad_norm": 4.412229537963867, "learning_rate": 4.99708891014915e-06, "loss": 0.19576852023601532, "step": 354 }, { "epoch": 0.09215991692627207, "grad_norm": 2.2090089321136475, "learning_rate": 4.997040604724514e-06, "loss": 0.1626296192407608, "step": 355 }, { "epoch": 0.09241952232606439, "grad_norm": 3.0865259170532227, "learning_rate": 4.9969919020745485e-06, "loss": 0.15128478407859802, "step": 356 }, { "epoch": 0.0926791277258567, "grad_norm": 4.046040058135986, "learning_rate": 4.996942802207866e-06, "loss": 0.1484414041042328, "step": 357 }, { "epoch": 0.09293873312564901, "grad_norm": 2.772932767868042, "learning_rate": 4.996893305133142e-06, "loss": 0.19554325938224792, "step": 358 }, { "epoch": 0.09319833852544133, "grad_norm": 2.76415753364563, "learning_rate": 4.996843410859129e-06, "loss": 0.14456886053085327, "step": 359 }, { "epoch": 0.09345794392523364, "grad_norm": 2.6407670974731445, "learning_rate": 4.996793119394648e-06, "loss": 0.14455553889274597, "step": 360 }, { "epoch": 0.09371754932502596, "grad_norm": 1.104627013206482, "learning_rate": 4.996742430748587e-06, "loss": 0.14846278727054596, "step": 361 }, { "epoch": 0.09397715472481828, "grad_norm": 5.3501105308532715, "learning_rate": 4.996691344929909e-06, "loss": 0.22468361258506775, "step": 362 }, { "epoch": 0.09423676012461059, "grad_norm": 4.837092876434326, "learning_rate": 4.996639861947643e-06, "loss": 0.1930350661277771, "step": 363 }, { "epoch": 0.09449636552440291, "grad_norm": 7.174378871917725, "learning_rate": 4.99658798181089e-06, "loss": 0.18148121237754822, "step": 364 }, { "epoch": 0.09475597092419523, "grad_norm": 2.593968152999878, "learning_rate": 4.996535704528823e-06, "loss": 0.16310881078243256, "step": 365 }, { "epoch": 0.09501557632398754, "grad_norm": 3.7978599071502686, "learning_rate": 4.996483030110682e-06, "loss": 0.18112844228744507, "step": 366 }, { "epoch": 0.09527518172377986, "grad_norm": 2.1288108825683594, "learning_rate": 4.996429958565779e-06, "loss": 0.14798244833946228, "step": 367 }, { "epoch": 0.09553478712357218, "grad_norm": 3.39198637008667, "learning_rate": 4.996376489903495e-06, "loss": 0.16301918029785156, "step": 368 }, { "epoch": 0.09579439252336448, "grad_norm": 1.6320899724960327, "learning_rate": 4.996322624133284e-06, "loss": 0.15145689249038696, "step": 369 }, { "epoch": 0.0960539979231568, "grad_norm": 3.0822412967681885, "learning_rate": 4.996268361264666e-06, "loss": 0.14336630702018738, "step": 370 }, { "epoch": 0.09631360332294912, "grad_norm": 1.076536774635315, "learning_rate": 4.996213701307235e-06, "loss": 0.14738903939723969, "step": 371 }, { "epoch": 0.09657320872274143, "grad_norm": 2.9792890548706055, "learning_rate": 4.9961586442706525e-06, "loss": 0.19246235489845276, "step": 372 }, { "epoch": 0.09683281412253375, "grad_norm": 4.58516788482666, "learning_rate": 4.9961031901646524e-06, "loss": 0.18802715837955475, "step": 373 }, { "epoch": 0.09709241952232607, "grad_norm": 1.985661268234253, "learning_rate": 4.996047338999037e-06, "loss": 0.1815195083618164, "step": 374 }, { "epoch": 0.09735202492211838, "grad_norm": 2.7883994579315186, "learning_rate": 4.99599109078368e-06, "loss": 0.15610450506210327, "step": 375 }, { "epoch": 0.0976116303219107, "grad_norm": 2.067955255508423, "learning_rate": 4.995934445528524e-06, "loss": 0.15325629711151123, "step": 376 }, { "epoch": 0.09787123572170302, "grad_norm": 3.1303582191467285, "learning_rate": 4.995877403243584e-06, "loss": 0.1672753244638443, "step": 377 }, { "epoch": 0.09813084112149532, "grad_norm": 4.221309185028076, "learning_rate": 4.995819963938942e-06, "loss": 0.15356671810150146, "step": 378 }, { "epoch": 0.09839044652128764, "grad_norm": 3.5819895267486572, "learning_rate": 4.995762127624753e-06, "loss": 0.15002581477165222, "step": 379 }, { "epoch": 0.09865005192107996, "grad_norm": 2.4365339279174805, "learning_rate": 4.995703894311242e-06, "loss": 0.15109136700630188, "step": 380 }, { "epoch": 0.09890965732087227, "grad_norm": 4.616072177886963, "learning_rate": 4.995645264008701e-06, "loss": 0.18770736455917358, "step": 381 }, { "epoch": 0.09916926272066459, "grad_norm": 1.9430228471755981, "learning_rate": 4.995586236727496e-06, "loss": 0.15813931822776794, "step": 382 }, { "epoch": 0.09942886812045691, "grad_norm": 4.074131011962891, "learning_rate": 4.995526812478061e-06, "loss": 0.21396809816360474, "step": 383 }, { "epoch": 0.09968847352024922, "grad_norm": 6.02095365524292, "learning_rate": 4.995466991270903e-06, "loss": 0.15067273378372192, "step": 384 }, { "epoch": 0.09994807892004154, "grad_norm": 3.214729070663452, "learning_rate": 4.9954067731165945e-06, "loss": 0.15123184025287628, "step": 385 }, { "epoch": 0.10020768431983386, "grad_norm": 3.016427755355835, "learning_rate": 4.995346158025781e-06, "loss": 0.14104118943214417, "step": 386 }, { "epoch": 0.10046728971962617, "grad_norm": 2.2259457111358643, "learning_rate": 4.995285146009179e-06, "loss": 0.1473507285118103, "step": 387 }, { "epoch": 0.10072689511941849, "grad_norm": 1.484101414680481, "learning_rate": 4.995223737077573e-06, "loss": 0.2019626796245575, "step": 388 }, { "epoch": 0.1009865005192108, "grad_norm": 2.0890631675720215, "learning_rate": 4.9951619312418196e-06, "loss": 0.1436842530965805, "step": 389 }, { "epoch": 0.10124610591900311, "grad_norm": 3.155684232711792, "learning_rate": 4.995099728512843e-06, "loss": 0.15940198302268982, "step": 390 }, { "epoch": 0.10150571131879543, "grad_norm": 2.0785977840423584, "learning_rate": 4.995037128901642e-06, "loss": 0.1525573432445526, "step": 391 }, { "epoch": 0.10176531671858775, "grad_norm": 4.246424674987793, "learning_rate": 4.99497413241928e-06, "loss": 0.1515767127275467, "step": 392 }, { "epoch": 0.10202492211838006, "grad_norm": 4.75434684753418, "learning_rate": 4.994910739076895e-06, "loss": 0.20019376277923584, "step": 393 }, { "epoch": 0.10228452751817238, "grad_norm": 2.3406307697296143, "learning_rate": 4.994846948885693e-06, "loss": 0.16949808597564697, "step": 394 }, { "epoch": 0.1025441329179647, "grad_norm": 6.900822639465332, "learning_rate": 4.99478276185695e-06, "loss": 0.17943888902664185, "step": 395 }, { "epoch": 0.102803738317757, "grad_norm": 1.0401290655136108, "learning_rate": 4.994718178002013e-06, "loss": 0.1376512348651886, "step": 396 }, { "epoch": 0.10306334371754933, "grad_norm": 4.441246509552002, "learning_rate": 4.9946531973323005e-06, "loss": 0.14881649613380432, "step": 397 }, { "epoch": 0.10332294911734165, "grad_norm": 2.1154730319976807, "learning_rate": 4.9945878198592975e-06, "loss": 0.1455685794353485, "step": 398 }, { "epoch": 0.10358255451713395, "grad_norm": 2.5731165409088135, "learning_rate": 4.994522045594563e-06, "loss": 0.1849280297756195, "step": 399 }, { "epoch": 0.10384215991692627, "grad_norm": 4.494977951049805, "learning_rate": 4.9944558745497236e-06, "loss": 0.1705615520477295, "step": 400 }, { "epoch": 0.1041017653167186, "grad_norm": 5.85746431350708, "learning_rate": 4.9943893067364765e-06, "loss": 0.17326027154922485, "step": 401 }, { "epoch": 0.1043613707165109, "grad_norm": 2.704498291015625, "learning_rate": 4.99432234216659e-06, "loss": 0.1750430315732956, "step": 402 }, { "epoch": 0.10462097611630322, "grad_norm": 3.0174617767333984, "learning_rate": 4.994254980851901e-06, "loss": 0.15335354208946228, "step": 403 }, { "epoch": 0.10488058151609553, "grad_norm": 5.7351813316345215, "learning_rate": 4.994187222804319e-06, "loss": 0.16483157873153687, "step": 404 }, { "epoch": 0.10514018691588785, "grad_norm": 3.678833484649658, "learning_rate": 4.99411906803582e-06, "loss": 0.1738872081041336, "step": 405 }, { "epoch": 0.10539979231568017, "grad_norm": 4.876923561096191, "learning_rate": 4.994050516558454e-06, "loss": 0.16672341525554657, "step": 406 }, { "epoch": 0.10565939771547248, "grad_norm": 6.078753471374512, "learning_rate": 4.993981568384339e-06, "loss": 0.16487985849380493, "step": 407 }, { "epoch": 0.1059190031152648, "grad_norm": 1.6971765756607056, "learning_rate": 4.993912223525661e-06, "loss": 0.1745944321155548, "step": 408 }, { "epoch": 0.10617860851505712, "grad_norm": 4.601200580596924, "learning_rate": 4.993842481994683e-06, "loss": 0.1782875657081604, "step": 409 }, { "epoch": 0.10643821391484942, "grad_norm": 3.748950958251953, "learning_rate": 4.99377234380373e-06, "loss": 0.18029116094112396, "step": 410 }, { "epoch": 0.10669781931464174, "grad_norm": 1.9940967559814453, "learning_rate": 4.9937018089652024e-06, "loss": 0.19976869225502014, "step": 411 }, { "epoch": 0.10695742471443406, "grad_norm": 5.4715166091918945, "learning_rate": 4.99363087749157e-06, "loss": 0.18043123185634613, "step": 412 }, { "epoch": 0.10721703011422637, "grad_norm": 5.725005626678467, "learning_rate": 4.993559549395369e-06, "loss": 0.20943757891654968, "step": 413 }, { "epoch": 0.10747663551401869, "grad_norm": 1.0150121450424194, "learning_rate": 4.993487824689211e-06, "loss": 0.15265914797782898, "step": 414 }, { "epoch": 0.10773624091381101, "grad_norm": 1.777919888496399, "learning_rate": 4.993415703385776e-06, "loss": 0.14379093050956726, "step": 415 }, { "epoch": 0.10799584631360332, "grad_norm": 3.899862051010132, "learning_rate": 4.99334318549781e-06, "loss": 0.15242598950862885, "step": 416 }, { "epoch": 0.10825545171339564, "grad_norm": 1.7420463562011719, "learning_rate": 4.993270271038136e-06, "loss": 0.1466905176639557, "step": 417 }, { "epoch": 0.10851505711318796, "grad_norm": 1.1775670051574707, "learning_rate": 4.993196960019642e-06, "loss": 0.14920246601104736, "step": 418 }, { "epoch": 0.10877466251298026, "grad_norm": 5.834907531738281, "learning_rate": 4.993123252455288e-06, "loss": 0.19962769746780396, "step": 419 }, { "epoch": 0.10903426791277258, "grad_norm": 2.0901741981506348, "learning_rate": 4.993049148358103e-06, "loss": 0.14886592328548431, "step": 420 }, { "epoch": 0.1092938733125649, "grad_norm": 2.8968021869659424, "learning_rate": 4.992974647741188e-06, "loss": 0.15550418198108673, "step": 421 }, { "epoch": 0.10955347871235721, "grad_norm": 4.452688694000244, "learning_rate": 4.992899750617712e-06, "loss": 0.21749162673950195, "step": 422 }, { "epoch": 0.10981308411214953, "grad_norm": 3.230570077896118, "learning_rate": 4.9928244570009165e-06, "loss": 0.15922047197818756, "step": 423 }, { "epoch": 0.11007268951194185, "grad_norm": 2.1715686321258545, "learning_rate": 4.99274876690411e-06, "loss": 0.1529265195131302, "step": 424 }, { "epoch": 0.11033229491173416, "grad_norm": 2.468291997909546, "learning_rate": 4.992672680340675e-06, "loss": 0.14936113357543945, "step": 425 }, { "epoch": 0.11059190031152648, "grad_norm": 1.4418796300888062, "learning_rate": 4.9925961973240595e-06, "loss": 0.14774450659751892, "step": 426 }, { "epoch": 0.1108515057113188, "grad_norm": 3.378425359725952, "learning_rate": 4.992519317867784e-06, "loss": 0.16288146376609802, "step": 427 }, { "epoch": 0.1111111111111111, "grad_norm": 4.88242769241333, "learning_rate": 4.992442041985442e-06, "loss": 0.181962788105011, "step": 428 }, { "epoch": 0.11137071651090343, "grad_norm": 2.1417438983917236, "learning_rate": 4.9923643696906915e-06, "loss": 0.14649531245231628, "step": 429 }, { "epoch": 0.11163032191069575, "grad_norm": 0.5396921634674072, "learning_rate": 4.992286300997263e-06, "loss": 0.15091271698474884, "step": 430 }, { "epoch": 0.11188992731048805, "grad_norm": 3.8184471130371094, "learning_rate": 4.992207835918959e-06, "loss": 0.186315655708313, "step": 431 }, { "epoch": 0.11214953271028037, "grad_norm": 1.9661232233047485, "learning_rate": 4.99212897446965e-06, "loss": 0.14573238790035248, "step": 432 }, { "epoch": 0.11240913811007269, "grad_norm": 2.1443963050842285, "learning_rate": 4.992049716663274e-06, "loss": 0.15495607256889343, "step": 433 }, { "epoch": 0.112668743509865, "grad_norm": 0.9811489582061768, "learning_rate": 4.991970062513845e-06, "loss": 0.148453950881958, "step": 434 }, { "epoch": 0.11292834890965732, "grad_norm": 1.9097024202346802, "learning_rate": 4.991890012035444e-06, "loss": 0.1999576985836029, "step": 435 }, { "epoch": 0.11318795430944964, "grad_norm": 1.7569769620895386, "learning_rate": 4.991809565242221e-06, "loss": 0.17953713238239288, "step": 436 }, { "epoch": 0.11344755970924195, "grad_norm": 6.228070259094238, "learning_rate": 4.991728722148398e-06, "loss": 0.17184211313724518, "step": 437 }, { "epoch": 0.11370716510903427, "grad_norm": 2.450096845626831, "learning_rate": 4.991647482768265e-06, "loss": 0.18668848276138306, "step": 438 }, { "epoch": 0.11396677050882659, "grad_norm": 3.8676464557647705, "learning_rate": 4.991565847116184e-06, "loss": 0.18482187390327454, "step": 439 }, { "epoch": 0.1142263759086189, "grad_norm": 2.2074882984161377, "learning_rate": 4.991483815206586e-06, "loss": 0.1461520493030548, "step": 440 }, { "epoch": 0.11448598130841121, "grad_norm": 2.1205031871795654, "learning_rate": 4.991401387053973e-06, "loss": 0.14704962074756622, "step": 441 }, { "epoch": 0.11474558670820353, "grad_norm": 6.2239990234375, "learning_rate": 4.991318562672917e-06, "loss": 0.17486253380775452, "step": 442 }, { "epoch": 0.11500519210799584, "grad_norm": 2.4148337841033936, "learning_rate": 4.991235342078058e-06, "loss": 0.1600492298603058, "step": 443 }, { "epoch": 0.11526479750778816, "grad_norm": 6.1362810134887695, "learning_rate": 4.9911517252841066e-06, "loss": 0.15022698044776917, "step": 444 }, { "epoch": 0.11552440290758048, "grad_norm": 4.900382995605469, "learning_rate": 4.991067712305848e-06, "loss": 0.2167195975780487, "step": 445 }, { "epoch": 0.11578400830737279, "grad_norm": 5.332601070404053, "learning_rate": 4.9909833031581294e-06, "loss": 0.16841955482959747, "step": 446 }, { "epoch": 0.11604361370716511, "grad_norm": 7.205552101135254, "learning_rate": 4.990898497855876e-06, "loss": 0.1864515244960785, "step": 447 }, { "epoch": 0.11630321910695743, "grad_norm": 3.095461368560791, "learning_rate": 4.990813296414077e-06, "loss": 0.19242852926254272, "step": 448 }, { "epoch": 0.11656282450674973, "grad_norm": 2.877354145050049, "learning_rate": 4.990727698847796e-06, "loss": 0.17614272236824036, "step": 449 }, { "epoch": 0.11682242990654206, "grad_norm": 5.348507881164551, "learning_rate": 4.990641705172163e-06, "loss": 0.17638251185417175, "step": 450 }, { "epoch": 0.11708203530633438, "grad_norm": 4.1287617683410645, "learning_rate": 4.99055531540238e-06, "loss": 0.1877962201833725, "step": 451 }, { "epoch": 0.11734164070612668, "grad_norm": 3.7971460819244385, "learning_rate": 4.99046852955372e-06, "loss": 0.18990248441696167, "step": 452 }, { "epoch": 0.117601246105919, "grad_norm": 2.6550660133361816, "learning_rate": 4.990381347641524e-06, "loss": 0.1364922821521759, "step": 453 }, { "epoch": 0.11786085150571132, "grad_norm": 2.7428431510925293, "learning_rate": 4.9902937696812034e-06, "loss": 0.1345965564250946, "step": 454 }, { "epoch": 0.11812045690550363, "grad_norm": 2.271375894546509, "learning_rate": 4.99020579568824e-06, "loss": 0.16789014637470245, "step": 455 }, { "epoch": 0.11838006230529595, "grad_norm": 2.2786216735839844, "learning_rate": 4.990117425678186e-06, "loss": 0.15005728602409363, "step": 456 }, { "epoch": 0.11863966770508827, "grad_norm": 2.026625156402588, "learning_rate": 4.990028659666663e-06, "loss": 0.147671639919281, "step": 457 }, { "epoch": 0.11889927310488058, "grad_norm": 1.1566710472106934, "learning_rate": 4.989939497669363e-06, "loss": 0.14535421133041382, "step": 458 }, { "epoch": 0.1191588785046729, "grad_norm": 3.114743709564209, "learning_rate": 4.989849939702047e-06, "loss": 0.18467804789543152, "step": 459 }, { "epoch": 0.11941848390446522, "grad_norm": 2.1399381160736084, "learning_rate": 4.989759985780548e-06, "loss": 0.1459164321422577, "step": 460 }, { "epoch": 0.11967808930425752, "grad_norm": 2.417593479156494, "learning_rate": 4.989669635920767e-06, "loss": 0.15952721238136292, "step": 461 }, { "epoch": 0.11993769470404984, "grad_norm": 2.811774253845215, "learning_rate": 4.989578890138676e-06, "loss": 0.12909725308418274, "step": 462 }, { "epoch": 0.12019730010384216, "grad_norm": 0.5662190318107605, "learning_rate": 4.989487748450318e-06, "loss": 0.1409587413072586, "step": 463 }, { "epoch": 0.12045690550363447, "grad_norm": 5.496500015258789, "learning_rate": 4.9893962108718015e-06, "loss": 0.21840360760688782, "step": 464 }, { "epoch": 0.12071651090342679, "grad_norm": 5.241122722625732, "learning_rate": 4.989304277419311e-06, "loss": 0.1986987590789795, "step": 465 }, { "epoch": 0.12097611630321911, "grad_norm": 4.754727363586426, "learning_rate": 4.989211948109099e-06, "loss": 0.1985604166984558, "step": 466 }, { "epoch": 0.12123572170301142, "grad_norm": 2.5238966941833496, "learning_rate": 4.989119222957483e-06, "loss": 0.18697571754455566, "step": 467 }, { "epoch": 0.12149532710280374, "grad_norm": 3.4993999004364014, "learning_rate": 4.98902610198086e-06, "loss": 0.19610612094402313, "step": 468 }, { "epoch": 0.12175493250259606, "grad_norm": 6.378687858581543, "learning_rate": 4.988932585195688e-06, "loss": 0.20859110355377197, "step": 469 }, { "epoch": 0.12201453790238836, "grad_norm": 1.1501224040985107, "learning_rate": 4.988838672618499e-06, "loss": 0.1382112205028534, "step": 470 }, { "epoch": 0.12227414330218069, "grad_norm": 2.730799913406372, "learning_rate": 4.988744364265896e-06, "loss": 0.18034978210926056, "step": 471 }, { "epoch": 0.122533748701973, "grad_norm": 7.077454566955566, "learning_rate": 4.988649660154551e-06, "loss": 0.17866864800453186, "step": 472 }, { "epoch": 0.12279335410176531, "grad_norm": 4.922679424285889, "learning_rate": 4.988554560301203e-06, "loss": 0.16310858726501465, "step": 473 }, { "epoch": 0.12305295950155763, "grad_norm": 4.067479133605957, "learning_rate": 4.988459064722664e-06, "loss": 0.11944504082202911, "step": 474 }, { "epoch": 0.12331256490134995, "grad_norm": 8.0326509475708, "learning_rate": 4.988363173435818e-06, "loss": 0.1855817288160324, "step": 475 }, { "epoch": 0.12357217030114226, "grad_norm": 4.580178260803223, "learning_rate": 4.9882668864576135e-06, "loss": 0.18757545948028564, "step": 476 }, { "epoch": 0.12383177570093458, "grad_norm": 2.075716257095337, "learning_rate": 4.988170203805074e-06, "loss": 0.15919554233551025, "step": 477 }, { "epoch": 0.1240913811007269, "grad_norm": 7.981110572814941, "learning_rate": 4.9880731254952885e-06, "loss": 0.16614550352096558, "step": 478 }, { "epoch": 0.1243509865005192, "grad_norm": 4.502553939819336, "learning_rate": 4.987975651545421e-06, "loss": 0.13948272168636322, "step": 479 }, { "epoch": 0.12461059190031153, "grad_norm": 3.1805334091186523, "learning_rate": 4.9878777819727e-06, "loss": 0.1602201610803604, "step": 480 }, { "epoch": 0.12487019730010385, "grad_norm": 5.331094264984131, "learning_rate": 4.9877795167944285e-06, "loss": 0.160407155752182, "step": 481 }, { "epoch": 0.12512980269989615, "grad_norm": 3.8394453525543213, "learning_rate": 4.987680856027976e-06, "loss": 0.1688137799501419, "step": 482 }, { "epoch": 0.12538940809968846, "grad_norm": 3.150787115097046, "learning_rate": 4.987581799690785e-06, "loss": 0.17186829447746277, "step": 483 }, { "epoch": 0.1256490134994808, "grad_norm": 2.5097851753234863, "learning_rate": 4.987482347800365e-06, "loss": 0.1340203881263733, "step": 484 }, { "epoch": 0.1259086188992731, "grad_norm": 6.324524402618408, "learning_rate": 4.987382500374298e-06, "loss": 0.2114974856376648, "step": 485 }, { "epoch": 0.1261682242990654, "grad_norm": 4.523403167724609, "learning_rate": 4.987282257430234e-06, "loss": 0.19433127343654633, "step": 486 }, { "epoch": 0.12642782969885774, "grad_norm": 1.2286063432693481, "learning_rate": 4.987181618985893e-06, "loss": 0.16938664019107819, "step": 487 }, { "epoch": 0.12668743509865005, "grad_norm": 3.1703169345855713, "learning_rate": 4.987080585059068e-06, "loss": 0.14563767611980438, "step": 488 }, { "epoch": 0.12694704049844235, "grad_norm": 2.252608060836792, "learning_rate": 4.9869791556676175e-06, "loss": 0.16045203804969788, "step": 489 }, { "epoch": 0.1272066458982347, "grad_norm": 3.392974376678467, "learning_rate": 4.986877330829472e-06, "loss": 0.14435595273971558, "step": 490 }, { "epoch": 0.127466251298027, "grad_norm": 0.9570261836051941, "learning_rate": 4.986775110562632e-06, "loss": 0.14443439245224, "step": 491 }, { "epoch": 0.1277258566978193, "grad_norm": 1.6745399236679077, "learning_rate": 4.9866724948851685e-06, "loss": 0.1856987476348877, "step": 492 }, { "epoch": 0.12798546209761164, "grad_norm": 1.0069934129714966, "learning_rate": 4.98656948381522e-06, "loss": 0.15421763062477112, "step": 493 }, { "epoch": 0.12824506749740394, "grad_norm": 6.357114315032959, "learning_rate": 4.986466077370998e-06, "loss": 0.1589915156364441, "step": 494 }, { "epoch": 0.12850467289719625, "grad_norm": 3.037384510040283, "learning_rate": 4.9863622755707815e-06, "loss": 0.19089731574058533, "step": 495 }, { "epoch": 0.12876427829698858, "grad_norm": 0.99639493227005, "learning_rate": 4.986258078432921e-06, "loss": 0.15376251935958862, "step": 496 }, { "epoch": 0.1290238836967809, "grad_norm": 2.4664080142974854, "learning_rate": 4.986153485975835e-06, "loss": 0.18619507551193237, "step": 497 }, { "epoch": 0.1292834890965732, "grad_norm": 2.6596877574920654, "learning_rate": 4.986048498218014e-06, "loss": 0.17035698890686035, "step": 498 }, { "epoch": 0.12954309449636553, "grad_norm": 1.4269568920135498, "learning_rate": 4.985943115178018e-06, "loss": 0.2043839544057846, "step": 499 }, { "epoch": 0.12980269989615784, "grad_norm": 4.461020469665527, "learning_rate": 4.985837336874475e-06, "loss": 0.15582388639450073, "step": 500 }, { "epoch": 0.13006230529595014, "grad_norm": 1.47158682346344, "learning_rate": 4.985731163326084e-06, "loss": 0.1902294158935547, "step": 501 }, { "epoch": 0.13032191069574248, "grad_norm": 1.6431536674499512, "learning_rate": 4.985624594551616e-06, "loss": 0.1602545529603958, "step": 502 }, { "epoch": 0.13058151609553478, "grad_norm": 1.7038798332214355, "learning_rate": 4.985517630569908e-06, "loss": 0.13633763790130615, "step": 503 }, { "epoch": 0.1308411214953271, "grad_norm": 3.4787087440490723, "learning_rate": 4.98541027139987e-06, "loss": 0.15286009013652802, "step": 504 }, { "epoch": 0.13110072689511942, "grad_norm": 2.130791425704956, "learning_rate": 4.98530251706048e-06, "loss": 0.17802028357982635, "step": 505 }, { "epoch": 0.13136033229491173, "grad_norm": 1.7750276327133179, "learning_rate": 4.9851943675707864e-06, "loss": 0.15955065190792084, "step": 506 }, { "epoch": 0.13161993769470404, "grad_norm": 4.169822692871094, "learning_rate": 4.985085822949908e-06, "loss": 0.15118028223514557, "step": 507 }, { "epoch": 0.13187954309449637, "grad_norm": 1.7643656730651855, "learning_rate": 4.984976883217032e-06, "loss": 0.16300350427627563, "step": 508 }, { "epoch": 0.13213914849428868, "grad_norm": 5.29390811920166, "learning_rate": 4.984867548391418e-06, "loss": 0.14642822742462158, "step": 509 }, { "epoch": 0.13239875389408098, "grad_norm": 4.0631842613220215, "learning_rate": 4.984757818492393e-06, "loss": 0.146174818277359, "step": 510 }, { "epoch": 0.13265835929387332, "grad_norm": 3.108771800994873, "learning_rate": 4.984647693539355e-06, "loss": 0.1441468894481659, "step": 511 }, { "epoch": 0.13291796469366562, "grad_norm": 4.632706642150879, "learning_rate": 4.984537173551772e-06, "loss": 0.14535614848136902, "step": 512 }, { "epoch": 0.13317757009345793, "grad_norm": 2.9308152198791504, "learning_rate": 4.98442625854918e-06, "loss": 0.1744910031557083, "step": 513 }, { "epoch": 0.13343717549325027, "grad_norm": 2.68670916557312, "learning_rate": 4.984314948551187e-06, "loss": 0.16832758486270905, "step": 514 }, { "epoch": 0.13369678089304257, "grad_norm": 6.6946702003479, "learning_rate": 4.98420324357747e-06, "loss": 0.17539334297180176, "step": 515 }, { "epoch": 0.13395638629283488, "grad_norm": 3.128406047821045, "learning_rate": 4.984091143647777e-06, "loss": 0.20086008310317993, "step": 516 }, { "epoch": 0.1342159916926272, "grad_norm": 4.33083963394165, "learning_rate": 4.983978648781923e-06, "loss": 0.16375479102134705, "step": 517 }, { "epoch": 0.13447559709241952, "grad_norm": 5.612628936767578, "learning_rate": 4.983865758999795e-06, "loss": 0.18000298738479614, "step": 518 }, { "epoch": 0.13473520249221183, "grad_norm": 1.755331039428711, "learning_rate": 4.98375247432135e-06, "loss": 0.13348037004470825, "step": 519 }, { "epoch": 0.13499480789200416, "grad_norm": 1.7834093570709229, "learning_rate": 4.983638794766613e-06, "loss": 0.16026104986667633, "step": 520 }, { "epoch": 0.13525441329179647, "grad_norm": 5.073459148406982, "learning_rate": 4.9835247203556815e-06, "loss": 0.18091410398483276, "step": 521 }, { "epoch": 0.13551401869158877, "grad_norm": 2.0713934898376465, "learning_rate": 4.9834102511087205e-06, "loss": 0.13549858331680298, "step": 522 }, { "epoch": 0.1357736240913811, "grad_norm": 1.389460802078247, "learning_rate": 4.983295387045965e-06, "loss": 0.18967565894126892, "step": 523 }, { "epoch": 0.1360332294911734, "grad_norm": 3.143690824508667, "learning_rate": 4.98318012818772e-06, "loss": 0.13708630204200745, "step": 524 }, { "epoch": 0.13629283489096572, "grad_norm": 2.7668254375457764, "learning_rate": 4.983064474554362e-06, "loss": 0.14407625794410706, "step": 525 }, { "epoch": 0.13655244029075805, "grad_norm": 3.611903190612793, "learning_rate": 4.9829484261663355e-06, "loss": 0.20191386342048645, "step": 526 }, { "epoch": 0.13681204569055036, "grad_norm": 4.17849588394165, "learning_rate": 4.982831983044155e-06, "loss": 0.14275625348091125, "step": 527 }, { "epoch": 0.13707165109034267, "grad_norm": 3.0404536724090576, "learning_rate": 4.982715145208405e-06, "loss": 0.15057146549224854, "step": 528 }, { "epoch": 0.137331256490135, "grad_norm": 3.7288873195648193, "learning_rate": 4.982597912679739e-06, "loss": 0.14510279893875122, "step": 529 }, { "epoch": 0.1375908618899273, "grad_norm": 4.980957508087158, "learning_rate": 4.9824802854788825e-06, "loss": 0.19459187984466553, "step": 530 }, { "epoch": 0.1378504672897196, "grad_norm": 1.6352490186691284, "learning_rate": 4.982362263626629e-06, "loss": 0.13356637954711914, "step": 531 }, { "epoch": 0.13811007268951195, "grad_norm": 1.2577710151672363, "learning_rate": 4.9822438471438404e-06, "loss": 0.13836967945098877, "step": 532 }, { "epoch": 0.13836967808930425, "grad_norm": 0.5862592458724976, "learning_rate": 4.982125036051452e-06, "loss": 0.13723421096801758, "step": 533 }, { "epoch": 0.13862928348909656, "grad_norm": 2.0376474857330322, "learning_rate": 4.982005830370466e-06, "loss": 0.18102195858955383, "step": 534 }, { "epoch": 0.1388888888888889, "grad_norm": 3.141864776611328, "learning_rate": 4.981886230121956e-06, "loss": 0.18107202649116516, "step": 535 }, { "epoch": 0.1391484942886812, "grad_norm": 1.1834663152694702, "learning_rate": 4.981766235327062e-06, "loss": 0.15370991826057434, "step": 536 }, { "epoch": 0.1394080996884735, "grad_norm": 1.5581823587417603, "learning_rate": 4.981645846007e-06, "loss": 0.13950267434120178, "step": 537 }, { "epoch": 0.13966770508826584, "grad_norm": 1.8585069179534912, "learning_rate": 4.98152506218305e-06, "loss": 0.18753600120544434, "step": 538 }, { "epoch": 0.13992731048805815, "grad_norm": 1.4040589332580566, "learning_rate": 4.981403883876564e-06, "loss": 0.1785471886396408, "step": 539 }, { "epoch": 0.14018691588785046, "grad_norm": 2.4022057056427, "learning_rate": 4.981282311108964e-06, "loss": 0.14117860794067383, "step": 540 }, { "epoch": 0.1404465212876428, "grad_norm": 2.096195697784424, "learning_rate": 4.98116034390174e-06, "loss": 0.18662813305854797, "step": 541 }, { "epoch": 0.1407061266874351, "grad_norm": 1.9186848402023315, "learning_rate": 4.981037982276456e-06, "loss": 0.16202738881111145, "step": 542 }, { "epoch": 0.1409657320872274, "grad_norm": 2.9408090114593506, "learning_rate": 4.980915226254739e-06, "loss": 0.14992378652095795, "step": 543 }, { "epoch": 0.14122533748701974, "grad_norm": 1.027977705001831, "learning_rate": 4.980792075858291e-06, "loss": 0.14621832966804504, "step": 544 }, { "epoch": 0.14148494288681204, "grad_norm": 2.035520315170288, "learning_rate": 4.9806685311088834e-06, "loss": 0.1541852355003357, "step": 545 }, { "epoch": 0.14174454828660435, "grad_norm": 2.5448660850524902, "learning_rate": 4.980544592028354e-06, "loss": 0.14050689339637756, "step": 546 }, { "epoch": 0.14200415368639668, "grad_norm": 0.9709540009498596, "learning_rate": 4.980420258638615e-06, "loss": 0.13829006254673004, "step": 547 }, { "epoch": 0.142263759086189, "grad_norm": 6.413013458251953, "learning_rate": 4.980295530961643e-06, "loss": 0.16098390519618988, "step": 548 }, { "epoch": 0.1425233644859813, "grad_norm": 1.2731047868728638, "learning_rate": 4.980170409019488e-06, "loss": 0.14939603209495544, "step": 549 }, { "epoch": 0.14278296988577363, "grad_norm": 3.0581297874450684, "learning_rate": 4.980044892834269e-06, "loss": 0.1349252462387085, "step": 550 }, { "epoch": 0.14304257528556594, "grad_norm": 3.008437156677246, "learning_rate": 4.979918982428174e-06, "loss": 0.1654202938079834, "step": 551 }, { "epoch": 0.14330218068535824, "grad_norm": 2.1671266555786133, "learning_rate": 4.9797926778234615e-06, "loss": 0.13941916823387146, "step": 552 }, { "epoch": 0.14356178608515058, "grad_norm": 2.2670559883117676, "learning_rate": 4.9796659790424595e-06, "loss": 0.13889601826667786, "step": 553 }, { "epoch": 0.14382139148494288, "grad_norm": 1.581849455833435, "learning_rate": 4.979538886107565e-06, "loss": 0.1367732733488083, "step": 554 }, { "epoch": 0.1440809968847352, "grad_norm": 1.79915189743042, "learning_rate": 4.979411399041244e-06, "loss": 0.1494360864162445, "step": 555 }, { "epoch": 0.14434060228452752, "grad_norm": 1.5659133195877075, "learning_rate": 4.979283517866036e-06, "loss": 0.14819888770580292, "step": 556 }, { "epoch": 0.14460020768431983, "grad_norm": 3.9636197090148926, "learning_rate": 4.979155242604545e-06, "loss": 0.18688413500785828, "step": 557 }, { "epoch": 0.14485981308411214, "grad_norm": 1.7161571979522705, "learning_rate": 4.979026573279447e-06, "loss": 0.16720141470432281, "step": 558 }, { "epoch": 0.14511941848390447, "grad_norm": 2.2171854972839355, "learning_rate": 4.97889750991349e-06, "loss": 0.13933922350406647, "step": 559 }, { "epoch": 0.14537902388369678, "grad_norm": 1.4050716161727905, "learning_rate": 4.978768052529488e-06, "loss": 0.1643497347831726, "step": 560 }, { "epoch": 0.14563862928348908, "grad_norm": 3.2890851497650146, "learning_rate": 4.978638201150327e-06, "loss": 0.15570186078548431, "step": 561 }, { "epoch": 0.14589823468328142, "grad_norm": 4.366910934448242, "learning_rate": 4.9785079557989605e-06, "loss": 0.1589299589395523, "step": 562 }, { "epoch": 0.14615784008307373, "grad_norm": 2.648056983947754, "learning_rate": 4.978377316498414e-06, "loss": 0.16382980346679688, "step": 563 }, { "epoch": 0.14641744548286603, "grad_norm": 1.4933481216430664, "learning_rate": 4.97824628327178e-06, "loss": 0.1370069980621338, "step": 564 }, { "epoch": 0.14667705088265837, "grad_norm": 0.9659265279769897, "learning_rate": 4.978114856142224e-06, "loss": 0.14779877662658691, "step": 565 }, { "epoch": 0.14693665628245067, "grad_norm": 3.0456080436706543, "learning_rate": 4.977983035132977e-06, "loss": 0.16018131375312805, "step": 566 }, { "epoch": 0.14719626168224298, "grad_norm": 2.12538743019104, "learning_rate": 4.977850820267346e-06, "loss": 0.15961076319217682, "step": 567 }, { "epoch": 0.1474558670820353, "grad_norm": 4.216616630554199, "learning_rate": 4.9777182115686995e-06, "loss": 0.14645355939865112, "step": 568 }, { "epoch": 0.14771547248182762, "grad_norm": 3.1143641471862793, "learning_rate": 4.977585209060481e-06, "loss": 0.16237510740756989, "step": 569 }, { "epoch": 0.14797507788161993, "grad_norm": 3.0953121185302734, "learning_rate": 4.977451812766203e-06, "loss": 0.20376275479793549, "step": 570 }, { "epoch": 0.14823468328141226, "grad_norm": 3.2839863300323486, "learning_rate": 4.977318022709445e-06, "loss": 0.19619110226631165, "step": 571 }, { "epoch": 0.14849428868120457, "grad_norm": 3.9088778495788574, "learning_rate": 4.977183838913862e-06, "loss": 0.15064027905464172, "step": 572 }, { "epoch": 0.14875389408099687, "grad_norm": 1.5366551876068115, "learning_rate": 4.97704926140317e-06, "loss": 0.1581503450870514, "step": 573 }, { "epoch": 0.1490134994807892, "grad_norm": 5.164812088012695, "learning_rate": 4.9769142902011615e-06, "loss": 0.17023497819900513, "step": 574 }, { "epoch": 0.14927310488058151, "grad_norm": 4.179591655731201, "learning_rate": 4.9767789253316964e-06, "loss": 0.19291044771671295, "step": 575 }, { "epoch": 0.14953271028037382, "grad_norm": 0.6784599423408508, "learning_rate": 4.976643166818704e-06, "loss": 0.14685983955860138, "step": 576 }, { "epoch": 0.14979231568016615, "grad_norm": 6.054105758666992, "learning_rate": 4.976507014686183e-06, "loss": 0.20440462231636047, "step": 577 }, { "epoch": 0.15005192107995846, "grad_norm": 1.4595752954483032, "learning_rate": 4.976370468958201e-06, "loss": 0.14591798186302185, "step": 578 }, { "epoch": 0.15031152647975077, "grad_norm": 0.955444872379303, "learning_rate": 4.976233529658898e-06, "loss": 0.14023679494857788, "step": 579 }, { "epoch": 0.1505711318795431, "grad_norm": 1.5688115358352661, "learning_rate": 4.9760961968124815e-06, "loss": 0.1552133858203888, "step": 580 }, { "epoch": 0.1508307372793354, "grad_norm": 3.047975778579712, "learning_rate": 4.975958470443228e-06, "loss": 0.14112047851085663, "step": 581 }, { "epoch": 0.15109034267912771, "grad_norm": 1.4476990699768066, "learning_rate": 4.9758203505754845e-06, "loss": 0.16851113736629486, "step": 582 }, { "epoch": 0.15134994807892005, "grad_norm": 2.5400917530059814, "learning_rate": 4.975681837233668e-06, "loss": 0.13776685297489166, "step": 583 }, { "epoch": 0.15160955347871236, "grad_norm": 4.478714942932129, "learning_rate": 4.975542930442263e-06, "loss": 0.14745232462882996, "step": 584 }, { "epoch": 0.15186915887850466, "grad_norm": 2.0910236835479736, "learning_rate": 4.975403630225827e-06, "loss": 0.1844441294670105, "step": 585 }, { "epoch": 0.152128764278297, "grad_norm": 2.323026180267334, "learning_rate": 4.975263936608984e-06, "loss": 0.1265145242214203, "step": 586 }, { "epoch": 0.1523883696780893, "grad_norm": 4.518466949462891, "learning_rate": 4.975123849616429e-06, "loss": 0.14541947841644287, "step": 587 }, { "epoch": 0.1526479750778816, "grad_norm": 2.3052117824554443, "learning_rate": 4.974983369272925e-06, "loss": 0.1359049528837204, "step": 588 }, { "epoch": 0.15290758047767394, "grad_norm": 4.905475616455078, "learning_rate": 4.9748424956033076e-06, "loss": 0.16395410895347595, "step": 589 }, { "epoch": 0.15316718587746625, "grad_norm": 4.931799411773682, "learning_rate": 4.9747012286324784e-06, "loss": 0.13912630081176758, "step": 590 }, { "epoch": 0.15342679127725856, "grad_norm": 2.4927752017974854, "learning_rate": 4.9745595683854115e-06, "loss": 0.1475859433412552, "step": 591 }, { "epoch": 0.1536863966770509, "grad_norm": 4.757664680480957, "learning_rate": 4.974417514887148e-06, "loss": 0.1699366718530655, "step": 592 }, { "epoch": 0.1539460020768432, "grad_norm": 6.130998611450195, "learning_rate": 4.9742750681628e-06, "loss": 0.17286336421966553, "step": 593 }, { "epoch": 0.1542056074766355, "grad_norm": 1.8089358806610107, "learning_rate": 4.97413222823755e-06, "loss": 0.1420123279094696, "step": 594 }, { "epoch": 0.15446521287642784, "grad_norm": 5.027434349060059, "learning_rate": 4.973988995136648e-06, "loss": 0.17128930985927582, "step": 595 }, { "epoch": 0.15472481827622014, "grad_norm": 2.6445248126983643, "learning_rate": 4.9738453688854145e-06, "loss": 0.13744625449180603, "step": 596 }, { "epoch": 0.15498442367601245, "grad_norm": 2.8907954692840576, "learning_rate": 4.973701349509238e-06, "loss": 0.16415587067604065, "step": 597 }, { "epoch": 0.15524402907580478, "grad_norm": 0.7700824737548828, "learning_rate": 4.973556937033579e-06, "loss": 0.12722402811050415, "step": 598 }, { "epoch": 0.1555036344755971, "grad_norm": 3.0540366172790527, "learning_rate": 4.973412131483966e-06, "loss": 0.1435597687959671, "step": 599 }, { "epoch": 0.1557632398753894, "grad_norm": 3.4935624599456787, "learning_rate": 4.973266932885999e-06, "loss": 0.14127638936042786, "step": 600 }, { "epoch": 0.15602284527518173, "grad_norm": 2.1813676357269287, "learning_rate": 4.973121341265343e-06, "loss": 0.1663215458393097, "step": 601 }, { "epoch": 0.15628245067497404, "grad_norm": 1.9661648273468018, "learning_rate": 4.9729753566477376e-06, "loss": 0.1761314421892166, "step": 602 }, { "epoch": 0.15654205607476634, "grad_norm": 2.795750141143799, "learning_rate": 4.972828979058988e-06, "loss": 0.19254279136657715, "step": 603 }, { "epoch": 0.15680166147455868, "grad_norm": 2.688838005065918, "learning_rate": 4.972682208524971e-06, "loss": 0.15442201495170593, "step": 604 }, { "epoch": 0.15706126687435099, "grad_norm": 1.5941044092178345, "learning_rate": 4.972535045071632e-06, "loss": 0.14471103250980377, "step": 605 }, { "epoch": 0.1573208722741433, "grad_norm": 2.432828426361084, "learning_rate": 4.972387488724986e-06, "loss": 0.15004578232765198, "step": 606 }, { "epoch": 0.15758047767393563, "grad_norm": 3.755484104156494, "learning_rate": 4.972239539511119e-06, "loss": 0.16443437337875366, "step": 607 }, { "epoch": 0.15784008307372793, "grad_norm": 1.3072099685668945, "learning_rate": 4.9720911974561835e-06, "loss": 0.1387033760547638, "step": 608 }, { "epoch": 0.15809968847352024, "grad_norm": 4.045420169830322, "learning_rate": 4.971942462586403e-06, "loss": 0.16992074251174927, "step": 609 }, { "epoch": 0.15835929387331257, "grad_norm": 4.015326023101807, "learning_rate": 4.971793334928071e-06, "loss": 0.13960367441177368, "step": 610 }, { "epoch": 0.15861889927310488, "grad_norm": 4.397986888885498, "learning_rate": 4.9716438145075505e-06, "loss": 0.17247521877288818, "step": 611 }, { "epoch": 0.1588785046728972, "grad_norm": 3.492053508758545, "learning_rate": 4.971493901351272e-06, "loss": 0.1603974997997284, "step": 612 }, { "epoch": 0.15913811007268952, "grad_norm": 1.5496989488601685, "learning_rate": 4.971343595485736e-06, "loss": 0.19659247994422913, "step": 613 }, { "epoch": 0.15939771547248183, "grad_norm": 3.425227165222168, "learning_rate": 4.971192896937516e-06, "loss": 0.16509142518043518, "step": 614 }, { "epoch": 0.15965732087227413, "grad_norm": 0.9043706655502319, "learning_rate": 4.97104180573325e-06, "loss": 0.13921719789505005, "step": 615 }, { "epoch": 0.15991692627206647, "grad_norm": 2.8447482585906982, "learning_rate": 4.970890321899647e-06, "loss": 0.1592404991388321, "step": 616 }, { "epoch": 0.16017653167185877, "grad_norm": 3.2333123683929443, "learning_rate": 4.970738445463488e-06, "loss": 0.18257874250411987, "step": 617 }, { "epoch": 0.16043613707165108, "grad_norm": 3.498009443283081, "learning_rate": 4.9705861764516195e-06, "loss": 0.175555020570755, "step": 618 }, { "epoch": 0.16069574247144341, "grad_norm": 2.212834119796753, "learning_rate": 4.97043351489096e-06, "loss": 0.13233807682991028, "step": 619 }, { "epoch": 0.16095534787123572, "grad_norm": 3.2205471992492676, "learning_rate": 4.970280460808495e-06, "loss": 0.1759558767080307, "step": 620 }, { "epoch": 0.16121495327102803, "grad_norm": 6.656906604766846, "learning_rate": 4.970127014231283e-06, "loss": 0.14745071530342102, "step": 621 }, { "epoch": 0.16147455867082036, "grad_norm": 2.7017369270324707, "learning_rate": 4.969973175186449e-06, "loss": 0.1595630794763565, "step": 622 }, { "epoch": 0.16173416407061267, "grad_norm": 5.2982048988342285, "learning_rate": 4.969818943701189e-06, "loss": 0.19442841410636902, "step": 623 }, { "epoch": 0.16199376947040497, "grad_norm": 5.368148326873779, "learning_rate": 4.969664319802766e-06, "loss": 0.14839662611484528, "step": 624 }, { "epoch": 0.1622533748701973, "grad_norm": 3.55342173576355, "learning_rate": 4.9695093035185154e-06, "loss": 0.15016070008277893, "step": 625 }, { "epoch": 0.16251298026998962, "grad_norm": 1.90584397315979, "learning_rate": 4.96935389487584e-06, "loss": 0.1303035318851471, "step": 626 }, { "epoch": 0.16277258566978192, "grad_norm": 3.082765817642212, "learning_rate": 4.969198093902212e-06, "loss": 0.1717076152563095, "step": 627 }, { "epoch": 0.16303219106957426, "grad_norm": 2.5999536514282227, "learning_rate": 4.969041900625174e-06, "loss": 0.1494203805923462, "step": 628 }, { "epoch": 0.16329179646936656, "grad_norm": 1.142769455909729, "learning_rate": 4.968885315072338e-06, "loss": 0.15643173456192017, "step": 629 }, { "epoch": 0.16355140186915887, "grad_norm": 1.630861759185791, "learning_rate": 4.968728337271383e-06, "loss": 0.16831478476524353, "step": 630 }, { "epoch": 0.1638110072689512, "grad_norm": 1.763482928276062, "learning_rate": 4.968570967250061e-06, "loss": 0.13365015387535095, "step": 631 }, { "epoch": 0.1640706126687435, "grad_norm": 1.3328490257263184, "learning_rate": 4.9684132050361896e-06, "loss": 0.14416541159152985, "step": 632 }, { "epoch": 0.16433021806853582, "grad_norm": 2.440983772277832, "learning_rate": 4.968255050657658e-06, "loss": 0.16182447969913483, "step": 633 }, { "epoch": 0.16458982346832815, "grad_norm": 1.5924861431121826, "learning_rate": 4.968096504142426e-06, "loss": 0.16647475957870483, "step": 634 }, { "epoch": 0.16484942886812046, "grad_norm": 4.762734889984131, "learning_rate": 4.96793756551852e-06, "loss": 0.14858025312423706, "step": 635 }, { "epoch": 0.16510903426791276, "grad_norm": 1.939932107925415, "learning_rate": 4.967778234814035e-06, "loss": 0.1422286033630371, "step": 636 }, { "epoch": 0.1653686396677051, "grad_norm": 1.3658967018127441, "learning_rate": 4.96761851205714e-06, "loss": 0.17028354108333588, "step": 637 }, { "epoch": 0.1656282450674974, "grad_norm": 3.734431028366089, "learning_rate": 4.9674583972760676e-06, "loss": 0.17426744103431702, "step": 638 }, { "epoch": 0.1658878504672897, "grad_norm": 2.310853958129883, "learning_rate": 4.967297890499124e-06, "loss": 0.19866672158241272, "step": 639 }, { "epoch": 0.16614745586708204, "grad_norm": 2.0920920372009277, "learning_rate": 4.967136991754683e-06, "loss": 0.12773120403289795, "step": 640 }, { "epoch": 0.16640706126687435, "grad_norm": 3.9262888431549072, "learning_rate": 4.966975701071188e-06, "loss": 0.13873258233070374, "step": 641 }, { "epoch": 0.16666666666666666, "grad_norm": 1.9454033374786377, "learning_rate": 4.96681401847715e-06, "loss": 0.15986597537994385, "step": 642 }, { "epoch": 0.166926272066459, "grad_norm": 1.7915219068527222, "learning_rate": 4.966651944001153e-06, "loss": 0.18865400552749634, "step": 643 }, { "epoch": 0.1671858774662513, "grad_norm": 5.264756202697754, "learning_rate": 4.966489477671846e-06, "loss": 0.1434047967195511, "step": 644 }, { "epoch": 0.1674454828660436, "grad_norm": 4.742119312286377, "learning_rate": 4.966326619517951e-06, "loss": 0.1649838089942932, "step": 645 }, { "epoch": 0.16770508826583594, "grad_norm": 1.4264631271362305, "learning_rate": 4.966163369568256e-06, "loss": 0.12892472743988037, "step": 646 }, { "epoch": 0.16796469366562825, "grad_norm": 6.642126083374023, "learning_rate": 4.965999727851621e-06, "loss": 0.19104740023612976, "step": 647 }, { "epoch": 0.16822429906542055, "grad_norm": 0.5464682579040527, "learning_rate": 4.965835694396973e-06, "loss": 0.14164039492607117, "step": 648 }, { "epoch": 0.16848390446521289, "grad_norm": 0.9597867727279663, "learning_rate": 4.965671269233312e-06, "loss": 0.16999022662639618, "step": 649 }, { "epoch": 0.1687435098650052, "grad_norm": 0.924707293510437, "learning_rate": 4.9655064523897015e-06, "loss": 0.15479272603988647, "step": 650 }, { "epoch": 0.1690031152647975, "grad_norm": 2.9756436347961426, "learning_rate": 4.965341243895279e-06, "loss": 0.2042239010334015, "step": 651 }, { "epoch": 0.16926272066458983, "grad_norm": 1.5740708112716675, "learning_rate": 4.9651756437792485e-06, "loss": 0.17972677946090698, "step": 652 }, { "epoch": 0.16952232606438214, "grad_norm": 1.2665008306503296, "learning_rate": 4.965009652070885e-06, "loss": 0.18270611763000488, "step": 653 }, { "epoch": 0.16978193146417445, "grad_norm": 0.6444963812828064, "learning_rate": 4.964843268799533e-06, "loss": 0.13824349641799927, "step": 654 }, { "epoch": 0.17004153686396678, "grad_norm": 2.5097715854644775, "learning_rate": 4.964676493994604e-06, "loss": 0.15573611855506897, "step": 655 }, { "epoch": 0.1703011422637591, "grad_norm": 2.147071123123169, "learning_rate": 4.964509327685579e-06, "loss": 0.12951400876045227, "step": 656 }, { "epoch": 0.1705607476635514, "grad_norm": 1.2416553497314453, "learning_rate": 4.964341769902012e-06, "loss": 0.17893768846988678, "step": 657 }, { "epoch": 0.17082035306334373, "grad_norm": 0.9456914067268372, "learning_rate": 4.964173820673521e-06, "loss": 0.12806962430477142, "step": 658 }, { "epoch": 0.17107995846313603, "grad_norm": 1.6513187885284424, "learning_rate": 4.964005480029797e-06, "loss": 0.1340130865573883, "step": 659 }, { "epoch": 0.17133956386292834, "grad_norm": 1.6864368915557861, "learning_rate": 4.963836748000597e-06, "loss": 0.1453533172607422, "step": 660 }, { "epoch": 0.17159916926272067, "grad_norm": 1.8603112697601318, "learning_rate": 4.963667624615751e-06, "loss": 0.1525903344154358, "step": 661 }, { "epoch": 0.17185877466251298, "grad_norm": 1.0712732076644897, "learning_rate": 4.9634981099051545e-06, "loss": 0.14858439564704895, "step": 662 }, { "epoch": 0.1721183800623053, "grad_norm": 3.6361632347106934, "learning_rate": 4.963328203898775e-06, "loss": 0.16014787554740906, "step": 663 }, { "epoch": 0.17237798546209762, "grad_norm": 3.1102547645568848, "learning_rate": 4.963157906626648e-06, "loss": 0.1675320416688919, "step": 664 }, { "epoch": 0.17263759086188993, "grad_norm": 2.1343917846679688, "learning_rate": 4.9629872181188774e-06, "loss": 0.15745848417282104, "step": 665 }, { "epoch": 0.17289719626168223, "grad_norm": 1.7633228302001953, "learning_rate": 4.962816138405637e-06, "loss": 0.15429875254631042, "step": 666 }, { "epoch": 0.17315680166147457, "grad_norm": 1.8543604612350464, "learning_rate": 4.9626446675171704e-06, "loss": 0.13552922010421753, "step": 667 }, { "epoch": 0.17341640706126688, "grad_norm": 3.9582462310791016, "learning_rate": 4.962472805483789e-06, "loss": 0.14810281991958618, "step": 668 }, { "epoch": 0.17367601246105918, "grad_norm": 1.3737919330596924, "learning_rate": 4.962300552335875e-06, "loss": 0.15984728932380676, "step": 669 }, { "epoch": 0.17393561786085152, "grad_norm": 4.697542667388916, "learning_rate": 4.962127908103878e-06, "loss": 0.17993132770061493, "step": 670 }, { "epoch": 0.17419522326064382, "grad_norm": 2.6036016941070557, "learning_rate": 4.961954872818318e-06, "loss": 0.16564688086509705, "step": 671 }, { "epoch": 0.17445482866043613, "grad_norm": 2.125357151031494, "learning_rate": 4.961781446509784e-06, "loss": 0.1890338659286499, "step": 672 }, { "epoch": 0.17471443406022846, "grad_norm": 0.6977387070655823, "learning_rate": 4.961607629208933e-06, "loss": 0.12493821233510971, "step": 673 }, { "epoch": 0.17497403946002077, "grad_norm": 3.2687573432922363, "learning_rate": 4.961433420946492e-06, "loss": 0.19068458676338196, "step": 674 }, { "epoch": 0.17523364485981308, "grad_norm": 2.8404998779296875, "learning_rate": 4.961258821753257e-06, "loss": 0.1412416249513626, "step": 675 }, { "epoch": 0.1754932502596054, "grad_norm": 3.4979522228240967, "learning_rate": 4.961083831660095e-06, "loss": 0.16022264957427979, "step": 676 }, { "epoch": 0.17575285565939772, "grad_norm": 3.7111127376556396, "learning_rate": 4.960908450697937e-06, "loss": 0.18936017155647278, "step": 677 }, { "epoch": 0.17601246105919002, "grad_norm": 2.4472579956054688, "learning_rate": 4.960732678897788e-06, "loss": 0.16227734088897705, "step": 678 }, { "epoch": 0.17627206645898236, "grad_norm": 7.6095380783081055, "learning_rate": 4.960556516290721e-06, "loss": 0.18758484721183777, "step": 679 }, { "epoch": 0.17653167185877466, "grad_norm": 1.8012237548828125, "learning_rate": 4.960379962907878e-06, "loss": 0.1512298732995987, "step": 680 }, { "epoch": 0.17679127725856697, "grad_norm": 2.5174663066864014, "learning_rate": 4.960203018780467e-06, "loss": 0.15519146621227264, "step": 681 }, { "epoch": 0.1770508826583593, "grad_norm": 6.0170674324035645, "learning_rate": 4.9600256839397695e-06, "loss": 0.18197883665561676, "step": 682 }, { "epoch": 0.1773104880581516, "grad_norm": 3.49804425239563, "learning_rate": 4.959847958417134e-06, "loss": 0.16996848583221436, "step": 683 }, { "epoch": 0.17757009345794392, "grad_norm": 4.097604274749756, "learning_rate": 4.959669842243979e-06, "loss": 0.1772782951593399, "step": 684 }, { "epoch": 0.17782969885773625, "grad_norm": 5.16591215133667, "learning_rate": 4.95949133545179e-06, "loss": 0.21095792949199677, "step": 685 }, { "epoch": 0.17808930425752856, "grad_norm": 4.162191390991211, "learning_rate": 4.959312438072123e-06, "loss": 0.17167843878269196, "step": 686 }, { "epoch": 0.17834890965732086, "grad_norm": 4.758688926696777, "learning_rate": 4.959133150136605e-06, "loss": 0.19457197189331055, "step": 687 }, { "epoch": 0.1786085150571132, "grad_norm": 4.722886085510254, "learning_rate": 4.958953471676927e-06, "loss": 0.16260871291160583, "step": 688 }, { "epoch": 0.1788681204569055, "grad_norm": 5.771823406219482, "learning_rate": 4.958773402724854e-06, "loss": 0.1666233241558075, "step": 689 }, { "epoch": 0.1791277258566978, "grad_norm": 5.729877471923828, "learning_rate": 4.958592943312218e-06, "loss": 0.1565927267074585, "step": 690 }, { "epoch": 0.17938733125649015, "grad_norm": 6.4813618659973145, "learning_rate": 4.958412093470919e-06, "loss": 0.196508526802063, "step": 691 }, { "epoch": 0.17964693665628245, "grad_norm": 2.405886173248291, "learning_rate": 4.958230853232928e-06, "loss": 0.17693713307380676, "step": 692 }, { "epoch": 0.17990654205607476, "grad_norm": 4.43494987487793, "learning_rate": 4.958049222630285e-06, "loss": 0.1884881854057312, "step": 693 }, { "epoch": 0.1801661474558671, "grad_norm": 11.46435832977295, "learning_rate": 4.957867201695096e-06, "loss": 0.20219624042510986, "step": 694 }, { "epoch": 0.1804257528556594, "grad_norm": 4.740798473358154, "learning_rate": 4.957684790459539e-06, "loss": 0.15706360340118408, "step": 695 }, { "epoch": 0.1806853582554517, "grad_norm": 1.739404320716858, "learning_rate": 4.957501988955861e-06, "loss": 0.18462762236595154, "step": 696 }, { "epoch": 0.18094496365524404, "grad_norm": 2.450059175491333, "learning_rate": 4.9573187972163765e-06, "loss": 0.1449810117483139, "step": 697 }, { "epoch": 0.18120456905503635, "grad_norm": 5.1547956466674805, "learning_rate": 4.9571352152734696e-06, "loss": 0.16985860466957092, "step": 698 }, { "epoch": 0.18146417445482865, "grad_norm": 4.633848667144775, "learning_rate": 4.956951243159593e-06, "loss": 0.1565624475479126, "step": 699 }, { "epoch": 0.181723779854621, "grad_norm": 4.121429920196533, "learning_rate": 4.956766880907269e-06, "loss": 0.14641474187374115, "step": 700 }, { "epoch": 0.1819833852544133, "grad_norm": 0.9638670682907104, "learning_rate": 4.956582128549089e-06, "loss": 0.1339174062013626, "step": 701 }, { "epoch": 0.1822429906542056, "grad_norm": 2.246594190597534, "learning_rate": 4.9563969861177126e-06, "loss": 0.17097648978233337, "step": 702 }, { "epoch": 0.18250259605399793, "grad_norm": 2.6169331073760986, "learning_rate": 4.9562114536458686e-06, "loss": 0.185060054063797, "step": 703 }, { "epoch": 0.18276220145379024, "grad_norm": 4.993634223937988, "learning_rate": 4.956025531166356e-06, "loss": 0.15587779879570007, "step": 704 }, { "epoch": 0.18302180685358255, "grad_norm": 1.4395376443862915, "learning_rate": 4.955839218712039e-06, "loss": 0.1613866090774536, "step": 705 }, { "epoch": 0.18328141225337488, "grad_norm": 2.0702309608459473, "learning_rate": 4.955652516315856e-06, "loss": 0.17242366075515747, "step": 706 }, { "epoch": 0.1835410176531672, "grad_norm": 1.2356623411178589, "learning_rate": 4.955465424010812e-06, "loss": 0.14399024844169617, "step": 707 }, { "epoch": 0.1838006230529595, "grad_norm": 2.8500096797943115, "learning_rate": 4.955277941829978e-06, "loss": 0.13614413142204285, "step": 708 }, { "epoch": 0.18406022845275183, "grad_norm": 4.79220724105835, "learning_rate": 4.955090069806499e-06, "loss": 0.17898398637771606, "step": 709 }, { "epoch": 0.18431983385254413, "grad_norm": 3.4305357933044434, "learning_rate": 4.954901807973586e-06, "loss": 0.12754201889038086, "step": 710 }, { "epoch": 0.18457943925233644, "grad_norm": 1.1793464422225952, "learning_rate": 4.954713156364517e-06, "loss": 0.13638617098331451, "step": 711 }, { "epoch": 0.18483904465212878, "grad_norm": 3.9561727046966553, "learning_rate": 4.954524115012645e-06, "loss": 0.14572401344776154, "step": 712 }, { "epoch": 0.18509865005192108, "grad_norm": 4.287055492401123, "learning_rate": 4.9543346839513856e-06, "loss": 0.14293047785758972, "step": 713 }, { "epoch": 0.1853582554517134, "grad_norm": 2.1153714656829834, "learning_rate": 4.954144863214226e-06, "loss": 0.14896288514137268, "step": 714 }, { "epoch": 0.18561786085150572, "grad_norm": 1.5844019651412964, "learning_rate": 4.953954652834724e-06, "loss": 0.15342575311660767, "step": 715 }, { "epoch": 0.18587746625129803, "grad_norm": 2.7455010414123535, "learning_rate": 4.953764052846502e-06, "loss": 0.15949508547782898, "step": 716 }, { "epoch": 0.18613707165109034, "grad_norm": 4.159176826477051, "learning_rate": 4.953573063283257e-06, "loss": 0.1455150991678238, "step": 717 }, { "epoch": 0.18639667705088267, "grad_norm": 2.115201473236084, "learning_rate": 4.953381684178748e-06, "loss": 0.1328771859407425, "step": 718 }, { "epoch": 0.18665628245067498, "grad_norm": 2.1978845596313477, "learning_rate": 4.9531899155668076e-06, "loss": 0.15137247741222382, "step": 719 }, { "epoch": 0.18691588785046728, "grad_norm": 2.5509495735168457, "learning_rate": 4.952997757481338e-06, "loss": 0.164193794131279, "step": 720 }, { "epoch": 0.18717549325025962, "grad_norm": 1.5931949615478516, "learning_rate": 4.952805209956306e-06, "loss": 0.1887739598751068, "step": 721 }, { "epoch": 0.18743509865005192, "grad_norm": 3.7209107875823975, "learning_rate": 4.95261227302575e-06, "loss": 0.20024161040782928, "step": 722 }, { "epoch": 0.18769470404984423, "grad_norm": 3.1177523136138916, "learning_rate": 4.952418946723778e-06, "loss": 0.16450265049934387, "step": 723 }, { "epoch": 0.18795430944963656, "grad_norm": 1.0524927377700806, "learning_rate": 4.952225231084565e-06, "loss": 0.15098100900650024, "step": 724 }, { "epoch": 0.18821391484942887, "grad_norm": 4.196286678314209, "learning_rate": 4.9520311261423556e-06, "loss": 0.16975346207618713, "step": 725 }, { "epoch": 0.18847352024922118, "grad_norm": 3.9763245582580566, "learning_rate": 4.951836631931461e-06, "loss": 0.1553436517715454, "step": 726 }, { "epoch": 0.1887331256490135, "grad_norm": 1.7729949951171875, "learning_rate": 4.951641748486268e-06, "loss": 0.13454832136631012, "step": 727 }, { "epoch": 0.18899273104880582, "grad_norm": 4.0420122146606445, "learning_rate": 4.951446475841223e-06, "loss": 0.1743011474609375, "step": 728 }, { "epoch": 0.18925233644859812, "grad_norm": 4.090044021606445, "learning_rate": 4.951250814030847e-06, "loss": 0.18805290758609772, "step": 729 }, { "epoch": 0.18951194184839046, "grad_norm": 2.2784101963043213, "learning_rate": 4.951054763089731e-06, "loss": 0.14412827789783478, "step": 730 }, { "epoch": 0.18977154724818276, "grad_norm": 2.905754566192627, "learning_rate": 4.95085832305253e-06, "loss": 0.19326438009738922, "step": 731 }, { "epoch": 0.19003115264797507, "grad_norm": 3.5752079486846924, "learning_rate": 4.950661493953969e-06, "loss": 0.18097391724586487, "step": 732 }, { "epoch": 0.1902907580477674, "grad_norm": 2.9049172401428223, "learning_rate": 4.950464275828845e-06, "loss": 0.13498471677303314, "step": 733 }, { "epoch": 0.1905503634475597, "grad_norm": 2.5811777114868164, "learning_rate": 4.9502666687120215e-06, "loss": 0.14239293336868286, "step": 734 }, { "epoch": 0.19080996884735202, "grad_norm": 1.9950052499771118, "learning_rate": 4.95006867263843e-06, "loss": 0.13774533569812775, "step": 735 }, { "epoch": 0.19106957424714435, "grad_norm": 3.5152084827423096, "learning_rate": 4.949870287643072e-06, "loss": 0.15667101740837097, "step": 736 }, { "epoch": 0.19132917964693666, "grad_norm": 2.306332588195801, "learning_rate": 4.949671513761018e-06, "loss": 0.1513390690088272, "step": 737 }, { "epoch": 0.19158878504672897, "grad_norm": 1.7253369092941284, "learning_rate": 4.949472351027405e-06, "loss": 0.16200250387191772, "step": 738 }, { "epoch": 0.1918483904465213, "grad_norm": 1.8478381633758545, "learning_rate": 4.949272799477442e-06, "loss": 0.15011471509933472, "step": 739 }, { "epoch": 0.1921079958463136, "grad_norm": 2.0375816822052, "learning_rate": 4.949072859146406e-06, "loss": 0.14094430208206177, "step": 740 }, { "epoch": 0.1923676012461059, "grad_norm": 2.9060585498809814, "learning_rate": 4.948872530069639e-06, "loss": 0.15726956725120544, "step": 741 }, { "epoch": 0.19262720664589825, "grad_norm": 1.7265502214431763, "learning_rate": 4.948671812282557e-06, "loss": 0.15707966685295105, "step": 742 }, { "epoch": 0.19288681204569055, "grad_norm": 2.6578004360198975, "learning_rate": 4.948470705820642e-06, "loss": 0.15178431570529938, "step": 743 }, { "epoch": 0.19314641744548286, "grad_norm": 1.1348544359207153, "learning_rate": 4.948269210719443e-06, "loss": 0.1465236246585846, "step": 744 }, { "epoch": 0.1934060228452752, "grad_norm": 2.782449960708618, "learning_rate": 4.948067327014583e-06, "loss": 0.13589602708816528, "step": 745 }, { "epoch": 0.1936656282450675, "grad_norm": 10.055377006530762, "learning_rate": 4.947865054741748e-06, "loss": 0.16256193816661835, "step": 746 }, { "epoch": 0.1939252336448598, "grad_norm": 1.3491908311843872, "learning_rate": 4.947662393936695e-06, "loss": 0.19980649650096893, "step": 747 }, { "epoch": 0.19418483904465214, "grad_norm": 5.729310989379883, "learning_rate": 4.947459344635251e-06, "loss": 0.16482749581336975, "step": 748 }, { "epoch": 0.19444444444444445, "grad_norm": 6.013210296630859, "learning_rate": 4.9472559068733104e-06, "loss": 0.1768849790096283, "step": 749 }, { "epoch": 0.19470404984423675, "grad_norm": 5.538680076599121, "learning_rate": 4.9470520806868356e-06, "loss": 0.15832722187042236, "step": 750 }, { "epoch": 0.1949636552440291, "grad_norm": 1.3268672227859497, "learning_rate": 4.946847866111859e-06, "loss": 0.14636918902397156, "step": 751 }, { "epoch": 0.1952232606438214, "grad_norm": 1.2359933853149414, "learning_rate": 4.9466432631844805e-06, "loss": 0.13430097699165344, "step": 752 }, { "epoch": 0.1954828660436137, "grad_norm": 5.895421504974365, "learning_rate": 4.946438271940869e-06, "loss": 0.1600664108991623, "step": 753 }, { "epoch": 0.19574247144340604, "grad_norm": 7.384987831115723, "learning_rate": 4.946232892417264e-06, "loss": 0.1505577713251114, "step": 754 }, { "epoch": 0.19600207684319834, "grad_norm": 5.4765167236328125, "learning_rate": 4.94602712464997e-06, "loss": 0.20106002688407898, "step": 755 }, { "epoch": 0.19626168224299065, "grad_norm": 3.2967705726623535, "learning_rate": 4.945820968675363e-06, "loss": 0.16605913639068604, "step": 756 }, { "epoch": 0.19652128764278298, "grad_norm": 5.763364315032959, "learning_rate": 4.945614424529886e-06, "loss": 0.15199044346809387, "step": 757 }, { "epoch": 0.1967808930425753, "grad_norm": 6.595897197723389, "learning_rate": 4.945407492250052e-06, "loss": 0.1997832953929901, "step": 758 }, { "epoch": 0.1970404984423676, "grad_norm": 4.341255187988281, "learning_rate": 4.945200171872442e-06, "loss": 0.1285090148448944, "step": 759 }, { "epoch": 0.19730010384215993, "grad_norm": 2.4566071033477783, "learning_rate": 4.944992463433704e-06, "loss": 0.14415864646434784, "step": 760 }, { "epoch": 0.19755970924195224, "grad_norm": 4.3827595710754395, "learning_rate": 4.944784366970558e-06, "loss": 0.14877426624298096, "step": 761 }, { "epoch": 0.19781931464174454, "grad_norm": 4.491208553314209, "learning_rate": 4.944575882519789e-06, "loss": 0.14183355867862701, "step": 762 }, { "epoch": 0.19807892004153688, "grad_norm": 5.5536956787109375, "learning_rate": 4.944367010118254e-06, "loss": 0.18165937066078186, "step": 763 }, { "epoch": 0.19833852544132918, "grad_norm": 2.603037118911743, "learning_rate": 4.944157749802876e-06, "loss": 0.16824065148830414, "step": 764 }, { "epoch": 0.1985981308411215, "grad_norm": 2.762375831604004, "learning_rate": 4.9439481016106465e-06, "loss": 0.16536185145378113, "step": 765 }, { "epoch": 0.19885773624091382, "grad_norm": 4.5985636711120605, "learning_rate": 4.943738065578628e-06, "loss": 0.21118640899658203, "step": 766 }, { "epoch": 0.19911734164070613, "grad_norm": 5.802451133728027, "learning_rate": 4.943527641743949e-06, "loss": 0.16924425959587097, "step": 767 }, { "epoch": 0.19937694704049844, "grad_norm": 3.6295971870422363, "learning_rate": 4.9433168301438084e-06, "loss": 0.15817837417125702, "step": 768 }, { "epoch": 0.19963655244029077, "grad_norm": 2.660010814666748, "learning_rate": 4.943105630815472e-06, "loss": 0.17399096488952637, "step": 769 }, { "epoch": 0.19989615784008308, "grad_norm": 4.252667427062988, "learning_rate": 4.942894043796276e-06, "loss": 0.15809698402881622, "step": 770 }, { "epoch": 0.20015576323987538, "grad_norm": 1.5269112586975098, "learning_rate": 4.942682069123623e-06, "loss": 0.1324625015258789, "step": 771 }, { "epoch": 0.20041536863966772, "grad_norm": 1.9175126552581787, "learning_rate": 4.942469706834987e-06, "loss": 0.1310434490442276, "step": 772 }, { "epoch": 0.20067497403946002, "grad_norm": 5.085960388183594, "learning_rate": 4.942256956967907e-06, "loss": 0.17377620935440063, "step": 773 }, { "epoch": 0.20093457943925233, "grad_norm": 6.399044513702393, "learning_rate": 4.942043819559993e-06, "loss": 0.17845852673053741, "step": 774 }, { "epoch": 0.20119418483904467, "grad_norm": 3.926325559616089, "learning_rate": 4.941830294648924e-06, "loss": 0.14725017547607422, "step": 775 }, { "epoch": 0.20145379023883697, "grad_norm": 3.9251203536987305, "learning_rate": 4.941616382272444e-06, "loss": 0.1422165036201477, "step": 776 }, { "epoch": 0.20171339563862928, "grad_norm": 3.2516562938690186, "learning_rate": 4.94140208246837e-06, "loss": 0.15738463401794434, "step": 777 }, { "epoch": 0.2019730010384216, "grad_norm": 3.1519386768341064, "learning_rate": 4.941187395274584e-06, "loss": 0.15681761503219604, "step": 778 }, { "epoch": 0.20223260643821392, "grad_norm": 3.257995128631592, "learning_rate": 4.940972320729036e-06, "loss": 0.1391572803258896, "step": 779 }, { "epoch": 0.20249221183800623, "grad_norm": 1.72589910030365, "learning_rate": 4.940756858869751e-06, "loss": 0.13426624238491058, "step": 780 }, { "epoch": 0.20275181723779856, "grad_norm": 4.021653175354004, "learning_rate": 4.940541009734814e-06, "loss": 0.187355637550354, "step": 781 }, { "epoch": 0.20301142263759087, "grad_norm": 4.592984199523926, "learning_rate": 4.940324773362383e-06, "loss": 0.1552489995956421, "step": 782 }, { "epoch": 0.20327102803738317, "grad_norm": 4.963769912719727, "learning_rate": 4.940108149790684e-06, "loss": 0.1768808364868164, "step": 783 }, { "epoch": 0.2035306334371755, "grad_norm": 1.1916147470474243, "learning_rate": 4.939891139058011e-06, "loss": 0.13103964924812317, "step": 784 }, { "epoch": 0.2037902388369678, "grad_norm": 2.0636675357818604, "learning_rate": 4.939673741202726e-06, "loss": 0.16796153783798218, "step": 785 }, { "epoch": 0.20404984423676012, "grad_norm": 3.914628028869629, "learning_rate": 4.939455956263262e-06, "loss": 0.12998783588409424, "step": 786 }, { "epoch": 0.20430944963655245, "grad_norm": 2.035728693008423, "learning_rate": 4.9392377842781156e-06, "loss": 0.15950550138950348, "step": 787 }, { "epoch": 0.20456905503634476, "grad_norm": 2.2426340579986572, "learning_rate": 4.939019225285856e-06, "loss": 0.15785431861877441, "step": 788 }, { "epoch": 0.20482866043613707, "grad_norm": 1.477671504020691, "learning_rate": 4.938800279325119e-06, "loss": 0.14006581902503967, "step": 789 }, { "epoch": 0.2050882658359294, "grad_norm": 1.0263320207595825, "learning_rate": 4.93858094643461e-06, "loss": 0.14119061827659607, "step": 790 }, { "epoch": 0.2053478712357217, "grad_norm": 0.9054765701293945, "learning_rate": 4.938361226653101e-06, "loss": 0.14179913699626923, "step": 791 }, { "epoch": 0.205607476635514, "grad_norm": 3.653961181640625, "learning_rate": 4.938141120019435e-06, "loss": 0.15734699368476868, "step": 792 }, { "epoch": 0.20586708203530635, "grad_norm": 2.386171579360962, "learning_rate": 4.9379206265725196e-06, "loss": 0.13512685894966125, "step": 793 }, { "epoch": 0.20612668743509865, "grad_norm": 2.7603976726531982, "learning_rate": 4.937699746351334e-06, "loss": 0.1334010660648346, "step": 794 }, { "epoch": 0.20638629283489096, "grad_norm": 2.7356669902801514, "learning_rate": 4.937478479394925e-06, "loss": 0.1532939076423645, "step": 795 }, { "epoch": 0.2066458982346833, "grad_norm": 3.4749488830566406, "learning_rate": 4.937256825742409e-06, "loss": 0.14083167910575867, "step": 796 }, { "epoch": 0.2069055036344756, "grad_norm": 3.0633370876312256, "learning_rate": 4.937034785432967e-06, "loss": 0.15827661752700806, "step": 797 }, { "epoch": 0.2071651090342679, "grad_norm": 3.037431240081787, "learning_rate": 4.936812358505851e-06, "loss": 0.14824652671813965, "step": 798 }, { "epoch": 0.20742471443406024, "grad_norm": 2.216768503189087, "learning_rate": 4.936589545000381e-06, "loss": 0.14746209979057312, "step": 799 }, { "epoch": 0.20768431983385255, "grad_norm": 2.949059247970581, "learning_rate": 4.936366344955947e-06, "loss": 0.16822436451911926, "step": 800 }, { "epoch": 0.20794392523364486, "grad_norm": 2.1467225551605225, "learning_rate": 4.936142758412004e-06, "loss": 0.1319306492805481, "step": 801 }, { "epoch": 0.2082035306334372, "grad_norm": 1.0873892307281494, "learning_rate": 4.935918785408076e-06, "loss": 0.15408186614513397, "step": 802 }, { "epoch": 0.2084631360332295, "grad_norm": 2.141500473022461, "learning_rate": 4.935694425983759e-06, "loss": 0.18059980869293213, "step": 803 }, { "epoch": 0.2087227414330218, "grad_norm": 3.568889856338501, "learning_rate": 4.935469680178713e-06, "loss": 0.2059852033853531, "step": 804 }, { "epoch": 0.2089823468328141, "grad_norm": 3.2200729846954346, "learning_rate": 4.9352445480326675e-06, "loss": 0.16590428352355957, "step": 805 }, { "epoch": 0.20924195223260644, "grad_norm": 2.5579166412353516, "learning_rate": 4.935019029585421e-06, "loss": 0.16735993325710297, "step": 806 }, { "epoch": 0.20950155763239875, "grad_norm": 4.513092994689941, "learning_rate": 4.934793124876842e-06, "loss": 0.1830074042081833, "step": 807 }, { "epoch": 0.20976116303219106, "grad_norm": 0.6792083978652954, "learning_rate": 4.934566833946863e-06, "loss": 0.13734349608421326, "step": 808 }, { "epoch": 0.2100207684319834, "grad_norm": 2.766061782836914, "learning_rate": 4.934340156835488e-06, "loss": 0.14618036150932312, "step": 809 }, { "epoch": 0.2102803738317757, "grad_norm": 4.117685794830322, "learning_rate": 4.934113093582787e-06, "loss": 0.1565437763929367, "step": 810 }, { "epoch": 0.210539979231568, "grad_norm": 0.9628108143806458, "learning_rate": 4.933885644228902e-06, "loss": 0.11678379029035568, "step": 811 }, { "epoch": 0.21079958463136034, "grad_norm": 2.4552359580993652, "learning_rate": 4.933657808814038e-06, "loss": 0.16932234168052673, "step": 812 }, { "epoch": 0.21105919003115264, "grad_norm": 3.7987799644470215, "learning_rate": 4.933429587378473e-06, "loss": 0.14328241348266602, "step": 813 }, { "epoch": 0.21131879543094495, "grad_norm": 1.1233631372451782, "learning_rate": 4.933200979962551e-06, "loss": 0.1340479999780655, "step": 814 }, { "epoch": 0.21157840083073728, "grad_norm": 0.9113669991493225, "learning_rate": 4.932971986606685e-06, "loss": 0.1265561282634735, "step": 815 }, { "epoch": 0.2118380062305296, "grad_norm": 0.7615392804145813, "learning_rate": 4.9327426073513555e-06, "loss": 0.13098861277103424, "step": 816 }, { "epoch": 0.2120976116303219, "grad_norm": 3.246051549911499, "learning_rate": 4.9325128422371095e-06, "loss": 0.12491073459386826, "step": 817 }, { "epoch": 0.21235721703011423, "grad_norm": 1.7256903648376465, "learning_rate": 4.932282691304567e-06, "loss": 0.12200640141963959, "step": 818 }, { "epoch": 0.21261682242990654, "grad_norm": 0.9348278045654297, "learning_rate": 4.932052154594413e-06, "loss": 0.1715504229068756, "step": 819 }, { "epoch": 0.21287642782969884, "grad_norm": 2.048565149307251, "learning_rate": 4.931821232147399e-06, "loss": 0.17338606715202332, "step": 820 }, { "epoch": 0.21313603322949118, "grad_norm": 1.932652235031128, "learning_rate": 4.931589924004349e-06, "loss": 0.1754382848739624, "step": 821 }, { "epoch": 0.21339563862928349, "grad_norm": 4.073315620422363, "learning_rate": 4.931358230206152e-06, "loss": 0.12805913388729095, "step": 822 }, { "epoch": 0.2136552440290758, "grad_norm": 1.3369531631469727, "learning_rate": 4.931126150793767e-06, "loss": 0.16010236740112305, "step": 823 }, { "epoch": 0.21391484942886813, "grad_norm": 0.8846494555473328, "learning_rate": 4.930893685808219e-06, "loss": 0.1318952739238739, "step": 824 }, { "epoch": 0.21417445482866043, "grad_norm": 1.488091230392456, "learning_rate": 4.930660835290603e-06, "loss": 0.14413076639175415, "step": 825 }, { "epoch": 0.21443406022845274, "grad_norm": 3.661111831665039, "learning_rate": 4.930427599282082e-06, "loss": 0.17507430911064148, "step": 826 }, { "epoch": 0.21469366562824507, "grad_norm": 1.4145803451538086, "learning_rate": 4.930193977823887e-06, "loss": 0.1366267204284668, "step": 827 }, { "epoch": 0.21495327102803738, "grad_norm": 0.7531731724739075, "learning_rate": 4.929959970957316e-06, "loss": 0.15719571709632874, "step": 828 }, { "epoch": 0.21521287642782969, "grad_norm": 2.9713339805603027, "learning_rate": 4.929725578723738e-06, "loss": 0.15650393068790436, "step": 829 }, { "epoch": 0.21547248182762202, "grad_norm": 0.993410587310791, "learning_rate": 4.929490801164585e-06, "loss": 0.15214085578918457, "step": 830 }, { "epoch": 0.21573208722741433, "grad_norm": 1.753170371055603, "learning_rate": 4.929255638321363e-06, "loss": 0.13656944036483765, "step": 831 }, { "epoch": 0.21599169262720663, "grad_norm": 0.8686413168907166, "learning_rate": 4.9290200902356424e-06, "loss": 0.13028010725975037, "step": 832 }, { "epoch": 0.21625129802699897, "grad_norm": 2.268303394317627, "learning_rate": 4.928784156949062e-06, "loss": 0.1405775249004364, "step": 833 }, { "epoch": 0.21651090342679127, "grad_norm": 1.3687409162521362, "learning_rate": 4.928547838503331e-06, "loss": 0.1609407365322113, "step": 834 }, { "epoch": 0.21677050882658358, "grad_norm": 0.957895040512085, "learning_rate": 4.928311134940224e-06, "loss": 0.1326204538345337, "step": 835 }, { "epoch": 0.21703011422637591, "grad_norm": 3.410414695739746, "learning_rate": 4.928074046301586e-06, "loss": 0.170960932970047, "step": 836 }, { "epoch": 0.21728971962616822, "grad_norm": 1.1303640604019165, "learning_rate": 4.9278365726293274e-06, "loss": 0.15914657711982727, "step": 837 }, { "epoch": 0.21754932502596053, "grad_norm": 1.1742103099822998, "learning_rate": 4.927598713965428e-06, "loss": 0.12458619475364685, "step": 838 }, { "epoch": 0.21780893042575286, "grad_norm": 1.5989551544189453, "learning_rate": 4.927360470351937e-06, "loss": 0.16476449370384216, "step": 839 }, { "epoch": 0.21806853582554517, "grad_norm": 1.8344848155975342, "learning_rate": 4.927121841830969e-06, "loss": 0.1299586445093155, "step": 840 }, { "epoch": 0.21832814122533747, "grad_norm": 1.6868021488189697, "learning_rate": 4.926882828444709e-06, "loss": 0.12542276084423065, "step": 841 }, { "epoch": 0.2185877466251298, "grad_norm": 0.8147488832473755, "learning_rate": 4.92664343023541e-06, "loss": 0.13611339032649994, "step": 842 }, { "epoch": 0.21884735202492211, "grad_norm": 2.206056594848633, "learning_rate": 4.926403647245391e-06, "loss": 0.15019986033439636, "step": 843 }, { "epoch": 0.21910695742471442, "grad_norm": 3.785151481628418, "learning_rate": 4.926163479517039e-06, "loss": 0.14262089133262634, "step": 844 }, { "epoch": 0.21936656282450676, "grad_norm": 3.1184628009796143, "learning_rate": 4.9259229270928115e-06, "loss": 0.15855172276496887, "step": 845 }, { "epoch": 0.21962616822429906, "grad_norm": 1.7713125944137573, "learning_rate": 4.925681990015233e-06, "loss": 0.15254972875118256, "step": 846 }, { "epoch": 0.21988577362409137, "grad_norm": 2.690967321395874, "learning_rate": 4.9254406683268955e-06, "loss": 0.17644383013248444, "step": 847 }, { "epoch": 0.2201453790238837, "grad_norm": 3.0537056922912598, "learning_rate": 4.925198962070459e-06, "loss": 0.17210426926612854, "step": 848 }, { "epoch": 0.220404984423676, "grad_norm": 2.5663623809814453, "learning_rate": 4.92495687128865e-06, "loss": 0.14925864338874817, "step": 849 }, { "epoch": 0.22066458982346832, "grad_norm": 3.4922914505004883, "learning_rate": 4.924714396024267e-06, "loss": 0.17693111300468445, "step": 850 }, { "epoch": 0.22092419522326065, "grad_norm": 1.9350910186767578, "learning_rate": 4.9244715363201735e-06, "loss": 0.16414660215377808, "step": 851 }, { "epoch": 0.22118380062305296, "grad_norm": 1.8907818794250488, "learning_rate": 4.924228292219301e-06, "loss": 0.14439579844474792, "step": 852 }, { "epoch": 0.22144340602284526, "grad_norm": 1.1170368194580078, "learning_rate": 4.923984663764649e-06, "loss": 0.12875288724899292, "step": 853 }, { "epoch": 0.2217030114226376, "grad_norm": 1.8917970657348633, "learning_rate": 4.923740650999287e-06, "loss": 0.14896252751350403, "step": 854 }, { "epoch": 0.2219626168224299, "grad_norm": 1.126955509185791, "learning_rate": 4.92349625396635e-06, "loss": 0.1325819343328476, "step": 855 }, { "epoch": 0.2222222222222222, "grad_norm": 1.8335586786270142, "learning_rate": 4.923251472709041e-06, "loss": 0.13986140489578247, "step": 856 }, { "epoch": 0.22248182762201454, "grad_norm": 3.7610185146331787, "learning_rate": 4.923006307270632e-06, "loss": 0.18367888033390045, "step": 857 }, { "epoch": 0.22274143302180685, "grad_norm": 2.180850028991699, "learning_rate": 4.922760757694464e-06, "loss": 0.13785520195960999, "step": 858 }, { "epoch": 0.22300103842159916, "grad_norm": 0.8006095290184021, "learning_rate": 4.922514824023944e-06, "loss": 0.13145022094249725, "step": 859 }, { "epoch": 0.2232606438213915, "grad_norm": 4.222301006317139, "learning_rate": 4.922268506302547e-06, "loss": 0.1489538699388504, "step": 860 }, { "epoch": 0.2235202492211838, "grad_norm": 1.8976268768310547, "learning_rate": 4.922021804573815e-06, "loss": 0.18318229913711548, "step": 861 }, { "epoch": 0.2237798546209761, "grad_norm": 0.9071184396743774, "learning_rate": 4.921774718881362e-06, "loss": 0.13417410850524902, "step": 862 }, { "epoch": 0.22403946002076844, "grad_norm": 2.700258731842041, "learning_rate": 4.921527249268866e-06, "loss": 0.17377758026123047, "step": 863 }, { "epoch": 0.22429906542056074, "grad_norm": 1.6791114807128906, "learning_rate": 4.921279395780073e-06, "loss": 0.1484023928642273, "step": 864 }, { "epoch": 0.22455867082035305, "grad_norm": 2.783618450164795, "learning_rate": 4.921031158458799e-06, "loss": 0.1384124457836151, "step": 865 }, { "epoch": 0.22481827622014539, "grad_norm": 2.0533876419067383, "learning_rate": 4.920782537348926e-06, "loss": 0.15300393104553223, "step": 866 }, { "epoch": 0.2250778816199377, "grad_norm": 2.4062116146087646, "learning_rate": 4.9205335324944045e-06, "loss": 0.14382776618003845, "step": 867 }, { "epoch": 0.22533748701973, "grad_norm": 4.05694580078125, "learning_rate": 4.9202841439392536e-06, "loss": 0.15210393071174622, "step": 868 }, { "epoch": 0.22559709241952233, "grad_norm": 1.6536246538162231, "learning_rate": 4.920034371727559e-06, "loss": 0.14975492656230927, "step": 869 }, { "epoch": 0.22585669781931464, "grad_norm": 3.2397425174713135, "learning_rate": 4.919784215903474e-06, "loss": 0.15697789192199707, "step": 870 }, { "epoch": 0.22611630321910695, "grad_norm": 1.227174997329712, "learning_rate": 4.9195336765112215e-06, "loss": 0.1648956537246704, "step": 871 }, { "epoch": 0.22637590861889928, "grad_norm": 2.5578699111938477, "learning_rate": 4.919282753595091e-06, "loss": 0.131670743227005, "step": 872 }, { "epoch": 0.2266355140186916, "grad_norm": 2.72446608543396, "learning_rate": 4.919031447199439e-06, "loss": 0.1782992035150528, "step": 873 }, { "epoch": 0.2268951194184839, "grad_norm": 1.4488283395767212, "learning_rate": 4.918779757368692e-06, "loss": 0.14786669611930847, "step": 874 }, { "epoch": 0.22715472481827623, "grad_norm": 2.2802796363830566, "learning_rate": 4.918527684147342e-06, "loss": 0.1287136971950531, "step": 875 }, { "epoch": 0.22741433021806853, "grad_norm": 1.2138924598693848, "learning_rate": 4.918275227579951e-06, "loss": 0.13669881224632263, "step": 876 }, { "epoch": 0.22767393561786084, "grad_norm": 5.877490520477295, "learning_rate": 4.918022387711145e-06, "loss": 0.1780470609664917, "step": 877 }, { "epoch": 0.22793354101765317, "grad_norm": 1.0639091730117798, "learning_rate": 4.917769164585623e-06, "loss": 0.126970574259758, "step": 878 }, { "epoch": 0.22819314641744548, "grad_norm": 4.263970375061035, "learning_rate": 4.917515558248148e-06, "loss": 0.15247632563114166, "step": 879 }, { "epoch": 0.2284527518172378, "grad_norm": 4.9708452224731445, "learning_rate": 4.9172615687435525e-06, "loss": 0.14825928211212158, "step": 880 }, { "epoch": 0.22871235721703012, "grad_norm": 2.1390464305877686, "learning_rate": 4.917007196116734e-06, "loss": 0.13674809038639069, "step": 881 }, { "epoch": 0.22897196261682243, "grad_norm": 2.7061538696289062, "learning_rate": 4.916752440412661e-06, "loss": 0.17681625485420227, "step": 882 }, { "epoch": 0.22923156801661473, "grad_norm": 1.6740623712539673, "learning_rate": 4.91649730167637e-06, "loss": 0.13375581800937653, "step": 883 }, { "epoch": 0.22949117341640707, "grad_norm": 4.32016658782959, "learning_rate": 4.916241779952961e-06, "loss": 0.17778728902339935, "step": 884 }, { "epoch": 0.22975077881619937, "grad_norm": 2.1698050498962402, "learning_rate": 4.915985875287607e-06, "loss": 0.12452610582113266, "step": 885 }, { "epoch": 0.23001038421599168, "grad_norm": 1.0098336935043335, "learning_rate": 4.915729587725544e-06, "loss": 0.14150330424308777, "step": 886 }, { "epoch": 0.23026998961578402, "grad_norm": 3.20552921295166, "learning_rate": 4.915472917312079e-06, "loss": 0.15385344624519348, "step": 887 }, { "epoch": 0.23052959501557632, "grad_norm": 3.499831438064575, "learning_rate": 4.915215864092584e-06, "loss": 0.13638803362846375, "step": 888 }, { "epoch": 0.23078920041536863, "grad_norm": 2.6660091876983643, "learning_rate": 4.914958428112502e-06, "loss": 0.17810535430908203, "step": 889 }, { "epoch": 0.23104880581516096, "grad_norm": 0.9728540182113647, "learning_rate": 4.91470060941734e-06, "loss": 0.13545551896095276, "step": 890 }, { "epoch": 0.23130841121495327, "grad_norm": 1.9537668228149414, "learning_rate": 4.914442408052676e-06, "loss": 0.12944886088371277, "step": 891 }, { "epoch": 0.23156801661474558, "grad_norm": 3.5156588554382324, "learning_rate": 4.9141838240641535e-06, "loss": 0.16896981000900269, "step": 892 }, { "epoch": 0.2318276220145379, "grad_norm": 1.2371561527252197, "learning_rate": 4.913924857497483e-06, "loss": 0.1273050308227539, "step": 893 }, { "epoch": 0.23208722741433022, "grad_norm": 2.6471943855285645, "learning_rate": 4.913665508398447e-06, "loss": 0.13835951685905457, "step": 894 }, { "epoch": 0.23234683281412252, "grad_norm": 2.589667320251465, "learning_rate": 4.913405776812888e-06, "loss": 0.1422118842601776, "step": 895 }, { "epoch": 0.23260643821391486, "grad_norm": 1.7657948732376099, "learning_rate": 4.913145662786725e-06, "loss": 0.12361162900924683, "step": 896 }, { "epoch": 0.23286604361370716, "grad_norm": 2.391376256942749, "learning_rate": 4.912885166365938e-06, "loss": 0.13661518692970276, "step": 897 }, { "epoch": 0.23312564901349947, "grad_norm": 2.198120594024658, "learning_rate": 4.912624287596577e-06, "loss": 0.17029233276844025, "step": 898 }, { "epoch": 0.2333852544132918, "grad_norm": 3.019800901412964, "learning_rate": 4.91236302652476e-06, "loss": 0.13875964283943176, "step": 899 }, { "epoch": 0.2336448598130841, "grad_norm": 3.803102970123291, "learning_rate": 4.9121013831966704e-06, "loss": 0.17425355315208435, "step": 900 }, { "epoch": 0.23390446521287642, "grad_norm": 1.347579836845398, "learning_rate": 4.911839357658562e-06, "loss": 0.12534457445144653, "step": 901 }, { "epoch": 0.23416407061266875, "grad_norm": 5.156105041503906, "learning_rate": 4.911576949956755e-06, "loss": 0.16193684935569763, "step": 902 }, { "epoch": 0.23442367601246106, "grad_norm": 4.078166484832764, "learning_rate": 4.9113141601376365e-06, "loss": 0.17908316850662231, "step": 903 }, { "epoch": 0.23468328141225336, "grad_norm": 1.0235815048217773, "learning_rate": 4.911050988247663e-06, "loss": 0.13715240359306335, "step": 904 }, { "epoch": 0.2349428868120457, "grad_norm": 3.0726516246795654, "learning_rate": 4.910787434333355e-06, "loss": 0.14025411009788513, "step": 905 }, { "epoch": 0.235202492211838, "grad_norm": 2.231998920440674, "learning_rate": 4.910523498441304e-06, "loss": 0.13066384196281433, "step": 906 }, { "epoch": 0.2354620976116303, "grad_norm": 1.9776486158370972, "learning_rate": 4.91025918061817e-06, "loss": 0.12724830210208893, "step": 907 }, { "epoch": 0.23572170301142265, "grad_norm": 3.761667490005493, "learning_rate": 4.909994480910674e-06, "loss": 0.14337143301963806, "step": 908 }, { "epoch": 0.23598130841121495, "grad_norm": 4.397122383117676, "learning_rate": 4.909729399365613e-06, "loss": 0.1424567550420761, "step": 909 }, { "epoch": 0.23624091381100726, "grad_norm": 4.802247047424316, "learning_rate": 4.909463936029843e-06, "loss": 0.14115869998931885, "step": 910 }, { "epoch": 0.2365005192107996, "grad_norm": 5.261049270629883, "learning_rate": 4.909198090950296e-06, "loss": 0.17856872081756592, "step": 911 }, { "epoch": 0.2367601246105919, "grad_norm": 2.3843486309051514, "learning_rate": 4.908931864173965e-06, "loss": 0.13401861488819122, "step": 912 }, { "epoch": 0.2370197300103842, "grad_norm": 2.460477113723755, "learning_rate": 4.908665255747914e-06, "loss": 0.1402890682220459, "step": 913 }, { "epoch": 0.23727933541017654, "grad_norm": 3.966076612472534, "learning_rate": 4.908398265719273e-06, "loss": 0.13222721219062805, "step": 914 }, { "epoch": 0.23753894080996885, "grad_norm": 2.691725969314575, "learning_rate": 4.908130894135238e-06, "loss": 0.14733314514160156, "step": 915 }, { "epoch": 0.23779854620976115, "grad_norm": 1.566828727722168, "learning_rate": 4.907863141043076e-06, "loss": 0.13062535226345062, "step": 916 }, { "epoch": 0.2380581516095535, "grad_norm": 1.0984591245651245, "learning_rate": 4.90759500649012e-06, "loss": 0.1523871272802353, "step": 917 }, { "epoch": 0.2383177570093458, "grad_norm": 1.8854873180389404, "learning_rate": 4.9073264905237685e-06, "loss": 0.1843504011631012, "step": 918 }, { "epoch": 0.2385773624091381, "grad_norm": 3.655935764312744, "learning_rate": 4.907057593191489e-06, "loss": 0.15056271851062775, "step": 919 }, { "epoch": 0.23883696780893043, "grad_norm": 3.526216745376587, "learning_rate": 4.906788314540819e-06, "loss": 0.14187651872634888, "step": 920 }, { "epoch": 0.23909657320872274, "grad_norm": 1.665218710899353, "learning_rate": 4.906518654619358e-06, "loss": 0.1305558681488037, "step": 921 }, { "epoch": 0.23935617860851505, "grad_norm": 4.923025608062744, "learning_rate": 4.906248613474776e-06, "loss": 0.19291704893112183, "step": 922 }, { "epoch": 0.23961578400830738, "grad_norm": 1.7190486192703247, "learning_rate": 4.905978191154811e-06, "loss": 0.17673823237419128, "step": 923 }, { "epoch": 0.2398753894080997, "grad_norm": 1.7269569635391235, "learning_rate": 4.905707387707267e-06, "loss": 0.1554998755455017, "step": 924 }, { "epoch": 0.240134994807892, "grad_norm": 1.3730390071868896, "learning_rate": 4.905436203180016e-06, "loss": 0.14975914359092712, "step": 925 }, { "epoch": 0.24039460020768433, "grad_norm": 0.7137861251831055, "learning_rate": 4.9051646376209976e-06, "loss": 0.13011422753334045, "step": 926 }, { "epoch": 0.24065420560747663, "grad_norm": 1.3315706253051758, "learning_rate": 4.904892691078218e-06, "loss": 0.14142368733882904, "step": 927 }, { "epoch": 0.24091381100726894, "grad_norm": 0.8496571183204651, "learning_rate": 4.904620363599752e-06, "loss": 0.1376720666885376, "step": 928 }, { "epoch": 0.24117341640706128, "grad_norm": 1.390945553779602, "learning_rate": 4.904347655233738e-06, "loss": 0.15224266052246094, "step": 929 }, { "epoch": 0.24143302180685358, "grad_norm": 0.7823232412338257, "learning_rate": 4.904074566028388e-06, "loss": 0.13157585263252258, "step": 930 }, { "epoch": 0.2416926272066459, "grad_norm": 3.280695915222168, "learning_rate": 4.903801096031976e-06, "loss": 0.1885198950767517, "step": 931 }, { "epoch": 0.24195223260643822, "grad_norm": 2.9356842041015625, "learning_rate": 4.903527245292847e-06, "loss": 0.1446102261543274, "step": 932 }, { "epoch": 0.24221183800623053, "grad_norm": 1.657981276512146, "learning_rate": 4.903253013859408e-06, "loss": 0.13414281606674194, "step": 933 }, { "epoch": 0.24247144340602284, "grad_norm": 2.496100902557373, "learning_rate": 4.9029784017801404e-06, "loss": 0.12847661972045898, "step": 934 }, { "epoch": 0.24273104880581517, "grad_norm": 1.1271154880523682, "learning_rate": 4.902703409103588e-06, "loss": 0.1303086280822754, "step": 935 }, { "epoch": 0.24299065420560748, "grad_norm": 1.7753052711486816, "learning_rate": 4.902428035878363e-06, "loss": 0.1352543979883194, "step": 936 }, { "epoch": 0.24325025960539978, "grad_norm": 1.3649976253509521, "learning_rate": 4.902152282153145e-06, "loss": 0.16458305716514587, "step": 937 }, { "epoch": 0.24350986500519212, "grad_norm": 5.245884418487549, "learning_rate": 4.901876147976682e-06, "loss": 0.17307788133621216, "step": 938 }, { "epoch": 0.24376947040498442, "grad_norm": 1.5592806339263916, "learning_rate": 4.901599633397787e-06, "loss": 0.12089499831199646, "step": 939 }, { "epoch": 0.24402907580477673, "grad_norm": 3.7820322513580322, "learning_rate": 4.901322738465341e-06, "loss": 0.1855798065662384, "step": 940 }, { "epoch": 0.24428868120456906, "grad_norm": 5.965107440948486, "learning_rate": 4.901045463228295e-06, "loss": 0.18418583273887634, "step": 941 }, { "epoch": 0.24454828660436137, "grad_norm": 1.6158292293548584, "learning_rate": 4.900767807735664e-06, "loss": 0.15727370977401733, "step": 942 }, { "epoch": 0.24480789200415368, "grad_norm": 1.3267608880996704, "learning_rate": 4.90048977203653e-06, "loss": 0.15350955724716187, "step": 943 }, { "epoch": 0.245067497403946, "grad_norm": 3.6074721813201904, "learning_rate": 4.900211356180044e-06, "loss": 0.13362711668014526, "step": 944 }, { "epoch": 0.24532710280373832, "grad_norm": 2.670017719268799, "learning_rate": 4.899932560215423e-06, "loss": 0.15869168937206268, "step": 945 }, { "epoch": 0.24558670820353062, "grad_norm": 2.145268201828003, "learning_rate": 4.899653384191952e-06, "loss": 0.1352510005235672, "step": 946 }, { "epoch": 0.24584631360332296, "grad_norm": 2.492349147796631, "learning_rate": 4.899373828158984e-06, "loss": 0.17063917219638824, "step": 947 }, { "epoch": 0.24610591900311526, "grad_norm": 1.0728085041046143, "learning_rate": 4.899093892165938e-06, "loss": 0.1394530087709427, "step": 948 }, { "epoch": 0.24636552440290757, "grad_norm": 1.665220856666565, "learning_rate": 4.898813576262299e-06, "loss": 0.13241969048976898, "step": 949 }, { "epoch": 0.2466251298026999, "grad_norm": 4.497960567474365, "learning_rate": 4.8985328804976205e-06, "loss": 0.1616322249174118, "step": 950 }, { "epoch": 0.2468847352024922, "grad_norm": 0.9416487812995911, "learning_rate": 4.898251804921523e-06, "loss": 0.1231742724776268, "step": 951 }, { "epoch": 0.24714434060228452, "grad_norm": 2.343282699584961, "learning_rate": 4.897970349583697e-06, "loss": 0.13204678893089294, "step": 952 }, { "epoch": 0.24740394600207685, "grad_norm": 0.9808669090270996, "learning_rate": 4.8976885145338935e-06, "loss": 0.14517545700073242, "step": 953 }, { "epoch": 0.24766355140186916, "grad_norm": 1.7403769493103027, "learning_rate": 4.897406299821936e-06, "loss": 0.17290954291820526, "step": 954 }, { "epoch": 0.24792315680166147, "grad_norm": 2.961521625518799, "learning_rate": 4.897123705497714e-06, "loss": 0.1891990303993225, "step": 955 }, { "epoch": 0.2481827622014538, "grad_norm": 2.6536762714385986, "learning_rate": 4.896840731611185e-06, "loss": 0.15550705790519714, "step": 956 }, { "epoch": 0.2484423676012461, "grad_norm": 0.8915792107582092, "learning_rate": 4.896557378212369e-06, "loss": 0.13660068809986115, "step": 957 }, { "epoch": 0.2487019730010384, "grad_norm": 1.891682505607605, "learning_rate": 4.8962736453513595e-06, "loss": 0.15443044900894165, "step": 958 }, { "epoch": 0.24896157840083075, "grad_norm": 1.7643574476242065, "learning_rate": 4.895989533078311e-06, "loss": 0.14859864115715027, "step": 959 }, { "epoch": 0.24922118380062305, "grad_norm": 1.283002257347107, "learning_rate": 4.895705041443452e-06, "loss": 0.13826695084571838, "step": 960 }, { "epoch": 0.24948078920041536, "grad_norm": 2.3037149906158447, "learning_rate": 4.89542017049707e-06, "loss": 0.14376771450042725, "step": 961 }, { "epoch": 0.2497403946002077, "grad_norm": 0.8567098379135132, "learning_rate": 4.895134920289526e-06, "loss": 0.1261465847492218, "step": 962 }, { "epoch": 0.25, "grad_norm": 1.2688931226730347, "learning_rate": 4.894849290871245e-06, "loss": 0.1437303125858307, "step": 963 }, { "epoch": 0.2502596053997923, "grad_norm": 2.1224403381347656, "learning_rate": 4.89456328229272e-06, "loss": 0.1396259367465973, "step": 964 }, { "epoch": 0.2505192107995846, "grad_norm": 1.5151273012161255, "learning_rate": 4.894276894604511e-06, "loss": 0.12890513241291046, "step": 965 }, { "epoch": 0.2507788161993769, "grad_norm": 2.3145270347595215, "learning_rate": 4.893990127857244e-06, "loss": 0.14605402946472168, "step": 966 }, { "epoch": 0.2510384215991693, "grad_norm": 3.730464458465576, "learning_rate": 4.893702982101614e-06, "loss": 0.1521889567375183, "step": 967 }, { "epoch": 0.2512980269989616, "grad_norm": 1.500139832496643, "learning_rate": 4.89341545738838e-06, "loss": 0.14458099007606506, "step": 968 }, { "epoch": 0.2515576323987539, "grad_norm": 3.9727609157562256, "learning_rate": 4.893127553768372e-06, "loss": 0.16171115636825562, "step": 969 }, { "epoch": 0.2518172377985462, "grad_norm": 1.5777660608291626, "learning_rate": 4.892839271292483e-06, "loss": 0.16718006134033203, "step": 970 }, { "epoch": 0.2520768431983385, "grad_norm": 3.9896559715270996, "learning_rate": 4.892550610011675e-06, "loss": 0.11836278438568115, "step": 971 }, { "epoch": 0.2523364485981308, "grad_norm": 3.93168568611145, "learning_rate": 4.8922615699769775e-06, "loss": 0.14107674360275269, "step": 972 }, { "epoch": 0.2525960539979232, "grad_norm": 1.7276928424835205, "learning_rate": 4.891972151239486e-06, "loss": 0.13266587257385254, "step": 973 }, { "epoch": 0.2528556593977155, "grad_norm": 2.7965874671936035, "learning_rate": 4.891682353850362e-06, "loss": 0.13357636332511902, "step": 974 }, { "epoch": 0.2531152647975078, "grad_norm": 3.680612087249756, "learning_rate": 4.8913921778608355e-06, "loss": 0.13599705696105957, "step": 975 }, { "epoch": 0.2533748701973001, "grad_norm": 2.278505325317383, "learning_rate": 4.891101623322205e-06, "loss": 0.1479111611843109, "step": 976 }, { "epoch": 0.2536344755970924, "grad_norm": 2.4172351360321045, "learning_rate": 4.890810690285831e-06, "loss": 0.12587815523147583, "step": 977 }, { "epoch": 0.2538940809968847, "grad_norm": 2.2321016788482666, "learning_rate": 4.8905193788031454e-06, "loss": 0.13813060522079468, "step": 978 }, { "epoch": 0.25415368639667707, "grad_norm": 3.682065486907959, "learning_rate": 4.890227688925645e-06, "loss": 0.15019947290420532, "step": 979 }, { "epoch": 0.2544132917964694, "grad_norm": 3.2526261806488037, "learning_rate": 4.8899356207048945e-06, "loss": 0.16593322157859802, "step": 980 }, { "epoch": 0.2546728971962617, "grad_norm": 1.4621394872665405, "learning_rate": 4.889643174192523e-06, "loss": 0.13640820980072021, "step": 981 }, { "epoch": 0.254932502596054, "grad_norm": 1.2493035793304443, "learning_rate": 4.8893503494402305e-06, "loss": 0.165280282497406, "step": 982 }, { "epoch": 0.2551921079958463, "grad_norm": 3.5254099369049072, "learning_rate": 4.889057146499781e-06, "loss": 0.18107448518276215, "step": 983 }, { "epoch": 0.2554517133956386, "grad_norm": 2.000248670578003, "learning_rate": 4.888763565423004e-06, "loss": 0.1553773730993271, "step": 984 }, { "epoch": 0.25571131879543096, "grad_norm": 2.7703044414520264, "learning_rate": 4.888469606261803e-06, "loss": 0.12448267638683319, "step": 985 }, { "epoch": 0.25597092419522327, "grad_norm": 1.0729777812957764, "learning_rate": 4.888175269068138e-06, "loss": 0.13377472758293152, "step": 986 }, { "epoch": 0.2562305295950156, "grad_norm": 1.8666757345199585, "learning_rate": 4.887880553894044e-06, "loss": 0.12081246078014374, "step": 987 }, { "epoch": 0.2564901349948079, "grad_norm": 7.569124221801758, "learning_rate": 4.887585460791618e-06, "loss": 0.164871484041214, "step": 988 }, { "epoch": 0.2567497403946002, "grad_norm": 0.741814136505127, "learning_rate": 4.887289989813029e-06, "loss": 0.12090761959552765, "step": 989 }, { "epoch": 0.2570093457943925, "grad_norm": 1.7262568473815918, "learning_rate": 4.886994141010505e-06, "loss": 0.1597064733505249, "step": 990 }, { "epoch": 0.25726895119418486, "grad_norm": 2.8165206909179688, "learning_rate": 4.886697914436348e-06, "loss": 0.16890820860862732, "step": 991 }, { "epoch": 0.25752855659397716, "grad_norm": 3.3078949451446533, "learning_rate": 4.8864013101429235e-06, "loss": 0.1362772434949875, "step": 992 }, { "epoch": 0.25778816199376947, "grad_norm": 2.3463597297668457, "learning_rate": 4.886104328182665e-06, "loss": 0.14096179604530334, "step": 993 }, { "epoch": 0.2580477673935618, "grad_norm": 2.379732608795166, "learning_rate": 4.885806968608072e-06, "loss": 0.15033206343650818, "step": 994 }, { "epoch": 0.2583073727933541, "grad_norm": 2.4489519596099854, "learning_rate": 4.8855092314717095e-06, "loss": 0.14095932245254517, "step": 995 }, { "epoch": 0.2585669781931464, "grad_norm": 0.908599853515625, "learning_rate": 4.885211116826212e-06, "loss": 0.1261143684387207, "step": 996 }, { "epoch": 0.25882658359293875, "grad_norm": 1.7936307191848755, "learning_rate": 4.884912624724279e-06, "loss": 0.14577056467533112, "step": 997 }, { "epoch": 0.25908618899273106, "grad_norm": 2.241117000579834, "learning_rate": 4.884613755218677e-06, "loss": 0.16279244422912598, "step": 998 }, { "epoch": 0.25934579439252337, "grad_norm": 2.0726845264434814, "learning_rate": 4.884314508362239e-06, "loss": 0.12949475646018982, "step": 999 }, { "epoch": 0.25960539979231567, "grad_norm": 2.734339714050293, "learning_rate": 4.884014884207865e-06, "loss": 0.14617286622524261, "step": 1000 }, { "epoch": 0.259865005192108, "grad_norm": 1.784023642539978, "learning_rate": 4.8837148828085225e-06, "loss": 0.1667255163192749, "step": 1001 }, { "epoch": 0.2601246105919003, "grad_norm": 1.8918349742889404, "learning_rate": 4.883414504217244e-06, "loss": 0.12686684727668762, "step": 1002 }, { "epoch": 0.26038421599169265, "grad_norm": 6.288288593292236, "learning_rate": 4.88311374848713e-06, "loss": 0.17132994532585144, "step": 1003 }, { "epoch": 0.26064382139148495, "grad_norm": 0.9001903533935547, "learning_rate": 4.882812615671347e-06, "loss": 0.14104677736759186, "step": 1004 }, { "epoch": 0.26090342679127726, "grad_norm": 1.7421331405639648, "learning_rate": 4.882511105823128e-06, "loss": 0.14698413014411926, "step": 1005 }, { "epoch": 0.26116303219106957, "grad_norm": 1.4160805940628052, "learning_rate": 4.882209218995775e-06, "loss": 0.12636375427246094, "step": 1006 }, { "epoch": 0.2614226375908619, "grad_norm": 3.9283599853515625, "learning_rate": 4.881906955242652e-06, "loss": 0.141691192984581, "step": 1007 }, { "epoch": 0.2616822429906542, "grad_norm": 3.2823827266693115, "learning_rate": 4.881604314617194e-06, "loss": 0.13517722487449646, "step": 1008 }, { "epoch": 0.26194184839044654, "grad_norm": 2.2568366527557373, "learning_rate": 4.8813012971729e-06, "loss": 0.16541410982608795, "step": 1009 }, { "epoch": 0.26220145379023885, "grad_norm": 1.0763739347457886, "learning_rate": 4.880997902963337e-06, "loss": 0.14652574062347412, "step": 1010 }, { "epoch": 0.26246105919003115, "grad_norm": 3.9545958042144775, "learning_rate": 4.880694132042138e-06, "loss": 0.16138926148414612, "step": 1011 }, { "epoch": 0.26272066458982346, "grad_norm": 3.5621063709259033, "learning_rate": 4.8803899844630034e-06, "loss": 0.13282789289951324, "step": 1012 }, { "epoch": 0.26298026998961577, "grad_norm": 1.4571083784103394, "learning_rate": 4.880085460279699e-06, "loss": 0.13031847774982452, "step": 1013 }, { "epoch": 0.2632398753894081, "grad_norm": 3.631215810775757, "learning_rate": 4.879780559546058e-06, "loss": 0.12625731527805328, "step": 1014 }, { "epoch": 0.26349948078920044, "grad_norm": 2.407233476638794, "learning_rate": 4.879475282315978e-06, "loss": 0.12351427972316742, "step": 1015 }, { "epoch": 0.26375908618899274, "grad_norm": 1.1784460544586182, "learning_rate": 4.879169628643428e-06, "loss": 0.1380614936351776, "step": 1016 }, { "epoch": 0.26401869158878505, "grad_norm": 2.5080859661102295, "learning_rate": 4.878863598582439e-06, "loss": 0.1483176052570343, "step": 1017 }, { "epoch": 0.26427829698857735, "grad_norm": 0.8743399381637573, "learning_rate": 4.87855719218711e-06, "loss": 0.12155675143003464, "step": 1018 }, { "epoch": 0.26453790238836966, "grad_norm": 1.5085209608078003, "learning_rate": 4.878250409511608e-06, "loss": 0.13253113627433777, "step": 1019 }, { "epoch": 0.26479750778816197, "grad_norm": 1.5428428649902344, "learning_rate": 4.877943250610163e-06, "loss": 0.13891932368278503, "step": 1020 }, { "epoch": 0.26505711318795433, "grad_norm": 2.3565218448638916, "learning_rate": 4.8776357155370745e-06, "loss": 0.177606001496315, "step": 1021 }, { "epoch": 0.26531671858774664, "grad_norm": 3.217137575149536, "learning_rate": 4.877327804346709e-06, "loss": 0.16179943084716797, "step": 1022 }, { "epoch": 0.26557632398753894, "grad_norm": 1.6290910243988037, "learning_rate": 4.877019517093497e-06, "loss": 0.12091101706027985, "step": 1023 }, { "epoch": 0.26583592938733125, "grad_norm": 1.1476656198501587, "learning_rate": 4.876710853831935e-06, "loss": 0.12292397022247314, "step": 1024 }, { "epoch": 0.26609553478712356, "grad_norm": 1.1496267318725586, "learning_rate": 4.8764018146165905e-06, "loss": 0.12632116675376892, "step": 1025 }, { "epoch": 0.26635514018691586, "grad_norm": 1.2367522716522217, "learning_rate": 4.876092399502093e-06, "loss": 0.12942549586296082, "step": 1026 }, { "epoch": 0.2666147455867082, "grad_norm": 0.9182763695716858, "learning_rate": 4.87578260854314e-06, "loss": 0.12146227061748505, "step": 1027 }, { "epoch": 0.26687435098650053, "grad_norm": 1.82649564743042, "learning_rate": 4.875472441794496e-06, "loss": 0.1580336093902588, "step": 1028 }, { "epoch": 0.26713395638629284, "grad_norm": 1.1542268991470337, "learning_rate": 4.875161899310992e-06, "loss": 0.13786175847053528, "step": 1029 }, { "epoch": 0.26739356178608514, "grad_norm": 1.3343849182128906, "learning_rate": 4.874850981147523e-06, "loss": 0.15198606252670288, "step": 1030 }, { "epoch": 0.26765316718587745, "grad_norm": 2.4400293827056885, "learning_rate": 4.874539687359054e-06, "loss": 0.1852383315563202, "step": 1031 }, { "epoch": 0.26791277258566976, "grad_norm": 2.644899606704712, "learning_rate": 4.874228018000613e-06, "loss": 0.16988393664360046, "step": 1032 }, { "epoch": 0.2681723779854621, "grad_norm": 1.1400643587112427, "learning_rate": 4.8739159731272975e-06, "loss": 0.13078570365905762, "step": 1033 }, { "epoch": 0.2684319833852544, "grad_norm": 1.3116586208343506, "learning_rate": 4.873603552794268e-06, "loss": 0.15267622470855713, "step": 1034 }, { "epoch": 0.26869158878504673, "grad_norm": 0.965375542640686, "learning_rate": 4.873290757056756e-06, "loss": 0.13368827104568481, "step": 1035 }, { "epoch": 0.26895119418483904, "grad_norm": 1.1705764532089233, "learning_rate": 4.872977585970054e-06, "loss": 0.13333182036876678, "step": 1036 }, { "epoch": 0.26921079958463134, "grad_norm": 1.4306097030639648, "learning_rate": 4.872664039589525e-06, "loss": 0.16461250185966492, "step": 1037 }, { "epoch": 0.26947040498442365, "grad_norm": 3.5182158946990967, "learning_rate": 4.872350117970596e-06, "loss": 0.16162890195846558, "step": 1038 }, { "epoch": 0.269730010384216, "grad_norm": 1.003766655921936, "learning_rate": 4.8720358211687615e-06, "loss": 0.14793376624584198, "step": 1039 }, { "epoch": 0.2699896157840083, "grad_norm": 1.832566499710083, "learning_rate": 4.871721149239583e-06, "loss": 0.16517090797424316, "step": 1040 }, { "epoch": 0.2702492211838006, "grad_norm": 1.070395588874817, "learning_rate": 4.8714061022386845e-06, "loss": 0.1489025205373764, "step": 1041 }, { "epoch": 0.27050882658359293, "grad_norm": 2.0684163570404053, "learning_rate": 4.8710906802217624e-06, "loss": 0.18328019976615906, "step": 1042 }, { "epoch": 0.27076843198338524, "grad_norm": 1.660422444343567, "learning_rate": 4.8707748832445745e-06, "loss": 0.1576867699623108, "step": 1043 }, { "epoch": 0.27102803738317754, "grad_norm": 1.0665993690490723, "learning_rate": 4.870458711362946e-06, "loss": 0.14136993885040283, "step": 1044 }, { "epoch": 0.2712876427829699, "grad_norm": 1.8713446855545044, "learning_rate": 4.8701421646327694e-06, "loss": 0.1479949802160263, "step": 1045 }, { "epoch": 0.2715472481827622, "grad_norm": 1.5788955688476562, "learning_rate": 4.869825243110003e-06, "loss": 0.17512555420398712, "step": 1046 }, { "epoch": 0.2718068535825545, "grad_norm": 3.6796443462371826, "learning_rate": 4.869507946850671e-06, "loss": 0.16290992498397827, "step": 1047 }, { "epoch": 0.2720664589823468, "grad_norm": 1.759730577468872, "learning_rate": 4.869190275910865e-06, "loss": 0.1630057990550995, "step": 1048 }, { "epoch": 0.27232606438213913, "grad_norm": 1.0356007814407349, "learning_rate": 4.868872230346741e-06, "loss": 0.14284703135490417, "step": 1049 }, { "epoch": 0.27258566978193144, "grad_norm": 1.1190060377120972, "learning_rate": 4.8685538102145215e-06, "loss": 0.1475752890110016, "step": 1050 }, { "epoch": 0.2728452751817238, "grad_norm": 2.2379589080810547, "learning_rate": 4.868235015570497e-06, "loss": 0.1264180839061737, "step": 1051 }, { "epoch": 0.2731048805815161, "grad_norm": 0.7045902013778687, "learning_rate": 4.867915846471024e-06, "loss": 0.12299661338329315, "step": 1052 }, { "epoch": 0.2733644859813084, "grad_norm": 3.521624803543091, "learning_rate": 4.867596302972523e-06, "loss": 0.14026854932308197, "step": 1053 }, { "epoch": 0.2736240913811007, "grad_norm": 4.410065650939941, "learning_rate": 4.867276385131482e-06, "loss": 0.14633844792842865, "step": 1054 }, { "epoch": 0.273883696780893, "grad_norm": 2.8417274951934814, "learning_rate": 4.866956093004454e-06, "loss": 0.13127964735031128, "step": 1055 }, { "epoch": 0.27414330218068533, "grad_norm": 3.1612768173217773, "learning_rate": 4.866635426648062e-06, "loss": 0.18252603709697723, "step": 1056 }, { "epoch": 0.2744029075804777, "grad_norm": 2.2525224685668945, "learning_rate": 4.866314386118991e-06, "loss": 0.1475117802619934, "step": 1057 }, { "epoch": 0.27466251298027, "grad_norm": 1.0536880493164062, "learning_rate": 4.865992971473993e-06, "loss": 0.16749230027198792, "step": 1058 }, { "epoch": 0.2749221183800623, "grad_norm": 1.9713375568389893, "learning_rate": 4.865671182769888e-06, "loss": 0.1426917314529419, "step": 1059 }, { "epoch": 0.2751817237798546, "grad_norm": 3.3237364292144775, "learning_rate": 4.865349020063561e-06, "loss": 0.16791020333766937, "step": 1060 }, { "epoch": 0.2754413291796469, "grad_norm": 2.721606969833374, "learning_rate": 4.865026483411961e-06, "loss": 0.16967391967773438, "step": 1061 }, { "epoch": 0.2757009345794392, "grad_norm": 3.785989999771118, "learning_rate": 4.864703572872107e-06, "loss": 0.16451959311962128, "step": 1062 }, { "epoch": 0.2759605399792316, "grad_norm": 1.8741284608840942, "learning_rate": 4.864380288501082e-06, "loss": 0.14216022193431854, "step": 1063 }, { "epoch": 0.2762201453790239, "grad_norm": 2.1408019065856934, "learning_rate": 4.864056630356034e-06, "loss": 0.1347057819366455, "step": 1064 }, { "epoch": 0.2764797507788162, "grad_norm": 4.472231388092041, "learning_rate": 4.86373259849418e-06, "loss": 0.1576007902622223, "step": 1065 }, { "epoch": 0.2767393561786085, "grad_norm": 2.976416826248169, "learning_rate": 4.8634081929728e-06, "loss": 0.13921162486076355, "step": 1066 }, { "epoch": 0.2769989615784008, "grad_norm": 1.6833510398864746, "learning_rate": 4.863083413849243e-06, "loss": 0.12910111248493195, "step": 1067 }, { "epoch": 0.2772585669781931, "grad_norm": 3.715768337249756, "learning_rate": 4.862758261180921e-06, "loss": 0.1726517677307129, "step": 1068 }, { "epoch": 0.2775181723779855, "grad_norm": 1.2956188917160034, "learning_rate": 4.8624327350253145e-06, "loss": 0.15573783218860626, "step": 1069 }, { "epoch": 0.2777777777777778, "grad_norm": 1.3909200429916382, "learning_rate": 4.862106835439968e-06, "loss": 0.14412976801395416, "step": 1070 }, { "epoch": 0.2780373831775701, "grad_norm": 1.9894278049468994, "learning_rate": 4.861780562482495e-06, "loss": 0.13076859712600708, "step": 1071 }, { "epoch": 0.2782969885773624, "grad_norm": 5.185438632965088, "learning_rate": 4.861453916210571e-06, "loss": 0.19495007395744324, "step": 1072 }, { "epoch": 0.2785565939771547, "grad_norm": 1.7705432176589966, "learning_rate": 4.861126896681941e-06, "loss": 0.14993008971214294, "step": 1073 }, { "epoch": 0.278816199376947, "grad_norm": 2.807716131210327, "learning_rate": 4.8607995039544135e-06, "loss": 0.1245112419128418, "step": 1074 }, { "epoch": 0.2790758047767394, "grad_norm": 2.8254213333129883, "learning_rate": 4.8604717380858655e-06, "loss": 0.12339312583208084, "step": 1075 }, { "epoch": 0.2793354101765317, "grad_norm": 1.1030006408691406, "learning_rate": 4.8601435991342375e-06, "loss": 0.1435021460056305, "step": 1076 }, { "epoch": 0.279595015576324, "grad_norm": 1.113705039024353, "learning_rate": 4.859815087157536e-06, "loss": 0.16104882955551147, "step": 1077 }, { "epoch": 0.2798546209761163, "grad_norm": 1.2494546175003052, "learning_rate": 4.859486202213836e-06, "loss": 0.1801396906375885, "step": 1078 }, { "epoch": 0.2801142263759086, "grad_norm": 1.5065213441848755, "learning_rate": 4.859156944361276e-06, "loss": 0.1593797206878662, "step": 1079 }, { "epoch": 0.2803738317757009, "grad_norm": 3.543828010559082, "learning_rate": 4.858827313658062e-06, "loss": 0.13623535633087158, "step": 1080 }, { "epoch": 0.28063343717549327, "grad_norm": 4.679182529449463, "learning_rate": 4.858497310162465e-06, "loss": 0.13786935806274414, "step": 1081 }, { "epoch": 0.2808930425752856, "grad_norm": 1.0602954626083374, "learning_rate": 4.858166933932821e-06, "loss": 0.13367372751235962, "step": 1082 }, { "epoch": 0.2811526479750779, "grad_norm": 2.873979330062866, "learning_rate": 4.857836185027534e-06, "loss": 0.1440383493900299, "step": 1083 }, { "epoch": 0.2814122533748702, "grad_norm": 6.110116004943848, "learning_rate": 4.857505063505073e-06, "loss": 0.18492993712425232, "step": 1084 }, { "epoch": 0.2816718587746625, "grad_norm": 2.4297046661376953, "learning_rate": 4.857173569423971e-06, "loss": 0.1661481261253357, "step": 1085 }, { "epoch": 0.2819314641744548, "grad_norm": 0.9589370489120483, "learning_rate": 4.856841702842831e-06, "loss": 0.1458682268857956, "step": 1086 }, { "epoch": 0.28219106957424717, "grad_norm": 5.129524230957031, "learning_rate": 4.856509463820318e-06, "loss": 0.18263503909111023, "step": 1087 }, { "epoch": 0.2824506749740395, "grad_norm": 3.375986337661743, "learning_rate": 4.856176852415164e-06, "loss": 0.12309233844280243, "step": 1088 }, { "epoch": 0.2827102803738318, "grad_norm": 1.474002718925476, "learning_rate": 4.855843868686168e-06, "loss": 0.14900609850883484, "step": 1089 }, { "epoch": 0.2829698857736241, "grad_norm": 0.9383192658424377, "learning_rate": 4.855510512692194e-06, "loss": 0.14221440255641937, "step": 1090 }, { "epoch": 0.2832294911734164, "grad_norm": 1.4423707723617554, "learning_rate": 4.8551767844921695e-06, "loss": 0.15397512912750244, "step": 1091 }, { "epoch": 0.2834890965732087, "grad_norm": 2.6023504734039307, "learning_rate": 4.854842684145093e-06, "loss": 0.11659656465053558, "step": 1092 }, { "epoch": 0.28374870197300106, "grad_norm": 2.7233097553253174, "learning_rate": 4.854508211710025e-06, "loss": 0.13128238916397095, "step": 1093 }, { "epoch": 0.28400830737279337, "grad_norm": 2.226788282394409, "learning_rate": 4.85417336724609e-06, "loss": 0.12753845751285553, "step": 1094 }, { "epoch": 0.2842679127725857, "grad_norm": 1.1188442707061768, "learning_rate": 4.853838150812485e-06, "loss": 0.1262504756450653, "step": 1095 }, { "epoch": 0.284527518172378, "grad_norm": 3.904909610748291, "learning_rate": 4.853502562468464e-06, "loss": 0.1471160650253296, "step": 1096 }, { "epoch": 0.2847871235721703, "grad_norm": 3.7450740337371826, "learning_rate": 4.853166602273356e-06, "loss": 0.1557050496339798, "step": 1097 }, { "epoch": 0.2850467289719626, "grad_norm": 1.794921636581421, "learning_rate": 4.852830270286548e-06, "loss": 0.15860247611999512, "step": 1098 }, { "epoch": 0.28530633437175495, "grad_norm": 1.6636310815811157, "learning_rate": 4.852493566567496e-06, "loss": 0.15499559044837952, "step": 1099 }, { "epoch": 0.28556593977154726, "grad_norm": 4.23613977432251, "learning_rate": 4.852156491175724e-06, "loss": 0.137344092130661, "step": 1100 }, { "epoch": 0.28582554517133957, "grad_norm": 2.5595293045043945, "learning_rate": 4.851819044170816e-06, "loss": 0.14860227704048157, "step": 1101 }, { "epoch": 0.2860851505711319, "grad_norm": 1.5013614892959595, "learning_rate": 4.851481225612426e-06, "loss": 0.14860540628433228, "step": 1102 }, { "epoch": 0.2863447559709242, "grad_norm": 1.3802638053894043, "learning_rate": 4.851143035560274e-06, "loss": 0.1301249861717224, "step": 1103 }, { "epoch": 0.2866043613707165, "grad_norm": 3.0389134883880615, "learning_rate": 4.850804474074143e-06, "loss": 0.13975027203559875, "step": 1104 }, { "epoch": 0.28686396677050885, "grad_norm": 1.8757734298706055, "learning_rate": 4.8504655412138816e-06, "loss": 0.14142143726348877, "step": 1105 }, { "epoch": 0.28712357217030116, "grad_norm": 3.0531458854675293, "learning_rate": 4.850126237039408e-06, "loss": 0.16048084199428558, "step": 1106 }, { "epoch": 0.28738317757009346, "grad_norm": 1.2869656085968018, "learning_rate": 4.849786561610702e-06, "loss": 0.12848493456840515, "step": 1107 }, { "epoch": 0.28764278296988577, "grad_norm": 2.4110817909240723, "learning_rate": 4.84944651498781e-06, "loss": 0.13020676374435425, "step": 1108 }, { "epoch": 0.2879023883696781, "grad_norm": 1.9734656810760498, "learning_rate": 4.849106097230845e-06, "loss": 0.137709379196167, "step": 1109 }, { "epoch": 0.2881619937694704, "grad_norm": 1.4477261304855347, "learning_rate": 4.848765308399987e-06, "loss": 0.150895893573761, "step": 1110 }, { "epoch": 0.28842159916926274, "grad_norm": 2.5002832412719727, "learning_rate": 4.848424148555476e-06, "loss": 0.15972110629081726, "step": 1111 }, { "epoch": 0.28868120456905505, "grad_norm": 0.8807445764541626, "learning_rate": 4.848082617757624e-06, "loss": 0.1278725117444992, "step": 1112 }, { "epoch": 0.28894080996884736, "grad_norm": 5.031692028045654, "learning_rate": 4.847740716066804e-06, "loss": 0.15422315895557404, "step": 1113 }, { "epoch": 0.28920041536863966, "grad_norm": 2.2700514793395996, "learning_rate": 4.847398443543459e-06, "loss": 0.13354602456092834, "step": 1114 }, { "epoch": 0.28946002076843197, "grad_norm": 1.0217019319534302, "learning_rate": 4.847055800248092e-06, "loss": 0.1587020456790924, "step": 1115 }, { "epoch": 0.2897196261682243, "grad_norm": 2.1552069187164307, "learning_rate": 4.846712786241275e-06, "loss": 0.1292034089565277, "step": 1116 }, { "epoch": 0.28997923156801664, "grad_norm": 2.4626693725585938, "learning_rate": 4.8463694015836465e-06, "loss": 0.15249082446098328, "step": 1117 }, { "epoch": 0.29023883696780894, "grad_norm": 2.0784294605255127, "learning_rate": 4.846025646335909e-06, "loss": 0.13872137665748596, "step": 1118 }, { "epoch": 0.29049844236760125, "grad_norm": 0.5407225489616394, "learning_rate": 4.845681520558829e-06, "loss": 0.12009967863559723, "step": 1119 }, { "epoch": 0.29075804776739356, "grad_norm": 1.4775320291519165, "learning_rate": 4.845337024313242e-06, "loss": 0.13873469829559326, "step": 1120 }, { "epoch": 0.29101765316718586, "grad_norm": 1.2001087665557861, "learning_rate": 4.844992157660044e-06, "loss": 0.12946678698062897, "step": 1121 }, { "epoch": 0.29127725856697817, "grad_norm": 1.3625552654266357, "learning_rate": 4.844646920660203e-06, "loss": 0.1495685577392578, "step": 1122 }, { "epoch": 0.29153686396677053, "grad_norm": 1.0010056495666504, "learning_rate": 4.844301313374747e-06, "loss": 0.13516896963119507, "step": 1123 }, { "epoch": 0.29179646936656284, "grad_norm": 2.278794527053833, "learning_rate": 4.843955335864772e-06, "loss": 0.12168221920728683, "step": 1124 }, { "epoch": 0.29205607476635514, "grad_norm": 0.7760934233665466, "learning_rate": 4.843608988191439e-06, "loss": 0.1402668058872223, "step": 1125 }, { "epoch": 0.29231568016614745, "grad_norm": 3.0485732555389404, "learning_rate": 4.843262270415974e-06, "loss": 0.15066450834274292, "step": 1126 }, { "epoch": 0.29257528556593976, "grad_norm": 2.3700551986694336, "learning_rate": 4.842915182599669e-06, "loss": 0.12599463760852814, "step": 1127 }, { "epoch": 0.29283489096573206, "grad_norm": 0.8695102334022522, "learning_rate": 4.842567724803881e-06, "loss": 0.1319536566734314, "step": 1128 }, { "epoch": 0.2930944963655244, "grad_norm": 2.8474037647247314, "learning_rate": 4.842219897090033e-06, "loss": 0.12553375959396362, "step": 1129 }, { "epoch": 0.29335410176531673, "grad_norm": 2.6172032356262207, "learning_rate": 4.8418716995196125e-06, "loss": 0.1537642776966095, "step": 1130 }, { "epoch": 0.29361370716510904, "grad_norm": 2.1750032901763916, "learning_rate": 4.841523132154173e-06, "loss": 0.15687911212444305, "step": 1131 }, { "epoch": 0.29387331256490135, "grad_norm": 2.399027109146118, "learning_rate": 4.841174195055333e-06, "loss": 0.11822056770324707, "step": 1132 }, { "epoch": 0.29413291796469365, "grad_norm": 4.245639801025391, "learning_rate": 4.8408248882847775e-06, "loss": 0.15645574033260345, "step": 1133 }, { "epoch": 0.29439252336448596, "grad_norm": 1.5319137573242188, "learning_rate": 4.8404752119042556e-06, "loss": 0.13296250998973846, "step": 1134 }, { "epoch": 0.2946521287642783, "grad_norm": 2.3213257789611816, "learning_rate": 4.8401251659755805e-06, "loss": 0.17153337597846985, "step": 1135 }, { "epoch": 0.2949117341640706, "grad_norm": 3.768371820449829, "learning_rate": 4.839774750560635e-06, "loss": 0.17187678813934326, "step": 1136 }, { "epoch": 0.29517133956386293, "grad_norm": 4.620259761810303, "learning_rate": 4.839423965721362e-06, "loss": 0.14530721306800842, "step": 1137 }, { "epoch": 0.29543094496365524, "grad_norm": 1.0519332885742188, "learning_rate": 4.839072811519773e-06, "loss": 0.15765774250030518, "step": 1138 }, { "epoch": 0.29569055036344755, "grad_norm": 1.7104848623275757, "learning_rate": 4.838721288017945e-06, "loss": 0.1302313208580017, "step": 1139 }, { "epoch": 0.29595015576323985, "grad_norm": 2.8678433895111084, "learning_rate": 4.838369395278019e-06, "loss": 0.1298464685678482, "step": 1140 }, { "epoch": 0.2962097611630322, "grad_norm": 3.3325984477996826, "learning_rate": 4.838017133362201e-06, "loss": 0.11966853588819504, "step": 1141 }, { "epoch": 0.2964693665628245, "grad_norm": 0.752282977104187, "learning_rate": 4.8376645023327625e-06, "loss": 0.1294786036014557, "step": 1142 }, { "epoch": 0.2967289719626168, "grad_norm": 1.1388508081436157, "learning_rate": 4.837311502252042e-06, "loss": 0.12969684600830078, "step": 1143 }, { "epoch": 0.29698857736240913, "grad_norm": 3.651008129119873, "learning_rate": 4.836958133182441e-06, "loss": 0.1748444139957428, "step": 1144 }, { "epoch": 0.29724818276220144, "grad_norm": 3.6618492603302, "learning_rate": 4.836604395186426e-06, "loss": 0.1757558286190033, "step": 1145 }, { "epoch": 0.29750778816199375, "grad_norm": 4.67765474319458, "learning_rate": 4.836250288326532e-06, "loss": 0.1455022692680359, "step": 1146 }, { "epoch": 0.2977673935617861, "grad_norm": 1.751347303390503, "learning_rate": 4.835895812665356e-06, "loss": 0.17372030019760132, "step": 1147 }, { "epoch": 0.2980269989615784, "grad_norm": 2.230644464492798, "learning_rate": 4.835540968265559e-06, "loss": 0.1212431788444519, "step": 1148 }, { "epoch": 0.2982866043613707, "grad_norm": 2.590515375137329, "learning_rate": 4.8351857551898734e-06, "loss": 0.1349814236164093, "step": 1149 }, { "epoch": 0.29854620976116303, "grad_norm": 4.870731830596924, "learning_rate": 4.83483017350109e-06, "loss": 0.18553702533245087, "step": 1150 }, { "epoch": 0.29880581516095533, "grad_norm": 8.301644325256348, "learning_rate": 4.834474223262068e-06, "loss": 0.18654093146324158, "step": 1151 }, { "epoch": 0.29906542056074764, "grad_norm": 2.4442553520202637, "learning_rate": 4.834117904535732e-06, "loss": 0.16780245304107666, "step": 1152 }, { "epoch": 0.29932502596054, "grad_norm": 6.275444030761719, "learning_rate": 4.83376121738507e-06, "loss": 0.14317497611045837, "step": 1153 }, { "epoch": 0.2995846313603323, "grad_norm": 6.201290130615234, "learning_rate": 4.833404161873137e-06, "loss": 0.14375914633274078, "step": 1154 }, { "epoch": 0.2998442367601246, "grad_norm": 4.955358505249023, "learning_rate": 4.8330467380630515e-06, "loss": 0.15549740195274353, "step": 1155 }, { "epoch": 0.3001038421599169, "grad_norm": 3.1143102645874023, "learning_rate": 4.8326889460179984e-06, "loss": 0.15798309445381165, "step": 1156 }, { "epoch": 0.30036344755970923, "grad_norm": 2.2452073097229004, "learning_rate": 4.832330785801227e-06, "loss": 0.15560847520828247, "step": 1157 }, { "epoch": 0.30062305295950154, "grad_norm": 4.34494686126709, "learning_rate": 4.8319722574760504e-06, "loss": 0.15959370136260986, "step": 1158 }, { "epoch": 0.3008826583592939, "grad_norm": 7.387198448181152, "learning_rate": 4.8316133611058504e-06, "loss": 0.14471286535263062, "step": 1159 }, { "epoch": 0.3011422637590862, "grad_norm": 3.3222920894622803, "learning_rate": 4.831254096754072e-06, "loss": 0.16793861985206604, "step": 1160 }, { "epoch": 0.3014018691588785, "grad_norm": 1.967986822128296, "learning_rate": 4.8308944644842225e-06, "loss": 0.14533104002475739, "step": 1161 }, { "epoch": 0.3016614745586708, "grad_norm": 0.8336844444274902, "learning_rate": 4.8305344643598774e-06, "loss": 0.13471782207489014, "step": 1162 }, { "epoch": 0.3019210799584631, "grad_norm": 2.281639337539673, "learning_rate": 4.830174096444678e-06, "loss": 0.15897879004478455, "step": 1163 }, { "epoch": 0.30218068535825543, "grad_norm": 3.940983533859253, "learning_rate": 4.8298133608023275e-06, "loss": 0.17698583006858826, "step": 1164 }, { "epoch": 0.3024402907580478, "grad_norm": 3.8960673809051514, "learning_rate": 4.829452257496596e-06, "loss": 0.13807019591331482, "step": 1165 }, { "epoch": 0.3026998961578401, "grad_norm": 2.3611483573913574, "learning_rate": 4.829090786591319e-06, "loss": 0.13074606657028198, "step": 1166 }, { "epoch": 0.3029595015576324, "grad_norm": 0.8580372929573059, "learning_rate": 4.828728948150395e-06, "loss": 0.12465144693851471, "step": 1167 }, { "epoch": 0.3032191069574247, "grad_norm": 1.0929538011550903, "learning_rate": 4.82836674223779e-06, "loss": 0.1326625943183899, "step": 1168 }, { "epoch": 0.303478712357217, "grad_norm": 1.5034730434417725, "learning_rate": 4.828004168917534e-06, "loss": 0.125589519739151, "step": 1169 }, { "epoch": 0.3037383177570093, "grad_norm": 2.5909583568573, "learning_rate": 4.8276412282537205e-06, "loss": 0.17029953002929688, "step": 1170 }, { "epoch": 0.3039979231568017, "grad_norm": 2.7083678245544434, "learning_rate": 4.8272779203105095e-06, "loss": 0.12284721434116364, "step": 1171 }, { "epoch": 0.304257528556594, "grad_norm": 1.9162516593933105, "learning_rate": 4.826914245152126e-06, "loss": 0.12509816884994507, "step": 1172 }, { "epoch": 0.3045171339563863, "grad_norm": 2.721066951751709, "learning_rate": 4.826550202842858e-06, "loss": 0.15269185602664948, "step": 1173 }, { "epoch": 0.3047767393561786, "grad_norm": 1.640368103981018, "learning_rate": 4.826185793447061e-06, "loss": 0.15299853682518005, "step": 1174 }, { "epoch": 0.3050363447559709, "grad_norm": 2.2917821407318115, "learning_rate": 4.825821017029154e-06, "loss": 0.15478214621543884, "step": 1175 }, { "epoch": 0.3052959501557632, "grad_norm": 1.2827141284942627, "learning_rate": 4.8254558736536216e-06, "loss": 0.13299506902694702, "step": 1176 }, { "epoch": 0.3055555555555556, "grad_norm": 1.4844963550567627, "learning_rate": 4.8250903633850125e-06, "loss": 0.13575094938278198, "step": 1177 }, { "epoch": 0.3058151609553479, "grad_norm": 1.8216427564620972, "learning_rate": 4.82472448628794e-06, "loss": 0.12491606175899506, "step": 1178 }, { "epoch": 0.3060747663551402, "grad_norm": 2.6060121059417725, "learning_rate": 4.824358242427083e-06, "loss": 0.16013772785663605, "step": 1179 }, { "epoch": 0.3063343717549325, "grad_norm": 0.9549803137779236, "learning_rate": 4.823991631867186e-06, "loss": 0.11886782944202423, "step": 1180 }, { "epoch": 0.3065939771547248, "grad_norm": 2.4394922256469727, "learning_rate": 4.8236246546730555e-06, "loss": 0.1605897843837738, "step": 1181 }, { "epoch": 0.3068535825545171, "grad_norm": 0.774270236492157, "learning_rate": 4.823257310909567e-06, "loss": 0.13725027441978455, "step": 1182 }, { "epoch": 0.3071131879543095, "grad_norm": 0.8005442023277283, "learning_rate": 4.822889600641657e-06, "loss": 0.12867426872253418, "step": 1183 }, { "epoch": 0.3073727933541018, "grad_norm": 2.9828004837036133, "learning_rate": 4.822521523934328e-06, "loss": 0.16995370388031006, "step": 1184 }, { "epoch": 0.3076323987538941, "grad_norm": 0.848332941532135, "learning_rate": 4.822153080852648e-06, "loss": 0.11246855556964874, "step": 1185 }, { "epoch": 0.3078920041536864, "grad_norm": 2.1935858726501465, "learning_rate": 4.82178427146175e-06, "loss": 0.164751797914505, "step": 1186 }, { "epoch": 0.3081516095534787, "grad_norm": 0.9114489555358887, "learning_rate": 4.82141509582683e-06, "loss": 0.1256294995546341, "step": 1187 }, { "epoch": 0.308411214953271, "grad_norm": 1.1428560018539429, "learning_rate": 4.821045554013151e-06, "loss": 0.1443510204553604, "step": 1188 }, { "epoch": 0.30867082035306337, "grad_norm": 1.039798378944397, "learning_rate": 4.820675646086037e-06, "loss": 0.12323667109012604, "step": 1189 }, { "epoch": 0.3089304257528557, "grad_norm": 1.855111002922058, "learning_rate": 4.8203053721108815e-06, "loss": 0.12394946068525314, "step": 1190 }, { "epoch": 0.309190031152648, "grad_norm": 1.05953848361969, "learning_rate": 4.819934732153141e-06, "loss": 0.1127692312002182, "step": 1191 }, { "epoch": 0.3094496365524403, "grad_norm": 1.5053564310073853, "learning_rate": 4.819563726278334e-06, "loss": 0.12069907784461975, "step": 1192 }, { "epoch": 0.3097092419522326, "grad_norm": 0.9713023900985718, "learning_rate": 4.819192354552047e-06, "loss": 0.12135916203260422, "step": 1193 }, { "epoch": 0.3099688473520249, "grad_norm": 2.0894203186035156, "learning_rate": 4.81882061703993e-06, "loss": 0.1624462604522705, "step": 1194 }, { "epoch": 0.31022845275181726, "grad_norm": 1.6126290559768677, "learning_rate": 4.818448513807698e-06, "loss": 0.1564202606678009, "step": 1195 }, { "epoch": 0.31048805815160957, "grad_norm": 0.7977124452590942, "learning_rate": 4.818076044921129e-06, "loss": 0.11904416978359222, "step": 1196 }, { "epoch": 0.3107476635514019, "grad_norm": 1.729129672050476, "learning_rate": 4.817703210446069e-06, "loss": 0.14569757878780365, "step": 1197 }, { "epoch": 0.3110072689511942, "grad_norm": 1.1132768392562866, "learning_rate": 4.817330010448426e-06, "loss": 0.1092890053987503, "step": 1198 }, { "epoch": 0.3112668743509865, "grad_norm": 2.137937545776367, "learning_rate": 4.816956444994172e-06, "loss": 0.15640507638454437, "step": 1199 }, { "epoch": 0.3115264797507788, "grad_norm": 1.605526328086853, "learning_rate": 4.816582514149344e-06, "loss": 0.12689423561096191, "step": 1200 }, { "epoch": 0.31178608515057116, "grad_norm": 1.2414095401763916, "learning_rate": 4.816208217980049e-06, "loss": 0.12139013409614563, "step": 1201 }, { "epoch": 0.31204569055036346, "grad_norm": 1.9485613107681274, "learning_rate": 4.815833556552448e-06, "loss": 0.12950636446475983, "step": 1202 }, { "epoch": 0.31230529595015577, "grad_norm": 0.9680469632148743, "learning_rate": 4.815458529932778e-06, "loss": 0.11622309684753418, "step": 1203 }, { "epoch": 0.3125649013499481, "grad_norm": 3.300347328186035, "learning_rate": 4.815083138187331e-06, "loss": 0.11541875451803207, "step": 1204 }, { "epoch": 0.3128245067497404, "grad_norm": 1.1451544761657715, "learning_rate": 4.8147073813824695e-06, "loss": 0.13052532076835632, "step": 1205 }, { "epoch": 0.3130841121495327, "grad_norm": 1.4349567890167236, "learning_rate": 4.814331259584619e-06, "loss": 0.1392824947834015, "step": 1206 }, { "epoch": 0.31334371754932505, "grad_norm": 1.551397442817688, "learning_rate": 4.8139547728602685e-06, "loss": 0.15960052609443665, "step": 1207 }, { "epoch": 0.31360332294911736, "grad_norm": 1.3938751220703125, "learning_rate": 4.8135779212759725e-06, "loss": 0.13569679856300354, "step": 1208 }, { "epoch": 0.31386292834890966, "grad_norm": 1.0428847074508667, "learning_rate": 4.81320070489835e-06, "loss": 0.11142078042030334, "step": 1209 }, { "epoch": 0.31412253374870197, "grad_norm": 1.878755807876587, "learning_rate": 4.812823123794084e-06, "loss": 0.14798250794410706, "step": 1210 }, { "epoch": 0.3143821391484943, "grad_norm": 1.406832218170166, "learning_rate": 4.8124451780299215e-06, "loss": 0.12235084921121597, "step": 1211 }, { "epoch": 0.3146417445482866, "grad_norm": 2.0433108806610107, "learning_rate": 4.812066867672676e-06, "loss": 0.15130537748336792, "step": 1212 }, { "epoch": 0.31490134994807895, "grad_norm": 2.5990915298461914, "learning_rate": 4.8116881927892225e-06, "loss": 0.15794870257377625, "step": 1213 }, { "epoch": 0.31516095534787125, "grad_norm": 1.810726523399353, "learning_rate": 4.811309153446503e-06, "loss": 0.17274168133735657, "step": 1214 }, { "epoch": 0.31542056074766356, "grad_norm": 1.5162819623947144, "learning_rate": 4.810929749711523e-06, "loss": 0.14503133296966553, "step": 1215 }, { "epoch": 0.31568016614745587, "grad_norm": 2.639643907546997, "learning_rate": 4.810549981651353e-06, "loss": 0.1663541942834854, "step": 1216 }, { "epoch": 0.31593977154724817, "grad_norm": 2.1491265296936035, "learning_rate": 4.810169849333126e-06, "loss": 0.14511092007160187, "step": 1217 }, { "epoch": 0.3161993769470405, "grad_norm": 1.0302734375, "learning_rate": 4.8097893528240415e-06, "loss": 0.12614251673221588, "step": 1218 }, { "epoch": 0.31645898234683284, "grad_norm": 2.476576089859009, "learning_rate": 4.809408492191362e-06, "loss": 0.175726979970932, "step": 1219 }, { "epoch": 0.31671858774662515, "grad_norm": 0.6385703682899475, "learning_rate": 4.809027267502415e-06, "loss": 0.12339694797992706, "step": 1220 }, { "epoch": 0.31697819314641745, "grad_norm": 0.8074113726615906, "learning_rate": 4.808645678824593e-06, "loss": 0.10954301059246063, "step": 1221 }, { "epoch": 0.31723779854620976, "grad_norm": 3.5567822456359863, "learning_rate": 4.8082637262253515e-06, "loss": 0.16413450241088867, "step": 1222 }, { "epoch": 0.31749740394600207, "grad_norm": 4.334929466247559, "learning_rate": 4.807881409772211e-06, "loss": 0.12371611595153809, "step": 1223 }, { "epoch": 0.3177570093457944, "grad_norm": 1.5111033916473389, "learning_rate": 4.807498729532758e-06, "loss": 0.16841164231300354, "step": 1224 }, { "epoch": 0.31801661474558673, "grad_norm": 2.4721288681030273, "learning_rate": 4.807115685574638e-06, "loss": 0.1511378437280655, "step": 1225 }, { "epoch": 0.31827622014537904, "grad_norm": 1.9976869821548462, "learning_rate": 4.806732277965569e-06, "loss": 0.12472950667142868, "step": 1226 }, { "epoch": 0.31853582554517135, "grad_norm": 1.7540998458862305, "learning_rate": 4.806348506773325e-06, "loss": 0.1691323071718216, "step": 1227 }, { "epoch": 0.31879543094496365, "grad_norm": 2.6258416175842285, "learning_rate": 4.805964372065748e-06, "loss": 0.1451742798089981, "step": 1228 }, { "epoch": 0.31905503634475596, "grad_norm": 2.0386033058166504, "learning_rate": 4.805579873910746e-06, "loss": 0.14644470810890198, "step": 1229 }, { "epoch": 0.31931464174454827, "grad_norm": 2.120169162750244, "learning_rate": 4.805195012376289e-06, "loss": 0.15577593445777893, "step": 1230 }, { "epoch": 0.31957424714434063, "grad_norm": 1.568080186843872, "learning_rate": 4.804809787530411e-06, "loss": 0.1565237045288086, "step": 1231 }, { "epoch": 0.31983385254413293, "grad_norm": 4.989901542663574, "learning_rate": 4.804424199441213e-06, "loss": 0.19405540823936462, "step": 1232 }, { "epoch": 0.32009345794392524, "grad_norm": 0.7814015746116638, "learning_rate": 4.804038248176855e-06, "loss": 0.11944444477558136, "step": 1233 }, { "epoch": 0.32035306334371755, "grad_norm": 1.38551926612854, "learning_rate": 4.803651933805567e-06, "loss": 0.1319662481546402, "step": 1234 }, { "epoch": 0.32061266874350985, "grad_norm": 0.9461792707443237, "learning_rate": 4.803265256395639e-06, "loss": 0.13011668622493744, "step": 1235 }, { "epoch": 0.32087227414330216, "grad_norm": 1.6176488399505615, "learning_rate": 4.802878216015428e-06, "loss": 0.1287858933210373, "step": 1236 }, { "epoch": 0.3211318795430945, "grad_norm": 2.565136432647705, "learning_rate": 4.802490812733352e-06, "loss": 0.1395796537399292, "step": 1237 }, { "epoch": 0.32139148494288683, "grad_norm": 2.1966874599456787, "learning_rate": 4.8021030466178965e-06, "loss": 0.15525230765342712, "step": 1238 }, { "epoch": 0.32165109034267914, "grad_norm": 3.5897068977355957, "learning_rate": 4.80171491773761e-06, "loss": 0.136399507522583, "step": 1239 }, { "epoch": 0.32191069574247144, "grad_norm": 4.817693710327148, "learning_rate": 4.801326426161103e-06, "loss": 0.16222134232521057, "step": 1240 }, { "epoch": 0.32217030114226375, "grad_norm": 2.423532724380493, "learning_rate": 4.800937571957053e-06, "loss": 0.12520800530910492, "step": 1241 }, { "epoch": 0.32242990654205606, "grad_norm": 1.3277840614318848, "learning_rate": 4.800548355194202e-06, "loss": 0.13568684458732605, "step": 1242 }, { "epoch": 0.3226895119418484, "grad_norm": 3.4006168842315674, "learning_rate": 4.800158775941351e-06, "loss": 0.13155308365821838, "step": 1243 }, { "epoch": 0.3229491173416407, "grad_norm": 3.6081559658050537, "learning_rate": 4.799768834267372e-06, "loss": 0.1444581151008606, "step": 1244 }, { "epoch": 0.32320872274143303, "grad_norm": 1.76287841796875, "learning_rate": 4.799378530241197e-06, "loss": 0.1292300969362259, "step": 1245 }, { "epoch": 0.32346832814122534, "grad_norm": 1.8288500308990479, "learning_rate": 4.7989878639318225e-06, "loss": 0.13797488808631897, "step": 1246 }, { "epoch": 0.32372793354101764, "grad_norm": 2.7724905014038086, "learning_rate": 4.798596835408309e-06, "loss": 0.13673292100429535, "step": 1247 }, { "epoch": 0.32398753894080995, "grad_norm": 2.931143283843994, "learning_rate": 4.79820544473978e-06, "loss": 0.12598995864391327, "step": 1248 }, { "epoch": 0.3242471443406023, "grad_norm": 2.1413769721984863, "learning_rate": 4.797813691995428e-06, "loss": 0.1410355120897293, "step": 1249 }, { "epoch": 0.3245067497403946, "grad_norm": 1.790583848953247, "learning_rate": 4.7974215772445035e-06, "loss": 0.16929729282855988, "step": 1250 }, { "epoch": 0.3247663551401869, "grad_norm": 0.7569156289100647, "learning_rate": 4.797029100556324e-06, "loss": 0.13195672631263733, "step": 1251 }, { "epoch": 0.32502596053997923, "grad_norm": 0.8179608583450317, "learning_rate": 4.796636262000269e-06, "loss": 0.12759675085544586, "step": 1252 }, { "epoch": 0.32528556593977154, "grad_norm": 2.193589687347412, "learning_rate": 4.796243061645786e-06, "loss": 0.13162371516227722, "step": 1253 }, { "epoch": 0.32554517133956384, "grad_norm": 1.2831205129623413, "learning_rate": 4.7958494995623824e-06, "loss": 0.16209332644939423, "step": 1254 }, { "epoch": 0.3258047767393562, "grad_norm": 1.1902552843093872, "learning_rate": 4.79545557581963e-06, "loss": 0.12981705367565155, "step": 1255 }, { "epoch": 0.3260643821391485, "grad_norm": 2.3497064113616943, "learning_rate": 4.795061290487168e-06, "loss": 0.13767890632152557, "step": 1256 }, { "epoch": 0.3263239875389408, "grad_norm": 3.076753854751587, "learning_rate": 4.794666643634695e-06, "loss": 0.13010655343532562, "step": 1257 }, { "epoch": 0.3265835929387331, "grad_norm": 1.3243298530578613, "learning_rate": 4.794271635331975e-06, "loss": 0.12392552196979523, "step": 1258 }, { "epoch": 0.32684319833852543, "grad_norm": 0.8884090185165405, "learning_rate": 4.793876265648839e-06, "loss": 0.1356445848941803, "step": 1259 }, { "epoch": 0.32710280373831774, "grad_norm": 2.5551235675811768, "learning_rate": 4.793480534655177e-06, "loss": 0.15771010518074036, "step": 1260 }, { "epoch": 0.3273624091381101, "grad_norm": 1.8328784704208374, "learning_rate": 4.793084442420946e-06, "loss": 0.1221461296081543, "step": 1261 }, { "epoch": 0.3276220145379024, "grad_norm": 2.518007278442383, "learning_rate": 4.792687989016166e-06, "loss": 0.15384525060653687, "step": 1262 }, { "epoch": 0.3278816199376947, "grad_norm": 1.2468063831329346, "learning_rate": 4.792291174510921e-06, "loss": 0.12704092264175415, "step": 1263 }, { "epoch": 0.328141225337487, "grad_norm": 0.7482008934020996, "learning_rate": 4.791893998975359e-06, "loss": 0.11943459510803223, "step": 1264 }, { "epoch": 0.3284008307372793, "grad_norm": 5.171235084533691, "learning_rate": 4.791496462479691e-06, "loss": 0.15804004669189453, "step": 1265 }, { "epoch": 0.32866043613707163, "grad_norm": 2.7809178829193115, "learning_rate": 4.791098565094192e-06, "loss": 0.1761743128299713, "step": 1266 }, { "epoch": 0.328920041536864, "grad_norm": 2.637300491333008, "learning_rate": 4.790700306889202e-06, "loss": 0.16140590608119965, "step": 1267 }, { "epoch": 0.3291796469366563, "grad_norm": 3.743769645690918, "learning_rate": 4.790301687935124e-06, "loss": 0.16317233443260193, "step": 1268 }, { "epoch": 0.3294392523364486, "grad_norm": 0.853007972240448, "learning_rate": 4.789902708302423e-06, "loss": 0.1235513687133789, "step": 1269 }, { "epoch": 0.3296988577362409, "grad_norm": 1.2829618453979492, "learning_rate": 4.789503368061632e-06, "loss": 0.1615789830684662, "step": 1270 }, { "epoch": 0.3299584631360332, "grad_norm": 1.9085426330566406, "learning_rate": 4.789103667283344e-06, "loss": 0.13886751234531403, "step": 1271 }, { "epoch": 0.3302180685358255, "grad_norm": 2.7287375926971436, "learning_rate": 4.788703606038216e-06, "loss": 0.15349465608596802, "step": 1272 }, { "epoch": 0.3304776739356179, "grad_norm": 1.1649278402328491, "learning_rate": 4.788303184396971e-06, "loss": 0.14382773637771606, "step": 1273 }, { "epoch": 0.3307372793354102, "grad_norm": 1.1768983602523804, "learning_rate": 4.787902402430395e-06, "loss": 0.12967854738235474, "step": 1274 }, { "epoch": 0.3309968847352025, "grad_norm": 1.1038200855255127, "learning_rate": 4.787501260209334e-06, "loss": 0.131271094083786, "step": 1275 }, { "epoch": 0.3312564901349948, "grad_norm": 1.0042675733566284, "learning_rate": 4.7870997578047035e-06, "loss": 0.1247377097606659, "step": 1276 }, { "epoch": 0.3315160955347871, "grad_norm": 8.061427116394043, "learning_rate": 4.7866978952874795e-06, "loss": 0.1590379774570465, "step": 1277 }, { "epoch": 0.3317757009345794, "grad_norm": 2.327861785888672, "learning_rate": 4.786295672728702e-06, "loss": 0.1306612193584442, "step": 1278 }, { "epoch": 0.3320353063343718, "grad_norm": 1.3335493803024292, "learning_rate": 4.785893090199475e-06, "loss": 0.13973119854927063, "step": 1279 }, { "epoch": 0.3322949117341641, "grad_norm": 1.7925026416778564, "learning_rate": 4.785490147770964e-06, "loss": 0.1469494253396988, "step": 1280 }, { "epoch": 0.3325545171339564, "grad_norm": 1.2801513671875, "learning_rate": 4.785086845514403e-06, "loss": 0.14951840043067932, "step": 1281 }, { "epoch": 0.3328141225337487, "grad_norm": 3.470902919769287, "learning_rate": 4.784683183501084e-06, "loss": 0.17273005843162537, "step": 1282 }, { "epoch": 0.333073727933541, "grad_norm": 1.9422553777694702, "learning_rate": 4.7842791618023666e-06, "loss": 0.15352077782154083, "step": 1283 }, { "epoch": 0.3333333333333333, "grad_norm": 1.6351075172424316, "learning_rate": 4.783874780489673e-06, "loss": 0.13277655839920044, "step": 1284 }, { "epoch": 0.3335929387331256, "grad_norm": 1.1446763277053833, "learning_rate": 4.783470039634486e-06, "loss": 0.14734745025634766, "step": 1285 }, { "epoch": 0.333852544132918, "grad_norm": 2.0374808311462402, "learning_rate": 4.783064939308357e-06, "loss": 0.18203315138816833, "step": 1286 }, { "epoch": 0.3341121495327103, "grad_norm": 0.9554693698883057, "learning_rate": 4.782659479582898e-06, "loss": 0.125034362077713, "step": 1287 }, { "epoch": 0.3343717549325026, "grad_norm": 1.2217227220535278, "learning_rate": 4.782253660529785e-06, "loss": 0.12495367228984833, "step": 1288 }, { "epoch": 0.3346313603322949, "grad_norm": 3.4374866485595703, "learning_rate": 4.781847482220757e-06, "loss": 0.19336797297000885, "step": 1289 }, { "epoch": 0.3348909657320872, "grad_norm": 1.843916654586792, "learning_rate": 4.781440944727617e-06, "loss": 0.150421142578125, "step": 1290 }, { "epoch": 0.3351505711318795, "grad_norm": 0.525956928730011, "learning_rate": 4.781034048122232e-06, "loss": 0.11838845908641815, "step": 1291 }, { "epoch": 0.3354101765316719, "grad_norm": 2.2368688583374023, "learning_rate": 4.780626792476532e-06, "loss": 0.17452669143676758, "step": 1292 }, { "epoch": 0.3356697819314642, "grad_norm": 2.224745273590088, "learning_rate": 4.78021917786251e-06, "loss": 0.12927159667015076, "step": 1293 }, { "epoch": 0.3359293873312565, "grad_norm": 0.9437210559844971, "learning_rate": 4.779811204352223e-06, "loss": 0.13391534984111786, "step": 1294 }, { "epoch": 0.3361889927310488, "grad_norm": 0.6494242548942566, "learning_rate": 4.779402872017792e-06, "loss": 0.11572268605232239, "step": 1295 }, { "epoch": 0.3364485981308411, "grad_norm": 0.9112452268600464, "learning_rate": 4.778994180931401e-06, "loss": 0.13275375962257385, "step": 1296 }, { "epoch": 0.3367082035306334, "grad_norm": 2.3811073303222656, "learning_rate": 4.778585131165295e-06, "loss": 0.14783212542533875, "step": 1297 }, { "epoch": 0.33696780893042577, "grad_norm": 2.030470848083496, "learning_rate": 4.778175722791788e-06, "loss": 0.08870754390954971, "step": 1298 }, { "epoch": 0.3372274143302181, "grad_norm": 3.063246011734009, "learning_rate": 4.777765955883251e-06, "loss": 0.1401691734790802, "step": 1299 }, { "epoch": 0.3374870197300104, "grad_norm": 2.403984785079956, "learning_rate": 4.777355830512122e-06, "loss": 0.18655511736869812, "step": 1300 }, { "epoch": 0.3377466251298027, "grad_norm": 3.0574798583984375, "learning_rate": 4.776945346750905e-06, "loss": 0.1302807480096817, "step": 1301 }, { "epoch": 0.338006230529595, "grad_norm": 1.8302069902420044, "learning_rate": 4.776534504672159e-06, "loss": 0.13279762864112854, "step": 1302 }, { "epoch": 0.3382658359293873, "grad_norm": 3.8734750747680664, "learning_rate": 4.7761233043485165e-06, "loss": 0.13238279521465302, "step": 1303 }, { "epoch": 0.33852544132917967, "grad_norm": 1.8024629354476929, "learning_rate": 4.7757117458526645e-06, "loss": 0.14472410082817078, "step": 1304 }, { "epoch": 0.338785046728972, "grad_norm": 1.6786203384399414, "learning_rate": 4.775299829257358e-06, "loss": 0.16379514336585999, "step": 1305 }, { "epoch": 0.3390446521287643, "grad_norm": 4.498153209686279, "learning_rate": 4.7748875546354155e-06, "loss": 0.17374038696289062, "step": 1306 }, { "epoch": 0.3393042575285566, "grad_norm": 1.3009097576141357, "learning_rate": 4.774474922059718e-06, "loss": 0.12988924980163574, "step": 1307 }, { "epoch": 0.3395638629283489, "grad_norm": 2.83691143989563, "learning_rate": 4.774061931603207e-06, "loss": 0.11916691064834595, "step": 1308 }, { "epoch": 0.3398234683281412, "grad_norm": 0.7402837872505188, "learning_rate": 4.773648583338892e-06, "loss": 0.127017542719841, "step": 1309 }, { "epoch": 0.34008307372793356, "grad_norm": 1.5330890417099, "learning_rate": 4.773234877339843e-06, "loss": 0.1294129192829132, "step": 1310 }, { "epoch": 0.34034267912772587, "grad_norm": 0.7729427814483643, "learning_rate": 4.772820813679194e-06, "loss": 0.12914878129959106, "step": 1311 }, { "epoch": 0.3406022845275182, "grad_norm": 3.3838863372802734, "learning_rate": 4.772406392430141e-06, "loss": 0.15667110681533813, "step": 1312 }, { "epoch": 0.3408618899273105, "grad_norm": 1.1086317300796509, "learning_rate": 4.771991613665945e-06, "loss": 0.1536780744791031, "step": 1313 }, { "epoch": 0.3411214953271028, "grad_norm": 0.943175196647644, "learning_rate": 4.771576477459929e-06, "loss": 0.12733261287212372, "step": 1314 }, { "epoch": 0.3413811007268951, "grad_norm": 0.8041452169418335, "learning_rate": 4.77116098388548e-06, "loss": 0.1323360800743103, "step": 1315 }, { "epoch": 0.34164070612668745, "grad_norm": 2.519228219985962, "learning_rate": 4.770745133016048e-06, "loss": 0.1535225808620453, "step": 1316 }, { "epoch": 0.34190031152647976, "grad_norm": 0.8277589082717896, "learning_rate": 4.770328924925145e-06, "loss": 0.10778428614139557, "step": 1317 }, { "epoch": 0.34215991692627207, "grad_norm": 1.2247110605239868, "learning_rate": 4.769912359686346e-06, "loss": 0.1436862051486969, "step": 1318 }, { "epoch": 0.3424195223260644, "grad_norm": 0.7524539828300476, "learning_rate": 4.7694954373732935e-06, "loss": 0.12064646929502487, "step": 1319 }, { "epoch": 0.3426791277258567, "grad_norm": 0.9462288618087769, "learning_rate": 4.769078158059687e-06, "loss": 0.13636374473571777, "step": 1320 }, { "epoch": 0.342938733125649, "grad_norm": 1.397720456123352, "learning_rate": 4.7686605218192935e-06, "loss": 0.15718896687030792, "step": 1321 }, { "epoch": 0.34319833852544135, "grad_norm": 1.202327847480774, "learning_rate": 4.768242528725941e-06, "loss": 0.15090316534042358, "step": 1322 }, { "epoch": 0.34345794392523366, "grad_norm": 1.0158737897872925, "learning_rate": 4.767824178853521e-06, "loss": 0.13194935023784637, "step": 1323 }, { "epoch": 0.34371754932502596, "grad_norm": 1.7510415315628052, "learning_rate": 4.767405472275988e-06, "loss": 0.13532494008541107, "step": 1324 }, { "epoch": 0.34397715472481827, "grad_norm": 0.9724754095077515, "learning_rate": 4.76698640906736e-06, "loss": 0.1219855397939682, "step": 1325 }, { "epoch": 0.3442367601246106, "grad_norm": 1.6009773015975952, "learning_rate": 4.766566989301718e-06, "loss": 0.16231125593185425, "step": 1326 }, { "epoch": 0.3444963655244029, "grad_norm": 1.8358746767044067, "learning_rate": 4.7661472130532056e-06, "loss": 0.1501385122537613, "step": 1327 }, { "epoch": 0.34475597092419524, "grad_norm": 1.4794831275939941, "learning_rate": 4.7657270803960285e-06, "loss": 0.13188841938972473, "step": 1328 }, { "epoch": 0.34501557632398755, "grad_norm": 3.348705530166626, "learning_rate": 4.765306591404459e-06, "loss": 0.1527765393257141, "step": 1329 }, { "epoch": 0.34527518172377986, "grad_norm": 0.6864472031593323, "learning_rate": 4.764885746152828e-06, "loss": 0.11759910732507706, "step": 1330 }, { "epoch": 0.34553478712357216, "grad_norm": 1.7684026956558228, "learning_rate": 4.764464544715532e-06, "loss": 0.13177308440208435, "step": 1331 }, { "epoch": 0.34579439252336447, "grad_norm": 1.560613751411438, "learning_rate": 4.764042987167029e-06, "loss": 0.146237313747406, "step": 1332 }, { "epoch": 0.3460539979231568, "grad_norm": 1.0644557476043701, "learning_rate": 4.763621073581842e-06, "loss": 0.1351773738861084, "step": 1333 }, { "epoch": 0.34631360332294914, "grad_norm": 1.3204313516616821, "learning_rate": 4.7631988040345554e-06, "loss": 0.13061848282814026, "step": 1334 }, { "epoch": 0.34657320872274144, "grad_norm": 1.4287225008010864, "learning_rate": 4.762776178599815e-06, "loss": 0.11928154528141022, "step": 1335 }, { "epoch": 0.34683281412253375, "grad_norm": 1.84377121925354, "learning_rate": 4.762353197352333e-06, "loss": 0.1562492549419403, "step": 1336 }, { "epoch": 0.34709241952232606, "grad_norm": 0.8110661506652832, "learning_rate": 4.761929860366883e-06, "loss": 0.13265369832515717, "step": 1337 }, { "epoch": 0.34735202492211836, "grad_norm": 0.7431944608688354, "learning_rate": 4.761506167718302e-06, "loss": 0.11106958985328674, "step": 1338 }, { "epoch": 0.34761163032191067, "grad_norm": 1.6822311878204346, "learning_rate": 4.761082119481486e-06, "loss": 0.13025081157684326, "step": 1339 }, { "epoch": 0.34787123572170303, "grad_norm": 3.363055467605591, "learning_rate": 4.7606577157314e-06, "loss": 0.15004131197929382, "step": 1340 }, { "epoch": 0.34813084112149534, "grad_norm": 1.5767459869384766, "learning_rate": 4.760232956543067e-06, "loss": 0.12689879536628723, "step": 1341 }, { "epoch": 0.34839044652128764, "grad_norm": 2.1230316162109375, "learning_rate": 4.759807841991576e-06, "loss": 0.1425706148147583, "step": 1342 }, { "epoch": 0.34865005192107995, "grad_norm": 2.1121644973754883, "learning_rate": 4.759382372152077e-06, "loss": 0.15262535214424133, "step": 1343 }, { "epoch": 0.34890965732087226, "grad_norm": 1.2953071594238281, "learning_rate": 4.7589565470997836e-06, "loss": 0.12977153062820435, "step": 1344 }, { "epoch": 0.34916926272066456, "grad_norm": 2.530599355697632, "learning_rate": 4.758530366909971e-06, "loss": 0.123292475938797, "step": 1345 }, { "epoch": 0.3494288681204569, "grad_norm": 1.4470173120498657, "learning_rate": 4.758103831657979e-06, "loss": 0.14258038997650146, "step": 1346 }, { "epoch": 0.34968847352024923, "grad_norm": 2.6269707679748535, "learning_rate": 4.75767694141921e-06, "loss": 0.15993264317512512, "step": 1347 }, { "epoch": 0.34994807892004154, "grad_norm": 2.4330787658691406, "learning_rate": 4.757249696269126e-06, "loss": 0.15390712022781372, "step": 1348 }, { "epoch": 0.35020768431983385, "grad_norm": 3.2105419635772705, "learning_rate": 4.756822096283257e-06, "loss": 0.16689282655715942, "step": 1349 }, { "epoch": 0.35046728971962615, "grad_norm": 3.724642276763916, "learning_rate": 4.7563941415371905e-06, "loss": 0.13237708806991577, "step": 1350 }, { "epoch": 0.35072689511941846, "grad_norm": 2.3832507133483887, "learning_rate": 4.755965832106581e-06, "loss": 0.1745269000530243, "step": 1351 }, { "epoch": 0.3509865005192108, "grad_norm": 1.6217633485794067, "learning_rate": 4.755537168067142e-06, "loss": 0.1631132960319519, "step": 1352 }, { "epoch": 0.3512461059190031, "grad_norm": 1.443334698677063, "learning_rate": 4.755108149494653e-06, "loss": 0.12856481969356537, "step": 1353 }, { "epoch": 0.35150571131879543, "grad_norm": 1.0513161420822144, "learning_rate": 4.754678776464954e-06, "loss": 0.13247381150722504, "step": 1354 }, { "epoch": 0.35176531671858774, "grad_norm": 1.925528883934021, "learning_rate": 4.754249049053948e-06, "loss": 0.159701406955719, "step": 1355 }, { "epoch": 0.35202492211838005, "grad_norm": 2.9159605503082275, "learning_rate": 4.753818967337602e-06, "loss": 0.15498420596122742, "step": 1356 }, { "epoch": 0.35228452751817235, "grad_norm": 2.0149927139282227, "learning_rate": 4.753388531391943e-06, "loss": 0.16328221559524536, "step": 1357 }, { "epoch": 0.3525441329179647, "grad_norm": 1.3578166961669922, "learning_rate": 4.752957741293064e-06, "loss": 0.14162370562553406, "step": 1358 }, { "epoch": 0.352803738317757, "grad_norm": 1.3818031549453735, "learning_rate": 4.752526597117117e-06, "loss": 0.13618120551109314, "step": 1359 }, { "epoch": 0.3530633437175493, "grad_norm": 1.9982399940490723, "learning_rate": 4.7520950989403205e-06, "loss": 0.15298154950141907, "step": 1360 }, { "epoch": 0.35332294911734163, "grad_norm": 0.864059567451477, "learning_rate": 4.751663246838953e-06, "loss": 0.14445117115974426, "step": 1361 }, { "epoch": 0.35358255451713394, "grad_norm": 0.9163714051246643, "learning_rate": 4.751231040889354e-06, "loss": 0.12458440661430359, "step": 1362 }, { "epoch": 0.35384215991692625, "grad_norm": 2.0528664588928223, "learning_rate": 4.7507984811679296e-06, "loss": 0.1491609811782837, "step": 1363 }, { "epoch": 0.3541017653167186, "grad_norm": 1.5978271961212158, "learning_rate": 4.750365567751146e-06, "loss": 0.163880854845047, "step": 1364 }, { "epoch": 0.3543613707165109, "grad_norm": 1.253422498703003, "learning_rate": 4.749932300715532e-06, "loss": 0.11688355356454849, "step": 1365 }, { "epoch": 0.3546209761163032, "grad_norm": 0.9675057530403137, "learning_rate": 4.74949868013768e-06, "loss": 0.13172397017478943, "step": 1366 }, { "epoch": 0.35488058151609553, "grad_norm": 1.3005214929580688, "learning_rate": 4.749064706094244e-06, "loss": 0.13720718026161194, "step": 1367 }, { "epoch": 0.35514018691588783, "grad_norm": 2.0793802738189697, "learning_rate": 4.74863037866194e-06, "loss": 0.11811662465333939, "step": 1368 }, { "epoch": 0.35539979231568014, "grad_norm": 4.162749767303467, "learning_rate": 4.748195697917547e-06, "loss": 0.15809419751167297, "step": 1369 }, { "epoch": 0.3556593977154725, "grad_norm": 1.371882438659668, "learning_rate": 4.747760663937908e-06, "loss": 0.15603958070278168, "step": 1370 }, { "epoch": 0.3559190031152648, "grad_norm": 3.6957037448883057, "learning_rate": 4.747325276799927e-06, "loss": 0.1268918365240097, "step": 1371 }, { "epoch": 0.3561786085150571, "grad_norm": 5.240827560424805, "learning_rate": 4.746889536580569e-06, "loss": 0.16172254085540771, "step": 1372 }, { "epoch": 0.3564382139148494, "grad_norm": 1.5669804811477661, "learning_rate": 4.7464534433568634e-06, "loss": 0.13249549269676208, "step": 1373 }, { "epoch": 0.35669781931464173, "grad_norm": 2.4164023399353027, "learning_rate": 4.7460169972059015e-06, "loss": 0.1206829696893692, "step": 1374 }, { "epoch": 0.35695742471443404, "grad_norm": 3.752718210220337, "learning_rate": 4.7455801982048374e-06, "loss": 0.1349913477897644, "step": 1375 }, { "epoch": 0.3572170301142264, "grad_norm": 2.6784324645996094, "learning_rate": 4.7451430464308875e-06, "loss": 0.12411671876907349, "step": 1376 }, { "epoch": 0.3574766355140187, "grad_norm": 1.4177985191345215, "learning_rate": 4.74470554196133e-06, "loss": 0.15651999413967133, "step": 1377 }, { "epoch": 0.357736240913811, "grad_norm": 1.2545338869094849, "learning_rate": 4.744267684873505e-06, "loss": 0.13325577974319458, "step": 1378 }, { "epoch": 0.3579958463136033, "grad_norm": 2.5165629386901855, "learning_rate": 4.743829475244816e-06, "loss": 0.12125176936388016, "step": 1379 }, { "epoch": 0.3582554517133956, "grad_norm": 2.3791534900665283, "learning_rate": 4.743390913152728e-06, "loss": 0.12560823559761047, "step": 1380 }, { "epoch": 0.35851505711318793, "grad_norm": 2.5755131244659424, "learning_rate": 4.74295199867477e-06, "loss": 0.13886865973472595, "step": 1381 }, { "epoch": 0.3587746625129803, "grad_norm": 1.9189479351043701, "learning_rate": 4.74251273188853e-06, "loss": 0.1425943225622177, "step": 1382 }, { "epoch": 0.3590342679127726, "grad_norm": 2.579998016357422, "learning_rate": 4.742073112871663e-06, "loss": 0.1336911916732788, "step": 1383 }, { "epoch": 0.3592938733125649, "grad_norm": 1.9923722743988037, "learning_rate": 4.741633141701881e-06, "loss": 0.15650781989097595, "step": 1384 }, { "epoch": 0.3595534787123572, "grad_norm": 1.7960238456726074, "learning_rate": 4.741192818456961e-06, "loss": 0.11986630409955978, "step": 1385 }, { "epoch": 0.3598130841121495, "grad_norm": 1.5295510292053223, "learning_rate": 4.740752143214744e-06, "loss": 0.14190787076950073, "step": 1386 }, { "epoch": 0.3600726895119418, "grad_norm": 1.8400228023529053, "learning_rate": 4.74031111605313e-06, "loss": 0.1543208807706833, "step": 1387 }, { "epoch": 0.3603322949117342, "grad_norm": 1.3779295682907104, "learning_rate": 4.739869737050084e-06, "loss": 0.1269652545452118, "step": 1388 }, { "epoch": 0.3605919003115265, "grad_norm": 0.9069055914878845, "learning_rate": 4.73942800628363e-06, "loss": 0.13519850373268127, "step": 1389 }, { "epoch": 0.3608515057113188, "grad_norm": 0.8462798595428467, "learning_rate": 4.7389859238318555e-06, "loss": 0.1331513226032257, "step": 1390 }, { "epoch": 0.3611111111111111, "grad_norm": 1.2632051706314087, "learning_rate": 4.738543489772912e-06, "loss": 0.1660267859697342, "step": 1391 }, { "epoch": 0.3613707165109034, "grad_norm": 2.1065638065338135, "learning_rate": 4.738100704185011e-06, "loss": 0.14416514337062836, "step": 1392 }, { "epoch": 0.3616303219106957, "grad_norm": 0.909235954284668, "learning_rate": 4.737657567146427e-06, "loss": 0.13579098880290985, "step": 1393 }, { "epoch": 0.3618899273104881, "grad_norm": 0.9398404955863953, "learning_rate": 4.737214078735496e-06, "loss": 0.13868805766105652, "step": 1394 }, { "epoch": 0.3621495327102804, "grad_norm": 3.7740931510925293, "learning_rate": 4.736770239030618e-06, "loss": 0.14058148860931396, "step": 1395 }, { "epoch": 0.3624091381100727, "grad_norm": 1.548064112663269, "learning_rate": 4.736326048110252e-06, "loss": 0.14769555628299713, "step": 1396 }, { "epoch": 0.362668743509865, "grad_norm": 2.043745517730713, "learning_rate": 4.735881506052922e-06, "loss": 0.16449350118637085, "step": 1397 }, { "epoch": 0.3629283489096573, "grad_norm": 1.1078661680221558, "learning_rate": 4.735436612937211e-06, "loss": 0.11628701537847519, "step": 1398 }, { "epoch": 0.3631879543094496, "grad_norm": 1.1248116493225098, "learning_rate": 4.734991368841769e-06, "loss": 0.12932687997817993, "step": 1399 }, { "epoch": 0.363447559709242, "grad_norm": 2.347850799560547, "learning_rate": 4.7345457738453025e-06, "loss": 0.12600383162498474, "step": 1400 }, { "epoch": 0.3637071651090343, "grad_norm": 1.4708212614059448, "learning_rate": 4.734099828026584e-06, "loss": 0.14416658878326416, "step": 1401 }, { "epoch": 0.3639667705088266, "grad_norm": 2.3146276473999023, "learning_rate": 4.733653531464445e-06, "loss": 0.13336297869682312, "step": 1402 }, { "epoch": 0.3642263759086189, "grad_norm": 1.5621916055679321, "learning_rate": 4.7332068842377815e-06, "loss": 0.13066285848617554, "step": 1403 }, { "epoch": 0.3644859813084112, "grad_norm": 1.240796685218811, "learning_rate": 4.73275988642555e-06, "loss": 0.1463553011417389, "step": 1404 }, { "epoch": 0.3647455867082035, "grad_norm": 1.9060711860656738, "learning_rate": 4.73231253810677e-06, "loss": 0.11645035445690155, "step": 1405 }, { "epoch": 0.36500519210799587, "grad_norm": 4.233433723449707, "learning_rate": 4.731864839360523e-06, "loss": 0.14258894324302673, "step": 1406 }, { "epoch": 0.3652647975077882, "grad_norm": 1.5146939754486084, "learning_rate": 4.73141679026595e-06, "loss": 0.12121175974607468, "step": 1407 }, { "epoch": 0.3655244029075805, "grad_norm": 1.501828670501709, "learning_rate": 4.730968390902257e-06, "loss": 0.1533154845237732, "step": 1408 }, { "epoch": 0.3657840083073728, "grad_norm": 2.1982975006103516, "learning_rate": 4.730519641348711e-06, "loss": 0.16902822256088257, "step": 1409 }, { "epoch": 0.3660436137071651, "grad_norm": 1.9126384258270264, "learning_rate": 4.73007054168464e-06, "loss": 0.1275428831577301, "step": 1410 }, { "epoch": 0.3663032191069574, "grad_norm": 1.5337151288986206, "learning_rate": 4.729621091989434e-06, "loss": 0.1185954138636589, "step": 1411 }, { "epoch": 0.36656282450674976, "grad_norm": 2.1951167583465576, "learning_rate": 4.729171292342549e-06, "loss": 0.13366955518722534, "step": 1412 }, { "epoch": 0.36682242990654207, "grad_norm": 0.793676495552063, "learning_rate": 4.728721142823495e-06, "loss": 0.12203957140445709, "step": 1413 }, { "epoch": 0.3670820353063344, "grad_norm": 1.845514178276062, "learning_rate": 4.728270643511849e-06, "loss": 0.14418502151966095, "step": 1414 }, { "epoch": 0.3673416407061267, "grad_norm": 1.0294474363327026, "learning_rate": 4.72781979448725e-06, "loss": 0.1196385994553566, "step": 1415 }, { "epoch": 0.367601246105919, "grad_norm": 1.2626398801803589, "learning_rate": 4.727368595829397e-06, "loss": 0.12390639632940292, "step": 1416 }, { "epoch": 0.3678608515057113, "grad_norm": 2.1953248977661133, "learning_rate": 4.726917047618053e-06, "loss": 0.13130202889442444, "step": 1417 }, { "epoch": 0.36812045690550366, "grad_norm": 1.514379620552063, "learning_rate": 4.72646514993304e-06, "loss": 0.1128387525677681, "step": 1418 }, { "epoch": 0.36838006230529596, "grad_norm": 2.0584287643432617, "learning_rate": 4.726012902854243e-06, "loss": 0.14301759004592896, "step": 1419 }, { "epoch": 0.36863966770508827, "grad_norm": 2.1545844078063965, "learning_rate": 4.72556030646161e-06, "loss": 0.13741028308868408, "step": 1420 }, { "epoch": 0.3688992731048806, "grad_norm": 1.2865989208221436, "learning_rate": 4.7251073608351485e-06, "loss": 0.12656445801258087, "step": 1421 }, { "epoch": 0.3691588785046729, "grad_norm": 1.4074679613113403, "learning_rate": 4.724654066054931e-06, "loss": 0.12105363607406616, "step": 1422 }, { "epoch": 0.3694184839044652, "grad_norm": 2.0726749897003174, "learning_rate": 4.724200422201087e-06, "loss": 0.15544670820236206, "step": 1423 }, { "epoch": 0.36967808930425755, "grad_norm": 2.0428576469421387, "learning_rate": 4.7237464293538115e-06, "loss": 0.12992776930332184, "step": 1424 }, { "epoch": 0.36993769470404986, "grad_norm": 1.627158522605896, "learning_rate": 4.72329208759336e-06, "loss": 0.15439775586128235, "step": 1425 }, { "epoch": 0.37019730010384216, "grad_norm": 1.2284611463546753, "learning_rate": 4.722837397000052e-06, "loss": 0.10911723226308823, "step": 1426 }, { "epoch": 0.37045690550363447, "grad_norm": 1.4007223844528198, "learning_rate": 4.722382357654262e-06, "loss": 0.12092658877372742, "step": 1427 }, { "epoch": 0.3707165109034268, "grad_norm": 2.3759348392486572, "learning_rate": 4.721926969636433e-06, "loss": 0.17014780640602112, "step": 1428 }, { "epoch": 0.3709761163032191, "grad_norm": 3.213531494140625, "learning_rate": 4.721471233027067e-06, "loss": 0.1499045193195343, "step": 1429 }, { "epoch": 0.37123572170301145, "grad_norm": 1.6337016820907593, "learning_rate": 4.721015147906727e-06, "loss": 0.12595447897911072, "step": 1430 }, { "epoch": 0.37149532710280375, "grad_norm": 1.7438609600067139, "learning_rate": 4.720558714356041e-06, "loss": 0.142084538936615, "step": 1431 }, { "epoch": 0.37175493250259606, "grad_norm": 3.4718892574310303, "learning_rate": 4.720101932455693e-06, "loss": 0.13195915520191193, "step": 1432 }, { "epoch": 0.37201453790238836, "grad_norm": 2.2047836780548096, "learning_rate": 4.719644802286433e-06, "loss": 0.1688711941242218, "step": 1433 }, { "epoch": 0.37227414330218067, "grad_norm": 3.6377577781677246, "learning_rate": 4.71918732392907e-06, "loss": 0.1463865339756012, "step": 1434 }, { "epoch": 0.372533748701973, "grad_norm": 1.816389560699463, "learning_rate": 4.718729497464477e-06, "loss": 0.1279807686805725, "step": 1435 }, { "epoch": 0.37279335410176534, "grad_norm": 3.070918560028076, "learning_rate": 4.7182713229735874e-06, "loss": 0.13572217524051666, "step": 1436 }, { "epoch": 0.37305295950155765, "grad_norm": 2.790269136428833, "learning_rate": 4.717812800537394e-06, "loss": 0.17263779044151306, "step": 1437 }, { "epoch": 0.37331256490134995, "grad_norm": 1.4880642890930176, "learning_rate": 4.717353930236956e-06, "loss": 0.1308063268661499, "step": 1438 }, { "epoch": 0.37357217030114226, "grad_norm": 2.310265064239502, "learning_rate": 4.716894712153388e-06, "loss": 0.18655800819396973, "step": 1439 }, { "epoch": 0.37383177570093457, "grad_norm": 2.0981521606445312, "learning_rate": 4.716435146367872e-06, "loss": 0.13961370289325714, "step": 1440 }, { "epoch": 0.37409138110072687, "grad_norm": 1.6500595808029175, "learning_rate": 4.715975232961647e-06, "loss": 0.15038394927978516, "step": 1441 }, { "epoch": 0.37435098650051923, "grad_norm": 1.0548943281173706, "learning_rate": 4.715514972016016e-06, "loss": 0.13236600160598755, "step": 1442 }, { "epoch": 0.37461059190031154, "grad_norm": 3.0242466926574707, "learning_rate": 4.715054363612343e-06, "loss": 0.15365581214427948, "step": 1443 }, { "epoch": 0.37487019730010385, "grad_norm": 2.245551347732544, "learning_rate": 4.714593407832051e-06, "loss": 0.134898841381073, "step": 1444 }, { "epoch": 0.37512980269989615, "grad_norm": 2.6329476833343506, "learning_rate": 4.714132104756628e-06, "loss": 0.13691319525241852, "step": 1445 }, { "epoch": 0.37538940809968846, "grad_norm": 1.6721844673156738, "learning_rate": 4.7136704544676215e-06, "loss": 0.1515313982963562, "step": 1446 }, { "epoch": 0.37564901349948077, "grad_norm": 0.6777632832527161, "learning_rate": 4.713208457046641e-06, "loss": 0.11897211521863937, "step": 1447 }, { "epoch": 0.37590861889927313, "grad_norm": 3.7318367958068848, "learning_rate": 4.7127461125753575e-06, "loss": 0.1645207405090332, "step": 1448 }, { "epoch": 0.37616822429906543, "grad_norm": 1.9850542545318604, "learning_rate": 4.7122834211355e-06, "loss": 0.11341743171215057, "step": 1449 }, { "epoch": 0.37642782969885774, "grad_norm": 1.4874833822250366, "learning_rate": 4.711820382808866e-06, "loss": 0.14707310497760773, "step": 1450 }, { "epoch": 0.37668743509865005, "grad_norm": 2.399528980255127, "learning_rate": 4.711356997677308e-06, "loss": 0.13901828229427338, "step": 1451 }, { "epoch": 0.37694704049844235, "grad_norm": 1.061990737915039, "learning_rate": 4.710893265822741e-06, "loss": 0.10402856767177582, "step": 1452 }, { "epoch": 0.37720664589823466, "grad_norm": 1.6792069673538208, "learning_rate": 4.710429187327144e-06, "loss": 0.1347983181476593, "step": 1453 }, { "epoch": 0.377466251298027, "grad_norm": 1.536779761314392, "learning_rate": 4.709964762272554e-06, "loss": 0.12350674718618393, "step": 1454 }, { "epoch": 0.37772585669781933, "grad_norm": 0.7335906028747559, "learning_rate": 4.709499990741071e-06, "loss": 0.12404444813728333, "step": 1455 }, { "epoch": 0.37798546209761164, "grad_norm": 0.742181658744812, "learning_rate": 4.709034872814856e-06, "loss": 0.10635783523321152, "step": 1456 }, { "epoch": 0.37824506749740394, "grad_norm": 2.274869918823242, "learning_rate": 4.708569408576133e-06, "loss": 0.14846938848495483, "step": 1457 }, { "epoch": 0.37850467289719625, "grad_norm": 2.077730417251587, "learning_rate": 4.708103598107182e-06, "loss": 0.11996392905712128, "step": 1458 }, { "epoch": 0.37876427829698855, "grad_norm": 1.5282491445541382, "learning_rate": 4.707637441490351e-06, "loss": 0.11977235972881317, "step": 1459 }, { "epoch": 0.3790238836967809, "grad_norm": 4.424641132354736, "learning_rate": 4.707170938808044e-06, "loss": 0.16003601253032684, "step": 1460 }, { "epoch": 0.3792834890965732, "grad_norm": 0.9236574172973633, "learning_rate": 4.706704090142728e-06, "loss": 0.11064255237579346, "step": 1461 }, { "epoch": 0.37954309449636553, "grad_norm": 2.1298882961273193, "learning_rate": 4.706236895576931e-06, "loss": 0.12600046396255493, "step": 1462 }, { "epoch": 0.37980269989615784, "grad_norm": 1.6705795526504517, "learning_rate": 4.705769355193244e-06, "loss": 0.119406558573246, "step": 1463 }, { "epoch": 0.38006230529595014, "grad_norm": 1.4448305368423462, "learning_rate": 4.705301469074314e-06, "loss": 0.143244206905365, "step": 1464 }, { "epoch": 0.38032191069574245, "grad_norm": 1.896626353263855, "learning_rate": 4.704833237302857e-06, "loss": 0.12086440622806549, "step": 1465 }, { "epoch": 0.3805815160955348, "grad_norm": 2.140542507171631, "learning_rate": 4.7043646599616425e-06, "loss": 0.12534910440444946, "step": 1466 }, { "epoch": 0.3808411214953271, "grad_norm": 2.5644004344940186, "learning_rate": 4.703895737133505e-06, "loss": 0.16075541079044342, "step": 1467 }, { "epoch": 0.3811007268951194, "grad_norm": 2.3108603954315186, "learning_rate": 4.703426468901339e-06, "loss": 0.1627943068742752, "step": 1468 }, { "epoch": 0.38136033229491173, "grad_norm": 1.176529049873352, "learning_rate": 4.702956855348102e-06, "loss": 0.13262374699115753, "step": 1469 }, { "epoch": 0.38161993769470404, "grad_norm": 0.8505261540412903, "learning_rate": 4.702486896556809e-06, "loss": 0.13569417595863342, "step": 1470 }, { "epoch": 0.38187954309449634, "grad_norm": 2.404855251312256, "learning_rate": 4.702016592610539e-06, "loss": 0.14877447485923767, "step": 1471 }, { "epoch": 0.3821391484942887, "grad_norm": 1.9436514377593994, "learning_rate": 4.70154594359243e-06, "loss": 0.15107308328151703, "step": 1472 }, { "epoch": 0.382398753894081, "grad_norm": 1.3538637161254883, "learning_rate": 4.701074949585684e-06, "loss": 0.14729925990104675, "step": 1473 }, { "epoch": 0.3826583592938733, "grad_norm": 3.4349451065063477, "learning_rate": 4.7006036106735594e-06, "loss": 0.12819021940231323, "step": 1474 }, { "epoch": 0.3829179646936656, "grad_norm": 1.973777174949646, "learning_rate": 4.700131926939379e-06, "loss": 0.13017433881759644, "step": 1475 }, { "epoch": 0.38317757009345793, "grad_norm": 2.0280046463012695, "learning_rate": 4.699659898466527e-06, "loss": 0.15285766124725342, "step": 1476 }, { "epoch": 0.38343717549325024, "grad_norm": 2.796659469604492, "learning_rate": 4.699187525338447e-06, "loss": 0.14462701976299286, "step": 1477 }, { "epoch": 0.3836967808930426, "grad_norm": 2.208179473876953, "learning_rate": 4.698714807638641e-06, "loss": 0.1592082381248474, "step": 1478 }, { "epoch": 0.3839563862928349, "grad_norm": 1.2631701231002808, "learning_rate": 4.698241745450678e-06, "loss": 0.14107976853847504, "step": 1479 }, { "epoch": 0.3842159916926272, "grad_norm": 1.945851445198059, "learning_rate": 4.697768338858183e-06, "loss": 0.12135645747184753, "step": 1480 }, { "epoch": 0.3844755970924195, "grad_norm": 1.352386474609375, "learning_rate": 4.697294587944844e-06, "loss": 0.15385675430297852, "step": 1481 }, { "epoch": 0.3847352024922118, "grad_norm": 2.6274733543395996, "learning_rate": 4.696820492794408e-06, "loss": 0.12927204370498657, "step": 1482 }, { "epoch": 0.38499480789200413, "grad_norm": 0.8683428168296814, "learning_rate": 4.6963460534906865e-06, "loss": 0.1313273161649704, "step": 1483 }, { "epoch": 0.3852544132917965, "grad_norm": 3.075589656829834, "learning_rate": 4.6958712701175465e-06, "loss": 0.12143807113170624, "step": 1484 }, { "epoch": 0.3855140186915888, "grad_norm": 1.0698949098587036, "learning_rate": 4.695396142758922e-06, "loss": 0.12656401097774506, "step": 1485 }, { "epoch": 0.3857736240913811, "grad_norm": 1.7052210569381714, "learning_rate": 4.694920671498802e-06, "loss": 0.13979217410087585, "step": 1486 }, { "epoch": 0.3860332294911734, "grad_norm": 1.5855833292007446, "learning_rate": 4.694444856421242e-06, "loss": 0.1325373500585556, "step": 1487 }, { "epoch": 0.3862928348909657, "grad_norm": 2.7482450008392334, "learning_rate": 4.693968697610351e-06, "loss": 0.14709851145744324, "step": 1488 }, { "epoch": 0.386552440290758, "grad_norm": 1.2809079885482788, "learning_rate": 4.693492195150306e-06, "loss": 0.15220460295677185, "step": 1489 }, { "epoch": 0.3868120456905504, "grad_norm": 2.767759084701538, "learning_rate": 4.693015349125341e-06, "loss": 0.12845417857170105, "step": 1490 }, { "epoch": 0.3870716510903427, "grad_norm": 1.7527660131454468, "learning_rate": 4.692538159619751e-06, "loss": 0.1518680602312088, "step": 1491 }, { "epoch": 0.387331256490135, "grad_norm": 0.7820326089859009, "learning_rate": 4.692060626717894e-06, "loss": 0.1174318715929985, "step": 1492 }, { "epoch": 0.3875908618899273, "grad_norm": 1.986935019493103, "learning_rate": 4.691582750504184e-06, "loss": 0.12848542630672455, "step": 1493 }, { "epoch": 0.3878504672897196, "grad_norm": 5.3867034912109375, "learning_rate": 4.691104531063101e-06, "loss": 0.16288971900939941, "step": 1494 }, { "epoch": 0.3881100726895119, "grad_norm": 1.3275002241134644, "learning_rate": 4.6906259684791816e-06, "loss": 0.12287956476211548, "step": 1495 }, { "epoch": 0.3883696780893043, "grad_norm": 2.020559310913086, "learning_rate": 4.690147062837026e-06, "loss": 0.12654995918273926, "step": 1496 }, { "epoch": 0.3886292834890966, "grad_norm": 1.679650068283081, "learning_rate": 4.689667814221293e-06, "loss": 0.12375159561634064, "step": 1497 }, { "epoch": 0.3888888888888889, "grad_norm": 1.9064768552780151, "learning_rate": 4.689188222716703e-06, "loss": 0.1310604214668274, "step": 1498 }, { "epoch": 0.3891484942886812, "grad_norm": 1.5633587837219238, "learning_rate": 4.688708288408036e-06, "loss": 0.15310269594192505, "step": 1499 }, { "epoch": 0.3894080996884735, "grad_norm": 1.6427903175354004, "learning_rate": 4.688228011380135e-06, "loss": 0.13615071773529053, "step": 1500 }, { "epoch": 0.3896677050882658, "grad_norm": 2.2933735847473145, "learning_rate": 4.687747391717901e-06, "loss": 0.13461431860923767, "step": 1501 }, { "epoch": 0.3899273104880582, "grad_norm": 2.1359658241271973, "learning_rate": 4.687266429506296e-06, "loss": 0.1329105645418167, "step": 1502 }, { "epoch": 0.3901869158878505, "grad_norm": 2.5290205478668213, "learning_rate": 4.686785124830345e-06, "loss": 0.12426373362541199, "step": 1503 }, { "epoch": 0.3904465212876428, "grad_norm": 2.599909782409668, "learning_rate": 4.686303477775129e-06, "loss": 0.1436503827571869, "step": 1504 }, { "epoch": 0.3907061266874351, "grad_norm": 1.5130817890167236, "learning_rate": 4.685821488425795e-06, "loss": 0.13315263390541077, "step": 1505 }, { "epoch": 0.3909657320872274, "grad_norm": 1.7371203899383545, "learning_rate": 4.6853391568675456e-06, "loss": 0.11865904927253723, "step": 1506 }, { "epoch": 0.3912253374870197, "grad_norm": 2.14774227142334, "learning_rate": 4.684856483185647e-06, "loss": 0.12161065638065338, "step": 1507 }, { "epoch": 0.39148494288681207, "grad_norm": 2.387582778930664, "learning_rate": 4.684373467465424e-06, "loss": 0.12842901051044464, "step": 1508 }, { "epoch": 0.3917445482866044, "grad_norm": 1.7759313583374023, "learning_rate": 4.683890109792264e-06, "loss": 0.17069095373153687, "step": 1509 }, { "epoch": 0.3920041536863967, "grad_norm": 3.070049285888672, "learning_rate": 4.683406410251612e-06, "loss": 0.1533936709165573, "step": 1510 }, { "epoch": 0.392263759086189, "grad_norm": 1.3580095767974854, "learning_rate": 4.682922368928977e-06, "loss": 0.11717816442251205, "step": 1511 }, { "epoch": 0.3925233644859813, "grad_norm": 1.468764305114746, "learning_rate": 4.682437985909924e-06, "loss": 0.1356249451637268, "step": 1512 }, { "epoch": 0.3927829698857736, "grad_norm": 0.9926334023475647, "learning_rate": 4.6819532612800824e-06, "loss": 0.139219269156456, "step": 1513 }, { "epoch": 0.39304257528556596, "grad_norm": 3.210310935974121, "learning_rate": 4.68146819512514e-06, "loss": 0.16247139871120453, "step": 1514 }, { "epoch": 0.39330218068535827, "grad_norm": 1.003624677658081, "learning_rate": 4.680982787530845e-06, "loss": 0.1515795886516571, "step": 1515 }, { "epoch": 0.3935617860851506, "grad_norm": 2.6400976181030273, "learning_rate": 4.680497038583008e-06, "loss": 0.15771576762199402, "step": 1516 }, { "epoch": 0.3938213914849429, "grad_norm": 2.1252801418304443, "learning_rate": 4.680010948367496e-06, "loss": 0.12223377823829651, "step": 1517 }, { "epoch": 0.3940809968847352, "grad_norm": 0.6977126598358154, "learning_rate": 4.67952451697024e-06, "loss": 0.12304096668958664, "step": 1518 }, { "epoch": 0.3943406022845275, "grad_norm": 2.9850411415100098, "learning_rate": 4.67903774447723e-06, "loss": 0.15147441625595093, "step": 1519 }, { "epoch": 0.39460020768431986, "grad_norm": 1.5552403926849365, "learning_rate": 4.678550630974515e-06, "loss": 0.18068768084049225, "step": 1520 }, { "epoch": 0.39485981308411217, "grad_norm": 0.9161517024040222, "learning_rate": 4.678063176548209e-06, "loss": 0.12283288687467575, "step": 1521 }, { "epoch": 0.39511941848390447, "grad_norm": 1.2256219387054443, "learning_rate": 4.677575381284478e-06, "loss": 0.12306894361972809, "step": 1522 }, { "epoch": 0.3953790238836968, "grad_norm": 2.104314088821411, "learning_rate": 4.6770872452695546e-06, "loss": 0.12645605206489563, "step": 1523 }, { "epoch": 0.3956386292834891, "grad_norm": 3.1677889823913574, "learning_rate": 4.6765987685897315e-06, "loss": 0.14779044687747955, "step": 1524 }, { "epoch": 0.3958982346832814, "grad_norm": 2.679760694503784, "learning_rate": 4.6761099513313595e-06, "loss": 0.15622612833976746, "step": 1525 }, { "epoch": 0.39615784008307375, "grad_norm": 1.9777947664260864, "learning_rate": 4.67562079358085e-06, "loss": 0.15053166449069977, "step": 1526 }, { "epoch": 0.39641744548286606, "grad_norm": 2.7598812580108643, "learning_rate": 4.675131295424675e-06, "loss": 0.12846025824546814, "step": 1527 }, { "epoch": 0.39667705088265837, "grad_norm": 1.8455476760864258, "learning_rate": 4.674641456949368e-06, "loss": 0.13724583387374878, "step": 1528 }, { "epoch": 0.3969366562824507, "grad_norm": 1.859091877937317, "learning_rate": 4.674151278241518e-06, "loss": 0.14729052782058716, "step": 1529 }, { "epoch": 0.397196261682243, "grad_norm": 2.2863168716430664, "learning_rate": 4.673660759387779e-06, "loss": 0.13723823428153992, "step": 1530 }, { "epoch": 0.3974558670820353, "grad_norm": 2.8980906009674072, "learning_rate": 4.673169900474865e-06, "loss": 0.1362212598323822, "step": 1531 }, { "epoch": 0.39771547248182765, "grad_norm": 0.7240413427352905, "learning_rate": 4.672678701589547e-06, "loss": 0.12598586082458496, "step": 1532 }, { "epoch": 0.39797507788161995, "grad_norm": 1.9976977109909058, "learning_rate": 4.672187162818658e-06, "loss": 0.11581826955080032, "step": 1533 }, { "epoch": 0.39823468328141226, "grad_norm": 2.283794403076172, "learning_rate": 4.671695284249091e-06, "loss": 0.10369665920734406, "step": 1534 }, { "epoch": 0.39849428868120457, "grad_norm": 1.0037140846252441, "learning_rate": 4.671203065967799e-06, "loss": 0.10809671878814697, "step": 1535 }, { "epoch": 0.3987538940809969, "grad_norm": 1.8738845586776733, "learning_rate": 4.670710508061794e-06, "loss": 0.14512574672698975, "step": 1536 }, { "epoch": 0.3990134994807892, "grad_norm": 2.2783303260803223, "learning_rate": 4.67021761061815e-06, "loss": 0.1314857453107834, "step": 1537 }, { "epoch": 0.39927310488058154, "grad_norm": 1.762107253074646, "learning_rate": 4.669724373724e-06, "loss": 0.15150567889213562, "step": 1538 }, { "epoch": 0.39953271028037385, "grad_norm": 1.8461923599243164, "learning_rate": 4.6692307974665355e-06, "loss": 0.15401792526245117, "step": 1539 }, { "epoch": 0.39979231568016615, "grad_norm": 0.9633582830429077, "learning_rate": 4.668736881933012e-06, "loss": 0.12109441310167313, "step": 1540 }, { "epoch": 0.40005192107995846, "grad_norm": 2.0014407634735107, "learning_rate": 4.66824262721074e-06, "loss": 0.13741225004196167, "step": 1541 }, { "epoch": 0.40031152647975077, "grad_norm": 1.8266340494155884, "learning_rate": 4.6677480333870935e-06, "loss": 0.12269404530525208, "step": 1542 }, { "epoch": 0.4005711318795431, "grad_norm": 0.7870243787765503, "learning_rate": 4.667253100549506e-06, "loss": 0.1336517333984375, "step": 1543 }, { "epoch": 0.40083073727933544, "grad_norm": 0.8818300366401672, "learning_rate": 4.666757828785469e-06, "loss": 0.13060863316059113, "step": 1544 }, { "epoch": 0.40109034267912774, "grad_norm": 1.1660722494125366, "learning_rate": 4.666262218182535e-06, "loss": 0.13153240084648132, "step": 1545 }, { "epoch": 0.40134994807892005, "grad_norm": 0.6566032767295837, "learning_rate": 4.665766268828319e-06, "loss": 0.12327892333269119, "step": 1546 }, { "epoch": 0.40160955347871236, "grad_norm": 1.4803661108016968, "learning_rate": 4.66526998081049e-06, "loss": 0.1295556277036667, "step": 1547 }, { "epoch": 0.40186915887850466, "grad_norm": 1.888271689414978, "learning_rate": 4.664773354216785e-06, "loss": 0.12518206238746643, "step": 1548 }, { "epoch": 0.40212876427829697, "grad_norm": 2.4807076454162598, "learning_rate": 4.664276389134991e-06, "loss": 0.12392231822013855, "step": 1549 }, { "epoch": 0.40238836967808933, "grad_norm": 2.1959269046783447, "learning_rate": 4.663779085652964e-06, "loss": 0.1445680558681488, "step": 1550 }, { "epoch": 0.40264797507788164, "grad_norm": 2.964811086654663, "learning_rate": 4.663281443858614e-06, "loss": 0.14798173308372498, "step": 1551 }, { "epoch": 0.40290758047767394, "grad_norm": 3.6826629638671875, "learning_rate": 4.662783463839913e-06, "loss": 0.12639454007148743, "step": 1552 }, { "epoch": 0.40316718587746625, "grad_norm": 2.401399850845337, "learning_rate": 4.662285145684892e-06, "loss": 0.1417941153049469, "step": 1553 }, { "epoch": 0.40342679127725856, "grad_norm": 1.66999351978302, "learning_rate": 4.661786489481644e-06, "loss": 0.16385123133659363, "step": 1554 }, { "epoch": 0.40368639667705086, "grad_norm": 5.988789081573486, "learning_rate": 4.661287495318317e-06, "loss": 0.16235390305519104, "step": 1555 }, { "epoch": 0.4039460020768432, "grad_norm": 4.478551387786865, "learning_rate": 4.660788163283125e-06, "loss": 0.18767330050468445, "step": 1556 }, { "epoch": 0.40420560747663553, "grad_norm": 2.6169919967651367, "learning_rate": 4.6602884934643364e-06, "loss": 0.16507738828659058, "step": 1557 }, { "epoch": 0.40446521287642784, "grad_norm": 1.2000595331192017, "learning_rate": 4.659788485950282e-06, "loss": 0.15008538961410522, "step": 1558 }, { "epoch": 0.40472481827622014, "grad_norm": 2.3288469314575195, "learning_rate": 4.659288140829351e-06, "loss": 0.1269841343164444, "step": 1559 }, { "epoch": 0.40498442367601245, "grad_norm": 1.2813397645950317, "learning_rate": 4.658787458189995e-06, "loss": 0.1420929878950119, "step": 1560 }, { "epoch": 0.40524402907580476, "grad_norm": 3.6806063652038574, "learning_rate": 4.658286438120721e-06, "loss": 0.16419994831085205, "step": 1561 }, { "epoch": 0.4055036344755971, "grad_norm": 2.1166670322418213, "learning_rate": 4.657785080710099e-06, "loss": 0.1634131669998169, "step": 1562 }, { "epoch": 0.4057632398753894, "grad_norm": 3.7444584369659424, "learning_rate": 4.657283386046757e-06, "loss": 0.12513726949691772, "step": 1563 }, { "epoch": 0.40602284527518173, "grad_norm": 2.1019349098205566, "learning_rate": 4.656781354219383e-06, "loss": 0.16035175323486328, "step": 1564 }, { "epoch": 0.40628245067497404, "grad_norm": 2.7557122707366943, "learning_rate": 4.656278985316725e-06, "loss": 0.12615787982940674, "step": 1565 }, { "epoch": 0.40654205607476634, "grad_norm": 1.4904687404632568, "learning_rate": 4.655776279427591e-06, "loss": 0.125169575214386, "step": 1566 }, { "epoch": 0.40680166147455865, "grad_norm": 2.419857978820801, "learning_rate": 4.655273236640847e-06, "loss": 0.1328289806842804, "step": 1567 }, { "epoch": 0.407061266874351, "grad_norm": 1.726430058479309, "learning_rate": 4.6547698570454206e-06, "loss": 0.14734140038490295, "step": 1568 }, { "epoch": 0.4073208722741433, "grad_norm": 3.730293035507202, "learning_rate": 4.654266140730297e-06, "loss": 0.1670660674571991, "step": 1569 }, { "epoch": 0.4075804776739356, "grad_norm": 0.8857203722000122, "learning_rate": 4.653762087784521e-06, "loss": 0.12000471353530884, "step": 1570 }, { "epoch": 0.40784008307372793, "grad_norm": 0.8043270111083984, "learning_rate": 4.653257698297198e-06, "loss": 0.12224400788545609, "step": 1571 }, { "epoch": 0.40809968847352024, "grad_norm": 2.350104331970215, "learning_rate": 4.652752972357492e-06, "loss": 0.16701142489910126, "step": 1572 }, { "epoch": 0.40835929387331255, "grad_norm": 1.0468060970306396, "learning_rate": 4.652247910054629e-06, "loss": 0.1361461877822876, "step": 1573 }, { "epoch": 0.4086188992731049, "grad_norm": 1.8565013408660889, "learning_rate": 4.6517425114778905e-06, "loss": 0.14571291208267212, "step": 1574 }, { "epoch": 0.4088785046728972, "grad_norm": 1.8932090997695923, "learning_rate": 4.65123677671662e-06, "loss": 0.11954417079687119, "step": 1575 }, { "epoch": 0.4091381100726895, "grad_norm": 1.9563794136047363, "learning_rate": 4.6507307058602205e-06, "loss": 0.13126176595687866, "step": 1576 }, { "epoch": 0.4093977154724818, "grad_norm": 0.7870089411735535, "learning_rate": 4.650224298998154e-06, "loss": 0.12527954578399658, "step": 1577 }, { "epoch": 0.40965732087227413, "grad_norm": 2.3742427825927734, "learning_rate": 4.649717556219939e-06, "loss": 0.1445164978504181, "step": 1578 }, { "epoch": 0.40991692627206644, "grad_norm": 1.3855946063995361, "learning_rate": 4.649210477615159e-06, "loss": 0.12748607993125916, "step": 1579 }, { "epoch": 0.4101765316718588, "grad_norm": 1.9591864347457886, "learning_rate": 4.648703063273452e-06, "loss": 0.13724759221076965, "step": 1580 }, { "epoch": 0.4104361370716511, "grad_norm": 2.344420909881592, "learning_rate": 4.6481953132845184e-06, "loss": 0.15642865002155304, "step": 1581 }, { "epoch": 0.4106957424714434, "grad_norm": 1.5851702690124512, "learning_rate": 4.647687227738116e-06, "loss": 0.15616728365421295, "step": 1582 }, { "epoch": 0.4109553478712357, "grad_norm": 0.8847413659095764, "learning_rate": 4.647178806724064e-06, "loss": 0.12521910667419434, "step": 1583 }, { "epoch": 0.411214953271028, "grad_norm": 1.191090703010559, "learning_rate": 4.646670050332238e-06, "loss": 0.12341044843196869, "step": 1584 }, { "epoch": 0.41147455867082033, "grad_norm": 3.523906946182251, "learning_rate": 4.646160958652575e-06, "loss": 0.14112496376037598, "step": 1585 }, { "epoch": 0.4117341640706127, "grad_norm": 1.9115209579467773, "learning_rate": 4.645651531775071e-06, "loss": 0.12012374401092529, "step": 1586 }, { "epoch": 0.411993769470405, "grad_norm": 3.671562671661377, "learning_rate": 4.64514176978978e-06, "loss": 0.15104393661022186, "step": 1587 }, { "epoch": 0.4122533748701973, "grad_norm": 2.3491554260253906, "learning_rate": 4.6446316727868175e-06, "loss": 0.17305901646614075, "step": 1588 }, { "epoch": 0.4125129802699896, "grad_norm": 3.199030637741089, "learning_rate": 4.644121240856357e-06, "loss": 0.14317789673805237, "step": 1589 }, { "epoch": 0.4127725856697819, "grad_norm": 2.8612680435180664, "learning_rate": 4.6436104740886305e-06, "loss": 0.11894748359918594, "step": 1590 }, { "epoch": 0.41303219106957423, "grad_norm": 3.8503124713897705, "learning_rate": 4.64309937257393e-06, "loss": 0.15931367874145508, "step": 1591 }, { "epoch": 0.4132917964693666, "grad_norm": 2.2471349239349365, "learning_rate": 4.642587936402607e-06, "loss": 0.1475498080253601, "step": 1592 }, { "epoch": 0.4135514018691589, "grad_norm": 1.1114046573638916, "learning_rate": 4.642076165665071e-06, "loss": 0.13553699851036072, "step": 1593 }, { "epoch": 0.4138110072689512, "grad_norm": 1.095144510269165, "learning_rate": 4.641564060451792e-06, "loss": 0.14516714215278625, "step": 1594 }, { "epoch": 0.4140706126687435, "grad_norm": 0.7800896763801575, "learning_rate": 4.641051620853298e-06, "loss": 0.13321009278297424, "step": 1595 }, { "epoch": 0.4143302180685358, "grad_norm": 4.073946475982666, "learning_rate": 4.640538846960177e-06, "loss": 0.1333150565624237, "step": 1596 }, { "epoch": 0.4145898234683281, "grad_norm": 2.4198122024536133, "learning_rate": 4.640025738863076e-06, "loss": 0.14259201288223267, "step": 1597 }, { "epoch": 0.4148494288681205, "grad_norm": 2.3850014209747314, "learning_rate": 4.639512296652699e-06, "loss": 0.1740029901266098, "step": 1598 }, { "epoch": 0.4151090342679128, "grad_norm": 1.5793436765670776, "learning_rate": 4.638998520419812e-06, "loss": 0.11468157172203064, "step": 1599 }, { "epoch": 0.4153686396677051, "grad_norm": 2.882397413253784, "learning_rate": 4.63848441025524e-06, "loss": 0.12174233794212341, "step": 1600 }, { "epoch": 0.4156282450674974, "grad_norm": 1.628192663192749, "learning_rate": 4.6379699662498646e-06, "loss": 0.14929580688476562, "step": 1601 }, { "epoch": 0.4158878504672897, "grad_norm": 2.1926796436309814, "learning_rate": 4.637455188494627e-06, "loss": 0.16523191332817078, "step": 1602 }, { "epoch": 0.416147455867082, "grad_norm": 0.6120513081550598, "learning_rate": 4.636940077080531e-06, "loss": 0.12378619611263275, "step": 1603 }, { "epoch": 0.4164070612668744, "grad_norm": 0.6803605556488037, "learning_rate": 4.6364246320986325e-06, "loss": 0.1328945904970169, "step": 1604 }, { "epoch": 0.4166666666666667, "grad_norm": 0.6439208388328552, "learning_rate": 4.6359088536400534e-06, "loss": 0.12745293974876404, "step": 1605 }, { "epoch": 0.416926272066459, "grad_norm": 0.7691042423248291, "learning_rate": 4.635392741795971e-06, "loss": 0.1398007869720459, "step": 1606 }, { "epoch": 0.4171858774662513, "grad_norm": 1.165950894355774, "learning_rate": 4.63487629665762e-06, "loss": 0.13984300196170807, "step": 1607 }, { "epoch": 0.4174454828660436, "grad_norm": 3.997809410095215, "learning_rate": 4.634359518316299e-06, "loss": 0.17149898409843445, "step": 1608 }, { "epoch": 0.4177050882658359, "grad_norm": 2.578307867050171, "learning_rate": 4.6338424068633615e-06, "loss": 0.11714284121990204, "step": 1609 }, { "epoch": 0.4179646936656282, "grad_norm": 3.972555160522461, "learning_rate": 4.6333249623902195e-06, "loss": 0.1276121735572815, "step": 1610 }, { "epoch": 0.4182242990654206, "grad_norm": 3.488574504852295, "learning_rate": 4.632807184988348e-06, "loss": 0.11247862130403519, "step": 1611 }, { "epoch": 0.4184839044652129, "grad_norm": 3.869182586669922, "learning_rate": 4.6322890747492765e-06, "loss": 0.14971667528152466, "step": 1612 }, { "epoch": 0.4187435098650052, "grad_norm": 2.0466461181640625, "learning_rate": 4.631770631764595e-06, "loss": 0.12305699288845062, "step": 1613 }, { "epoch": 0.4190031152647975, "grad_norm": 2.305476188659668, "learning_rate": 4.631251856125952e-06, "loss": 0.16865047812461853, "step": 1614 }, { "epoch": 0.4192627206645898, "grad_norm": 1.3293534517288208, "learning_rate": 4.630732747925057e-06, "loss": 0.1500069499015808, "step": 1615 }, { "epoch": 0.4195223260643821, "grad_norm": 1.6980661153793335, "learning_rate": 4.630213307253674e-06, "loss": 0.13390029966831207, "step": 1616 }, { "epoch": 0.4197819314641745, "grad_norm": 2.940758466720581, "learning_rate": 4.6296935342036305e-06, "loss": 0.15559431910514832, "step": 1617 }, { "epoch": 0.4200415368639668, "grad_norm": 2.146955728530884, "learning_rate": 4.629173428866808e-06, "loss": 0.12069999426603317, "step": 1618 }, { "epoch": 0.4203011422637591, "grad_norm": 1.888954997062683, "learning_rate": 4.628652991335153e-06, "loss": 0.15717819333076477, "step": 1619 }, { "epoch": 0.4205607476635514, "grad_norm": 1.8183586597442627, "learning_rate": 4.628132221700663e-06, "loss": 0.12213359773159027, "step": 1620 }, { "epoch": 0.4208203530633437, "grad_norm": 2.7128472328186035, "learning_rate": 4.627611120055399e-06, "loss": 0.12050091475248337, "step": 1621 }, { "epoch": 0.421079958463136, "grad_norm": 3.3753607273101807, "learning_rate": 4.627089686491482e-06, "loss": 0.12044312059879303, "step": 1622 }, { "epoch": 0.42133956386292837, "grad_norm": 1.3205150365829468, "learning_rate": 4.626567921101087e-06, "loss": 0.14429731667041779, "step": 1623 }, { "epoch": 0.4215991692627207, "grad_norm": 1.8206908702850342, "learning_rate": 4.626045823976453e-06, "loss": 0.11958081275224686, "step": 1624 }, { "epoch": 0.421858774662513, "grad_norm": 1.324662446975708, "learning_rate": 4.625523395209871e-06, "loss": 0.11394717544317245, "step": 1625 }, { "epoch": 0.4221183800623053, "grad_norm": 1.2552069425582886, "learning_rate": 4.625000634893698e-06, "loss": 0.1268482655286789, "step": 1626 }, { "epoch": 0.4223779854620976, "grad_norm": 2.6682803630828857, "learning_rate": 4.624477543120344e-06, "loss": 0.14854300022125244, "step": 1627 }, { "epoch": 0.4226375908618899, "grad_norm": 1.3079814910888672, "learning_rate": 4.62395411998228e-06, "loss": 0.12413150072097778, "step": 1628 }, { "epoch": 0.42289719626168226, "grad_norm": 1.528245449066162, "learning_rate": 4.623430365572036e-06, "loss": 0.11091380566358566, "step": 1629 }, { "epoch": 0.42315680166147457, "grad_norm": 2.819509267807007, "learning_rate": 4.622906279982199e-06, "loss": 0.17403563857078552, "step": 1630 }, { "epoch": 0.4234164070612669, "grad_norm": 2.7458794116973877, "learning_rate": 4.6223818633054155e-06, "loss": 0.13740909099578857, "step": 1631 }, { "epoch": 0.4236760124610592, "grad_norm": 1.1958874464035034, "learning_rate": 4.62185711563439e-06, "loss": 0.1285683512687683, "step": 1632 }, { "epoch": 0.4239356178608515, "grad_norm": 2.134711980819702, "learning_rate": 4.621332037061887e-06, "loss": 0.12240105867385864, "step": 1633 }, { "epoch": 0.4241952232606438, "grad_norm": 1.4752388000488281, "learning_rate": 4.620806627680728e-06, "loss": 0.13686582446098328, "step": 1634 }, { "epoch": 0.42445482866043616, "grad_norm": 2.4647676944732666, "learning_rate": 4.6202808875837915e-06, "loss": 0.16619843244552612, "step": 1635 }, { "epoch": 0.42471443406022846, "grad_norm": 1.8004182577133179, "learning_rate": 4.619754816864019e-06, "loss": 0.16461658477783203, "step": 1636 }, { "epoch": 0.42497403946002077, "grad_norm": 1.4343823194503784, "learning_rate": 4.619228415614406e-06, "loss": 0.1457020342350006, "step": 1637 }, { "epoch": 0.4252336448598131, "grad_norm": 1.5011452436447144, "learning_rate": 4.61870168392801e-06, "loss": 0.13116741180419922, "step": 1638 }, { "epoch": 0.4254932502596054, "grad_norm": 1.1605595350265503, "learning_rate": 4.6181746218979424e-06, "loss": 0.12268711626529694, "step": 1639 }, { "epoch": 0.4257528556593977, "grad_norm": 2.1007378101348877, "learning_rate": 4.6176472296173775e-06, "loss": 0.13378837704658508, "step": 1640 }, { "epoch": 0.42601246105919005, "grad_norm": 1.9360883235931396, "learning_rate": 4.617119507179546e-06, "loss": 0.1186615526676178, "step": 1641 }, { "epoch": 0.42627206645898236, "grad_norm": 0.9854421019554138, "learning_rate": 4.616591454677737e-06, "loss": 0.11864715814590454, "step": 1642 }, { "epoch": 0.42653167185877466, "grad_norm": 0.7631796598434448, "learning_rate": 4.616063072205298e-06, "loss": 0.12444348633289337, "step": 1643 }, { "epoch": 0.42679127725856697, "grad_norm": 2.2226226329803467, "learning_rate": 4.615534359855636e-06, "loss": 0.14650854468345642, "step": 1644 }, { "epoch": 0.4270508826583593, "grad_norm": 2.7060956954956055, "learning_rate": 4.615005317722213e-06, "loss": 0.13839823007583618, "step": 1645 }, { "epoch": 0.4273104880581516, "grad_norm": 1.948121190071106, "learning_rate": 4.614475945898554e-06, "loss": 0.1178799495100975, "step": 1646 }, { "epoch": 0.42757009345794394, "grad_norm": 1.592331886291504, "learning_rate": 4.613946244478238e-06, "loss": 0.11673509329557419, "step": 1647 }, { "epoch": 0.42782969885773625, "grad_norm": 1.7326288223266602, "learning_rate": 4.613416213554904e-06, "loss": 0.11017502844333649, "step": 1648 }, { "epoch": 0.42808930425752856, "grad_norm": 1.1660014390945435, "learning_rate": 4.61288585322225e-06, "loss": 0.10690521448850632, "step": 1649 }, { "epoch": 0.42834890965732086, "grad_norm": 0.9269199967384338, "learning_rate": 4.6123551635740336e-06, "loss": 0.12094193696975708, "step": 1650 }, { "epoch": 0.42860851505711317, "grad_norm": 1.7536085844039917, "learning_rate": 4.611824144704064e-06, "loss": 0.13681961596012115, "step": 1651 }, { "epoch": 0.4288681204569055, "grad_norm": 2.176046848297119, "learning_rate": 4.611292796706218e-06, "loss": 0.1341073215007782, "step": 1652 }, { "epoch": 0.42912772585669784, "grad_norm": 1.0455307960510254, "learning_rate": 4.610761119674422e-06, "loss": 0.12000992894172668, "step": 1653 }, { "epoch": 0.42938733125649015, "grad_norm": 1.1335786581039429, "learning_rate": 4.610229113702666e-06, "loss": 0.12651735544204712, "step": 1654 }, { "epoch": 0.42964693665628245, "grad_norm": 1.4616669416427612, "learning_rate": 4.609696778884997e-06, "loss": 0.12092135101556778, "step": 1655 }, { "epoch": 0.42990654205607476, "grad_norm": 2.289724826812744, "learning_rate": 4.6091641153155175e-06, "loss": 0.1071176528930664, "step": 1656 }, { "epoch": 0.43016614745586707, "grad_norm": 2.8594372272491455, "learning_rate": 4.608631123088393e-06, "loss": 0.14174067974090576, "step": 1657 }, { "epoch": 0.43042575285565937, "grad_norm": 1.4866361618041992, "learning_rate": 4.6080978022978425e-06, "loss": 0.1270064115524292, "step": 1658 }, { "epoch": 0.43068535825545173, "grad_norm": 1.693389892578125, "learning_rate": 4.607564153038146e-06, "loss": 0.13494643568992615, "step": 1659 }, { "epoch": 0.43094496365524404, "grad_norm": 1.9224916696548462, "learning_rate": 4.607030175403638e-06, "loss": 0.13548538088798523, "step": 1660 }, { "epoch": 0.43120456905503635, "grad_norm": 3.4889438152313232, "learning_rate": 4.606495869488716e-06, "loss": 0.1613515317440033, "step": 1661 }, { "epoch": 0.43146417445482865, "grad_norm": 1.7161250114440918, "learning_rate": 4.605961235387834e-06, "loss": 0.14636941254138947, "step": 1662 }, { "epoch": 0.43172377985462096, "grad_norm": 2.922800302505493, "learning_rate": 4.605426273195499e-06, "loss": 0.1376442313194275, "step": 1663 }, { "epoch": 0.43198338525441327, "grad_norm": 1.9369187355041504, "learning_rate": 4.604890983006285e-06, "loss": 0.12634941935539246, "step": 1664 }, { "epoch": 0.4322429906542056, "grad_norm": 2.6223673820495605, "learning_rate": 4.604355364914816e-06, "loss": 0.15573447942733765, "step": 1665 }, { "epoch": 0.43250259605399793, "grad_norm": 3.1921610832214355, "learning_rate": 4.603819419015776e-06, "loss": 0.15582603216171265, "step": 1666 }, { "epoch": 0.43276220145379024, "grad_norm": 2.921555995941162, "learning_rate": 4.603283145403911e-06, "loss": 0.12828461825847626, "step": 1667 }, { "epoch": 0.43302180685358255, "grad_norm": 1.310411810874939, "learning_rate": 4.602746544174021e-06, "loss": 0.12082026153802872, "step": 1668 }, { "epoch": 0.43328141225337485, "grad_norm": 1.908447265625, "learning_rate": 4.602209615420963e-06, "loss": 0.143108069896698, "step": 1669 }, { "epoch": 0.43354101765316716, "grad_norm": 1.2595326900482178, "learning_rate": 4.601672359239654e-06, "loss": 0.11604511737823486, "step": 1670 }, { "epoch": 0.4338006230529595, "grad_norm": 1.8179048299789429, "learning_rate": 4.6011347757250716e-06, "loss": 0.1450304090976715, "step": 1671 }, { "epoch": 0.43406022845275183, "grad_norm": 4.754559516906738, "learning_rate": 4.600596864972246e-06, "loss": 0.15986484289169312, "step": 1672 }, { "epoch": 0.43431983385254413, "grad_norm": 0.7483028173446655, "learning_rate": 4.600058627076266e-06, "loss": 0.11514708399772644, "step": 1673 }, { "epoch": 0.43457943925233644, "grad_norm": 3.1556525230407715, "learning_rate": 4.599520062132283e-06, "loss": 0.13916637003421783, "step": 1674 }, { "epoch": 0.43483904465212875, "grad_norm": 4.48134708404541, "learning_rate": 4.598981170235501e-06, "loss": 0.14192742109298706, "step": 1675 }, { "epoch": 0.43509865005192105, "grad_norm": 1.4108861684799194, "learning_rate": 4.598441951481184e-06, "loss": 0.11871317028999329, "step": 1676 }, { "epoch": 0.4353582554517134, "grad_norm": 1.130672812461853, "learning_rate": 4.597902405964654e-06, "loss": 0.11918944120407104, "step": 1677 }, { "epoch": 0.4356178608515057, "grad_norm": 2.9317407608032227, "learning_rate": 4.59736253378129e-06, "loss": 0.13373243808746338, "step": 1678 }, { "epoch": 0.43587746625129803, "grad_norm": 3.7107677459716797, "learning_rate": 4.596822335026529e-06, "loss": 0.12475594878196716, "step": 1679 }, { "epoch": 0.43613707165109034, "grad_norm": 3.858133554458618, "learning_rate": 4.596281809795866e-06, "loss": 0.1422741711139679, "step": 1680 }, { "epoch": 0.43639667705088264, "grad_norm": 1.3897037506103516, "learning_rate": 4.5957409581848535e-06, "loss": 0.127621591091156, "step": 1681 }, { "epoch": 0.43665628245067495, "grad_norm": 2.2220139503479004, "learning_rate": 4.595199780289101e-06, "loss": 0.16581612825393677, "step": 1682 }, { "epoch": 0.4369158878504673, "grad_norm": 4.627699851989746, "learning_rate": 4.5946582762042765e-06, "loss": 0.14272958040237427, "step": 1683 }, { "epoch": 0.4371754932502596, "grad_norm": 2.9615073204040527, "learning_rate": 4.594116446026106e-06, "loss": 0.13385385274887085, "step": 1684 }, { "epoch": 0.4374350986500519, "grad_norm": 2.9923272132873535, "learning_rate": 4.5935742898503725e-06, "loss": 0.12892863154411316, "step": 1685 }, { "epoch": 0.43769470404984423, "grad_norm": 1.3536230325698853, "learning_rate": 4.5930318077729165e-06, "loss": 0.1236502155661583, "step": 1686 }, { "epoch": 0.43795430944963654, "grad_norm": 2.176347255706787, "learning_rate": 4.592488999889638e-06, "loss": 0.12535397708415985, "step": 1687 }, { "epoch": 0.43821391484942884, "grad_norm": 2.114151954650879, "learning_rate": 4.5919458662964905e-06, "loss": 0.1166197657585144, "step": 1688 }, { "epoch": 0.4384735202492212, "grad_norm": 3.3243510723114014, "learning_rate": 4.591402407089489e-06, "loss": 0.12087231874465942, "step": 1689 }, { "epoch": 0.4387331256490135, "grad_norm": 3.424987316131592, "learning_rate": 4.590858622364705e-06, "loss": 0.12581633031368256, "step": 1690 }, { "epoch": 0.4389927310488058, "grad_norm": 1.4123549461364746, "learning_rate": 4.590314512218266e-06, "loss": 0.14953193068504333, "step": 1691 }, { "epoch": 0.4392523364485981, "grad_norm": 2.6847777366638184, "learning_rate": 4.58977007674636e-06, "loss": 0.1300140917301178, "step": 1692 }, { "epoch": 0.43951194184839043, "grad_norm": 4.345132350921631, "learning_rate": 4.589225316045228e-06, "loss": 0.13191547989845276, "step": 1693 }, { "epoch": 0.43977154724818274, "grad_norm": 4.2772345542907715, "learning_rate": 4.588680230211174e-06, "loss": 0.13087238371372223, "step": 1694 }, { "epoch": 0.4400311526479751, "grad_norm": 1.3113048076629639, "learning_rate": 4.588134819340554e-06, "loss": 0.12196694314479828, "step": 1695 }, { "epoch": 0.4402907580477674, "grad_norm": 1.578352451324463, "learning_rate": 4.587589083529788e-06, "loss": 0.16865167021751404, "step": 1696 }, { "epoch": 0.4405503634475597, "grad_norm": 3.6769416332244873, "learning_rate": 4.587043022875345e-06, "loss": 0.13941112160682678, "step": 1697 }, { "epoch": 0.440809968847352, "grad_norm": 4.794406890869141, "learning_rate": 4.586496637473759e-06, "loss": 0.13517311215400696, "step": 1698 }, { "epoch": 0.4410695742471443, "grad_norm": 2.5716986656188965, "learning_rate": 4.585949927421618e-06, "loss": 0.13883155584335327, "step": 1699 }, { "epoch": 0.44132917964693663, "grad_norm": 2.2271599769592285, "learning_rate": 4.5854028928155674e-06, "loss": 0.11053328216075897, "step": 1700 }, { "epoch": 0.441588785046729, "grad_norm": 0.77818763256073, "learning_rate": 4.584855533752311e-06, "loss": 0.11940284818410873, "step": 1701 }, { "epoch": 0.4418483904465213, "grad_norm": 2.3182625770568848, "learning_rate": 4.584307850328608e-06, "loss": 0.12244823575019836, "step": 1702 }, { "epoch": 0.4421079958463136, "grad_norm": 4.535281658172607, "learning_rate": 4.5837598426412775e-06, "loss": 0.12172901630401611, "step": 1703 }, { "epoch": 0.4423676012461059, "grad_norm": 3.0191352367401123, "learning_rate": 4.583211510787194e-06, "loss": 0.12324754893779755, "step": 1704 }, { "epoch": 0.4426272066458982, "grad_norm": 0.6674609780311584, "learning_rate": 4.582662854863291e-06, "loss": 0.11657015979290009, "step": 1705 }, { "epoch": 0.4428868120456905, "grad_norm": 2.150773048400879, "learning_rate": 4.582113874966558e-06, "loss": 0.1732957661151886, "step": 1706 }, { "epoch": 0.4431464174454829, "grad_norm": 2.24819016456604, "learning_rate": 4.581564571194041e-06, "loss": 0.13516610860824585, "step": 1707 }, { "epoch": 0.4434060228452752, "grad_norm": 4.660454750061035, "learning_rate": 4.581014943642845e-06, "loss": 0.1715250313282013, "step": 1708 }, { "epoch": 0.4436656282450675, "grad_norm": 4.2910284996032715, "learning_rate": 4.580464992410132e-06, "loss": 0.14790776371955872, "step": 1709 }, { "epoch": 0.4439252336448598, "grad_norm": 7.591300964355469, "learning_rate": 4.57991471759312e-06, "loss": 0.17323195934295654, "step": 1710 }, { "epoch": 0.4441848390446521, "grad_norm": 4.092546463012695, "learning_rate": 4.579364119289086e-06, "loss": 0.1557813584804535, "step": 1711 }, { "epoch": 0.4444444444444444, "grad_norm": 2.4102864265441895, "learning_rate": 4.578813197595363e-06, "loss": 0.15222591161727905, "step": 1712 }, { "epoch": 0.4447040498442368, "grad_norm": 1.6867069005966187, "learning_rate": 4.578261952609339e-06, "loss": 0.13253360986709595, "step": 1713 }, { "epoch": 0.4449636552440291, "grad_norm": 2.484297275543213, "learning_rate": 4.577710384428466e-06, "loss": 0.13509710133075714, "step": 1714 }, { "epoch": 0.4452232606438214, "grad_norm": 8.154130935668945, "learning_rate": 4.577158493150244e-06, "loss": 0.16969677805900574, "step": 1715 }, { "epoch": 0.4454828660436137, "grad_norm": 8.56702709197998, "learning_rate": 4.576606278872238e-06, "loss": 0.15467995405197144, "step": 1716 }, { "epoch": 0.445742471443406, "grad_norm": 3.878664493560791, "learning_rate": 4.576053741692065e-06, "loss": 0.1445457488298416, "step": 1717 }, { "epoch": 0.4460020768431983, "grad_norm": 1.536787509918213, "learning_rate": 4.575500881707402e-06, "loss": 0.14411714673042297, "step": 1718 }, { "epoch": 0.4462616822429907, "grad_norm": 2.212897300720215, "learning_rate": 4.574947699015981e-06, "loss": 0.15296661853790283, "step": 1719 }, { "epoch": 0.446521287642783, "grad_norm": 4.747832775115967, "learning_rate": 4.574394193715593e-06, "loss": 0.17123346030712128, "step": 1720 }, { "epoch": 0.4467808930425753, "grad_norm": 5.353792667388916, "learning_rate": 4.573840365904084e-06, "loss": 0.13824281096458435, "step": 1721 }, { "epoch": 0.4470404984423676, "grad_norm": 5.595029830932617, "learning_rate": 4.57328621567936e-06, "loss": 0.1605689525604248, "step": 1722 }, { "epoch": 0.4473001038421599, "grad_norm": 2.6258039474487305, "learning_rate": 4.57273174313938e-06, "loss": 0.12569542229175568, "step": 1723 }, { "epoch": 0.4475597092419522, "grad_norm": 2.9106035232543945, "learning_rate": 4.5721769483821625e-06, "loss": 0.18376091122627258, "step": 1724 }, { "epoch": 0.44781931464174457, "grad_norm": 2.916879415512085, "learning_rate": 4.571621831505783e-06, "loss": 0.1382085382938385, "step": 1725 }, { "epoch": 0.4480789200415369, "grad_norm": 4.750349521636963, "learning_rate": 4.571066392608374e-06, "loss": 0.17275121808052063, "step": 1726 }, { "epoch": 0.4483385254413292, "grad_norm": 3.083491325378418, "learning_rate": 4.5705106317881225e-06, "loss": 0.1632036715745926, "step": 1727 }, { "epoch": 0.4485981308411215, "grad_norm": 3.7095131874084473, "learning_rate": 4.569954549143276e-06, "loss": 0.15552903711795807, "step": 1728 }, { "epoch": 0.4488577362409138, "grad_norm": 1.1219167709350586, "learning_rate": 4.569398144772137e-06, "loss": 0.11966659128665924, "step": 1729 }, { "epoch": 0.4491173416407061, "grad_norm": 1.409401535987854, "learning_rate": 4.568841418773064e-06, "loss": 0.12199245393276215, "step": 1730 }, { "epoch": 0.44937694704049846, "grad_norm": 2.0045056343078613, "learning_rate": 4.568284371244475e-06, "loss": 0.1417345106601715, "step": 1731 }, { "epoch": 0.44963655244029077, "grad_norm": 4.046040058135986, "learning_rate": 4.567727002284842e-06, "loss": 0.15857544541358948, "step": 1732 }, { "epoch": 0.4498961578400831, "grad_norm": 2.5595920085906982, "learning_rate": 4.567169311992696e-06, "loss": 0.13453473150730133, "step": 1733 }, { "epoch": 0.4501557632398754, "grad_norm": 3.9344425201416016, "learning_rate": 4.566611300466622e-06, "loss": 0.1515001356601715, "step": 1734 }, { "epoch": 0.4504153686396677, "grad_norm": 0.992333173751831, "learning_rate": 4.566052967805266e-06, "loss": 0.1208975687623024, "step": 1735 }, { "epoch": 0.45067497403946, "grad_norm": 2.427849292755127, "learning_rate": 4.565494314107327e-06, "loss": 0.11826874315738678, "step": 1736 }, { "epoch": 0.45093457943925236, "grad_norm": 2.7974116802215576, "learning_rate": 4.564935339471563e-06, "loss": 0.13365446031093597, "step": 1737 }, { "epoch": 0.45119418483904467, "grad_norm": 1.5536149740219116, "learning_rate": 4.564376043996787e-06, "loss": 0.13476040959358215, "step": 1738 }, { "epoch": 0.45145379023883697, "grad_norm": 2.8034873008728027, "learning_rate": 4.5638164277818705e-06, "loss": 0.12558692693710327, "step": 1739 }, { "epoch": 0.4517133956386293, "grad_norm": 1.1533253192901611, "learning_rate": 4.56325649092574e-06, "loss": 0.14368772506713867, "step": 1740 }, { "epoch": 0.4519730010384216, "grad_norm": 2.240077257156372, "learning_rate": 4.56269623352738e-06, "loss": 0.13400670886039734, "step": 1741 }, { "epoch": 0.4522326064382139, "grad_norm": 1.7022830247879028, "learning_rate": 4.562135655685831e-06, "loss": 0.13596117496490479, "step": 1742 }, { "epoch": 0.45249221183800625, "grad_norm": 1.080570101737976, "learning_rate": 4.561574757500192e-06, "loss": 0.1286035031080246, "step": 1743 }, { "epoch": 0.45275181723779856, "grad_norm": 1.7854830026626587, "learning_rate": 4.5610135390696135e-06, "loss": 0.12896966934204102, "step": 1744 }, { "epoch": 0.45301142263759087, "grad_norm": 2.2614738941192627, "learning_rate": 4.560452000493309e-06, "loss": 0.10973799228668213, "step": 1745 }, { "epoch": 0.4532710280373832, "grad_norm": 1.3039029836654663, "learning_rate": 4.559890141870545e-06, "loss": 0.14571671187877655, "step": 1746 }, { "epoch": 0.4535306334371755, "grad_norm": 1.1131136417388916, "learning_rate": 4.559327963300643e-06, "loss": 0.12212885916233063, "step": 1747 }, { "epoch": 0.4537902388369678, "grad_norm": 2.1384804248809814, "learning_rate": 4.558765464882987e-06, "loss": 0.19145464897155762, "step": 1748 }, { "epoch": 0.45404984423676015, "grad_norm": 2.6245298385620117, "learning_rate": 4.558202646717011e-06, "loss": 0.12006436288356781, "step": 1749 }, { "epoch": 0.45430944963655245, "grad_norm": 1.6786103248596191, "learning_rate": 4.557639508902211e-06, "loss": 0.12325862050056458, "step": 1750 }, { "epoch": 0.45456905503634476, "grad_norm": 1.1814885139465332, "learning_rate": 4.557076051538134e-06, "loss": 0.11767333745956421, "step": 1751 }, { "epoch": 0.45482866043613707, "grad_norm": 1.2700996398925781, "learning_rate": 4.556512274724387e-06, "loss": 0.14106649160385132, "step": 1752 }, { "epoch": 0.4550882658359294, "grad_norm": 3.041506290435791, "learning_rate": 4.555948178560635e-06, "loss": 0.10314895957708359, "step": 1753 }, { "epoch": 0.4553478712357217, "grad_norm": 1.3904231786727905, "learning_rate": 4.555383763146596e-06, "loss": 0.12951233983039856, "step": 1754 }, { "epoch": 0.45560747663551404, "grad_norm": 1.4796894788742065, "learning_rate": 4.554819028582044e-06, "loss": 0.11966925859451294, "step": 1755 }, { "epoch": 0.45586708203530635, "grad_norm": 1.5372731685638428, "learning_rate": 4.5542539749668134e-06, "loss": 0.1223294734954834, "step": 1756 }, { "epoch": 0.45612668743509865, "grad_norm": 2.129375457763672, "learning_rate": 4.553688602400792e-06, "loss": 0.1415310800075531, "step": 1757 }, { "epoch": 0.45638629283489096, "grad_norm": 1.630109190940857, "learning_rate": 4.5531229109839244e-06, "loss": 0.14075665175914764, "step": 1758 }, { "epoch": 0.45664589823468327, "grad_norm": 1.6748883724212646, "learning_rate": 4.552556900816212e-06, "loss": 0.1195715069770813, "step": 1759 }, { "epoch": 0.4569055036344756, "grad_norm": 1.8601678609848022, "learning_rate": 4.551990571997713e-06, "loss": 0.11719448864459991, "step": 1760 }, { "epoch": 0.45716510903426794, "grad_norm": 2.0383265018463135, "learning_rate": 4.551423924628541e-06, "loss": 0.11340485513210297, "step": 1761 }, { "epoch": 0.45742471443406024, "grad_norm": 1.1788403987884521, "learning_rate": 4.550856958808867e-06, "loss": 0.14531102776527405, "step": 1762 }, { "epoch": 0.45768431983385255, "grad_norm": 1.6508508920669556, "learning_rate": 4.550289674638915e-06, "loss": 0.14885476231575012, "step": 1763 }, { "epoch": 0.45794392523364486, "grad_norm": 1.530574917793274, "learning_rate": 4.549722072218971e-06, "loss": 0.12988042831420898, "step": 1764 }, { "epoch": 0.45820353063343716, "grad_norm": 3.4717020988464355, "learning_rate": 4.5491541516493725e-06, "loss": 0.1278572678565979, "step": 1765 }, { "epoch": 0.45846313603322947, "grad_norm": 1.2402560710906982, "learning_rate": 4.548585913030515e-06, "loss": 0.12435615062713623, "step": 1766 }, { "epoch": 0.45872274143302183, "grad_norm": 1.604663372039795, "learning_rate": 4.548017356462851e-06, "loss": 0.13121247291564941, "step": 1767 }, { "epoch": 0.45898234683281414, "grad_norm": 1.9529272317886353, "learning_rate": 4.547448482046887e-06, "loss": 0.17189496755599976, "step": 1768 }, { "epoch": 0.45924195223260644, "grad_norm": 2.200683116912842, "learning_rate": 4.546879289883187e-06, "loss": 0.1514635533094406, "step": 1769 }, { "epoch": 0.45950155763239875, "grad_norm": 2.3669373989105225, "learning_rate": 4.5463097800723725e-06, "loss": 0.154879629611969, "step": 1770 }, { "epoch": 0.45976116303219106, "grad_norm": 1.4020566940307617, "learning_rate": 4.545739952715118e-06, "loss": 0.11697366833686829, "step": 1771 }, { "epoch": 0.46002076843198336, "grad_norm": 1.089242696762085, "learning_rate": 4.5451698079121575e-06, "loss": 0.13697996735572815, "step": 1772 }, { "epoch": 0.4602803738317757, "grad_norm": 2.0645477771759033, "learning_rate": 4.544599345764279e-06, "loss": 0.13536815345287323, "step": 1773 }, { "epoch": 0.46053997923156803, "grad_norm": 2.2729411125183105, "learning_rate": 4.544028566372326e-06, "loss": 0.1337997019290924, "step": 1774 }, { "epoch": 0.46079958463136034, "grad_norm": 1.3547574281692505, "learning_rate": 4.543457469837202e-06, "loss": 0.13051819801330566, "step": 1775 }, { "epoch": 0.46105919003115264, "grad_norm": 2.398087739944458, "learning_rate": 4.542886056259861e-06, "loss": 0.11951974034309387, "step": 1776 }, { "epoch": 0.46131879543094495, "grad_norm": 1.6879342794418335, "learning_rate": 4.542314325741316e-06, "loss": 0.12417396903038025, "step": 1777 }, { "epoch": 0.46157840083073726, "grad_norm": 1.2597341537475586, "learning_rate": 4.541742278382638e-06, "loss": 0.1413659006357193, "step": 1778 }, { "epoch": 0.4618380062305296, "grad_norm": 1.024936318397522, "learning_rate": 4.541169914284951e-06, "loss": 0.1300259381532669, "step": 1779 }, { "epoch": 0.4620976116303219, "grad_norm": 2.128288984298706, "learning_rate": 4.5405972335494345e-06, "loss": 0.12893760204315186, "step": 1780 }, { "epoch": 0.46235721703011423, "grad_norm": 1.383245825767517, "learning_rate": 4.540024236277327e-06, "loss": 0.1455286592245102, "step": 1781 }, { "epoch": 0.46261682242990654, "grad_norm": 1.7414774894714355, "learning_rate": 4.539450922569921e-06, "loss": 0.15617258846759796, "step": 1782 }, { "epoch": 0.46287642782969884, "grad_norm": 1.9982081651687622, "learning_rate": 4.538877292528564e-06, "loss": 0.1486908346414566, "step": 1783 }, { "epoch": 0.46313603322949115, "grad_norm": 0.9898884296417236, "learning_rate": 4.538303346254662e-06, "loss": 0.11138419806957245, "step": 1784 }, { "epoch": 0.4633956386292835, "grad_norm": 1.0198776721954346, "learning_rate": 4.537729083849675e-06, "loss": 0.1342882513999939, "step": 1785 }, { "epoch": 0.4636552440290758, "grad_norm": 1.4686819314956665, "learning_rate": 4.53715450541512e-06, "loss": 0.1323826014995575, "step": 1786 }, { "epoch": 0.4639148494288681, "grad_norm": 2.2554898262023926, "learning_rate": 4.5365796110525674e-06, "loss": 0.1278790980577469, "step": 1787 }, { "epoch": 0.46417445482866043, "grad_norm": 2.5155763626098633, "learning_rate": 4.5360044008636485e-06, "loss": 0.14516526460647583, "step": 1788 }, { "epoch": 0.46443406022845274, "grad_norm": 1.1039870977401733, "learning_rate": 4.535428874950045e-06, "loss": 0.14517179131507874, "step": 1789 }, { "epoch": 0.46469366562824505, "grad_norm": 1.8181204795837402, "learning_rate": 4.5348530334134975e-06, "loss": 0.13055585324764252, "step": 1790 }, { "epoch": 0.4649532710280374, "grad_norm": 3.6424407958984375, "learning_rate": 4.534276876355801e-06, "loss": 0.13318926095962524, "step": 1791 }, { "epoch": 0.4652128764278297, "grad_norm": 1.4786808490753174, "learning_rate": 4.533700403878806e-06, "loss": 0.15607786178588867, "step": 1792 }, { "epoch": 0.465472481827622, "grad_norm": 2.9111955165863037, "learning_rate": 4.533123616084423e-06, "loss": 0.1278553009033203, "step": 1793 }, { "epoch": 0.4657320872274143, "grad_norm": 2.0760834217071533, "learning_rate": 4.5325465130746125e-06, "loss": 0.15321241319179535, "step": 1794 }, { "epoch": 0.46599169262720663, "grad_norm": 2.1516520977020264, "learning_rate": 4.531969094951393e-06, "loss": 0.15321381390094757, "step": 1795 }, { "epoch": 0.46625129802699894, "grad_norm": 1.9027191400527954, "learning_rate": 4.5313913618168394e-06, "loss": 0.1224946528673172, "step": 1796 }, { "epoch": 0.4665109034267913, "grad_norm": 1.1397899389266968, "learning_rate": 4.530813313773081e-06, "loss": 0.12374964356422424, "step": 1797 }, { "epoch": 0.4667705088265836, "grad_norm": 0.9498236775398254, "learning_rate": 4.530234950922304e-06, "loss": 0.11376450210809708, "step": 1798 }, { "epoch": 0.4670301142263759, "grad_norm": 5.7746500968933105, "learning_rate": 4.52965627336675e-06, "loss": 0.157751202583313, "step": 1799 }, { "epoch": 0.4672897196261682, "grad_norm": 1.6855329275131226, "learning_rate": 4.529077281208716e-06, "loss": 0.11908071488142014, "step": 1800 }, { "epoch": 0.4675493250259605, "grad_norm": 1.0469341278076172, "learning_rate": 4.528497974550553e-06, "loss": 0.12086302042007446, "step": 1801 }, { "epoch": 0.46780893042575283, "grad_norm": 1.842877984046936, "learning_rate": 4.527918353494672e-06, "loss": 0.12028856575489044, "step": 1802 }, { "epoch": 0.4680685358255452, "grad_norm": 1.1352031230926514, "learning_rate": 4.527338418143533e-06, "loss": 0.1436455249786377, "step": 1803 }, { "epoch": 0.4683281412253375, "grad_norm": 1.5549169778823853, "learning_rate": 4.526758168599658e-06, "loss": 0.11683764308691025, "step": 1804 }, { "epoch": 0.4685877466251298, "grad_norm": 4.235115051269531, "learning_rate": 4.526177604965622e-06, "loss": 0.13589338958263397, "step": 1805 }, { "epoch": 0.4688473520249221, "grad_norm": 1.6850557327270508, "learning_rate": 4.525596727344054e-06, "loss": 0.12112563848495483, "step": 1806 }, { "epoch": 0.4691069574247144, "grad_norm": 2.2262167930603027, "learning_rate": 4.52501553583764e-06, "loss": 0.135565847158432, "step": 1807 }, { "epoch": 0.46936656282450673, "grad_norm": 2.552520513534546, "learning_rate": 4.524434030549122e-06, "loss": 0.12720969319343567, "step": 1808 }, { "epoch": 0.4696261682242991, "grad_norm": 5.42121696472168, "learning_rate": 4.523852211581297e-06, "loss": 0.15359577536582947, "step": 1809 }, { "epoch": 0.4698857736240914, "grad_norm": 1.253149390220642, "learning_rate": 4.523270079037016e-06, "loss": 0.11058098077774048, "step": 1810 }, { "epoch": 0.4701453790238837, "grad_norm": 0.9539614915847778, "learning_rate": 4.522687633019189e-06, "loss": 0.11006379872560501, "step": 1811 }, { "epoch": 0.470404984423676, "grad_norm": 2.816179037094116, "learning_rate": 4.522104873630776e-06, "loss": 0.13404101133346558, "step": 1812 }, { "epoch": 0.4706645898234683, "grad_norm": 2.7834038734436035, "learning_rate": 4.521521800974798e-06, "loss": 0.15591706335544586, "step": 1813 }, { "epoch": 0.4709241952232606, "grad_norm": 1.9530858993530273, "learning_rate": 4.520938415154328e-06, "loss": 0.1496381163597107, "step": 1814 }, { "epoch": 0.471183800623053, "grad_norm": 0.6597929000854492, "learning_rate": 4.520354716272495e-06, "loss": 0.11875079572200775, "step": 1815 }, { "epoch": 0.4714434060228453, "grad_norm": 2.6142678260803223, "learning_rate": 4.519770704432484e-06, "loss": 0.13155284523963928, "step": 1816 }, { "epoch": 0.4717030114226376, "grad_norm": 1.7818530797958374, "learning_rate": 4.519186379737535e-06, "loss": 0.13918998837471008, "step": 1817 }, { "epoch": 0.4719626168224299, "grad_norm": 2.6070566177368164, "learning_rate": 4.518601742290943e-06, "loss": 0.14336302876472473, "step": 1818 }, { "epoch": 0.4722222222222222, "grad_norm": 1.7153091430664062, "learning_rate": 4.518016792196059e-06, "loss": 0.15349169075489044, "step": 1819 }, { "epoch": 0.4724818276220145, "grad_norm": 2.8545031547546387, "learning_rate": 4.5174315295562885e-06, "loss": 0.1637812852859497, "step": 1820 }, { "epoch": 0.4727414330218069, "grad_norm": 1.2636572122573853, "learning_rate": 4.516845954475091e-06, "loss": 0.1250220686197281, "step": 1821 }, { "epoch": 0.4730010384215992, "grad_norm": 1.198575735092163, "learning_rate": 4.516260067055985e-06, "loss": 0.11448890715837479, "step": 1822 }, { "epoch": 0.4732606438213915, "grad_norm": 1.0601954460144043, "learning_rate": 4.515673867402541e-06, "loss": 0.12754875421524048, "step": 1823 }, { "epoch": 0.4735202492211838, "grad_norm": 1.5954453945159912, "learning_rate": 4.515087355618387e-06, "loss": 0.1483316719532013, "step": 1824 }, { "epoch": 0.4737798546209761, "grad_norm": 2.736315965652466, "learning_rate": 4.514500531807201e-06, "loss": 0.11778105795383453, "step": 1825 }, { "epoch": 0.4740394600207684, "grad_norm": 2.4077601432800293, "learning_rate": 4.513913396072725e-06, "loss": 0.1463821828365326, "step": 1826 }, { "epoch": 0.4742990654205608, "grad_norm": 1.5257127285003662, "learning_rate": 4.513325948518748e-06, "loss": 0.11738690733909607, "step": 1827 }, { "epoch": 0.4745586708203531, "grad_norm": 3.0432658195495605, "learning_rate": 4.512738189249119e-06, "loss": 0.12899631261825562, "step": 1828 }, { "epoch": 0.4748182762201454, "grad_norm": 1.963335394859314, "learning_rate": 4.512150118367739e-06, "loss": 0.13969311118125916, "step": 1829 }, { "epoch": 0.4750778816199377, "grad_norm": 1.572925329208374, "learning_rate": 4.511561735978564e-06, "loss": 0.13658727705478668, "step": 1830 }, { "epoch": 0.47533748701973, "grad_norm": 1.5065006017684937, "learning_rate": 4.510973042185611e-06, "loss": 0.12296489626169205, "step": 1831 }, { "epoch": 0.4755970924195223, "grad_norm": 1.4822821617126465, "learning_rate": 4.5103840370929446e-06, "loss": 0.13091956079006195, "step": 1832 }, { "epoch": 0.47585669781931467, "grad_norm": 1.5760189294815063, "learning_rate": 4.509794720804688e-06, "loss": 0.13587278127670288, "step": 1833 }, { "epoch": 0.476116303219107, "grad_norm": 1.0058956146240234, "learning_rate": 4.509205093425018e-06, "loss": 0.11011812835931778, "step": 1834 }, { "epoch": 0.4763759086188993, "grad_norm": 0.8850865364074707, "learning_rate": 4.508615155058168e-06, "loss": 0.13186320662498474, "step": 1835 }, { "epoch": 0.4766355140186916, "grad_norm": 1.809670090675354, "learning_rate": 4.508024905808426e-06, "loss": 0.13151708245277405, "step": 1836 }, { "epoch": 0.4768951194184839, "grad_norm": 1.5583473443984985, "learning_rate": 4.507434345780133e-06, "loss": 0.13265091180801392, "step": 1837 }, { "epoch": 0.4771547248182762, "grad_norm": 1.9397122859954834, "learning_rate": 4.506843475077689e-06, "loss": 0.14261949062347412, "step": 1838 }, { "epoch": 0.47741433021806856, "grad_norm": 1.558482050895691, "learning_rate": 4.5062522938055444e-06, "loss": 0.15367679297924042, "step": 1839 }, { "epoch": 0.47767393561786087, "grad_norm": 1.3327289819717407, "learning_rate": 4.505660802068207e-06, "loss": 0.1315830498933792, "step": 1840 }, { "epoch": 0.4779335410176532, "grad_norm": 2.041689872741699, "learning_rate": 4.50506899997024e-06, "loss": 0.11022370308637619, "step": 1841 }, { "epoch": 0.4781931464174455, "grad_norm": 1.3487364053726196, "learning_rate": 4.504476887616258e-06, "loss": 0.14281488955020905, "step": 1842 }, { "epoch": 0.4784527518172378, "grad_norm": 2.1992015838623047, "learning_rate": 4.503884465110935e-06, "loss": 0.161794975399971, "step": 1843 }, { "epoch": 0.4787123572170301, "grad_norm": 1.7328448295593262, "learning_rate": 4.503291732558998e-06, "loss": 0.13862216472625732, "step": 1844 }, { "epoch": 0.47897196261682246, "grad_norm": 2.4418721199035645, "learning_rate": 4.502698690065227e-06, "loss": 0.1278693526983261, "step": 1845 }, { "epoch": 0.47923156801661476, "grad_norm": 1.8263416290283203, "learning_rate": 4.5021053377344596e-06, "loss": 0.13039623200893402, "step": 1846 }, { "epoch": 0.47949117341640707, "grad_norm": 2.4038076400756836, "learning_rate": 4.501511675671586e-06, "loss": 0.1264672577381134, "step": 1847 }, { "epoch": 0.4797507788161994, "grad_norm": 1.8041454553604126, "learning_rate": 4.500917703981553e-06, "loss": 0.16111324727535248, "step": 1848 }, { "epoch": 0.4800103842159917, "grad_norm": 1.2357579469680786, "learning_rate": 4.5003234227693596e-06, "loss": 0.12493743002414703, "step": 1849 }, { "epoch": 0.480269989615784, "grad_norm": 1.4858084917068481, "learning_rate": 4.499728832140063e-06, "loss": 0.14188063144683838, "step": 1850 }, { "epoch": 0.48052959501557635, "grad_norm": 2.249511241912842, "learning_rate": 4.499133932198772e-06, "loss": 0.1499459147453308, "step": 1851 }, { "epoch": 0.48078920041536866, "grad_norm": 1.9945855140686035, "learning_rate": 4.498538723050651e-06, "loss": 0.1452936828136444, "step": 1852 }, { "epoch": 0.48104880581516096, "grad_norm": 3.2174489498138428, "learning_rate": 4.4979432048009215e-06, "loss": 0.10732902586460114, "step": 1853 }, { "epoch": 0.48130841121495327, "grad_norm": 2.7485804557800293, "learning_rate": 4.497347377554855e-06, "loss": 0.12716594338417053, "step": 1854 }, { "epoch": 0.4815680166147456, "grad_norm": 2.9535024166107178, "learning_rate": 4.4967512414177816e-06, "loss": 0.11613470315933228, "step": 1855 }, { "epoch": 0.4818276220145379, "grad_norm": 1.2572975158691406, "learning_rate": 4.496154796495083e-06, "loss": 0.12405359745025635, "step": 1856 }, { "epoch": 0.48208722741433024, "grad_norm": 2.1951651573181152, "learning_rate": 4.495558042892199e-06, "loss": 0.1334218680858612, "step": 1857 }, { "epoch": 0.48234683281412255, "grad_norm": 1.205174446105957, "learning_rate": 4.49496098071462e-06, "loss": 0.12222971022129059, "step": 1858 }, { "epoch": 0.48260643821391486, "grad_norm": 1.2937294244766235, "learning_rate": 4.494363610067895e-06, "loss": 0.13910125195980072, "step": 1859 }, { "epoch": 0.48286604361370716, "grad_norm": 3.245677947998047, "learning_rate": 4.493765931057624e-06, "loss": 0.10896074771881104, "step": 1860 }, { "epoch": 0.48312564901349947, "grad_norm": 3.4530513286590576, "learning_rate": 4.493167943789463e-06, "loss": 0.13808558881282806, "step": 1861 }, { "epoch": 0.4833852544132918, "grad_norm": 0.7433202266693115, "learning_rate": 4.492569648369124e-06, "loss": 0.12159725278615952, "step": 1862 }, { "epoch": 0.48364485981308414, "grad_norm": 3.2931768894195557, "learning_rate": 4.4919710449023695e-06, "loss": 0.13806161284446716, "step": 1863 }, { "epoch": 0.48390446521287644, "grad_norm": 4.434366703033447, "learning_rate": 4.491372133495021e-06, "loss": 0.15912756323814392, "step": 1864 }, { "epoch": 0.48416407061266875, "grad_norm": 3.820781707763672, "learning_rate": 4.49077291425295e-06, "loss": 0.1527813822031021, "step": 1865 }, { "epoch": 0.48442367601246106, "grad_norm": 2.182077169418335, "learning_rate": 4.490173387282087e-06, "loss": 0.12693214416503906, "step": 1866 }, { "epoch": 0.48468328141225336, "grad_norm": 3.944525718688965, "learning_rate": 4.489573552688414e-06, "loss": 0.11391797661781311, "step": 1867 }, { "epoch": 0.48494288681204567, "grad_norm": 3.8817100524902344, "learning_rate": 4.488973410577968e-06, "loss": 0.13540999591350555, "step": 1868 }, { "epoch": 0.48520249221183803, "grad_norm": 2.975600242614746, "learning_rate": 4.48837296105684e-06, "loss": 0.12012440711259842, "step": 1869 }, { "epoch": 0.48546209761163034, "grad_norm": 1.9135080575942993, "learning_rate": 4.4877722042311765e-06, "loss": 0.14273720979690552, "step": 1870 }, { "epoch": 0.48572170301142265, "grad_norm": 2.907864809036255, "learning_rate": 4.487171140207176e-06, "loss": 0.14129340648651123, "step": 1871 }, { "epoch": 0.48598130841121495, "grad_norm": 2.729306697845459, "learning_rate": 4.4865697690910945e-06, "loss": 0.14184294641017914, "step": 1872 }, { "epoch": 0.48624091381100726, "grad_norm": 2.874668598175049, "learning_rate": 4.48596809098924e-06, "loss": 0.14474241435527802, "step": 1873 }, { "epoch": 0.48650051921079956, "grad_norm": 1.4771251678466797, "learning_rate": 4.485366106007975e-06, "loss": 0.12957651913166046, "step": 1874 }, { "epoch": 0.4867601246105919, "grad_norm": 1.3953096866607666, "learning_rate": 4.484763814253717e-06, "loss": 0.1523284912109375, "step": 1875 }, { "epoch": 0.48701973001038423, "grad_norm": 1.4436582326889038, "learning_rate": 4.484161215832938e-06, "loss": 0.1193288266658783, "step": 1876 }, { "epoch": 0.48727933541017654, "grad_norm": 5.009468078613281, "learning_rate": 4.483558310852163e-06, "loss": 0.1537945568561554, "step": 1877 }, { "epoch": 0.48753894080996885, "grad_norm": 3.5555946826934814, "learning_rate": 4.482955099417971e-06, "loss": 0.10814721137285233, "step": 1878 }, { "epoch": 0.48779854620976115, "grad_norm": 1.8423430919647217, "learning_rate": 4.482351581636998e-06, "loss": 0.09365713596343994, "step": 1879 }, { "epoch": 0.48805815160955346, "grad_norm": 1.9577182531356812, "learning_rate": 4.48174775761593e-06, "loss": 0.13206204771995544, "step": 1880 }, { "epoch": 0.4883177570093458, "grad_norm": 2.370722770690918, "learning_rate": 4.48114362746151e-06, "loss": 0.14106649160385132, "step": 1881 }, { "epoch": 0.4885773624091381, "grad_norm": 3.489910364151001, "learning_rate": 4.4805391912805366e-06, "loss": 0.1312439888715744, "step": 1882 }, { "epoch": 0.48883696780893043, "grad_norm": 2.3507049083709717, "learning_rate": 4.4799344491798564e-06, "loss": 0.12666870653629303, "step": 1883 }, { "epoch": 0.48909657320872274, "grad_norm": 2.2838008403778076, "learning_rate": 4.4793294012663765e-06, "loss": 0.16065439581871033, "step": 1884 }, { "epoch": 0.48935617860851505, "grad_norm": 1.4251174926757812, "learning_rate": 4.478724047647054e-06, "loss": 0.13918066024780273, "step": 1885 }, { "epoch": 0.48961578400830735, "grad_norm": 3.5822508335113525, "learning_rate": 4.478118388428904e-06, "loss": 0.13122577965259552, "step": 1886 }, { "epoch": 0.4898753894080997, "grad_norm": 0.9258812069892883, "learning_rate": 4.47751242371899e-06, "loss": 0.12649771571159363, "step": 1887 }, { "epoch": 0.490134994807892, "grad_norm": 1.9192157983779907, "learning_rate": 4.476906153624435e-06, "loss": 0.1086912676692009, "step": 1888 }, { "epoch": 0.49039460020768433, "grad_norm": 1.920163631439209, "learning_rate": 4.476299578252414e-06, "loss": 0.13736256957054138, "step": 1889 }, { "epoch": 0.49065420560747663, "grad_norm": 3.039689779281616, "learning_rate": 4.475692697710153e-06, "loss": 0.14245477318763733, "step": 1890 }, { "epoch": 0.49091381100726894, "grad_norm": 2.6105499267578125, "learning_rate": 4.475085512104937e-06, "loss": 0.13097722828388214, "step": 1891 }, { "epoch": 0.49117341640706125, "grad_norm": 3.1602158546447754, "learning_rate": 4.474478021544102e-06, "loss": 0.14704734086990356, "step": 1892 }, { "epoch": 0.4914330218068536, "grad_norm": 0.7411168217658997, "learning_rate": 4.473870226135036e-06, "loss": 0.13819274306297302, "step": 1893 }, { "epoch": 0.4916926272066459, "grad_norm": 1.678052544593811, "learning_rate": 4.473262125985188e-06, "loss": 0.13161206245422363, "step": 1894 }, { "epoch": 0.4919522326064382, "grad_norm": 1.6172904968261719, "learning_rate": 4.472653721202052e-06, "loss": 0.12925076484680176, "step": 1895 }, { "epoch": 0.49221183800623053, "grad_norm": 3.4782943725585938, "learning_rate": 4.472045011893182e-06, "loss": 0.18291963636875153, "step": 1896 }, { "epoch": 0.49247144340602284, "grad_norm": 1.9833647012710571, "learning_rate": 4.471435998166185e-06, "loss": 0.12682737410068512, "step": 1897 }, { "epoch": 0.49273104880581514, "grad_norm": 1.4531958103179932, "learning_rate": 4.470826680128717e-06, "loss": 0.13076822459697723, "step": 1898 }, { "epoch": 0.4929906542056075, "grad_norm": 2.180894613265991, "learning_rate": 4.470217057888495e-06, "loss": 0.1341886818408966, "step": 1899 }, { "epoch": 0.4932502596053998, "grad_norm": 1.2759350538253784, "learning_rate": 4.469607131553284e-06, "loss": 0.13045895099639893, "step": 1900 }, { "epoch": 0.4935098650051921, "grad_norm": 1.6635873317718506, "learning_rate": 4.468996901230907e-06, "loss": 0.11302178353071213, "step": 1901 }, { "epoch": 0.4937694704049844, "grad_norm": 3.481642961502075, "learning_rate": 4.468386367029236e-06, "loss": 0.1595846712589264, "step": 1902 }, { "epoch": 0.49402907580477673, "grad_norm": 3.0783374309539795, "learning_rate": 4.467775529056203e-06, "loss": 0.12742136418819427, "step": 1903 }, { "epoch": 0.49428868120456904, "grad_norm": 1.984643816947937, "learning_rate": 4.467164387419787e-06, "loss": 0.1002914309501648, "step": 1904 }, { "epoch": 0.4945482866043614, "grad_norm": 0.9775622487068176, "learning_rate": 4.466552942228026e-06, "loss": 0.11737390607595444, "step": 1905 }, { "epoch": 0.4948078920041537, "grad_norm": 2.762321949005127, "learning_rate": 4.465941193589008e-06, "loss": 0.11303883790969849, "step": 1906 }, { "epoch": 0.495067497403946, "grad_norm": 2.2537105083465576, "learning_rate": 4.465329141610877e-06, "loss": 0.1379968225955963, "step": 1907 }, { "epoch": 0.4953271028037383, "grad_norm": 4.0398077964782715, "learning_rate": 4.46471678640183e-06, "loss": 0.1344737410545349, "step": 1908 }, { "epoch": 0.4955867082035306, "grad_norm": 3.6244781017303467, "learning_rate": 4.4641041280701164e-06, "loss": 0.13157936930656433, "step": 1909 }, { "epoch": 0.49584631360332293, "grad_norm": 2.1568305492401123, "learning_rate": 4.463491166724041e-06, "loss": 0.13017085194587708, "step": 1910 }, { "epoch": 0.4961059190031153, "grad_norm": 3.4219624996185303, "learning_rate": 4.462877902471961e-06, "loss": 0.11928484588861465, "step": 1911 }, { "epoch": 0.4963655244029076, "grad_norm": 4.338759899139404, "learning_rate": 4.462264335422288e-06, "loss": 0.12400345504283905, "step": 1912 }, { "epoch": 0.4966251298026999, "grad_norm": 3.2041923999786377, "learning_rate": 4.4616504656834855e-06, "loss": 0.11529414355754852, "step": 1913 }, { "epoch": 0.4968847352024922, "grad_norm": 1.1186758279800415, "learning_rate": 4.461036293364073e-06, "loss": 0.12444297969341278, "step": 1914 }, { "epoch": 0.4971443406022845, "grad_norm": 2.2396717071533203, "learning_rate": 4.46042181857262e-06, "loss": 0.10437227040529251, "step": 1915 }, { "epoch": 0.4974039460020768, "grad_norm": 4.060431957244873, "learning_rate": 4.459807041417755e-06, "loss": 0.1352619230747223, "step": 1916 }, { "epoch": 0.4976635514018692, "grad_norm": 3.9089112281799316, "learning_rate": 4.4591919620081526e-06, "loss": 0.13260382413864136, "step": 1917 }, { "epoch": 0.4979231568016615, "grad_norm": 1.9383431673049927, "learning_rate": 4.458576580452547e-06, "loss": 0.12443087995052338, "step": 1918 }, { "epoch": 0.4981827622014538, "grad_norm": 1.5974781513214111, "learning_rate": 4.457960896859724e-06, "loss": 0.12067008018493652, "step": 1919 }, { "epoch": 0.4984423676012461, "grad_norm": 2.462010622024536, "learning_rate": 4.4573449113385214e-06, "loss": 0.1360504925251007, "step": 1920 }, { "epoch": 0.4987019730010384, "grad_norm": 4.124441623687744, "learning_rate": 4.456728623997831e-06, "loss": 0.14130184054374695, "step": 1921 }, { "epoch": 0.4989615784008307, "grad_norm": 0.854224443435669, "learning_rate": 4.456112034946599e-06, "loss": 0.12208720296621323, "step": 1922 }, { "epoch": 0.4992211838006231, "grad_norm": 1.7405366897583008, "learning_rate": 4.455495144293824e-06, "loss": 0.13688631355762482, "step": 1923 }, { "epoch": 0.4994807892004154, "grad_norm": 2.0130069255828857, "learning_rate": 4.454877952148558e-06, "loss": 0.15188643336296082, "step": 1924 }, { "epoch": 0.4997403946002077, "grad_norm": 1.6722965240478516, "learning_rate": 4.454260458619907e-06, "loss": 0.14078710973262787, "step": 1925 }, { "epoch": 0.5, "grad_norm": 2.468806028366089, "learning_rate": 4.45364266381703e-06, "loss": 0.11940490454435349, "step": 1926 }, { "epoch": 0.5002596053997923, "grad_norm": 1.7768560647964478, "learning_rate": 4.453024567849136e-06, "loss": 0.12459972500801086, "step": 1927 }, { "epoch": 0.5005192107995846, "grad_norm": 1.5876544713974, "learning_rate": 4.452406170825494e-06, "loss": 0.13611580431461334, "step": 1928 }, { "epoch": 0.5007788161993769, "grad_norm": 2.0420243740081787, "learning_rate": 4.451787472855421e-06, "loss": 0.11474601924419403, "step": 1929 }, { "epoch": 0.5010384215991692, "grad_norm": 1.2113434076309204, "learning_rate": 4.451168474048288e-06, "loss": 0.14830663800239563, "step": 1930 }, { "epoch": 0.5012980269989615, "grad_norm": 2.0357184410095215, "learning_rate": 4.450549174513521e-06, "loss": 0.1554143726825714, "step": 1931 }, { "epoch": 0.5015576323987538, "grad_norm": 2.7715845108032227, "learning_rate": 4.449929574360597e-06, "loss": 0.17486847937107086, "step": 1932 }, { "epoch": 0.5018172377985463, "grad_norm": 2.816990852355957, "learning_rate": 4.4493096736990476e-06, "loss": 0.11129427701234818, "step": 1933 }, { "epoch": 0.5020768431983386, "grad_norm": 2.147299289703369, "learning_rate": 4.448689472638457e-06, "loss": 0.11990490555763245, "step": 1934 }, { "epoch": 0.5023364485981309, "grad_norm": 1.2219841480255127, "learning_rate": 4.448068971288462e-06, "loss": 0.12832990288734436, "step": 1935 }, { "epoch": 0.5025960539979232, "grad_norm": 2.7420129776000977, "learning_rate": 4.447448169758754e-06, "loss": 0.11303611099720001, "step": 1936 }, { "epoch": 0.5028556593977155, "grad_norm": 1.7368181943893433, "learning_rate": 4.446827068159076e-06, "loss": 0.11821766197681427, "step": 1937 }, { "epoch": 0.5031152647975078, "grad_norm": 2.1837103366851807, "learning_rate": 4.446205666599224e-06, "loss": 0.14808952808380127, "step": 1938 }, { "epoch": 0.5033748701973001, "grad_norm": 1.5232473611831665, "learning_rate": 4.445583965189049e-06, "loss": 0.13881638646125793, "step": 1939 }, { "epoch": 0.5036344755970924, "grad_norm": 1.0080803632736206, "learning_rate": 4.444961964038452e-06, "loss": 0.11653783172369003, "step": 1940 }, { "epoch": 0.5038940809968847, "grad_norm": 2.21138596534729, "learning_rate": 4.444339663257389e-06, "loss": 0.1312830001115799, "step": 1941 }, { "epoch": 0.504153686396677, "grad_norm": 2.1427361965179443, "learning_rate": 4.443717062955869e-06, "loss": 0.14355391263961792, "step": 1942 }, { "epoch": 0.5044132917964693, "grad_norm": 3.1383249759674072, "learning_rate": 4.443094163243954e-06, "loss": 0.11533109843730927, "step": 1943 }, { "epoch": 0.5046728971962616, "grad_norm": 3.5757033824920654, "learning_rate": 4.442470964231758e-06, "loss": 0.13788817822933197, "step": 1944 }, { "epoch": 0.504932502596054, "grad_norm": 1.1262321472167969, "learning_rate": 4.441847466029447e-06, "loss": 0.11506059020757675, "step": 1945 }, { "epoch": 0.5051921079958464, "grad_norm": 2.1199936866760254, "learning_rate": 4.441223668747241e-06, "loss": 0.12557296454906464, "step": 1946 }, { "epoch": 0.5054517133956387, "grad_norm": 0.7014800310134888, "learning_rate": 4.440599572495416e-06, "loss": 0.11949744820594788, "step": 1947 }, { "epoch": 0.505711318795431, "grad_norm": 0.7791844606399536, "learning_rate": 4.439975177384296e-06, "loss": 0.11472035944461823, "step": 1948 }, { "epoch": 0.5059709241952233, "grad_norm": 1.7042746543884277, "learning_rate": 4.439350483524259e-06, "loss": 0.11486458033323288, "step": 1949 }, { "epoch": 0.5062305295950156, "grad_norm": 1.529409408569336, "learning_rate": 4.4387254910257386e-06, "loss": 0.11097574234008789, "step": 1950 }, { "epoch": 0.5064901349948079, "grad_norm": 1.8089920282363892, "learning_rate": 4.438100199999217e-06, "loss": 0.10878445208072662, "step": 1951 }, { "epoch": 0.5067497403946002, "grad_norm": 1.5710757970809937, "learning_rate": 4.437474610555234e-06, "loss": 0.13982723653316498, "step": 1952 }, { "epoch": 0.5070093457943925, "grad_norm": 2.258061647415161, "learning_rate": 4.436848722804376e-06, "loss": 0.12547507882118225, "step": 1953 }, { "epoch": 0.5072689511941848, "grad_norm": 3.010901927947998, "learning_rate": 4.436222536857289e-06, "loss": 0.11954014003276825, "step": 1954 }, { "epoch": 0.5075285565939771, "grad_norm": 2.044177293777466, "learning_rate": 4.435596052824666e-06, "loss": 0.10939747840166092, "step": 1955 }, { "epoch": 0.5077881619937694, "grad_norm": 2.2968413829803467, "learning_rate": 4.434969270817257e-06, "loss": 0.12712454795837402, "step": 1956 }, { "epoch": 0.5080477673935618, "grad_norm": 2.0171823501586914, "learning_rate": 4.4343421909458605e-06, "loss": 0.1434367299079895, "step": 1957 }, { "epoch": 0.5083073727933541, "grad_norm": 3.052140235900879, "learning_rate": 4.433714813321331e-06, "loss": 0.1606782078742981, "step": 1958 }, { "epoch": 0.5085669781931464, "grad_norm": 2.520799160003662, "learning_rate": 4.433087138054575e-06, "loss": 0.10835136473178864, "step": 1959 }, { "epoch": 0.5088265835929388, "grad_norm": 1.8891929388046265, "learning_rate": 4.4324591652565505e-06, "loss": 0.1379459649324417, "step": 1960 }, { "epoch": 0.5090861889927311, "grad_norm": 2.6394922733306885, "learning_rate": 4.4318308950382685e-06, "loss": 0.13399936258792877, "step": 1961 }, { "epoch": 0.5093457943925234, "grad_norm": 0.9530012607574463, "learning_rate": 4.431202327510793e-06, "loss": 0.12035687267780304, "step": 1962 }, { "epoch": 0.5096053997923157, "grad_norm": 2.469759464263916, "learning_rate": 4.43057346278524e-06, "loss": 0.1272672414779663, "step": 1963 }, { "epoch": 0.509865005192108, "grad_norm": 1.7734384536743164, "learning_rate": 4.42994430097278e-06, "loss": 0.1443936824798584, "step": 1964 }, { "epoch": 0.5101246105919003, "grad_norm": 1.871090054512024, "learning_rate": 4.429314842184632e-06, "loss": 0.09173734486103058, "step": 1965 }, { "epoch": 0.5103842159916926, "grad_norm": 2.382906913757324, "learning_rate": 4.428685086532072e-06, "loss": 0.14291918277740479, "step": 1966 }, { "epoch": 0.5106438213914849, "grad_norm": 2.237011671066284, "learning_rate": 4.428055034126426e-06, "loss": 0.13037864863872528, "step": 1967 }, { "epoch": 0.5109034267912772, "grad_norm": 1.7603659629821777, "learning_rate": 4.427424685079071e-06, "loss": 0.12805113196372986, "step": 1968 }, { "epoch": 0.5111630321910696, "grad_norm": 1.740740418434143, "learning_rate": 4.42679403950144e-06, "loss": 0.09633849561214447, "step": 1969 }, { "epoch": 0.5114226375908619, "grad_norm": 2.1428520679473877, "learning_rate": 4.426163097505017e-06, "loss": 0.1307034194469452, "step": 1970 }, { "epoch": 0.5116822429906542, "grad_norm": 1.7647929191589355, "learning_rate": 4.425531859201337e-06, "loss": 0.13376913964748383, "step": 1971 }, { "epoch": 0.5119418483904465, "grad_norm": 1.5858731269836426, "learning_rate": 4.424900324701989e-06, "loss": 0.11125736683607101, "step": 1972 }, { "epoch": 0.5122014537902388, "grad_norm": 1.3216211795806885, "learning_rate": 4.424268494118613e-06, "loss": 0.1225164532661438, "step": 1973 }, { "epoch": 0.5124610591900312, "grad_norm": 1.8562582731246948, "learning_rate": 4.423636367562903e-06, "loss": 0.1232914999127388, "step": 1974 }, { "epoch": 0.5127206645898235, "grad_norm": 1.781930685043335, "learning_rate": 4.423003945146605e-06, "loss": 0.11063579469919205, "step": 1975 }, { "epoch": 0.5129802699896158, "grad_norm": 1.8141449689865112, "learning_rate": 4.422371226981517e-06, "loss": 0.10518123209476471, "step": 1976 }, { "epoch": 0.5132398753894081, "grad_norm": 2.7011165618896484, "learning_rate": 4.421738213179487e-06, "loss": 0.13707835972309113, "step": 1977 }, { "epoch": 0.5134994807892004, "grad_norm": 1.7765963077545166, "learning_rate": 4.421104903852419e-06, "loss": 0.13731467723846436, "step": 1978 }, { "epoch": 0.5137590861889927, "grad_norm": 1.482124924659729, "learning_rate": 4.420471299112267e-06, "loss": 0.1191546767950058, "step": 1979 }, { "epoch": 0.514018691588785, "grad_norm": 1.9222257137298584, "learning_rate": 4.419837399071038e-06, "loss": 0.10738889873027802, "step": 1980 }, { "epoch": 0.5142782969885774, "grad_norm": 1.3221479654312134, "learning_rate": 4.419203203840792e-06, "loss": 0.11712304502725601, "step": 1981 }, { "epoch": 0.5145379023883697, "grad_norm": 1.0940661430358887, "learning_rate": 4.418568713533638e-06, "loss": 0.1150471419095993, "step": 1982 }, { "epoch": 0.514797507788162, "grad_norm": 1.8003743886947632, "learning_rate": 4.4179339282617416e-06, "loss": 0.14471985399723053, "step": 1983 }, { "epoch": 0.5150571131879543, "grad_norm": 2.172306537628174, "learning_rate": 4.417298848137317e-06, "loss": 0.09311144053936005, "step": 1984 }, { "epoch": 0.5153167185877466, "grad_norm": 2.7082552909851074, "learning_rate": 4.416663473272632e-06, "loss": 0.13607631623744965, "step": 1985 }, { "epoch": 0.5155763239875389, "grad_norm": 2.55436372756958, "learning_rate": 4.416027803780006e-06, "loss": 0.13135886192321777, "step": 1986 }, { "epoch": 0.5158359293873312, "grad_norm": 2.1390280723571777, "learning_rate": 4.415391839771812e-06, "loss": 0.1307382583618164, "step": 1987 }, { "epoch": 0.5160955347871236, "grad_norm": 0.896160900592804, "learning_rate": 4.414755581360473e-06, "loss": 0.1226753368973732, "step": 1988 }, { "epoch": 0.5163551401869159, "grad_norm": 3.122188091278076, "learning_rate": 4.414119028658465e-06, "loss": 0.12794815003871918, "step": 1989 }, { "epoch": 0.5166147455867082, "grad_norm": 2.076775550842285, "learning_rate": 4.4134821817783155e-06, "loss": 0.11526815593242645, "step": 1990 }, { "epoch": 0.5168743509865005, "grad_norm": 2.6055855751037598, "learning_rate": 4.4128450408326065e-06, "loss": 0.1350288689136505, "step": 1991 }, { "epoch": 0.5171339563862928, "grad_norm": 2.4762368202209473, "learning_rate": 4.412207605933967e-06, "loss": 0.11765296757221222, "step": 1992 }, { "epoch": 0.5173935617860852, "grad_norm": 2.35624361038208, "learning_rate": 4.411569877195082e-06, "loss": 0.12954331934452057, "step": 1993 }, { "epoch": 0.5176531671858775, "grad_norm": 3.375237464904785, "learning_rate": 4.410931854728689e-06, "loss": 0.12527023255825043, "step": 1994 }, { "epoch": 0.5179127725856698, "grad_norm": 4.435391426086426, "learning_rate": 4.410293538647573e-06, "loss": 0.13530436158180237, "step": 1995 }, { "epoch": 0.5181723779854621, "grad_norm": 2.552717924118042, "learning_rate": 4.409654929064575e-06, "loss": 0.1179015263915062, "step": 1996 }, { "epoch": 0.5184319833852544, "grad_norm": 1.2941659688949585, "learning_rate": 4.4090160260925874e-06, "loss": 0.11554919183254242, "step": 1997 }, { "epoch": 0.5186915887850467, "grad_norm": 4.227990627288818, "learning_rate": 4.408376829844553e-06, "loss": 0.13423903286457062, "step": 1998 }, { "epoch": 0.518951194184839, "grad_norm": 6.037315845489502, "learning_rate": 4.407737340433466e-06, "loss": 0.1465374380350113, "step": 1999 }, { "epoch": 0.5192107995846313, "grad_norm": 1.7316973209381104, "learning_rate": 4.407097557972375e-06, "loss": 0.12840908765792847, "step": 2000 }, { "epoch": 0.5194704049844237, "grad_norm": 1.854560136795044, "learning_rate": 4.406457482574379e-06, "loss": 0.12981760501861572, "step": 2001 }, { "epoch": 0.519730010384216, "grad_norm": 2.938807487487793, "learning_rate": 4.405817114352627e-06, "loss": 0.16080373525619507, "step": 2002 }, { "epoch": 0.5199896157840083, "grad_norm": 3.4121806621551514, "learning_rate": 4.405176453420322e-06, "loss": 0.13336443901062012, "step": 2003 }, { "epoch": 0.5202492211838006, "grad_norm": 3.703376054763794, "learning_rate": 4.4045354998907196e-06, "loss": 0.13430200517177582, "step": 2004 }, { "epoch": 0.520508826583593, "grad_norm": 1.2569366693496704, "learning_rate": 4.4038942538771235e-06, "loss": 0.10145886242389679, "step": 2005 }, { "epoch": 0.5207684319833853, "grad_norm": 2.234372854232788, "learning_rate": 4.403252715492893e-06, "loss": 0.11406318843364716, "step": 2006 }, { "epoch": 0.5210280373831776, "grad_norm": 0.8912186622619629, "learning_rate": 4.402610884851437e-06, "loss": 0.10464617609977722, "step": 2007 }, { "epoch": 0.5212876427829699, "grad_norm": 1.7320895195007324, "learning_rate": 4.401968762066216e-06, "loss": 0.1130923181772232, "step": 2008 }, { "epoch": 0.5215472481827622, "grad_norm": 0.9011538624763489, "learning_rate": 4.4013263472507434e-06, "loss": 0.11032798886299133, "step": 2009 }, { "epoch": 0.5218068535825545, "grad_norm": 1.252747893333435, "learning_rate": 4.400683640518583e-06, "loss": 0.1201297864317894, "step": 2010 }, { "epoch": 0.5220664589823468, "grad_norm": 2.4780335426330566, "learning_rate": 4.40004064198335e-06, "loss": 0.1099986582994461, "step": 2011 }, { "epoch": 0.5223260643821391, "grad_norm": 1.3060675859451294, "learning_rate": 4.399397351758714e-06, "loss": 0.14261916279792786, "step": 2012 }, { "epoch": 0.5225856697819314, "grad_norm": 2.6087863445281982, "learning_rate": 4.39875376995839e-06, "loss": 0.12939313054084778, "step": 2013 }, { "epoch": 0.5228452751817237, "grad_norm": 0.923251211643219, "learning_rate": 4.3981098966961524e-06, "loss": 0.11858774721622467, "step": 2014 }, { "epoch": 0.523104880581516, "grad_norm": 1.5857617855072021, "learning_rate": 4.397465732085822e-06, "loss": 0.14296379685401917, "step": 2015 }, { "epoch": 0.5233644859813084, "grad_norm": 2.963228225708008, "learning_rate": 4.396821276241272e-06, "loss": 0.12140292674303055, "step": 2016 }, { "epoch": 0.5236240913811008, "grad_norm": 2.6673927307128906, "learning_rate": 4.396176529276426e-06, "loss": 0.13788719475269318, "step": 2017 }, { "epoch": 0.5238836967808931, "grad_norm": 3.0063421726226807, "learning_rate": 4.3955314913052636e-06, "loss": 0.1229124516248703, "step": 2018 }, { "epoch": 0.5241433021806854, "grad_norm": 2.716844081878662, "learning_rate": 4.394886162441811e-06, "loss": 0.10648428648710251, "step": 2019 }, { "epoch": 0.5244029075804777, "grad_norm": 1.9179636240005493, "learning_rate": 4.394240542800148e-06, "loss": 0.12284885346889496, "step": 2020 }, { "epoch": 0.52466251298027, "grad_norm": 1.2216860055923462, "learning_rate": 4.393594632494404e-06, "loss": 0.11456058919429779, "step": 2021 }, { "epoch": 0.5249221183800623, "grad_norm": 1.8390909433364868, "learning_rate": 4.3929484316387636e-06, "loss": 0.12740540504455566, "step": 2022 }, { "epoch": 0.5251817237798546, "grad_norm": 2.586827278137207, "learning_rate": 4.392301940347458e-06, "loss": 0.11237775534391403, "step": 2023 }, { "epoch": 0.5254413291796469, "grad_norm": 2.311187267303467, "learning_rate": 4.391655158734774e-06, "loss": 0.14509980380535126, "step": 2024 }, { "epoch": 0.5257009345794392, "grad_norm": 2.522839307785034, "learning_rate": 4.391008086915046e-06, "loss": 0.13751477003097534, "step": 2025 }, { "epoch": 0.5259605399792315, "grad_norm": 2.786626100540161, "learning_rate": 4.390360725002663e-06, "loss": 0.11725039780139923, "step": 2026 }, { "epoch": 0.5262201453790238, "grad_norm": 2.1277992725372314, "learning_rate": 4.389713073112063e-06, "loss": 0.12178555876016617, "step": 2027 }, { "epoch": 0.5264797507788161, "grad_norm": 2.5246241092681885, "learning_rate": 4.389065131357736e-06, "loss": 0.1378745138645172, "step": 2028 }, { "epoch": 0.5267393561786086, "grad_norm": 0.7892370820045471, "learning_rate": 4.388416899854224e-06, "loss": 0.12761279940605164, "step": 2029 }, { "epoch": 0.5269989615784009, "grad_norm": 3.154615879058838, "learning_rate": 4.387768378716117e-06, "loss": 0.13568700850009918, "step": 2030 }, { "epoch": 0.5272585669781932, "grad_norm": 2.569096088409424, "learning_rate": 4.387119568058063e-06, "loss": 0.12676620483398438, "step": 2031 }, { "epoch": 0.5275181723779855, "grad_norm": 2.84428071975708, "learning_rate": 4.386470467994753e-06, "loss": 0.09830192476511002, "step": 2032 }, { "epoch": 0.5277777777777778, "grad_norm": 2.1550843715667725, "learning_rate": 4.385821078640935e-06, "loss": 0.13930505514144897, "step": 2033 }, { "epoch": 0.5280373831775701, "grad_norm": 1.0575652122497559, "learning_rate": 4.385171400111405e-06, "loss": 0.1097402274608612, "step": 2034 }, { "epoch": 0.5282969885773624, "grad_norm": 1.2621275186538696, "learning_rate": 4.3845214325210124e-06, "loss": 0.12112028151750565, "step": 2035 }, { "epoch": 0.5285565939771547, "grad_norm": 2.0469377040863037, "learning_rate": 4.383871175984656e-06, "loss": 0.11580029129981995, "step": 2036 }, { "epoch": 0.528816199376947, "grad_norm": 2.0660300254821777, "learning_rate": 4.383220630617287e-06, "loss": 0.13090553879737854, "step": 2037 }, { "epoch": 0.5290758047767393, "grad_norm": 1.7514721155166626, "learning_rate": 4.3825697965339055e-06, "loss": 0.12423159182071686, "step": 2038 }, { "epoch": 0.5293354101765316, "grad_norm": 2.1482203006744385, "learning_rate": 4.3819186738495645e-06, "loss": 0.12890899181365967, "step": 2039 }, { "epoch": 0.5295950155763239, "grad_norm": 1.1787562370300293, "learning_rate": 4.381267262679369e-06, "loss": 0.12147702276706696, "step": 2040 }, { "epoch": 0.5298546209761164, "grad_norm": 1.7558724880218506, "learning_rate": 4.380615563138472e-06, "loss": 0.1308945268392563, "step": 2041 }, { "epoch": 0.5301142263759087, "grad_norm": 4.501769542694092, "learning_rate": 4.3799635753420794e-06, "loss": 0.13367673754692078, "step": 2042 }, { "epoch": 0.530373831775701, "grad_norm": 1.3863239288330078, "learning_rate": 4.379311299405449e-06, "loss": 0.11402938514947891, "step": 2043 }, { "epoch": 0.5306334371754933, "grad_norm": 1.4369868040084839, "learning_rate": 4.378658735443886e-06, "loss": 0.11966346204280853, "step": 2044 }, { "epoch": 0.5308930425752856, "grad_norm": 2.3310484886169434, "learning_rate": 4.37800588357275e-06, "loss": 0.11253201961517334, "step": 2045 }, { "epoch": 0.5311526479750779, "grad_norm": 3.7285380363464355, "learning_rate": 4.37735274390745e-06, "loss": 0.13668575882911682, "step": 2046 }, { "epoch": 0.5314122533748702, "grad_norm": 3.967851161956787, "learning_rate": 4.376699316563447e-06, "loss": 0.12113263458013535, "step": 2047 }, { "epoch": 0.5316718587746625, "grad_norm": 2.144009828567505, "learning_rate": 4.3760456016562505e-06, "loss": 0.10719842463731766, "step": 2048 }, { "epoch": 0.5319314641744548, "grad_norm": 1.2421785593032837, "learning_rate": 4.375391599301424e-06, "loss": 0.108690544962883, "step": 2049 }, { "epoch": 0.5321910695742471, "grad_norm": 3.8489036560058594, "learning_rate": 4.374737309614579e-06, "loss": 0.13155698776245117, "step": 2050 }, { "epoch": 0.5324506749740394, "grad_norm": 4.703084468841553, "learning_rate": 4.374082732711379e-06, "loss": 0.12924854457378387, "step": 2051 }, { "epoch": 0.5327102803738317, "grad_norm": 1.3749008178710938, "learning_rate": 4.373427868707539e-06, "loss": 0.10274747014045715, "step": 2052 }, { "epoch": 0.5329698857736241, "grad_norm": 3.348554849624634, "learning_rate": 4.3727727177188226e-06, "loss": 0.12195485830307007, "step": 2053 }, { "epoch": 0.5332294911734164, "grad_norm": 2.617170572280884, "learning_rate": 4.372117279861047e-06, "loss": 0.10578452795743942, "step": 2054 }, { "epoch": 0.5334890965732088, "grad_norm": 2.7889182567596436, "learning_rate": 4.371461555250077e-06, "loss": 0.11448205262422562, "step": 2055 }, { "epoch": 0.5337487019730011, "grad_norm": 4.151893615722656, "learning_rate": 4.370805544001831e-06, "loss": 0.12974178791046143, "step": 2056 }, { "epoch": 0.5340083073727934, "grad_norm": 4.399490833282471, "learning_rate": 4.370149246232277e-06, "loss": 0.12768813967704773, "step": 2057 }, { "epoch": 0.5342679127725857, "grad_norm": 2.5324907302856445, "learning_rate": 4.3694926620574315e-06, "loss": 0.13365080952644348, "step": 2058 }, { "epoch": 0.534527518172378, "grad_norm": 3.7318274974823, "learning_rate": 4.3688357915933656e-06, "loss": 0.1370311677455902, "step": 2059 }, { "epoch": 0.5347871235721703, "grad_norm": 2.096196413040161, "learning_rate": 4.368178634956199e-06, "loss": 0.14124169945716858, "step": 2060 }, { "epoch": 0.5350467289719626, "grad_norm": 2.891901969909668, "learning_rate": 4.3675211922621e-06, "loss": 0.11941219866275787, "step": 2061 }, { "epoch": 0.5353063343717549, "grad_norm": 1.8914992809295654, "learning_rate": 4.36686346362729e-06, "loss": 0.10752731561660767, "step": 2062 }, { "epoch": 0.5355659397715472, "grad_norm": 2.8582468032836914, "learning_rate": 4.366205449168042e-06, "loss": 0.11159883439540863, "step": 2063 }, { "epoch": 0.5358255451713395, "grad_norm": 1.2420744895935059, "learning_rate": 4.3655471490006766e-06, "loss": 0.10649193823337555, "step": 2064 }, { "epoch": 0.5360851505711319, "grad_norm": 0.6561457514762878, "learning_rate": 4.364888563241566e-06, "loss": 0.11421777307987213, "step": 2065 }, { "epoch": 0.5363447559709242, "grad_norm": 0.7036945819854736, "learning_rate": 4.364229692007134e-06, "loss": 0.1302507370710373, "step": 2066 }, { "epoch": 0.5366043613707165, "grad_norm": 1.8413666486740112, "learning_rate": 4.363570535413852e-06, "loss": 0.10922195762395859, "step": 2067 }, { "epoch": 0.5368639667705088, "grad_norm": 3.784742832183838, "learning_rate": 4.362911093578246e-06, "loss": 0.12804514169692993, "step": 2068 }, { "epoch": 0.5371235721703012, "grad_norm": 3.5305991172790527, "learning_rate": 4.362251366616891e-06, "loss": 0.1631297767162323, "step": 2069 }, { "epoch": 0.5373831775700935, "grad_norm": 2.4734113216400146, "learning_rate": 4.361591354646409e-06, "loss": 0.13001318275928497, "step": 2070 }, { "epoch": 0.5376427829698858, "grad_norm": 3.7674312591552734, "learning_rate": 4.360931057783475e-06, "loss": 0.13384321331977844, "step": 2071 }, { "epoch": 0.5379023883696781, "grad_norm": 4.660126209259033, "learning_rate": 4.360270476144817e-06, "loss": 0.1241922676563263, "step": 2072 }, { "epoch": 0.5381619937694704, "grad_norm": 1.034399390220642, "learning_rate": 4.359609609847208e-06, "loss": 0.11901848018169403, "step": 2073 }, { "epoch": 0.5384215991692627, "grad_norm": 1.1042118072509766, "learning_rate": 4.358948459007475e-06, "loss": 0.11260054260492325, "step": 2074 }, { "epoch": 0.538681204569055, "grad_norm": 2.808044195175171, "learning_rate": 4.358287023742495e-06, "loss": 0.09756219387054443, "step": 2075 }, { "epoch": 0.5389408099688473, "grad_norm": 1.4872417449951172, "learning_rate": 4.357625304169194e-06, "loss": 0.1266084909439087, "step": 2076 }, { "epoch": 0.5392004153686397, "grad_norm": 2.0710337162017822, "learning_rate": 4.356963300404549e-06, "loss": 0.10715897381305695, "step": 2077 }, { "epoch": 0.539460020768432, "grad_norm": 1.4954478740692139, "learning_rate": 4.356301012565588e-06, "loss": 0.12809789180755615, "step": 2078 }, { "epoch": 0.5397196261682243, "grad_norm": 2.3204147815704346, "learning_rate": 4.355638440769385e-06, "loss": 0.1491343379020691, "step": 2079 }, { "epoch": 0.5399792315680166, "grad_norm": 1.8930323123931885, "learning_rate": 4.354975585133072e-06, "loss": 0.11800997704267502, "step": 2080 }, { "epoch": 0.5402388369678089, "grad_norm": 1.7854437828063965, "learning_rate": 4.354312445773825e-06, "loss": 0.14169636368751526, "step": 2081 }, { "epoch": 0.5404984423676013, "grad_norm": 2.7621541023254395, "learning_rate": 4.353649022808869e-06, "loss": 0.1373385190963745, "step": 2082 }, { "epoch": 0.5407580477673936, "grad_norm": 3.005800485610962, "learning_rate": 4.352985316355487e-06, "loss": 0.14274069666862488, "step": 2083 }, { "epoch": 0.5410176531671859, "grad_norm": 2.6733767986297607, "learning_rate": 4.3523213265310036e-06, "loss": 0.10640892386436462, "step": 2084 }, { "epoch": 0.5412772585669782, "grad_norm": 0.8661907911300659, "learning_rate": 4.351657053452798e-06, "loss": 0.13440477848052979, "step": 2085 }, { "epoch": 0.5415368639667705, "grad_norm": 3.382884979248047, "learning_rate": 4.350992497238298e-06, "loss": 0.1298668384552002, "step": 2086 }, { "epoch": 0.5417964693665628, "grad_norm": 2.13779878616333, "learning_rate": 4.350327658004983e-06, "loss": 0.12927764654159546, "step": 2087 }, { "epoch": 0.5420560747663551, "grad_norm": 2.7751612663269043, "learning_rate": 4.349662535870381e-06, "loss": 0.12739020586013794, "step": 2088 }, { "epoch": 0.5423156801661475, "grad_norm": 4.383551597595215, "learning_rate": 4.348997130952069e-06, "loss": 0.13656356930732727, "step": 2089 }, { "epoch": 0.5425752855659398, "grad_norm": 2.118856906890869, "learning_rate": 4.348331443367677e-06, "loss": 0.1385429948568344, "step": 2090 }, { "epoch": 0.5428348909657321, "grad_norm": 1.2350395917892456, "learning_rate": 4.3476654732348826e-06, "loss": 0.10839395970106125, "step": 2091 }, { "epoch": 0.5430944963655244, "grad_norm": 3.091099739074707, "learning_rate": 4.346999220671414e-06, "loss": 0.14865411818027496, "step": 2092 }, { "epoch": 0.5433541017653167, "grad_norm": 2.320262908935547, "learning_rate": 4.34633268579505e-06, "loss": 0.11238642036914825, "step": 2093 }, { "epoch": 0.543613707165109, "grad_norm": 1.9771019220352173, "learning_rate": 4.345665868723618e-06, "loss": 0.13177867233753204, "step": 2094 }, { "epoch": 0.5438733125649013, "grad_norm": 3.7930452823638916, "learning_rate": 4.3449987695749964e-06, "loss": 0.10819742828607559, "step": 2095 }, { "epoch": 0.5441329179646937, "grad_norm": 2.1141343116760254, "learning_rate": 4.344331388467113e-06, "loss": 0.14215229451656342, "step": 2096 }, { "epoch": 0.544392523364486, "grad_norm": 1.38864266872406, "learning_rate": 4.343663725517946e-06, "loss": 0.11482000350952148, "step": 2097 }, { "epoch": 0.5446521287642783, "grad_norm": 2.8833425045013428, "learning_rate": 4.342995780845522e-06, "loss": 0.16835349798202515, "step": 2098 }, { "epoch": 0.5449117341640706, "grad_norm": 3.9514379501342773, "learning_rate": 4.342327554567918e-06, "loss": 0.13029497861862183, "step": 2099 }, { "epoch": 0.5451713395638629, "grad_norm": 1.9632902145385742, "learning_rate": 4.341659046803263e-06, "loss": 0.12261711061000824, "step": 2100 }, { "epoch": 0.5454309449636553, "grad_norm": 1.4149627685546875, "learning_rate": 4.340990257669732e-06, "loss": 0.127326101064682, "step": 2101 }, { "epoch": 0.5456905503634476, "grad_norm": 2.017209768295288, "learning_rate": 4.340321187285553e-06, "loss": 0.12983286380767822, "step": 2102 }, { "epoch": 0.5459501557632399, "grad_norm": 3.6134674549102783, "learning_rate": 4.339651835769e-06, "loss": 0.1348467469215393, "step": 2103 }, { "epoch": 0.5462097611630322, "grad_norm": 1.4373763799667358, "learning_rate": 4.338982203238403e-06, "loss": 0.11949199438095093, "step": 2104 }, { "epoch": 0.5464693665628245, "grad_norm": 1.7733126878738403, "learning_rate": 4.338312289812134e-06, "loss": 0.13232815265655518, "step": 2105 }, { "epoch": 0.5467289719626168, "grad_norm": 3.35530424118042, "learning_rate": 4.33764209560862e-06, "loss": 0.14628398418426514, "step": 2106 }, { "epoch": 0.5469885773624091, "grad_norm": 1.2984098196029663, "learning_rate": 4.336971620746334e-06, "loss": 0.11212940514087677, "step": 2107 }, { "epoch": 0.5472481827622014, "grad_norm": 1.6937847137451172, "learning_rate": 4.336300865343805e-06, "loss": 0.13971492648124695, "step": 2108 }, { "epoch": 0.5475077881619937, "grad_norm": 1.9679608345031738, "learning_rate": 4.335629829519602e-06, "loss": 0.12790578603744507, "step": 2109 }, { "epoch": 0.547767393561786, "grad_norm": 1.4468743801116943, "learning_rate": 4.334958513392353e-06, "loss": 0.11880370229482651, "step": 2110 }, { "epoch": 0.5480269989615784, "grad_norm": 0.989520788192749, "learning_rate": 4.33428691708073e-06, "loss": 0.11742885410785675, "step": 2111 }, { "epoch": 0.5482866043613707, "grad_norm": 1.917098879814148, "learning_rate": 4.333615040703454e-06, "loss": 0.10485367476940155, "step": 2112 }, { "epoch": 0.5485462097611631, "grad_norm": 2.83607816696167, "learning_rate": 4.332942884379299e-06, "loss": 0.11933189630508423, "step": 2113 }, { "epoch": 0.5488058151609554, "grad_norm": 1.939396619796753, "learning_rate": 4.332270448227087e-06, "loss": 0.10933391749858856, "step": 2114 }, { "epoch": 0.5490654205607477, "grad_norm": 0.9068745970726013, "learning_rate": 4.331597732365689e-06, "loss": 0.11379389464855194, "step": 2115 }, { "epoch": 0.54932502596054, "grad_norm": 2.4952597618103027, "learning_rate": 4.330924736914026e-06, "loss": 0.12805037200450897, "step": 2116 }, { "epoch": 0.5495846313603323, "grad_norm": 1.429154634475708, "learning_rate": 4.3302514619910665e-06, "loss": 0.11638383567333221, "step": 2117 }, { "epoch": 0.5498442367601246, "grad_norm": 2.752803087234497, "learning_rate": 4.329577907715833e-06, "loss": 0.1364152431488037, "step": 2118 }, { "epoch": 0.5501038421599169, "grad_norm": 2.7975101470947266, "learning_rate": 4.328904074207392e-06, "loss": 0.12187162041664124, "step": 2119 }, { "epoch": 0.5503634475597092, "grad_norm": 6.622096061706543, "learning_rate": 4.328229961584864e-06, "loss": 0.1638190746307373, "step": 2120 }, { "epoch": 0.5506230529595015, "grad_norm": 1.6123067140579224, "learning_rate": 4.327555569967414e-06, "loss": 0.12634778022766113, "step": 2121 }, { "epoch": 0.5508826583592938, "grad_norm": 1.5435256958007812, "learning_rate": 4.326880899474261e-06, "loss": 0.13004334270954132, "step": 2122 }, { "epoch": 0.5511422637590861, "grad_norm": 2.2713680267333984, "learning_rate": 4.32620595022467e-06, "loss": 0.1227455884218216, "step": 2123 }, { "epoch": 0.5514018691588785, "grad_norm": 4.47537899017334, "learning_rate": 4.325530722337959e-06, "loss": 0.144060879945755, "step": 2124 }, { "epoch": 0.5516614745586709, "grad_norm": 2.8858590126037598, "learning_rate": 4.32485521593349e-06, "loss": 0.12076052278280258, "step": 2125 }, { "epoch": 0.5519210799584632, "grad_norm": 1.1193526983261108, "learning_rate": 4.324179431130679e-06, "loss": 0.11494635045528412, "step": 2126 }, { "epoch": 0.5521806853582555, "grad_norm": 1.6349329948425293, "learning_rate": 4.323503368048987e-06, "loss": 0.1155320256948471, "step": 2127 }, { "epoch": 0.5524402907580478, "grad_norm": 1.4362112283706665, "learning_rate": 4.32282702680793e-06, "loss": 0.09523142129182816, "step": 2128 }, { "epoch": 0.5526998961578401, "grad_norm": 3.8995649814605713, "learning_rate": 4.322150407527067e-06, "loss": 0.12975825369358063, "step": 2129 }, { "epoch": 0.5529595015576324, "grad_norm": 4.156457424163818, "learning_rate": 4.321473510326009e-06, "loss": 0.12531018257141113, "step": 2130 }, { "epoch": 0.5532191069574247, "grad_norm": 2.6728711128234863, "learning_rate": 4.320796335324416e-06, "loss": 0.13313569128513336, "step": 2131 }, { "epoch": 0.553478712357217, "grad_norm": 1.3468784093856812, "learning_rate": 4.3201188826419985e-06, "loss": 0.1234951764345169, "step": 2132 }, { "epoch": 0.5537383177570093, "grad_norm": 1.2520360946655273, "learning_rate": 4.319441152398513e-06, "loss": 0.11340320110321045, "step": 2133 }, { "epoch": 0.5539979231568016, "grad_norm": 2.655923366546631, "learning_rate": 4.318763144713767e-06, "loss": 0.13454023003578186, "step": 2134 }, { "epoch": 0.5542575285565939, "grad_norm": 3.8653361797332764, "learning_rate": 4.318084859707617e-06, "loss": 0.12859714031219482, "step": 2135 }, { "epoch": 0.5545171339563862, "grad_norm": 1.4903894662857056, "learning_rate": 4.317406297499968e-06, "loss": 0.11845198273658752, "step": 2136 }, { "epoch": 0.5547767393561787, "grad_norm": 1.5234030485153198, "learning_rate": 4.316727458210774e-06, "loss": 0.11210314929485321, "step": 2137 }, { "epoch": 0.555036344755971, "grad_norm": 1.8951869010925293, "learning_rate": 4.31604834196004e-06, "loss": 0.11291594803333282, "step": 2138 }, { "epoch": 0.5552959501557633, "grad_norm": 2.3458166122436523, "learning_rate": 4.315368948867816e-06, "loss": 0.11166614294052124, "step": 2139 }, { "epoch": 0.5555555555555556, "grad_norm": 1.9185532331466675, "learning_rate": 4.3146892790542035e-06, "loss": 0.12326475977897644, "step": 2140 }, { "epoch": 0.5558151609553479, "grad_norm": 1.6358262300491333, "learning_rate": 4.314009332639355e-06, "loss": 0.1234857439994812, "step": 2141 }, { "epoch": 0.5560747663551402, "grad_norm": 1.4346277713775635, "learning_rate": 4.313329109743466e-06, "loss": 0.12519392371177673, "step": 2142 }, { "epoch": 0.5563343717549325, "grad_norm": 1.4741350412368774, "learning_rate": 4.312648610486786e-06, "loss": 0.12273754924535751, "step": 2143 }, { "epoch": 0.5565939771547248, "grad_norm": 2.257161855697632, "learning_rate": 4.311967834989612e-06, "loss": 0.13979530334472656, "step": 2144 }, { "epoch": 0.5568535825545171, "grad_norm": 1.697346568107605, "learning_rate": 4.311286783372289e-06, "loss": 0.13014495372772217, "step": 2145 }, { "epoch": 0.5571131879543094, "grad_norm": 1.1103930473327637, "learning_rate": 4.310605455755211e-06, "loss": 0.13168886303901672, "step": 2146 }, { "epoch": 0.5573727933541017, "grad_norm": 0.9980549812316895, "learning_rate": 4.3099238522588225e-06, "loss": 0.11820510029792786, "step": 2147 }, { "epoch": 0.557632398753894, "grad_norm": 1.3823119401931763, "learning_rate": 4.309241973003614e-06, "loss": 0.10716646909713745, "step": 2148 }, { "epoch": 0.5578920041536864, "grad_norm": 2.3531696796417236, "learning_rate": 4.3085598181101256e-06, "loss": 0.14623235166072845, "step": 2149 }, { "epoch": 0.5581516095534788, "grad_norm": 1.633825421333313, "learning_rate": 4.307877387698949e-06, "loss": 0.11733352392911911, "step": 2150 }, { "epoch": 0.5584112149532711, "grad_norm": 2.855511426925659, "learning_rate": 4.30719468189072e-06, "loss": 0.13357219099998474, "step": 2151 }, { "epoch": 0.5586708203530634, "grad_norm": 0.8708691596984863, "learning_rate": 4.306511700806127e-06, "loss": 0.10855606198310852, "step": 2152 }, { "epoch": 0.5589304257528557, "grad_norm": 2.0919299125671387, "learning_rate": 4.305828444565904e-06, "loss": 0.10199739784002304, "step": 2153 }, { "epoch": 0.559190031152648, "grad_norm": 1.3372478485107422, "learning_rate": 4.305144913290837e-06, "loss": 0.11746694147586823, "step": 2154 }, { "epoch": 0.5594496365524403, "grad_norm": 3.0707285404205322, "learning_rate": 4.304461107101756e-06, "loss": 0.15860377252101898, "step": 2155 }, { "epoch": 0.5597092419522326, "grad_norm": 2.6211204528808594, "learning_rate": 4.303777026119545e-06, "loss": 0.14127126336097717, "step": 2156 }, { "epoch": 0.5599688473520249, "grad_norm": 3.374558687210083, "learning_rate": 4.3030926704651314e-06, "loss": 0.13813725113868713, "step": 2157 }, { "epoch": 0.5602284527518172, "grad_norm": 0.9651496410369873, "learning_rate": 4.302408040259495e-06, "loss": 0.12327460199594498, "step": 2158 }, { "epoch": 0.5604880581516095, "grad_norm": 2.12308669090271, "learning_rate": 4.301723135623662e-06, "loss": 0.15020817518234253, "step": 2159 }, { "epoch": 0.5607476635514018, "grad_norm": 2.2843079566955566, "learning_rate": 4.30103795667871e-06, "loss": 0.07900737226009369, "step": 2160 }, { "epoch": 0.5610072689511942, "grad_norm": 3.197741746902466, "learning_rate": 4.300352503545761e-06, "loss": 0.12169955670833588, "step": 2161 }, { "epoch": 0.5612668743509865, "grad_norm": 3.659212827682495, "learning_rate": 4.299666776345987e-06, "loss": 0.14319851994514465, "step": 2162 }, { "epoch": 0.5615264797507789, "grad_norm": 1.3658626079559326, "learning_rate": 4.298980775200609e-06, "loss": 0.10691912472248077, "step": 2163 }, { "epoch": 0.5617860851505712, "grad_norm": 1.237369418144226, "learning_rate": 4.298294500230898e-06, "loss": 0.11639857292175293, "step": 2164 }, { "epoch": 0.5620456905503635, "grad_norm": 2.8609883785247803, "learning_rate": 4.29760795155817e-06, "loss": 0.13444390892982483, "step": 2165 }, { "epoch": 0.5623052959501558, "grad_norm": 3.020962715148926, "learning_rate": 4.296921129303792e-06, "loss": 0.12683890759944916, "step": 2166 }, { "epoch": 0.5625649013499481, "grad_norm": 1.5287212133407593, "learning_rate": 4.296234033589178e-06, "loss": 0.12296915054321289, "step": 2167 }, { "epoch": 0.5628245067497404, "grad_norm": 2.255314350128174, "learning_rate": 4.295546664535791e-06, "loss": 0.13913112878799438, "step": 2168 }, { "epoch": 0.5630841121495327, "grad_norm": 2.207930326461792, "learning_rate": 4.2948590222651424e-06, "loss": 0.11930154263973236, "step": 2169 }, { "epoch": 0.563343717549325, "grad_norm": 3.0515079498291016, "learning_rate": 4.294171106898791e-06, "loss": 0.1297917515039444, "step": 2170 }, { "epoch": 0.5636033229491173, "grad_norm": 1.8180814981460571, "learning_rate": 4.293482918558344e-06, "loss": 0.11300523579120636, "step": 2171 }, { "epoch": 0.5638629283489096, "grad_norm": 1.4614008665084839, "learning_rate": 4.2927944573654595e-06, "loss": 0.12885867059230804, "step": 2172 }, { "epoch": 0.564122533748702, "grad_norm": 1.3427081108093262, "learning_rate": 4.292105723441839e-06, "loss": 0.12142626196146011, "step": 2173 }, { "epoch": 0.5643821391484943, "grad_norm": 2.9374067783355713, "learning_rate": 4.291416716909237e-06, "loss": 0.12211351841688156, "step": 2174 }, { "epoch": 0.5646417445482866, "grad_norm": 1.0308665037155151, "learning_rate": 4.290727437889453e-06, "loss": 0.11771275103092194, "step": 2175 }, { "epoch": 0.564901349948079, "grad_norm": 1.0950430631637573, "learning_rate": 4.290037886504335e-06, "loss": 0.11026161909103394, "step": 2176 }, { "epoch": 0.5651609553478713, "grad_norm": 2.6204285621643066, "learning_rate": 4.289348062875782e-06, "loss": 0.12188730388879776, "step": 2177 }, { "epoch": 0.5654205607476636, "grad_norm": 1.6774868965148926, "learning_rate": 4.288657967125738e-06, "loss": 0.14366570115089417, "step": 2178 }, { "epoch": 0.5656801661474559, "grad_norm": 2.4073078632354736, "learning_rate": 4.2879675993761954e-06, "loss": 0.1425931453704834, "step": 2179 }, { "epoch": 0.5659397715472482, "grad_norm": 1.4544689655303955, "learning_rate": 4.287276959749196e-06, "loss": 0.10923303663730621, "step": 2180 }, { "epoch": 0.5661993769470405, "grad_norm": 1.2043136358261108, "learning_rate": 4.286586048366829e-06, "loss": 0.10238893330097198, "step": 2181 }, { "epoch": 0.5664589823468328, "grad_norm": 3.2969696521759033, "learning_rate": 4.285894865351233e-06, "loss": 0.10633926093578339, "step": 2182 }, { "epoch": 0.5667185877466251, "grad_norm": 1.9208431243896484, "learning_rate": 4.2852034108245895e-06, "loss": 0.1143711730837822, "step": 2183 }, { "epoch": 0.5669781931464174, "grad_norm": 1.4393250942230225, "learning_rate": 4.284511684909136e-06, "loss": 0.11301346123218536, "step": 2184 }, { "epoch": 0.5672377985462098, "grad_norm": 2.4894676208496094, "learning_rate": 4.2838196877271536e-06, "loss": 0.12578783929347992, "step": 2185 }, { "epoch": 0.5674974039460021, "grad_norm": 2.325291156768799, "learning_rate": 4.283127419400969e-06, "loss": 0.12507450580596924, "step": 2186 }, { "epoch": 0.5677570093457944, "grad_norm": 0.8934677243232727, "learning_rate": 4.28243488005296e-06, "loss": 0.11344558000564575, "step": 2187 }, { "epoch": 0.5680166147455867, "grad_norm": 3.8562538623809814, "learning_rate": 4.281742069805553e-06, "loss": 0.08667925000190735, "step": 2188 }, { "epoch": 0.568276220145379, "grad_norm": 2.99410343170166, "learning_rate": 4.281048988781221e-06, "loss": 0.13000695407390594, "step": 2189 }, { "epoch": 0.5685358255451713, "grad_norm": 2.9198098182678223, "learning_rate": 4.280355637102483e-06, "loss": 0.0764036551117897, "step": 2190 }, { "epoch": 0.5687954309449637, "grad_norm": 2.1498281955718994, "learning_rate": 4.279662014891909e-06, "loss": 0.11167917400598526, "step": 2191 }, { "epoch": 0.569055036344756, "grad_norm": 2.6819119453430176, "learning_rate": 4.2789681222721155e-06, "loss": 0.1394347846508026, "step": 2192 }, { "epoch": 0.5693146417445483, "grad_norm": 1.3877924680709839, "learning_rate": 4.278273959365768e-06, "loss": 0.1137663871049881, "step": 2193 }, { "epoch": 0.5695742471443406, "grad_norm": 2.731618642807007, "learning_rate": 4.277579526295575e-06, "loss": 0.10878664255142212, "step": 2194 }, { "epoch": 0.5698338525441329, "grad_norm": 2.867351531982422, "learning_rate": 4.2768848231843e-06, "loss": 0.1474630981683731, "step": 2195 }, { "epoch": 0.5700934579439252, "grad_norm": 2.9996864795684814, "learning_rate": 4.27618985015475e-06, "loss": 0.1377643644809723, "step": 2196 }, { "epoch": 0.5703530633437176, "grad_norm": 2.840428352355957, "learning_rate": 4.275494607329778e-06, "loss": 0.12502261996269226, "step": 2197 }, { "epoch": 0.5706126687435099, "grad_norm": 1.6733899116516113, "learning_rate": 4.27479909483229e-06, "loss": 0.10288174450397491, "step": 2198 }, { "epoch": 0.5708722741433022, "grad_norm": 3.3504300117492676, "learning_rate": 4.274103312785235e-06, "loss": 0.13188232481479645, "step": 2199 }, { "epoch": 0.5711318795430945, "grad_norm": 5.169712066650391, "learning_rate": 4.27340726131161e-06, "loss": 0.1347419172525406, "step": 2200 }, { "epoch": 0.5713914849428868, "grad_norm": 4.17756986618042, "learning_rate": 4.272710940534465e-06, "loss": 0.13448107242584229, "step": 2201 }, { "epoch": 0.5716510903426791, "grad_norm": 3.177748441696167, "learning_rate": 4.27201435057689e-06, "loss": 0.122717946767807, "step": 2202 }, { "epoch": 0.5719106957424714, "grad_norm": 0.8651571869850159, "learning_rate": 4.271317491562028e-06, "loss": 0.10863283276557922, "step": 2203 }, { "epoch": 0.5721703011422637, "grad_norm": 2.3702030181884766, "learning_rate": 4.270620363613066e-06, "loss": 0.11849527060985565, "step": 2204 }, { "epoch": 0.572429906542056, "grad_norm": 3.357276678085327, "learning_rate": 4.269922966853242e-06, "loss": 0.09374142438173294, "step": 2205 }, { "epoch": 0.5726895119418484, "grad_norm": 2.299835681915283, "learning_rate": 4.269225301405838e-06, "loss": 0.11981119215488434, "step": 2206 }, { "epoch": 0.5729491173416407, "grad_norm": 3.443274736404419, "learning_rate": 4.268527367394188e-06, "loss": 0.14671403169631958, "step": 2207 }, { "epoch": 0.573208722741433, "grad_norm": 3.392484426498413, "learning_rate": 4.267829164941669e-06, "loss": 0.11726045608520508, "step": 2208 }, { "epoch": 0.5734683281412254, "grad_norm": 1.3552230596542358, "learning_rate": 4.267130694171706e-06, "loss": 0.11538594961166382, "step": 2209 }, { "epoch": 0.5737279335410177, "grad_norm": 3.2078304290771484, "learning_rate": 4.266431955207774e-06, "loss": 0.1420484483242035, "step": 2210 }, { "epoch": 0.57398753894081, "grad_norm": 2.1697819232940674, "learning_rate": 4.265732948173395e-06, "loss": 0.12354709953069687, "step": 2211 }, { "epoch": 0.5742471443406023, "grad_norm": 2.703402280807495, "learning_rate": 4.265033673192136e-06, "loss": 0.12435738742351532, "step": 2212 }, { "epoch": 0.5745067497403946, "grad_norm": 2.266188144683838, "learning_rate": 4.264334130387614e-06, "loss": 0.12426462769508362, "step": 2213 }, { "epoch": 0.5747663551401869, "grad_norm": 3.6043877601623535, "learning_rate": 4.26363431988349e-06, "loss": 0.15253397822380066, "step": 2214 }, { "epoch": 0.5750259605399792, "grad_norm": 1.95622718334198, "learning_rate": 4.262934241803477e-06, "loss": 0.11189554631710052, "step": 2215 }, { "epoch": 0.5752855659397715, "grad_norm": 1.4457260370254517, "learning_rate": 4.2622338962713315e-06, "loss": 0.11617730557918549, "step": 2216 }, { "epoch": 0.5755451713395638, "grad_norm": 2.267808198928833, "learning_rate": 4.261533283410858e-06, "loss": 0.12755466997623444, "step": 2217 }, { "epoch": 0.5758047767393562, "grad_norm": 2.921065092086792, "learning_rate": 4.26083240334591e-06, "loss": 0.12394343316555023, "step": 2218 }, { "epoch": 0.5760643821391485, "grad_norm": 2.4210052490234375, "learning_rate": 4.260131256200387e-06, "loss": 0.12910938262939453, "step": 2219 }, { "epoch": 0.5763239875389408, "grad_norm": 3.146411418914795, "learning_rate": 4.259429842098235e-06, "loss": 0.12575724720954895, "step": 2220 }, { "epoch": 0.5765835929387332, "grad_norm": 1.6064035892486572, "learning_rate": 4.258728161163448e-06, "loss": 0.11328618973493576, "step": 2221 }, { "epoch": 0.5768431983385255, "grad_norm": 1.0409741401672363, "learning_rate": 4.258026213520067e-06, "loss": 0.12155638635158539, "step": 2222 }, { "epoch": 0.5771028037383178, "grad_norm": 1.5508233308792114, "learning_rate": 4.257323999292182e-06, "loss": 0.14239317178726196, "step": 2223 }, { "epoch": 0.5773624091381101, "grad_norm": 3.1871824264526367, "learning_rate": 4.256621518603926e-06, "loss": 0.11207517981529236, "step": 2224 }, { "epoch": 0.5776220145379024, "grad_norm": 2.975775718688965, "learning_rate": 4.255918771579482e-06, "loss": 0.1400042176246643, "step": 2225 }, { "epoch": 0.5778816199376947, "grad_norm": 2.211725950241089, "learning_rate": 4.255215758343082e-06, "loss": 0.11753467470407486, "step": 2226 }, { "epoch": 0.578141225337487, "grad_norm": 1.059409737586975, "learning_rate": 4.2545124790189995e-06, "loss": 0.11646062135696411, "step": 2227 }, { "epoch": 0.5784008307372793, "grad_norm": 1.361619472503662, "learning_rate": 4.2538089337315594e-06, "loss": 0.1297234445810318, "step": 2228 }, { "epoch": 0.5786604361370716, "grad_norm": 1.0642213821411133, "learning_rate": 4.253105122605134e-06, "loss": 0.11346648633480072, "step": 2229 }, { "epoch": 0.5789200415368639, "grad_norm": 2.4296586513519287, "learning_rate": 4.252401045764138e-06, "loss": 0.12607160210609436, "step": 2230 }, { "epoch": 0.5791796469366562, "grad_norm": 2.9506053924560547, "learning_rate": 4.251696703333037e-06, "loss": 0.12216562032699585, "step": 2231 }, { "epoch": 0.5794392523364486, "grad_norm": 0.8522006869316101, "learning_rate": 4.250992095436345e-06, "loss": 0.11770258098840714, "step": 2232 }, { "epoch": 0.579698857736241, "grad_norm": 1.0607692003250122, "learning_rate": 4.250287222198618e-06, "loss": 0.10498513281345367, "step": 2233 }, { "epoch": 0.5799584631360333, "grad_norm": 1.7082756757736206, "learning_rate": 4.249582083744461e-06, "loss": 0.11693581193685532, "step": 2234 }, { "epoch": 0.5802180685358256, "grad_norm": 4.534714698791504, "learning_rate": 4.248876680198529e-06, "loss": 0.13195160031318665, "step": 2235 }, { "epoch": 0.5804776739356179, "grad_norm": 1.308016061782837, "learning_rate": 4.248171011685519e-06, "loss": 0.11695890873670578, "step": 2236 }, { "epoch": 0.5807372793354102, "grad_norm": 2.1208269596099854, "learning_rate": 4.247465078330177e-06, "loss": 0.11030103266239166, "step": 2237 }, { "epoch": 0.5809968847352025, "grad_norm": 2.697556257247925, "learning_rate": 4.246758880257299e-06, "loss": 0.13701485097408295, "step": 2238 }, { "epoch": 0.5812564901349948, "grad_norm": 1.5041069984436035, "learning_rate": 4.24605241759172e-06, "loss": 0.1248963475227356, "step": 2239 }, { "epoch": 0.5815160955347871, "grad_norm": 1.2163828611373901, "learning_rate": 4.245345690458329e-06, "loss": 0.10255341976881027, "step": 2240 }, { "epoch": 0.5817757009345794, "grad_norm": 3.1718907356262207, "learning_rate": 4.244638698982059e-06, "loss": 0.10177428275346756, "step": 2241 }, { "epoch": 0.5820353063343717, "grad_norm": 2.7345938682556152, "learning_rate": 4.243931443287889e-06, "loss": 0.11321385204792023, "step": 2242 }, { "epoch": 0.582294911734164, "grad_norm": 1.1579710245132446, "learning_rate": 4.243223923500846e-06, "loss": 0.11686702817678452, "step": 2243 }, { "epoch": 0.5825545171339563, "grad_norm": 1.7346750497817993, "learning_rate": 4.2425161397460055e-06, "loss": 0.11356289684772491, "step": 2244 }, { "epoch": 0.5828141225337488, "grad_norm": 3.525043487548828, "learning_rate": 4.241808092148485e-06, "loss": 0.1339268535375595, "step": 2245 }, { "epoch": 0.5830737279335411, "grad_norm": 2.9872496128082275, "learning_rate": 4.241099780833451e-06, "loss": 0.10608460009098053, "step": 2246 }, { "epoch": 0.5833333333333334, "grad_norm": 0.8797601461410522, "learning_rate": 4.240391205926118e-06, "loss": 0.11561189591884613, "step": 2247 }, { "epoch": 0.5835929387331257, "grad_norm": 1.6043577194213867, "learning_rate": 4.239682367551745e-06, "loss": 0.12474356591701508, "step": 2248 }, { "epoch": 0.583852544132918, "grad_norm": 1.7437303066253662, "learning_rate": 4.238973265835639e-06, "loss": 0.11304054409265518, "step": 2249 }, { "epoch": 0.5841121495327103, "grad_norm": 1.3913551568984985, "learning_rate": 4.238263900903152e-06, "loss": 0.12407448142766953, "step": 2250 }, { "epoch": 0.5843717549325026, "grad_norm": 1.8681401014328003, "learning_rate": 4.2375542728796835e-06, "loss": 0.11831870675086975, "step": 2251 }, { "epoch": 0.5846313603322949, "grad_norm": 1.4162297248840332, "learning_rate": 4.236844381890681e-06, "loss": 0.10934621840715408, "step": 2252 }, { "epoch": 0.5848909657320872, "grad_norm": 0.8600590229034424, "learning_rate": 4.236134228061636e-06, "loss": 0.12987995147705078, "step": 2253 }, { "epoch": 0.5851505711318795, "grad_norm": 1.1710655689239502, "learning_rate": 4.235423811518087e-06, "loss": 0.10330867767333984, "step": 2254 }, { "epoch": 0.5854101765316718, "grad_norm": 3.1489923000335693, "learning_rate": 4.2347131323856195e-06, "loss": 0.13777324557304382, "step": 2255 }, { "epoch": 0.5856697819314641, "grad_norm": 2.009965181350708, "learning_rate": 4.234002190789867e-06, "loss": 0.11270011216402054, "step": 2256 }, { "epoch": 0.5859293873312564, "grad_norm": 2.0490670204162598, "learning_rate": 4.233290986856504e-06, "loss": 0.10597583651542664, "step": 2257 }, { "epoch": 0.5861889927310489, "grad_norm": 1.6791428327560425, "learning_rate": 4.23257952071126e-06, "loss": 0.11464180052280426, "step": 2258 }, { "epoch": 0.5864485981308412, "grad_norm": 4.225561141967773, "learning_rate": 4.231867792479902e-06, "loss": 0.10397031158208847, "step": 2259 }, { "epoch": 0.5867082035306335, "grad_norm": 1.7615004777908325, "learning_rate": 4.2311558022882485e-06, "loss": 0.12244241684675217, "step": 2260 }, { "epoch": 0.5869678089304258, "grad_norm": 3.1589231491088867, "learning_rate": 4.2304435502621635e-06, "loss": 0.11538337171077728, "step": 2261 }, { "epoch": 0.5872274143302181, "grad_norm": 2.22286057472229, "learning_rate": 4.229731036527556e-06, "loss": 0.11631645262241364, "step": 2262 }, { "epoch": 0.5874870197300104, "grad_norm": 1.633072853088379, "learning_rate": 4.229018261210383e-06, "loss": 0.1228964552283287, "step": 2263 }, { "epoch": 0.5877466251298027, "grad_norm": 2.681211233139038, "learning_rate": 4.2283052244366465e-06, "loss": 0.12480559945106506, "step": 2264 }, { "epoch": 0.588006230529595, "grad_norm": 2.6344692707061768, "learning_rate": 4.227591926332394e-06, "loss": 0.12976062297821045, "step": 2265 }, { "epoch": 0.5882658359293873, "grad_norm": 2.8916916847229004, "learning_rate": 4.226878367023724e-06, "loss": 0.11782172322273254, "step": 2266 }, { "epoch": 0.5885254413291796, "grad_norm": 3.936310052871704, "learning_rate": 4.226164546636772e-06, "loss": 0.12713642418384552, "step": 2267 }, { "epoch": 0.5887850467289719, "grad_norm": 2.7574009895324707, "learning_rate": 4.2254504652977285e-06, "loss": 0.12550535798072815, "step": 2268 }, { "epoch": 0.5890446521287642, "grad_norm": 3.495481491088867, "learning_rate": 4.224736123132826e-06, "loss": 0.09393130987882614, "step": 2269 }, { "epoch": 0.5893042575285566, "grad_norm": 3.750777006149292, "learning_rate": 4.224021520268345e-06, "loss": 0.1378643661737442, "step": 2270 }, { "epoch": 0.589563862928349, "grad_norm": 0.9809849858283997, "learning_rate": 4.2233066568306084e-06, "loss": 0.11916075646877289, "step": 2271 }, { "epoch": 0.5898234683281413, "grad_norm": 3.8551735877990723, "learning_rate": 4.22259153294599e-06, "loss": 0.1358720362186432, "step": 2272 }, { "epoch": 0.5900830737279336, "grad_norm": 1.484987497329712, "learning_rate": 4.221876148740906e-06, "loss": 0.11999393999576569, "step": 2273 }, { "epoch": 0.5903426791277259, "grad_norm": 1.811470627784729, "learning_rate": 4.221160504341821e-06, "loss": 0.12664394080638885, "step": 2274 }, { "epoch": 0.5906022845275182, "grad_norm": 1.6705725193023682, "learning_rate": 4.220444599875243e-06, "loss": 0.1087401881814003, "step": 2275 }, { "epoch": 0.5908618899273105, "grad_norm": 2.092379331588745, "learning_rate": 4.219728435467729e-06, "loss": 0.11489003896713257, "step": 2276 }, { "epoch": 0.5911214953271028, "grad_norm": 1.9150776863098145, "learning_rate": 4.21901201124588e-06, "loss": 0.13344964385032654, "step": 2277 }, { "epoch": 0.5913811007268951, "grad_norm": 3.1250340938568115, "learning_rate": 4.2182953273363445e-06, "loss": 0.1072918176651001, "step": 2278 }, { "epoch": 0.5916407061266874, "grad_norm": 3.7952752113342285, "learning_rate": 4.217578383865813e-06, "loss": 0.08687102794647217, "step": 2279 }, { "epoch": 0.5919003115264797, "grad_norm": 2.5307466983795166, "learning_rate": 4.216861180961026e-06, "loss": 0.1240190863609314, "step": 2280 }, { "epoch": 0.592159916926272, "grad_norm": 2.171010971069336, "learning_rate": 4.21614371874877e-06, "loss": 0.11636444181203842, "step": 2281 }, { "epoch": 0.5924195223260644, "grad_norm": 1.921120285987854, "learning_rate": 4.215425997355875e-06, "loss": 0.11280882358551025, "step": 2282 }, { "epoch": 0.5926791277258567, "grad_norm": 2.9581398963928223, "learning_rate": 4.214708016909216e-06, "loss": 0.0860479399561882, "step": 2283 }, { "epoch": 0.592938733125649, "grad_norm": 4.1804399490356445, "learning_rate": 4.213989777535719e-06, "loss": 0.09837237745523453, "step": 2284 }, { "epoch": 0.5931983385254413, "grad_norm": 3.187270164489746, "learning_rate": 4.2132712793623485e-06, "loss": 0.11252858489751816, "step": 2285 }, { "epoch": 0.5934579439252337, "grad_norm": 3.304765224456787, "learning_rate": 4.2125525225161205e-06, "loss": 0.11543834209442139, "step": 2286 }, { "epoch": 0.593717549325026, "grad_norm": 3.383683681488037, "learning_rate": 4.211833507124095e-06, "loss": 0.11023540794849396, "step": 2287 }, { "epoch": 0.5939771547248183, "grad_norm": 3.667628049850464, "learning_rate": 4.211114233313377e-06, "loss": 0.139484703540802, "step": 2288 }, { "epoch": 0.5942367601246106, "grad_norm": 1.4934271574020386, "learning_rate": 4.2103947012111176e-06, "loss": 0.11713646352291107, "step": 2289 }, { "epoch": 0.5944963655244029, "grad_norm": 2.4228463172912598, "learning_rate": 4.209674910944513e-06, "loss": 0.12451548874378204, "step": 2290 }, { "epoch": 0.5947559709241952, "grad_norm": 3.7152745723724365, "learning_rate": 4.208954862640807e-06, "loss": 0.10904008150100708, "step": 2291 }, { "epoch": 0.5950155763239875, "grad_norm": 3.717928409576416, "learning_rate": 4.208234556427288e-06, "loss": 0.12097376585006714, "step": 2292 }, { "epoch": 0.5952751817237798, "grad_norm": 3.8814215660095215, "learning_rate": 4.207513992431286e-06, "loss": 0.15042659640312195, "step": 2293 }, { "epoch": 0.5955347871235722, "grad_norm": 1.2686723470687866, "learning_rate": 4.206793170780185e-06, "loss": 0.10346663743257523, "step": 2294 }, { "epoch": 0.5957943925233645, "grad_norm": 3.8376212120056152, "learning_rate": 4.206072091601406e-06, "loss": 0.15665006637573242, "step": 2295 }, { "epoch": 0.5960539979231568, "grad_norm": 3.9705750942230225, "learning_rate": 4.2053507550224224e-06, "loss": 0.1358690857887268, "step": 2296 }, { "epoch": 0.5963136033229491, "grad_norm": 2.9574036598205566, "learning_rate": 4.204629161170747e-06, "loss": 0.11357700824737549, "step": 2297 }, { "epoch": 0.5965732087227414, "grad_norm": 2.533025026321411, "learning_rate": 4.2039073101739434e-06, "loss": 0.10390818119049072, "step": 2298 }, { "epoch": 0.5968328141225338, "grad_norm": 1.102702260017395, "learning_rate": 4.203185202159617e-06, "loss": 0.11690355837345123, "step": 2299 }, { "epoch": 0.5970924195223261, "grad_norm": 1.3239059448242188, "learning_rate": 4.202462837255421e-06, "loss": 0.11238324642181396, "step": 2300 }, { "epoch": 0.5973520249221184, "grad_norm": 1.0512887239456177, "learning_rate": 4.201740215589051e-06, "loss": 0.10998168587684631, "step": 2301 }, { "epoch": 0.5976116303219107, "grad_norm": 6.410540580749512, "learning_rate": 4.201017337288253e-06, "loss": 0.1334044635295868, "step": 2302 }, { "epoch": 0.597871235721703, "grad_norm": 1.5934723615646362, "learning_rate": 4.2002942024808125e-06, "loss": 0.12156113982200623, "step": 2303 }, { "epoch": 0.5981308411214953, "grad_norm": 1.2137682437896729, "learning_rate": 4.199570811294564e-06, "loss": 0.10290854424238205, "step": 2304 }, { "epoch": 0.5983904465212876, "grad_norm": 0.6572914123535156, "learning_rate": 4.198847163857388e-06, "loss": 0.1283690333366394, "step": 2305 }, { "epoch": 0.59865005192108, "grad_norm": 3.3792014122009277, "learning_rate": 4.1981232602972075e-06, "loss": 0.11569494754076004, "step": 2306 }, { "epoch": 0.5989096573208723, "grad_norm": 1.8306158781051636, "learning_rate": 4.197399100741991e-06, "loss": 0.11737659573554993, "step": 2307 }, { "epoch": 0.5991692627206646, "grad_norm": 3.408560276031494, "learning_rate": 4.196674685319755e-06, "loss": 0.14002437889575958, "step": 2308 }, { "epoch": 0.5994288681204569, "grad_norm": 1.3444292545318604, "learning_rate": 4.19595001415856e-06, "loss": 0.11294954270124435, "step": 2309 }, { "epoch": 0.5996884735202492, "grad_norm": 1.35906982421875, "learning_rate": 4.19522508738651e-06, "loss": 0.12370140850543976, "step": 2310 }, { "epoch": 0.5999480789200415, "grad_norm": 1.8983994722366333, "learning_rate": 4.194499905131755e-06, "loss": 0.11366012692451477, "step": 2311 }, { "epoch": 0.6002076843198338, "grad_norm": 2.0271718502044678, "learning_rate": 4.1937744675224925e-06, "loss": 0.11412826180458069, "step": 2312 }, { "epoch": 0.6004672897196262, "grad_norm": 1.1730523109436035, "learning_rate": 4.193048774686961e-06, "loss": 0.10677607357501984, "step": 2313 }, { "epoch": 0.6007268951194185, "grad_norm": 2.5546109676361084, "learning_rate": 4.192322826753449e-06, "loss": 0.09707493335008621, "step": 2314 }, { "epoch": 0.6009865005192108, "grad_norm": 1.7067372798919678, "learning_rate": 4.1915966238502855e-06, "loss": 0.1209525614976883, "step": 2315 }, { "epoch": 0.6012461059190031, "grad_norm": 1.9321461915969849, "learning_rate": 4.1908701661058465e-06, "loss": 0.08502374589443207, "step": 2316 }, { "epoch": 0.6015057113187954, "grad_norm": 3.017226219177246, "learning_rate": 4.190143453648554e-06, "loss": 0.10894308984279633, "step": 2317 }, { "epoch": 0.6017653167185878, "grad_norm": 2.8239166736602783, "learning_rate": 4.1894164866068745e-06, "loss": 0.1046709418296814, "step": 2318 }, { "epoch": 0.6020249221183801, "grad_norm": 1.5676780939102173, "learning_rate": 4.188689265109318e-06, "loss": 0.11466994136571884, "step": 2319 }, { "epoch": 0.6022845275181724, "grad_norm": 2.570040702819824, "learning_rate": 4.187961789284441e-06, "loss": 0.1420571357011795, "step": 2320 }, { "epoch": 0.6025441329179647, "grad_norm": 2.720642328262329, "learning_rate": 4.187234059260845e-06, "loss": 0.10498727858066559, "step": 2321 }, { "epoch": 0.602803738317757, "grad_norm": 3.711198329925537, "learning_rate": 4.186506075167174e-06, "loss": 0.1534785032272339, "step": 2322 }, { "epoch": 0.6030633437175493, "grad_norm": 1.3554033041000366, "learning_rate": 4.1857778371321225e-06, "loss": 0.12449641525745392, "step": 2323 }, { "epoch": 0.6033229491173416, "grad_norm": 2.8238413333892822, "learning_rate": 4.185049345284423e-06, "loss": 0.14308591187000275, "step": 2324 }, { "epoch": 0.6035825545171339, "grad_norm": 1.347179889678955, "learning_rate": 4.184320599752858e-06, "loss": 0.11114032566547394, "step": 2325 }, { "epoch": 0.6038421599169262, "grad_norm": 2.5697762966156006, "learning_rate": 4.183591600666253e-06, "loss": 0.11080712080001831, "step": 2326 }, { "epoch": 0.6041017653167186, "grad_norm": 2.4372801780700684, "learning_rate": 4.182862348153478e-06, "loss": 0.0959460437297821, "step": 2327 }, { "epoch": 0.6043613707165109, "grad_norm": 2.595355272293091, "learning_rate": 4.182132842343448e-06, "loss": 0.15289925038814545, "step": 2328 }, { "epoch": 0.6046209761163032, "grad_norm": 4.183606147766113, "learning_rate": 4.181403083365122e-06, "loss": 0.08405144512653351, "step": 2329 }, { "epoch": 0.6048805815160956, "grad_norm": 2.2183749675750732, "learning_rate": 4.180673071347506e-06, "loss": 0.11384738236665726, "step": 2330 }, { "epoch": 0.6051401869158879, "grad_norm": 3.2060112953186035, "learning_rate": 4.17994280641965e-06, "loss": 0.13906967639923096, "step": 2331 }, { "epoch": 0.6053997923156802, "grad_norm": 1.0025744438171387, "learning_rate": 4.179212288710646e-06, "loss": 0.12102891504764557, "step": 2332 }, { "epoch": 0.6056593977154725, "grad_norm": 2.626760959625244, "learning_rate": 4.178481518349634e-06, "loss": 0.12581641972064972, "step": 2333 }, { "epoch": 0.6059190031152648, "grad_norm": 1.1600189208984375, "learning_rate": 4.1777504954657975e-06, "loss": 0.13473238050937653, "step": 2334 }, { "epoch": 0.6061786085150571, "grad_norm": 2.3669872283935547, "learning_rate": 4.1770192201883656e-06, "loss": 0.12646004557609558, "step": 2335 }, { "epoch": 0.6064382139148494, "grad_norm": 4.320856094360352, "learning_rate": 4.17628769264661e-06, "loss": 0.14535047113895416, "step": 2336 }, { "epoch": 0.6066978193146417, "grad_norm": 2.88010573387146, "learning_rate": 4.175555912969846e-06, "loss": 0.13658291101455688, "step": 2337 }, { "epoch": 0.606957424714434, "grad_norm": 1.1078611612319946, "learning_rate": 4.17482388128744e-06, "loss": 0.10554593801498413, "step": 2338 }, { "epoch": 0.6072170301142263, "grad_norm": 2.4437055587768555, "learning_rate": 4.174091597728793e-06, "loss": 0.13931569457054138, "step": 2339 }, { "epoch": 0.6074766355140186, "grad_norm": 2.1738455295562744, "learning_rate": 4.173359062423361e-06, "loss": 0.11755320429801941, "step": 2340 }, { "epoch": 0.607736240913811, "grad_norm": 2.3801932334899902, "learning_rate": 4.172626275500637e-06, "loss": 0.12046996504068375, "step": 2341 }, { "epoch": 0.6079958463136034, "grad_norm": 2.8265511989593506, "learning_rate": 4.171893237090161e-06, "loss": 0.11911826580762863, "step": 2342 }, { "epoch": 0.6082554517133957, "grad_norm": 2.2794792652130127, "learning_rate": 4.171159947321519e-06, "loss": 0.11173705756664276, "step": 2343 }, { "epoch": 0.608515057113188, "grad_norm": 1.376367449760437, "learning_rate": 4.170426406324337e-06, "loss": 0.10909318178892136, "step": 2344 }, { "epoch": 0.6087746625129803, "grad_norm": 1.4286894798278809, "learning_rate": 4.16969261422829e-06, "loss": 0.09919454157352448, "step": 2345 }, { "epoch": 0.6090342679127726, "grad_norm": 1.388911247253418, "learning_rate": 4.168958571163097e-06, "loss": 0.12158100306987762, "step": 2346 }, { "epoch": 0.6092938733125649, "grad_norm": 1.6035038232803345, "learning_rate": 4.168224277258517e-06, "loss": 0.11383134871721268, "step": 2347 }, { "epoch": 0.6095534787123572, "grad_norm": 2.8715193271636963, "learning_rate": 4.167489732644358e-06, "loss": 0.16745486855506897, "step": 2348 }, { "epoch": 0.6098130841121495, "grad_norm": 2.8412091732025146, "learning_rate": 4.166754937450471e-06, "loss": 0.11758622527122498, "step": 2349 }, { "epoch": 0.6100726895119418, "grad_norm": 2.3371050357818604, "learning_rate": 4.1660198918067504e-06, "loss": 0.09434132277965546, "step": 2350 }, { "epoch": 0.6103322949117341, "grad_norm": 2.2229762077331543, "learning_rate": 4.165284595843135e-06, "loss": 0.10177873075008392, "step": 2351 }, { "epoch": 0.6105919003115264, "grad_norm": 2.1999220848083496, "learning_rate": 4.1645490496896085e-06, "loss": 0.10282264649868011, "step": 2352 }, { "epoch": 0.6108515057113187, "grad_norm": 3.9747776985168457, "learning_rate": 4.1638132534762e-06, "loss": 0.1089913547039032, "step": 2353 }, { "epoch": 0.6111111111111112, "grad_norm": 1.7181849479675293, "learning_rate": 4.163077207332979e-06, "loss": 0.12135862559080124, "step": 2354 }, { "epoch": 0.6113707165109035, "grad_norm": 1.875899076461792, "learning_rate": 4.1623409113900636e-06, "loss": 0.1272238940000534, "step": 2355 }, { "epoch": 0.6116303219106958, "grad_norm": 1.3249993324279785, "learning_rate": 4.161604365777613e-06, "loss": 0.11552923917770386, "step": 2356 }, { "epoch": 0.6118899273104881, "grad_norm": 4.224564075469971, "learning_rate": 4.160867570625833e-06, "loss": 0.10426849126815796, "step": 2357 }, { "epoch": 0.6121495327102804, "grad_norm": 0.9638420939445496, "learning_rate": 4.16013052606497e-06, "loss": 0.11694213002920151, "step": 2358 }, { "epoch": 0.6124091381100727, "grad_norm": 2.882052421569824, "learning_rate": 4.159393232225318e-06, "loss": 0.11749567091464996, "step": 2359 }, { "epoch": 0.612668743509865, "grad_norm": 2.406954050064087, "learning_rate": 4.1586556892372135e-06, "loss": 0.13103163242340088, "step": 2360 }, { "epoch": 0.6129283489096573, "grad_norm": 1.1595637798309326, "learning_rate": 4.157917897231037e-06, "loss": 0.10853689908981323, "step": 2361 }, { "epoch": 0.6131879543094496, "grad_norm": 2.2944419384002686, "learning_rate": 4.157179856337214e-06, "loss": 0.13170044124126434, "step": 2362 }, { "epoch": 0.6134475597092419, "grad_norm": 2.9355766773223877, "learning_rate": 4.156441566686212e-06, "loss": 0.113438680768013, "step": 2363 }, { "epoch": 0.6137071651090342, "grad_norm": 2.8988492488861084, "learning_rate": 4.155703028408544e-06, "loss": 0.13944846391677856, "step": 2364 }, { "epoch": 0.6139667705088265, "grad_norm": 2.2344796657562256, "learning_rate": 4.1549642416347676e-06, "loss": 0.13175362348556519, "step": 2365 }, { "epoch": 0.614226375908619, "grad_norm": 1.701129674911499, "learning_rate": 4.154225206495482e-06, "loss": 0.10322223603725433, "step": 2366 }, { "epoch": 0.6144859813084113, "grad_norm": 1.8029052019119263, "learning_rate": 4.153485923121332e-06, "loss": 0.11527562141418457, "step": 2367 }, { "epoch": 0.6147455867082036, "grad_norm": 3.6440863609313965, "learning_rate": 4.152746391643008e-06, "loss": 0.1169605478644371, "step": 2368 }, { "epoch": 0.6150051921079959, "grad_norm": 3.5458128452301025, "learning_rate": 4.152006612191239e-06, "loss": 0.10360264778137207, "step": 2369 }, { "epoch": 0.6152647975077882, "grad_norm": 2.0945305824279785, "learning_rate": 4.151266584896803e-06, "loss": 0.11647142469882965, "step": 2370 }, { "epoch": 0.6155244029075805, "grad_norm": 1.0273116827011108, "learning_rate": 4.1505263098905196e-06, "loss": 0.11938557028770447, "step": 2371 }, { "epoch": 0.6157840083073728, "grad_norm": 1.114499568939209, "learning_rate": 4.149785787303253e-06, "loss": 0.12572288513183594, "step": 2372 }, { "epoch": 0.6160436137071651, "grad_norm": 2.367978572845459, "learning_rate": 4.14904501726591e-06, "loss": 0.12678153812885284, "step": 2373 }, { "epoch": 0.6163032191069574, "grad_norm": 3.1378676891326904, "learning_rate": 4.148303999909441e-06, "loss": 0.11887888610363007, "step": 2374 }, { "epoch": 0.6165628245067497, "grad_norm": 2.181852102279663, "learning_rate": 4.147562735364842e-06, "loss": 0.11122570931911469, "step": 2375 }, { "epoch": 0.616822429906542, "grad_norm": 2.3761627674102783, "learning_rate": 4.146821223763151e-06, "loss": 0.11882372200489044, "step": 2376 }, { "epoch": 0.6170820353063343, "grad_norm": 1.6931493282318115, "learning_rate": 4.146079465235452e-06, "loss": 0.09717366099357605, "step": 2377 }, { "epoch": 0.6173416407061267, "grad_norm": 1.1681188344955444, "learning_rate": 4.145337459912868e-06, "loss": 0.12633436918258667, "step": 2378 }, { "epoch": 0.617601246105919, "grad_norm": 1.030105710029602, "learning_rate": 4.144595207926571e-06, "loss": 0.10326629877090454, "step": 2379 }, { "epoch": 0.6178608515057114, "grad_norm": 1.6095588207244873, "learning_rate": 4.143852709407773e-06, "loss": 0.09341633319854736, "step": 2380 }, { "epoch": 0.6181204569055037, "grad_norm": 0.9776865839958191, "learning_rate": 4.143109964487731e-06, "loss": 0.11419764161109924, "step": 2381 }, { "epoch": 0.618380062305296, "grad_norm": 0.9041070342063904, "learning_rate": 4.1423669732977455e-06, "loss": 0.10252851247787476, "step": 2382 }, { "epoch": 0.6186396677050883, "grad_norm": 1.7873661518096924, "learning_rate": 4.1416237359691595e-06, "loss": 0.11233393847942352, "step": 2383 }, { "epoch": 0.6188992731048806, "grad_norm": 2.6447999477386475, "learning_rate": 4.140880252633362e-06, "loss": 0.12011236697435379, "step": 2384 }, { "epoch": 0.6191588785046729, "grad_norm": 3.026615858078003, "learning_rate": 4.140136523421782e-06, "loss": 0.14149153232574463, "step": 2385 }, { "epoch": 0.6194184839044652, "grad_norm": 0.8039629459381104, "learning_rate": 4.139392548465895e-06, "loss": 0.11459983885288239, "step": 2386 }, { "epoch": 0.6196780893042575, "grad_norm": 3.1240475177764893, "learning_rate": 4.138648327897216e-06, "loss": 0.09144440293312073, "step": 2387 }, { "epoch": 0.6199376947040498, "grad_norm": 2.0224359035491943, "learning_rate": 4.137903861847311e-06, "loss": 0.1150861531496048, "step": 2388 }, { "epoch": 0.6201973001038421, "grad_norm": 1.0786503553390503, "learning_rate": 4.137159150447782e-06, "loss": 0.11729926615953445, "step": 2389 }, { "epoch": 0.6204569055036345, "grad_norm": 3.1036078929901123, "learning_rate": 4.136414193830276e-06, "loss": 0.11207859963178635, "step": 2390 }, { "epoch": 0.6207165109034268, "grad_norm": 3.2007956504821777, "learning_rate": 4.135668992126485e-06, "loss": 0.10987761616706848, "step": 2391 }, { "epoch": 0.6209761163032191, "grad_norm": 4.018417835235596, "learning_rate": 4.134923545468144e-06, "loss": 0.13675448298454285, "step": 2392 }, { "epoch": 0.6212357217030114, "grad_norm": 0.9679316282272339, "learning_rate": 4.134177853987032e-06, "loss": 0.11765818297863007, "step": 2393 }, { "epoch": 0.6214953271028038, "grad_norm": 1.7428497076034546, "learning_rate": 4.133431917814967e-06, "loss": 0.11078911274671555, "step": 2394 }, { "epoch": 0.6217549325025961, "grad_norm": 1.6291974782943726, "learning_rate": 4.132685737083817e-06, "loss": 0.10993702709674835, "step": 2395 }, { "epoch": 0.6220145379023884, "grad_norm": 2.340385675430298, "learning_rate": 4.131939311925486e-06, "loss": 0.08498582243919373, "step": 2396 }, { "epoch": 0.6222741433021807, "grad_norm": 2.7378902435302734, "learning_rate": 4.131192642471929e-06, "loss": 0.10263147950172424, "step": 2397 }, { "epoch": 0.622533748701973, "grad_norm": 1.025571584701538, "learning_rate": 4.130445728855137e-06, "loss": 0.11767937242984772, "step": 2398 }, { "epoch": 0.6227933541017653, "grad_norm": 1.544297695159912, "learning_rate": 4.129698571207148e-06, "loss": 0.12394748628139496, "step": 2399 }, { "epoch": 0.6230529595015576, "grad_norm": 2.760021924972534, "learning_rate": 4.128951169660043e-06, "loss": 0.10886208713054657, "step": 2400 }, { "epoch": 0.6233125649013499, "grad_norm": 2.2488019466400146, "learning_rate": 4.1282035243459446e-06, "loss": 0.11791838705539703, "step": 2401 }, { "epoch": 0.6235721703011423, "grad_norm": 2.7218148708343506, "learning_rate": 4.1274556353970196e-06, "loss": 0.12210136651992798, "step": 2402 }, { "epoch": 0.6238317757009346, "grad_norm": 1.3816282749176025, "learning_rate": 4.126707502945478e-06, "loss": 0.12047232687473297, "step": 2403 }, { "epoch": 0.6240913811007269, "grad_norm": 2.041569471359253, "learning_rate": 4.125959127123572e-06, "loss": 0.10834428668022156, "step": 2404 }, { "epoch": 0.6243509865005192, "grad_norm": 0.9850131869316101, "learning_rate": 4.125210508063599e-06, "loss": 0.10825909674167633, "step": 2405 }, { "epoch": 0.6246105919003115, "grad_norm": 1.0824365615844727, "learning_rate": 4.1244616458978945e-06, "loss": 0.10837885737419128, "step": 2406 }, { "epoch": 0.6248701973001038, "grad_norm": 1.134917974472046, "learning_rate": 4.1237125407588434e-06, "loss": 0.10834470391273499, "step": 2407 }, { "epoch": 0.6251298026998962, "grad_norm": 2.2793707847595215, "learning_rate": 4.1229631927788684e-06, "loss": 0.11769978702068329, "step": 2408 }, { "epoch": 0.6253894080996885, "grad_norm": 3.2852747440338135, "learning_rate": 4.122213602090438e-06, "loss": 0.12018882483243942, "step": 2409 }, { "epoch": 0.6256490134994808, "grad_norm": 1.7046610116958618, "learning_rate": 4.121463768826061e-06, "loss": 0.11403338611125946, "step": 2410 }, { "epoch": 0.6259086188992731, "grad_norm": 3.4018361568450928, "learning_rate": 4.120713693118293e-06, "loss": 0.13713809847831726, "step": 2411 }, { "epoch": 0.6261682242990654, "grad_norm": 2.5978028774261475, "learning_rate": 4.119963375099729e-06, "loss": 0.13654664158821106, "step": 2412 }, { "epoch": 0.6264278296988577, "grad_norm": 2.8088572025299072, "learning_rate": 4.119212814903009e-06, "loss": 0.12104000151157379, "step": 2413 }, { "epoch": 0.6266874350986501, "grad_norm": 1.6388564109802246, "learning_rate": 4.118462012660813e-06, "loss": 0.11306895315647125, "step": 2414 }, { "epoch": 0.6269470404984424, "grad_norm": 2.3611748218536377, "learning_rate": 4.117710968505867e-06, "loss": 0.10922945290803909, "step": 2415 }, { "epoch": 0.6272066458982347, "grad_norm": 1.7859631776809692, "learning_rate": 4.116959682570938e-06, "loss": 0.1223631203174591, "step": 2416 }, { "epoch": 0.627466251298027, "grad_norm": 1.8853248357772827, "learning_rate": 4.1162081549888375e-06, "loss": 0.12096904963254929, "step": 2417 }, { "epoch": 0.6277258566978193, "grad_norm": 2.236530065536499, "learning_rate": 4.1154563858924164e-06, "loss": 0.10088566690683365, "step": 2418 }, { "epoch": 0.6279854620976116, "grad_norm": 3.543083667755127, "learning_rate": 4.114704375414572e-06, "loss": 0.13697628676891327, "step": 2419 }, { "epoch": 0.6282450674974039, "grad_norm": 2.1882221698760986, "learning_rate": 4.1139521236882405e-06, "loss": 0.10456030070781708, "step": 2420 }, { "epoch": 0.6285046728971962, "grad_norm": 1.3018145561218262, "learning_rate": 4.113199630846405e-06, "loss": 0.11882414668798447, "step": 2421 }, { "epoch": 0.6287642782969886, "grad_norm": 0.6613242030143738, "learning_rate": 4.112446897022086e-06, "loss": 0.11184876412153244, "step": 2422 }, { "epoch": 0.6290238836967809, "grad_norm": 4.881644248962402, "learning_rate": 4.111693922348353e-06, "loss": 0.11252328753471375, "step": 2423 }, { "epoch": 0.6292834890965732, "grad_norm": 2.2172446250915527, "learning_rate": 4.110940706958313e-06, "loss": 0.10170918703079224, "step": 2424 }, { "epoch": 0.6295430944963655, "grad_norm": 2.4434800148010254, "learning_rate": 4.110187250985117e-06, "loss": 0.11507900804281235, "step": 2425 }, { "epoch": 0.6298026998961579, "grad_norm": 1.724881649017334, "learning_rate": 4.10943355456196e-06, "loss": 0.1269783079624176, "step": 2426 }, { "epoch": 0.6300623052959502, "grad_norm": 2.330186367034912, "learning_rate": 4.108679617822077e-06, "loss": 0.11942926049232483, "step": 2427 }, { "epoch": 0.6303219106957425, "grad_norm": 1.327824592590332, "learning_rate": 4.107925440898747e-06, "loss": 0.11843203008174896, "step": 2428 }, { "epoch": 0.6305815160955348, "grad_norm": 1.8957284688949585, "learning_rate": 4.107171023925292e-06, "loss": 0.1379353106021881, "step": 2429 }, { "epoch": 0.6308411214953271, "grad_norm": 1.65714430809021, "learning_rate": 4.106416367035075e-06, "loss": 0.12410624325275421, "step": 2430 }, { "epoch": 0.6311007268951194, "grad_norm": 4.442312717437744, "learning_rate": 4.1056614703615015e-06, "loss": 0.10728806257247925, "step": 2431 }, { "epoch": 0.6313603322949117, "grad_norm": 3.529125452041626, "learning_rate": 4.1049063340380205e-06, "loss": 0.07954699546098709, "step": 2432 }, { "epoch": 0.631619937694704, "grad_norm": 1.260228157043457, "learning_rate": 4.104150958198123e-06, "loss": 0.11348994076251984, "step": 2433 }, { "epoch": 0.6318795430944963, "grad_norm": 4.352293968200684, "learning_rate": 4.103395342975343e-06, "loss": 0.12059508264064789, "step": 2434 }, { "epoch": 0.6321391484942886, "grad_norm": 1.647815227508545, "learning_rate": 4.102639488503254e-06, "loss": 0.11213189363479614, "step": 2435 }, { "epoch": 0.632398753894081, "grad_norm": 3.547651767730713, "learning_rate": 4.101883394915475e-06, "loss": 0.08065926283597946, "step": 2436 }, { "epoch": 0.6326583592938733, "grad_norm": 3.98990797996521, "learning_rate": 4.101127062345666e-06, "loss": 0.119209423661232, "step": 2437 }, { "epoch": 0.6329179646936657, "grad_norm": 2.6829049587249756, "learning_rate": 4.100370490927529e-06, "loss": 0.13653402030467987, "step": 2438 }, { "epoch": 0.633177570093458, "grad_norm": 4.183703899383545, "learning_rate": 4.0996136807948085e-06, "loss": 0.11767058819532394, "step": 2439 }, { "epoch": 0.6334371754932503, "grad_norm": 2.5404274463653564, "learning_rate": 4.0988566320812915e-06, "loss": 0.11223548650741577, "step": 2440 }, { "epoch": 0.6336967808930426, "grad_norm": 2.866048812866211, "learning_rate": 4.098099344920806e-06, "loss": 0.10415387898683548, "step": 2441 }, { "epoch": 0.6339563862928349, "grad_norm": 1.8117631673812866, "learning_rate": 4.097341819447224e-06, "loss": 0.13174556195735931, "step": 2442 }, { "epoch": 0.6342159916926272, "grad_norm": 1.814422369003296, "learning_rate": 4.096584055794459e-06, "loss": 0.09014864265918732, "step": 2443 }, { "epoch": 0.6344755970924195, "grad_norm": 3.29309344291687, "learning_rate": 4.095826054096465e-06, "loss": 0.11192329972982407, "step": 2444 }, { "epoch": 0.6347352024922118, "grad_norm": 4.500725746154785, "learning_rate": 4.09506781448724e-06, "loss": 0.13279010355472565, "step": 2445 }, { "epoch": 0.6349948078920041, "grad_norm": 2.7972471714019775, "learning_rate": 4.094309337100823e-06, "loss": 0.10974034667015076, "step": 2446 }, { "epoch": 0.6352544132917964, "grad_norm": 1.3126033544540405, "learning_rate": 4.093550622071296e-06, "loss": 0.10777920484542847, "step": 2447 }, { "epoch": 0.6355140186915887, "grad_norm": 3.723698139190674, "learning_rate": 4.092791669532784e-06, "loss": 0.11174140870571136, "step": 2448 }, { "epoch": 0.635773624091381, "grad_norm": 0.9178189635276794, "learning_rate": 4.0920324796194485e-06, "loss": 0.11727041006088257, "step": 2449 }, { "epoch": 0.6360332294911735, "grad_norm": 3.5303566455841064, "learning_rate": 4.091273052465501e-06, "loss": 0.12337598949670792, "step": 2450 }, { "epoch": 0.6362928348909658, "grad_norm": 1.3588613271713257, "learning_rate": 4.090513388205187e-06, "loss": 0.10547979176044464, "step": 2451 }, { "epoch": 0.6365524402907581, "grad_norm": 4.355074405670166, "learning_rate": 4.0897534869728e-06, "loss": 0.09528762102127075, "step": 2452 }, { "epoch": 0.6368120456905504, "grad_norm": 2.9978537559509277, "learning_rate": 4.088993348902673e-06, "loss": 0.10381124913692474, "step": 2453 }, { "epoch": 0.6370716510903427, "grad_norm": 3.737379312515259, "learning_rate": 4.0882329741291815e-06, "loss": 0.14686045050621033, "step": 2454 }, { "epoch": 0.637331256490135, "grad_norm": 1.9994206428527832, "learning_rate": 4.087472362786741e-06, "loss": 0.11348957568407059, "step": 2455 }, { "epoch": 0.6375908618899273, "grad_norm": 2.5210046768188477, "learning_rate": 4.086711515009811e-06, "loss": 0.11681589484214783, "step": 2456 }, { "epoch": 0.6378504672897196, "grad_norm": 2.05819034576416, "learning_rate": 4.085950430932892e-06, "loss": 0.12991201877593994, "step": 2457 }, { "epoch": 0.6381100726895119, "grad_norm": 1.0770989656448364, "learning_rate": 4.085189110690526e-06, "loss": 0.11144650727510452, "step": 2458 }, { "epoch": 0.6383696780893042, "grad_norm": 2.2090022563934326, "learning_rate": 4.084427554417296e-06, "loss": 0.10390417277812958, "step": 2459 }, { "epoch": 0.6386292834890965, "grad_norm": 3.060878038406372, "learning_rate": 4.08366576224783e-06, "loss": 0.10989449173212051, "step": 2460 }, { "epoch": 0.6388888888888888, "grad_norm": 3.1655969619750977, "learning_rate": 4.082903734316793e-06, "loss": 0.13803818821907043, "step": 2461 }, { "epoch": 0.6391484942886813, "grad_norm": 2.6853346824645996, "learning_rate": 4.082141470758896e-06, "loss": 0.12468606978654861, "step": 2462 }, { "epoch": 0.6394080996884736, "grad_norm": 2.7663180828094482, "learning_rate": 4.081378971708889e-06, "loss": 0.12181016057729721, "step": 2463 }, { "epoch": 0.6396677050882659, "grad_norm": 3.825321912765503, "learning_rate": 4.080616237301563e-06, "loss": 0.13339991867542267, "step": 2464 }, { "epoch": 0.6399273104880582, "grad_norm": 1.3441987037658691, "learning_rate": 4.079853267671755e-06, "loss": 0.10383746027946472, "step": 2465 }, { "epoch": 0.6401869158878505, "grad_norm": 3.4119555950164795, "learning_rate": 4.079090062954339e-06, "loss": 0.08956315368413925, "step": 2466 }, { "epoch": 0.6404465212876428, "grad_norm": 2.0832273960113525, "learning_rate": 4.078326623284231e-06, "loss": 0.11877720057964325, "step": 2467 }, { "epoch": 0.6407061266874351, "grad_norm": 3.350811004638672, "learning_rate": 4.077562948796391e-06, "loss": 0.11189152300357819, "step": 2468 }, { "epoch": 0.6409657320872274, "grad_norm": 2.32316255569458, "learning_rate": 4.076799039625819e-06, "loss": 0.07190349698066711, "step": 2469 }, { "epoch": 0.6412253374870197, "grad_norm": 2.7619259357452393, "learning_rate": 4.076034895907556e-06, "loss": 0.13035717606544495, "step": 2470 }, { "epoch": 0.641484942886812, "grad_norm": 1.388921856880188, "learning_rate": 4.075270517776686e-06, "loss": 0.10642531514167786, "step": 2471 }, { "epoch": 0.6417445482866043, "grad_norm": 1.8489024639129639, "learning_rate": 4.0745059053683325e-06, "loss": 0.0847519114613533, "step": 2472 }, { "epoch": 0.6420041536863966, "grad_norm": 1.8042129278182983, "learning_rate": 4.073741058817663e-06, "loss": 0.11271613836288452, "step": 2473 }, { "epoch": 0.642263759086189, "grad_norm": 1.2772904634475708, "learning_rate": 4.072975978259883e-06, "loss": 0.1084975153207779, "step": 2474 }, { "epoch": 0.6425233644859814, "grad_norm": 2.42950701713562, "learning_rate": 4.072210663830243e-06, "loss": 0.08131487667560577, "step": 2475 }, { "epoch": 0.6427829698857737, "grad_norm": 2.313732624053955, "learning_rate": 4.0714451156640314e-06, "loss": 0.1310541033744812, "step": 2476 }, { "epoch": 0.643042575285566, "grad_norm": 1.2544584274291992, "learning_rate": 4.070679333896582e-06, "loss": 0.11456809192895889, "step": 2477 }, { "epoch": 0.6433021806853583, "grad_norm": 2.6339547634124756, "learning_rate": 4.069913318663264e-06, "loss": 0.12518420815467834, "step": 2478 }, { "epoch": 0.6435617860851506, "grad_norm": 1.6885697841644287, "learning_rate": 4.069147070099495e-06, "loss": 0.12253459542989731, "step": 2479 }, { "epoch": 0.6438213914849429, "grad_norm": 1.6657235622406006, "learning_rate": 4.068380588340728e-06, "loss": 0.10880666971206665, "step": 2480 }, { "epoch": 0.6440809968847352, "grad_norm": 1.9761037826538086, "learning_rate": 4.067613873522459e-06, "loss": 0.11919811367988586, "step": 2481 }, { "epoch": 0.6443406022845275, "grad_norm": 2.406336784362793, "learning_rate": 4.066846925780226e-06, "loss": 0.09984996914863586, "step": 2482 }, { "epoch": 0.6446002076843198, "grad_norm": 1.148956298828125, "learning_rate": 4.06607974524961e-06, "loss": 0.11249673366546631, "step": 2483 }, { "epoch": 0.6448598130841121, "grad_norm": 1.7696391344070435, "learning_rate": 4.065312332066229e-06, "loss": 0.11770498752593994, "step": 2484 }, { "epoch": 0.6451194184839044, "grad_norm": 1.644737720489502, "learning_rate": 4.064544686365744e-06, "loss": 0.11295656859874725, "step": 2485 }, { "epoch": 0.6453790238836968, "grad_norm": 3.11092209815979, "learning_rate": 4.063776808283857e-06, "loss": 0.11827775835990906, "step": 2486 }, { "epoch": 0.6456386292834891, "grad_norm": 1.1422346830368042, "learning_rate": 4.0630086979563125e-06, "loss": 0.10149000585079193, "step": 2487 }, { "epoch": 0.6458982346832814, "grad_norm": 2.187117099761963, "learning_rate": 4.062240355518895e-06, "loss": 0.10991968214511871, "step": 2488 }, { "epoch": 0.6461578400830738, "grad_norm": 0.9940130114555359, "learning_rate": 4.061471781107427e-06, "loss": 0.1160832941532135, "step": 2489 }, { "epoch": 0.6464174454828661, "grad_norm": 3.0254876613616943, "learning_rate": 4.060702974857778e-06, "loss": 0.1453057825565338, "step": 2490 }, { "epoch": 0.6466770508826584, "grad_norm": 3.246220111846924, "learning_rate": 4.059933936905854e-06, "loss": 0.12292148172855377, "step": 2491 }, { "epoch": 0.6469366562824507, "grad_norm": 0.8818775415420532, "learning_rate": 4.059164667387603e-06, "loss": 0.11847320199012756, "step": 2492 }, { "epoch": 0.647196261682243, "grad_norm": 1.44447922706604, "learning_rate": 4.058395166439015e-06, "loss": 0.10148157924413681, "step": 2493 }, { "epoch": 0.6474558670820353, "grad_norm": 2.438692808151245, "learning_rate": 4.05762543419612e-06, "loss": 0.138850137591362, "step": 2494 }, { "epoch": 0.6477154724818276, "grad_norm": 1.8276991844177246, "learning_rate": 4.056855470794988e-06, "loss": 0.12443864345550537, "step": 2495 }, { "epoch": 0.6479750778816199, "grad_norm": 1.5183658599853516, "learning_rate": 4.0560852763717335e-06, "loss": 0.11023460328578949, "step": 2496 }, { "epoch": 0.6482346832814122, "grad_norm": 1.0700637102127075, "learning_rate": 4.0553148510625065e-06, "loss": 0.10295599699020386, "step": 2497 }, { "epoch": 0.6484942886812046, "grad_norm": 1.8653136491775513, "learning_rate": 4.054544195003502e-06, "loss": 0.10565727204084396, "step": 2498 }, { "epoch": 0.6487538940809969, "grad_norm": 1.4582246541976929, "learning_rate": 4.053773308330953e-06, "loss": 0.12902173399925232, "step": 2499 }, { "epoch": 0.6490134994807892, "grad_norm": 1.0809376239776611, "learning_rate": 4.053002191181137e-06, "loss": 0.11358675360679626, "step": 2500 }, { "epoch": 0.6492731048805815, "grad_norm": 3.1224491596221924, "learning_rate": 4.052230843690367e-06, "loss": 0.08695870637893677, "step": 2501 }, { "epoch": 0.6495327102803738, "grad_norm": 2.3361921310424805, "learning_rate": 4.051459265995001e-06, "loss": 0.1271822452545166, "step": 2502 }, { "epoch": 0.6497923156801662, "grad_norm": 2.515500068664551, "learning_rate": 4.050687458231437e-06, "loss": 0.08345888555049896, "step": 2503 }, { "epoch": 0.6500519210799585, "grad_norm": 1.2188727855682373, "learning_rate": 4.049915420536112e-06, "loss": 0.09783542156219482, "step": 2504 }, { "epoch": 0.6503115264797508, "grad_norm": 0.7751520276069641, "learning_rate": 4.049143153045505e-06, "loss": 0.10997267067432404, "step": 2505 }, { "epoch": 0.6505711318795431, "grad_norm": 1.3248153924942017, "learning_rate": 4.048370655896135e-06, "loss": 0.10554583370685577, "step": 2506 }, { "epoch": 0.6508307372793354, "grad_norm": 1.1231352090835571, "learning_rate": 4.047597929224561e-06, "loss": 0.11933591961860657, "step": 2507 }, { "epoch": 0.6510903426791277, "grad_norm": 2.3145244121551514, "learning_rate": 4.0468249731673845e-06, "loss": 0.11104726791381836, "step": 2508 }, { "epoch": 0.65134994807892, "grad_norm": 3.3399431705474854, "learning_rate": 4.046051787861247e-06, "loss": 0.13958856463432312, "step": 2509 }, { "epoch": 0.6516095534787124, "grad_norm": 1.0548938512802124, "learning_rate": 4.045278373442828e-06, "loss": 0.11223109066486359, "step": 2510 }, { "epoch": 0.6518691588785047, "grad_norm": 1.6667516231536865, "learning_rate": 4.04450473004885e-06, "loss": 0.11011253297328949, "step": 2511 }, { "epoch": 0.652128764278297, "grad_norm": 2.3357956409454346, "learning_rate": 4.043730857816077e-06, "loss": 0.12060566246509552, "step": 2512 }, { "epoch": 0.6523883696780893, "grad_norm": 0.8708771467208862, "learning_rate": 4.042956756881311e-06, "loss": 0.11442548036575317, "step": 2513 }, { "epoch": 0.6526479750778816, "grad_norm": 2.7329909801483154, "learning_rate": 4.042182427381395e-06, "loss": 0.11335065215826035, "step": 2514 }, { "epoch": 0.6529075804776739, "grad_norm": 2.1977109909057617, "learning_rate": 4.041407869453212e-06, "loss": 0.10962718725204468, "step": 2515 }, { "epoch": 0.6531671858774662, "grad_norm": 2.5048158168792725, "learning_rate": 4.0406330832336885e-06, "loss": 0.11601860821247101, "step": 2516 }, { "epoch": 0.6534267912772586, "grad_norm": 2.5361897945404053, "learning_rate": 4.039858068859787e-06, "loss": 0.12525197863578796, "step": 2517 }, { "epoch": 0.6536863966770509, "grad_norm": 0.9308905601501465, "learning_rate": 4.039082826468511e-06, "loss": 0.11331382393836975, "step": 2518 }, { "epoch": 0.6539460020768432, "grad_norm": 1.328322172164917, "learning_rate": 4.03830735619691e-06, "loss": 0.11468847841024399, "step": 2519 }, { "epoch": 0.6542056074766355, "grad_norm": 2.1684155464172363, "learning_rate": 4.037531658182065e-06, "loss": 0.11607905477285385, "step": 2520 }, { "epoch": 0.6544652128764278, "grad_norm": 1.9438318014144897, "learning_rate": 4.036755732561103e-06, "loss": 0.10457207262516022, "step": 2521 }, { "epoch": 0.6547248182762202, "grad_norm": 1.215828537940979, "learning_rate": 4.035979579471192e-06, "loss": 0.12406733632087708, "step": 2522 }, { "epoch": 0.6549844236760125, "grad_norm": 1.2657678127288818, "learning_rate": 4.035203199049536e-06, "loss": 0.10011093318462372, "step": 2523 }, { "epoch": 0.6552440290758048, "grad_norm": 1.740829586982727, "learning_rate": 4.0344265914333805e-06, "loss": 0.0986378937959671, "step": 2524 }, { "epoch": 0.6555036344755971, "grad_norm": 3.664623975753784, "learning_rate": 4.033649756760014e-06, "loss": 0.09761594235897064, "step": 2525 }, { "epoch": 0.6557632398753894, "grad_norm": 2.3912603855133057, "learning_rate": 4.032872695166763e-06, "loss": 0.13632738590240479, "step": 2526 }, { "epoch": 0.6560228452751817, "grad_norm": 1.2451223134994507, "learning_rate": 4.032095406790992e-06, "loss": 0.10751524567604065, "step": 2527 }, { "epoch": 0.656282450674974, "grad_norm": 3.647181510925293, "learning_rate": 4.031317891770109e-06, "loss": 0.13492053747177124, "step": 2528 }, { "epoch": 0.6565420560747663, "grad_norm": 2.8514559268951416, "learning_rate": 4.030540150241563e-06, "loss": 0.11911842226982117, "step": 2529 }, { "epoch": 0.6568016614745587, "grad_norm": 1.3528244495391846, "learning_rate": 4.029762182342836e-06, "loss": 0.112612783908844, "step": 2530 }, { "epoch": 0.657061266874351, "grad_norm": 1.0197975635528564, "learning_rate": 4.02898398821146e-06, "loss": 0.12258186936378479, "step": 2531 }, { "epoch": 0.6573208722741433, "grad_norm": 3.8307137489318848, "learning_rate": 4.028205567984999e-06, "loss": 0.11409826576709747, "step": 2532 }, { "epoch": 0.6575804776739356, "grad_norm": 1.8107401132583618, "learning_rate": 4.0274269218010606e-06, "loss": 0.11756095290184021, "step": 2533 }, { "epoch": 0.657840083073728, "grad_norm": 1.4164175987243652, "learning_rate": 4.026648049797292e-06, "loss": 0.1083962544798851, "step": 2534 }, { "epoch": 0.6580996884735203, "grad_norm": 2.3728556632995605, "learning_rate": 4.025868952111379e-06, "loss": 0.13030444085597992, "step": 2535 }, { "epoch": 0.6583592938733126, "grad_norm": 1.8345351219177246, "learning_rate": 4.025089628881049e-06, "loss": 0.11407865583896637, "step": 2536 }, { "epoch": 0.6586188992731049, "grad_norm": 1.8815163373947144, "learning_rate": 4.024310080244069e-06, "loss": 0.1164185181260109, "step": 2537 }, { "epoch": 0.6588785046728972, "grad_norm": 2.582616090774536, "learning_rate": 4.023530306338244e-06, "loss": 0.1335882544517517, "step": 2538 }, { "epoch": 0.6591381100726895, "grad_norm": 2.236693859100342, "learning_rate": 4.02275030730142e-06, "loss": 0.12160185724496841, "step": 2539 }, { "epoch": 0.6593977154724818, "grad_norm": 0.9494959712028503, "learning_rate": 4.021970083271486e-06, "loss": 0.11365482956171036, "step": 2540 }, { "epoch": 0.6596573208722741, "grad_norm": 1.511730432510376, "learning_rate": 4.021189634386363e-06, "loss": 0.1085856705904007, "step": 2541 }, { "epoch": 0.6599169262720664, "grad_norm": 2.8628976345062256, "learning_rate": 4.0204089607840204e-06, "loss": 0.12283358722925186, "step": 2542 }, { "epoch": 0.6601765316718587, "grad_norm": 1.3843765258789062, "learning_rate": 4.019628062602462e-06, "loss": 0.12155701220035553, "step": 2543 }, { "epoch": 0.660436137071651, "grad_norm": 1.3274738788604736, "learning_rate": 4.018846939979732e-06, "loss": 0.11254669725894928, "step": 2544 }, { "epoch": 0.6606957424714434, "grad_norm": 1.2014065980911255, "learning_rate": 4.018065593053917e-06, "loss": 0.09983876347541809, "step": 2545 }, { "epoch": 0.6609553478712358, "grad_norm": 1.5050395727157593, "learning_rate": 4.01728402196314e-06, "loss": 0.11755560338497162, "step": 2546 }, { "epoch": 0.6612149532710281, "grad_norm": 1.860675573348999, "learning_rate": 4.016502226845564e-06, "loss": 0.10690553486347198, "step": 2547 }, { "epoch": 0.6614745586708204, "grad_norm": 1.609434962272644, "learning_rate": 4.015720207839394e-06, "loss": 0.10502642393112183, "step": 2548 }, { "epoch": 0.6617341640706127, "grad_norm": 1.5401028394699097, "learning_rate": 4.014937965082871e-06, "loss": 0.10312378406524658, "step": 2549 }, { "epoch": 0.661993769470405, "grad_norm": 1.7936851978302002, "learning_rate": 4.014155498714281e-06, "loss": 0.10618877410888672, "step": 2550 }, { "epoch": 0.6622533748701973, "grad_norm": 1.8816550970077515, "learning_rate": 4.0133728088719425e-06, "loss": 0.10584039986133575, "step": 2551 }, { "epoch": 0.6625129802699896, "grad_norm": 3.1434319019317627, "learning_rate": 4.0125898956942195e-06, "loss": 0.10907801985740662, "step": 2552 }, { "epoch": 0.6627725856697819, "grad_norm": 0.980100691318512, "learning_rate": 4.011806759319513e-06, "loss": 0.11558353155851364, "step": 2553 }, { "epoch": 0.6630321910695742, "grad_norm": 3.1917734146118164, "learning_rate": 4.011023399886262e-06, "loss": 0.09929382055997849, "step": 2554 }, { "epoch": 0.6632917964693665, "grad_norm": 2.7325525283813477, "learning_rate": 4.010239817532947e-06, "loss": 0.1261742264032364, "step": 2555 }, { "epoch": 0.6635514018691588, "grad_norm": 2.719470500946045, "learning_rate": 4.009456012398089e-06, "loss": 0.11954813450574875, "step": 2556 }, { "epoch": 0.6638110072689511, "grad_norm": 3.175814628601074, "learning_rate": 4.008671984620246e-06, "loss": 0.1286437213420868, "step": 2557 }, { "epoch": 0.6640706126687436, "grad_norm": 3.203566551208496, "learning_rate": 4.007887734338016e-06, "loss": 0.12528444826602936, "step": 2558 }, { "epoch": 0.6643302180685359, "grad_norm": 2.511192798614502, "learning_rate": 4.007103261690035e-06, "loss": 0.08088205754756927, "step": 2559 }, { "epoch": 0.6645898234683282, "grad_norm": 2.499647855758667, "learning_rate": 4.006318566814982e-06, "loss": 0.1188487559556961, "step": 2560 }, { "epoch": 0.6648494288681205, "grad_norm": 2.1262450218200684, "learning_rate": 4.005533649851573e-06, "loss": 0.09296159446239471, "step": 2561 }, { "epoch": 0.6651090342679128, "grad_norm": 4.788316249847412, "learning_rate": 4.004748510938562e-06, "loss": 0.15512686967849731, "step": 2562 }, { "epoch": 0.6653686396677051, "grad_norm": 2.9229769706726074, "learning_rate": 4.003963150214743e-06, "loss": 0.11328081786632538, "step": 2563 }, { "epoch": 0.6656282450674974, "grad_norm": 1.0438770055770874, "learning_rate": 4.003177567818952e-06, "loss": 0.09974609315395355, "step": 2564 }, { "epoch": 0.6658878504672897, "grad_norm": 3.929334878921509, "learning_rate": 4.002391763890062e-06, "loss": 0.11989568918943405, "step": 2565 }, { "epoch": 0.666147455867082, "grad_norm": 3.453537940979004, "learning_rate": 4.001605738566983e-06, "loss": 0.10123199969530106, "step": 2566 }, { "epoch": 0.6664070612668743, "grad_norm": 3.170307159423828, "learning_rate": 4.000819491988667e-06, "loss": 0.090296670794487, "step": 2567 }, { "epoch": 0.6666666666666666, "grad_norm": 1.5022077560424805, "learning_rate": 4.000033024294105e-06, "loss": 0.10458296537399292, "step": 2568 }, { "epoch": 0.6669262720664589, "grad_norm": 1.5640987157821655, "learning_rate": 3.999246335622327e-06, "loss": 0.10872894525527954, "step": 2569 }, { "epoch": 0.6671858774662512, "grad_norm": 1.6579768657684326, "learning_rate": 3.9984594261124e-06, "loss": 0.1081191822886467, "step": 2570 }, { "epoch": 0.6674454828660437, "grad_norm": 1.0389959812164307, "learning_rate": 3.997672295903433e-06, "loss": 0.1148475855588913, "step": 2571 }, { "epoch": 0.667705088265836, "grad_norm": 3.9437546730041504, "learning_rate": 3.996884945134572e-06, "loss": 0.09181588143110275, "step": 2572 }, { "epoch": 0.6679646936656283, "grad_norm": 3.722720146179199, "learning_rate": 3.996097373945003e-06, "loss": 0.1291065216064453, "step": 2573 }, { "epoch": 0.6682242990654206, "grad_norm": 3.9985880851745605, "learning_rate": 3.995309582473949e-06, "loss": 0.08930882066488266, "step": 2574 }, { "epoch": 0.6684839044652129, "grad_norm": 2.8860883712768555, "learning_rate": 3.994521570860676e-06, "loss": 0.1267857700586319, "step": 2575 }, { "epoch": 0.6687435098650052, "grad_norm": 2.4862866401672363, "learning_rate": 3.993733339244484e-06, "loss": 0.11788765341043472, "step": 2576 }, { "epoch": 0.6690031152647975, "grad_norm": 2.969118356704712, "learning_rate": 3.992944887764715e-06, "loss": 0.11653921008110046, "step": 2577 }, { "epoch": 0.6692627206645898, "grad_norm": 1.814808964729309, "learning_rate": 3.9921562165607494e-06, "loss": 0.12251631915569305, "step": 2578 }, { "epoch": 0.6695223260643821, "grad_norm": 2.297724962234497, "learning_rate": 3.991367325772007e-06, "loss": 0.10372909903526306, "step": 2579 }, { "epoch": 0.6697819314641744, "grad_norm": 2.529895067214966, "learning_rate": 3.990578215537945e-06, "loss": 0.09937204420566559, "step": 2580 }, { "epoch": 0.6700415368639667, "grad_norm": 3.00490403175354, "learning_rate": 3.98978888599806e-06, "loss": 0.13423582911491394, "step": 2581 }, { "epoch": 0.670301142263759, "grad_norm": 4.630871295928955, "learning_rate": 3.988999337291887e-06, "loss": 0.1140131950378418, "step": 2582 }, { "epoch": 0.6705607476635514, "grad_norm": 2.874410390853882, "learning_rate": 3.988209569559001e-06, "loss": 0.07859215140342712, "step": 2583 }, { "epoch": 0.6708203530633438, "grad_norm": 3.455836296081543, "learning_rate": 3.987419582939014e-06, "loss": 0.1089957058429718, "step": 2584 }, { "epoch": 0.6710799584631361, "grad_norm": 1.953749418258667, "learning_rate": 3.986629377571579e-06, "loss": 0.1298307180404663, "step": 2585 }, { "epoch": 0.6713395638629284, "grad_norm": 2.031511068344116, "learning_rate": 3.985838953596383e-06, "loss": 0.11880451440811157, "step": 2586 }, { "epoch": 0.6715991692627207, "grad_norm": 2.653212070465088, "learning_rate": 3.985048311153158e-06, "loss": 0.1482664793729782, "step": 2587 }, { "epoch": 0.671858774662513, "grad_norm": 3.464606523513794, "learning_rate": 3.98425745038167e-06, "loss": 0.12799178063869476, "step": 2588 }, { "epoch": 0.6721183800623053, "grad_norm": 2.4021310806274414, "learning_rate": 3.983466371421727e-06, "loss": 0.10042113065719604, "step": 2589 }, { "epoch": 0.6723779854620976, "grad_norm": 1.7456700801849365, "learning_rate": 3.982675074413171e-06, "loss": 0.10036252439022064, "step": 2590 }, { "epoch": 0.6726375908618899, "grad_norm": 2.956198215484619, "learning_rate": 3.981883559495886e-06, "loss": 0.12006856501102448, "step": 2591 }, { "epoch": 0.6728971962616822, "grad_norm": 2.796142578125, "learning_rate": 3.9810918268097945e-06, "loss": 0.1098552793264389, "step": 2592 }, { "epoch": 0.6731568016614745, "grad_norm": 5.42109489440918, "learning_rate": 3.980299876494857e-06, "loss": 0.12207676470279694, "step": 2593 }, { "epoch": 0.6734164070612668, "grad_norm": 2.1521902084350586, "learning_rate": 3.979507708691071e-06, "loss": 0.11325608193874359, "step": 2594 }, { "epoch": 0.6736760124610592, "grad_norm": 2.7029225826263428, "learning_rate": 3.978715323538474e-06, "loss": 0.11115796118974686, "step": 2595 }, { "epoch": 0.6739356178608515, "grad_norm": 3.2586941719055176, "learning_rate": 3.977922721177141e-06, "loss": 0.10404950380325317, "step": 2596 }, { "epoch": 0.6741952232606438, "grad_norm": 2.387798547744751, "learning_rate": 3.977129901747187e-06, "loss": 0.112629234790802, "step": 2597 }, { "epoch": 0.6744548286604362, "grad_norm": 2.84840726852417, "learning_rate": 3.976336865388765e-06, "loss": 0.12621358036994934, "step": 2598 }, { "epoch": 0.6747144340602285, "grad_norm": 3.2210731506347656, "learning_rate": 3.975543612242063e-06, "loss": 0.12330032885074615, "step": 2599 }, { "epoch": 0.6749740394600208, "grad_norm": 3.0510337352752686, "learning_rate": 3.974750142447312e-06, "loss": 0.11289796978235245, "step": 2600 }, { "epoch": 0.6752336448598131, "grad_norm": 1.3065929412841797, "learning_rate": 3.97395645614478e-06, "loss": 0.09689034521579742, "step": 2601 }, { "epoch": 0.6754932502596054, "grad_norm": 2.035322904586792, "learning_rate": 3.97316255347477e-06, "loss": 0.11441610008478165, "step": 2602 }, { "epoch": 0.6757528556593977, "grad_norm": 2.4474737644195557, "learning_rate": 3.972368434577628e-06, "loss": 0.11457526683807373, "step": 2603 }, { "epoch": 0.67601246105919, "grad_norm": 3.2529900074005127, "learning_rate": 3.971574099593735e-06, "loss": 0.1098138689994812, "step": 2604 }, { "epoch": 0.6762720664589823, "grad_norm": 2.2140188217163086, "learning_rate": 3.970779548663512e-06, "loss": 0.11324682831764221, "step": 2605 }, { "epoch": 0.6765316718587746, "grad_norm": 3.971703290939331, "learning_rate": 3.969984781927417e-06, "loss": 0.12519720196723938, "step": 2606 }, { "epoch": 0.676791277258567, "grad_norm": 2.3497772216796875, "learning_rate": 3.969189799525946e-06, "loss": 0.11125382781028748, "step": 2607 }, { "epoch": 0.6770508826583593, "grad_norm": 3.0356247425079346, "learning_rate": 3.9683946015996335e-06, "loss": 0.13027819991111755, "step": 2608 }, { "epoch": 0.6773104880581516, "grad_norm": 1.7666560411453247, "learning_rate": 3.967599188289054e-06, "loss": 0.09091038256883621, "step": 2609 }, { "epoch": 0.677570093457944, "grad_norm": 2.2616565227508545, "learning_rate": 3.966803559734818e-06, "loss": 0.12095437198877335, "step": 2610 }, { "epoch": 0.6778296988577363, "grad_norm": 2.0910024642944336, "learning_rate": 3.966007716077572e-06, "loss": 0.12436547130346298, "step": 2611 }, { "epoch": 0.6780893042575286, "grad_norm": 4.177242755889893, "learning_rate": 3.965211657458006e-06, "loss": 0.11328330636024475, "step": 2612 }, { "epoch": 0.6783489096573209, "grad_norm": 2.606191873550415, "learning_rate": 3.964415384016843e-06, "loss": 0.11344914138317108, "step": 2613 }, { "epoch": 0.6786085150571132, "grad_norm": 2.51713490486145, "learning_rate": 3.963618895894847e-06, "loss": 0.106154665350914, "step": 2614 }, { "epoch": 0.6788681204569055, "grad_norm": 2.5472183227539062, "learning_rate": 3.962822193232818e-06, "loss": 0.10488792508840561, "step": 2615 }, { "epoch": 0.6791277258566978, "grad_norm": 4.685958385467529, "learning_rate": 3.9620252761715955e-06, "loss": 0.1235046535730362, "step": 2616 }, { "epoch": 0.6793873312564901, "grad_norm": 1.7311919927597046, "learning_rate": 3.961228144852056e-06, "loss": 0.12205274403095245, "step": 2617 }, { "epoch": 0.6796469366562824, "grad_norm": 1.3667131662368774, "learning_rate": 3.960430799415114e-06, "loss": 0.11766530573368073, "step": 2618 }, { "epoch": 0.6799065420560748, "grad_norm": 2.9251410961151123, "learning_rate": 3.959633240001721e-06, "loss": 0.08090227842330933, "step": 2619 }, { "epoch": 0.6801661474558671, "grad_norm": 1.7775018215179443, "learning_rate": 3.958835466752868e-06, "loss": 0.09813238680362701, "step": 2620 }, { "epoch": 0.6804257528556594, "grad_norm": 3.7087953090667725, "learning_rate": 3.958037479809583e-06, "loss": 0.12118583917617798, "step": 2621 }, { "epoch": 0.6806853582554517, "grad_norm": 3.149655818939209, "learning_rate": 3.957239279312934e-06, "loss": 0.09863680601119995, "step": 2622 }, { "epoch": 0.680944963655244, "grad_norm": 3.1960041522979736, "learning_rate": 3.9564408654040206e-06, "loss": 0.1142585426568985, "step": 2623 }, { "epoch": 0.6812045690550363, "grad_norm": 3.171276569366455, "learning_rate": 3.955642238223987e-06, "loss": 0.11762286722660065, "step": 2624 }, { "epoch": 0.6814641744548287, "grad_norm": 3.3923473358154297, "learning_rate": 3.954843397914011e-06, "loss": 0.10101041942834854, "step": 2625 }, { "epoch": 0.681723779854621, "grad_norm": 2.9447014331817627, "learning_rate": 3.954044344615309e-06, "loss": 0.09653384238481522, "step": 2626 }, { "epoch": 0.6819833852544133, "grad_norm": 1.9956095218658447, "learning_rate": 3.953245078469137e-06, "loss": 0.10890171676874161, "step": 2627 }, { "epoch": 0.6822429906542056, "grad_norm": 2.051238536834717, "learning_rate": 3.952445599616784e-06, "loss": 0.10529213398694992, "step": 2628 }, { "epoch": 0.6825025960539979, "grad_norm": 3.331047296524048, "learning_rate": 3.951645908199583e-06, "loss": 0.11239215731620789, "step": 2629 }, { "epoch": 0.6827622014537902, "grad_norm": 1.6540536880493164, "learning_rate": 3.950846004358899e-06, "loss": 0.11072152853012085, "step": 2630 }, { "epoch": 0.6830218068535826, "grad_norm": 2.8678295612335205, "learning_rate": 3.950045888236137e-06, "loss": 0.11786500364542007, "step": 2631 }, { "epoch": 0.6832814122533749, "grad_norm": 0.8013585209846497, "learning_rate": 3.949245559972739e-06, "loss": 0.12149693071842194, "step": 2632 }, { "epoch": 0.6835410176531672, "grad_norm": 1.2956955432891846, "learning_rate": 3.9484450197101856e-06, "loss": 0.10475534200668335, "step": 2633 }, { "epoch": 0.6838006230529595, "grad_norm": 2.934264659881592, "learning_rate": 3.947644267589993e-06, "loss": 0.12441268563270569, "step": 2634 }, { "epoch": 0.6840602284527518, "grad_norm": 2.363128185272217, "learning_rate": 3.946843303753716e-06, "loss": 0.11504265666007996, "step": 2635 }, { "epoch": 0.6843198338525441, "grad_norm": 3.328765630722046, "learning_rate": 3.946042128342948e-06, "loss": 0.12339580804109573, "step": 2636 }, { "epoch": 0.6845794392523364, "grad_norm": 1.3328173160552979, "learning_rate": 3.945240741499317e-06, "loss": 0.11678212881088257, "step": 2637 }, { "epoch": 0.6848390446521287, "grad_norm": 2.3660476207733154, "learning_rate": 3.94443914336449e-06, "loss": 0.08063194900751114, "step": 2638 }, { "epoch": 0.685098650051921, "grad_norm": 2.600555419921875, "learning_rate": 3.9436373340801724e-06, "loss": 0.1096864864230156, "step": 2639 }, { "epoch": 0.6853582554517134, "grad_norm": 3.0013985633850098, "learning_rate": 3.9428353137881034e-06, "loss": 0.12625908851623535, "step": 2640 }, { "epoch": 0.6856178608515057, "grad_norm": 3.3900694847106934, "learning_rate": 3.942033082630064e-06, "loss": 0.11090248823165894, "step": 2641 }, { "epoch": 0.685877466251298, "grad_norm": 3.1159591674804688, "learning_rate": 3.94123064074787e-06, "loss": 0.11254864931106567, "step": 2642 }, { "epoch": 0.6861370716510904, "grad_norm": 2.64243221282959, "learning_rate": 3.940427988283374e-06, "loss": 0.09506280720233917, "step": 2643 }, { "epoch": 0.6863966770508827, "grad_norm": 2.895455837249756, "learning_rate": 3.939625125378467e-06, "loss": 0.08755673468112946, "step": 2644 }, { "epoch": 0.686656282450675, "grad_norm": 3.283888816833496, "learning_rate": 3.938822052175076e-06, "loss": 0.1212029755115509, "step": 2645 }, { "epoch": 0.6869158878504673, "grad_norm": 2.1119210720062256, "learning_rate": 3.9380187688151664e-06, "loss": 0.1002790555357933, "step": 2646 }, { "epoch": 0.6871754932502596, "grad_norm": 3.4089527130126953, "learning_rate": 3.9372152754407415e-06, "loss": 0.10760033875703812, "step": 2647 }, { "epoch": 0.6874350986500519, "grad_norm": 3.030773878097534, "learning_rate": 3.936411572193839e-06, "loss": 0.0960741639137268, "step": 2648 }, { "epoch": 0.6876947040498442, "grad_norm": 1.7637555599212646, "learning_rate": 3.9356076592165355e-06, "loss": 0.12618905305862427, "step": 2649 }, { "epoch": 0.6879543094496365, "grad_norm": 2.2217659950256348, "learning_rate": 3.934803536650945e-06, "loss": 0.11070047318935394, "step": 2650 }, { "epoch": 0.6882139148494288, "grad_norm": 4.477643013000488, "learning_rate": 3.933999204639218e-06, "loss": 0.12691615521907806, "step": 2651 }, { "epoch": 0.6884735202492211, "grad_norm": 2.253926992416382, "learning_rate": 3.933194663323541e-06, "loss": 0.09993799030780792, "step": 2652 }, { "epoch": 0.6887331256490135, "grad_norm": 2.337456226348877, "learning_rate": 3.93238991284614e-06, "loss": 0.10869361460208893, "step": 2653 }, { "epoch": 0.6889927310488058, "grad_norm": 4.360559463500977, "learning_rate": 3.931584953349274e-06, "loss": 0.10963881015777588, "step": 2654 }, { "epoch": 0.6892523364485982, "grad_norm": 1.6033985614776611, "learning_rate": 3.930779784975244e-06, "loss": 0.10224384814500809, "step": 2655 }, { "epoch": 0.6895119418483905, "grad_norm": 2.630526304244995, "learning_rate": 3.929974407866384e-06, "loss": 0.10832780599594116, "step": 2656 }, { "epoch": 0.6897715472481828, "grad_norm": 3.8921124935150146, "learning_rate": 3.929168822165067e-06, "loss": 0.11654221266508102, "step": 2657 }, { "epoch": 0.6900311526479751, "grad_norm": 3.119436025619507, "learning_rate": 3.9283630280137e-06, "loss": 0.0849187970161438, "step": 2658 }, { "epoch": 0.6902907580477674, "grad_norm": 2.6142091751098633, "learning_rate": 3.927557025554733e-06, "loss": 0.09063668549060822, "step": 2659 }, { "epoch": 0.6905503634475597, "grad_norm": 3.1451215744018555, "learning_rate": 3.926750814930645e-06, "loss": 0.08614689856767654, "step": 2660 }, { "epoch": 0.690809968847352, "grad_norm": 2.2006826400756836, "learning_rate": 3.925944396283957e-06, "loss": 0.12453324347734451, "step": 2661 }, { "epoch": 0.6910695742471443, "grad_norm": 3.414482355117798, "learning_rate": 3.9251377697572254e-06, "loss": 0.11563445627689362, "step": 2662 }, { "epoch": 0.6913291796469366, "grad_norm": 1.4121438264846802, "learning_rate": 3.9243309354930426e-06, "loss": 0.11474259942770004, "step": 2663 }, { "epoch": 0.6915887850467289, "grad_norm": 1.9253084659576416, "learning_rate": 3.9235238936340405e-06, "loss": 0.09055926650762558, "step": 2664 }, { "epoch": 0.6918483904465212, "grad_norm": 1.5148428678512573, "learning_rate": 3.922716644322883e-06, "loss": 0.10200954973697662, "step": 2665 }, { "epoch": 0.6921079958463136, "grad_norm": 3.0517616271972656, "learning_rate": 3.921909187702275e-06, "loss": 0.08936239778995514, "step": 2666 }, { "epoch": 0.692367601246106, "grad_norm": 3.3581254482269287, "learning_rate": 3.921101523914956e-06, "loss": 0.09772644937038422, "step": 2667 }, { "epoch": 0.6926272066458983, "grad_norm": 1.8416236639022827, "learning_rate": 3.920293653103701e-06, "loss": 0.10802595317363739, "step": 2668 }, { "epoch": 0.6928868120456906, "grad_norm": 1.813161015510559, "learning_rate": 3.9194855754113245e-06, "loss": 0.1437072455883026, "step": 2669 }, { "epoch": 0.6931464174454829, "grad_norm": 3.2023050785064697, "learning_rate": 3.918677290980678e-06, "loss": 0.1249716654419899, "step": 2670 }, { "epoch": 0.6934060228452752, "grad_norm": 2.8135323524475098, "learning_rate": 3.917868799954644e-06, "loss": 0.08249448984861374, "step": 2671 }, { "epoch": 0.6936656282450675, "grad_norm": 1.044467568397522, "learning_rate": 3.9170601024761465e-06, "loss": 0.10782447457313538, "step": 2672 }, { "epoch": 0.6939252336448598, "grad_norm": 2.407503366470337, "learning_rate": 3.916251198688145e-06, "loss": 0.10207574814558029, "step": 2673 }, { "epoch": 0.6941848390446521, "grad_norm": 2.5268709659576416, "learning_rate": 3.915442088733636e-06, "loss": 0.08146600425243378, "step": 2674 }, { "epoch": 0.6944444444444444, "grad_norm": 4.809574604034424, "learning_rate": 3.9146327727556504e-06, "loss": 0.12743082642555237, "step": 2675 }, { "epoch": 0.6947040498442367, "grad_norm": 2.1691954135894775, "learning_rate": 3.913823250897257e-06, "loss": 0.1136183887720108, "step": 2676 }, { "epoch": 0.694963655244029, "grad_norm": 3.182213544845581, "learning_rate": 3.913013523301561e-06, "loss": 0.11043563485145569, "step": 2677 }, { "epoch": 0.6952232606438213, "grad_norm": 2.8202497959136963, "learning_rate": 3.912203590111703e-06, "loss": 0.13589802384376526, "step": 2678 }, { "epoch": 0.6954828660436138, "grad_norm": 1.5701074600219727, "learning_rate": 3.911393451470863e-06, "loss": 0.10338018089532852, "step": 2679 }, { "epoch": 0.6957424714434061, "grad_norm": 1.9492064714431763, "learning_rate": 3.910583107522251e-06, "loss": 0.11314769089221954, "step": 2680 }, { "epoch": 0.6960020768431984, "grad_norm": 1.0480645895004272, "learning_rate": 3.90977255840912e-06, "loss": 0.12417177855968475, "step": 2681 }, { "epoch": 0.6962616822429907, "grad_norm": 3.4050750732421875, "learning_rate": 3.9089618042747564e-06, "loss": 0.11466772109270096, "step": 2682 }, { "epoch": 0.696521287642783, "grad_norm": 2.232011079788208, "learning_rate": 3.908150845262483e-06, "loss": 0.09762606769800186, "step": 2683 }, { "epoch": 0.6967808930425753, "grad_norm": 3.14617919921875, "learning_rate": 3.907339681515657e-06, "loss": 0.09132993221282959, "step": 2684 }, { "epoch": 0.6970404984423676, "grad_norm": 2.96366810798645, "learning_rate": 3.906528313177676e-06, "loss": 0.11632534116506577, "step": 2685 }, { "epoch": 0.6973001038421599, "grad_norm": 3.9088239669799805, "learning_rate": 3.90571674039197e-06, "loss": 0.08656269311904907, "step": 2686 }, { "epoch": 0.6975597092419522, "grad_norm": 1.1608498096466064, "learning_rate": 3.904904963302007e-06, "loss": 0.09911273419857025, "step": 2687 }, { "epoch": 0.6978193146417445, "grad_norm": 3.6878600120544434, "learning_rate": 3.904092982051289e-06, "loss": 0.07655616849660873, "step": 2688 }, { "epoch": 0.6980789200415368, "grad_norm": 3.1951823234558105, "learning_rate": 3.903280796783358e-06, "loss": 0.1028485894203186, "step": 2689 }, { "epoch": 0.6983385254413291, "grad_norm": 3.1112592220306396, "learning_rate": 3.902468407641788e-06, "loss": 0.11205048114061356, "step": 2690 }, { "epoch": 0.6985981308411215, "grad_norm": 2.7462382316589355, "learning_rate": 3.901655814770193e-06, "loss": 0.12625761330127716, "step": 2691 }, { "epoch": 0.6988577362409139, "grad_norm": 3.1726489067077637, "learning_rate": 3.900843018312218e-06, "loss": 0.09526357054710388, "step": 2692 }, { "epoch": 0.6991173416407062, "grad_norm": 1.6048794984817505, "learning_rate": 3.90003001841155e-06, "loss": 0.12058861553668976, "step": 2693 }, { "epoch": 0.6993769470404985, "grad_norm": 2.443143606185913, "learning_rate": 3.899216815211905e-06, "loss": 0.11622850596904755, "step": 2694 }, { "epoch": 0.6996365524402908, "grad_norm": 1.3951971530914307, "learning_rate": 3.898403408857042e-06, "loss": 0.10427868366241455, "step": 2695 }, { "epoch": 0.6998961578400831, "grad_norm": 1.499351143836975, "learning_rate": 3.897589799490749e-06, "loss": 0.11002891510725021, "step": 2696 }, { "epoch": 0.7001557632398754, "grad_norm": 1.0499008893966675, "learning_rate": 3.8967759872568584e-06, "loss": 0.09546752274036407, "step": 2697 }, { "epoch": 0.7004153686396677, "grad_norm": 2.8151957988739014, "learning_rate": 3.89596197229923e-06, "loss": 0.09896614402532578, "step": 2698 }, { "epoch": 0.70067497403946, "grad_norm": 1.2806812524795532, "learning_rate": 3.895147754761763e-06, "loss": 0.1064363420009613, "step": 2699 }, { "epoch": 0.7009345794392523, "grad_norm": 0.9766316413879395, "learning_rate": 3.894333334788395e-06, "loss": 0.12207628786563873, "step": 2700 }, { "epoch": 0.7011941848390446, "grad_norm": 2.745859146118164, "learning_rate": 3.893518712523094e-06, "loss": 0.10138404369354248, "step": 2701 }, { "epoch": 0.7014537902388369, "grad_norm": 3.0323498249053955, "learning_rate": 3.8927038881098675e-06, "loss": 0.09598201513290405, "step": 2702 }, { "epoch": 0.7017133956386293, "grad_norm": 2.790902853012085, "learning_rate": 3.891888861692759e-06, "loss": 0.10267733037471771, "step": 2703 }, { "epoch": 0.7019730010384216, "grad_norm": 2.3155696392059326, "learning_rate": 3.891073633415845e-06, "loss": 0.10589943081140518, "step": 2704 }, { "epoch": 0.702232606438214, "grad_norm": 2.697683334350586, "learning_rate": 3.89025820342324e-06, "loss": 0.135154590010643, "step": 2705 }, { "epoch": 0.7024922118380063, "grad_norm": 2.2222185134887695, "learning_rate": 3.889442571859093e-06, "loss": 0.1032896339893341, "step": 2706 }, { "epoch": 0.7027518172377986, "grad_norm": 1.591623306274414, "learning_rate": 3.8886267388675885e-06, "loss": 0.11180000007152557, "step": 2707 }, { "epoch": 0.7030114226375909, "grad_norm": 1.0104601383209229, "learning_rate": 3.887810704592949e-06, "loss": 0.11815489083528519, "step": 2708 }, { "epoch": 0.7032710280373832, "grad_norm": 0.9492846131324768, "learning_rate": 3.886994469179429e-06, "loss": 0.11300129443407059, "step": 2709 }, { "epoch": 0.7035306334371755, "grad_norm": 3.2905354499816895, "learning_rate": 3.886178032771319e-06, "loss": 0.08354391157627106, "step": 2710 }, { "epoch": 0.7037902388369678, "grad_norm": 1.3063640594482422, "learning_rate": 3.8853613955129495e-06, "loss": 0.09289152175188065, "step": 2711 }, { "epoch": 0.7040498442367601, "grad_norm": 1.3409162759780884, "learning_rate": 3.88454455754868e-06, "loss": 0.09768141061067581, "step": 2712 }, { "epoch": 0.7043094496365524, "grad_norm": 1.5081431865692139, "learning_rate": 3.883727519022912e-06, "loss": 0.11043613404035568, "step": 2713 }, { "epoch": 0.7045690550363447, "grad_norm": 3.3246800899505615, "learning_rate": 3.882910280080077e-06, "loss": 0.09746856987476349, "step": 2714 }, { "epoch": 0.7048286604361371, "grad_norm": 1.7346038818359375, "learning_rate": 3.882092840864645e-06, "loss": 0.09843267500400543, "step": 2715 }, { "epoch": 0.7050882658359294, "grad_norm": 0.9429997205734253, "learning_rate": 3.88127520152112e-06, "loss": 0.12445928901433945, "step": 2716 }, { "epoch": 0.7053478712357217, "grad_norm": 3.6501567363739014, "learning_rate": 3.880457362194041e-06, "loss": 0.09593748301267624, "step": 2717 }, { "epoch": 0.705607476635514, "grad_norm": 1.7931749820709229, "learning_rate": 3.879639323027986e-06, "loss": 0.10112689435482025, "step": 2718 }, { "epoch": 0.7058670820353063, "grad_norm": 2.3948798179626465, "learning_rate": 3.878821084167563e-06, "loss": 0.09171615540981293, "step": 2719 }, { "epoch": 0.7061266874350987, "grad_norm": 2.6039113998413086, "learning_rate": 3.8780026457574195e-06, "loss": 0.10850902646780014, "step": 2720 }, { "epoch": 0.706386292834891, "grad_norm": 1.6677395105361938, "learning_rate": 3.877184007942236e-06, "loss": 0.10340455919504166, "step": 2721 }, { "epoch": 0.7066458982346833, "grad_norm": 1.287675380706787, "learning_rate": 3.8763651708667285e-06, "loss": 0.10742147266864777, "step": 2722 }, { "epoch": 0.7069055036344756, "grad_norm": 0.9757656455039978, "learning_rate": 3.8755461346756505e-06, "loss": 0.11343758553266525, "step": 2723 }, { "epoch": 0.7071651090342679, "grad_norm": 2.0869743824005127, "learning_rate": 3.874726899513786e-06, "loss": 0.07621191442012787, "step": 2724 }, { "epoch": 0.7074247144340602, "grad_norm": 2.225184917449951, "learning_rate": 3.873907465525959e-06, "loss": 0.10701709985733032, "step": 2725 }, { "epoch": 0.7076843198338525, "grad_norm": 0.9851828813552856, "learning_rate": 3.873087832857026e-06, "loss": 0.10371214151382446, "step": 2726 }, { "epoch": 0.7079439252336449, "grad_norm": 2.771995782852173, "learning_rate": 3.8722680016518796e-06, "loss": 0.09991337358951569, "step": 2727 }, { "epoch": 0.7082035306334372, "grad_norm": 1.51675283908844, "learning_rate": 3.8714479720554475e-06, "loss": 0.09132906794548035, "step": 2728 }, { "epoch": 0.7084631360332295, "grad_norm": 2.024951696395874, "learning_rate": 3.870627744212691e-06, "loss": 0.08495685458183289, "step": 2729 }, { "epoch": 0.7087227414330218, "grad_norm": 0.9430338144302368, "learning_rate": 3.869807318268609e-06, "loss": 0.11164765059947968, "step": 2730 }, { "epoch": 0.7089823468328141, "grad_norm": 1.341825008392334, "learning_rate": 3.868986694368232e-06, "loss": 0.1044355034828186, "step": 2731 }, { "epoch": 0.7092419522326064, "grad_norm": 3.114211320877075, "learning_rate": 3.8681658726566305e-06, "loss": 0.10809481143951416, "step": 2732 }, { "epoch": 0.7095015576323987, "grad_norm": 1.608161449432373, "learning_rate": 3.8673448532789044e-06, "loss": 0.08855509757995605, "step": 2733 }, { "epoch": 0.7097611630321911, "grad_norm": 1.727489948272705, "learning_rate": 3.8665236363801915e-06, "loss": 0.09785249084234238, "step": 2734 }, { "epoch": 0.7100207684319834, "grad_norm": 2.8596251010894775, "learning_rate": 3.865702222105666e-06, "loss": 0.06305670738220215, "step": 2735 }, { "epoch": 0.7102803738317757, "grad_norm": 2.7338664531707764, "learning_rate": 3.8648806106005335e-06, "loss": 0.12600263953208923, "step": 2736 }, { "epoch": 0.710539979231568, "grad_norm": 2.2918648719787598, "learning_rate": 3.864058802010036e-06, "loss": 0.1046922579407692, "step": 2737 }, { "epoch": 0.7107995846313603, "grad_norm": 4.732938766479492, "learning_rate": 3.86323679647945e-06, "loss": 0.10324355959892273, "step": 2738 }, { "epoch": 0.7110591900311527, "grad_norm": 1.2945938110351562, "learning_rate": 3.8624145941540896e-06, "loss": 0.10020795464515686, "step": 2739 }, { "epoch": 0.711318795430945, "grad_norm": 2.175574779510498, "learning_rate": 3.861592195179299e-06, "loss": 0.10870735347270966, "step": 2740 }, { "epoch": 0.7115784008307373, "grad_norm": 1.0714497566223145, "learning_rate": 3.860769599700461e-06, "loss": 0.10265032947063446, "step": 2741 }, { "epoch": 0.7118380062305296, "grad_norm": 3.5055174827575684, "learning_rate": 3.859946807862991e-06, "loss": 0.12934212386608124, "step": 2742 }, { "epoch": 0.7120976116303219, "grad_norm": 3.107671022415161, "learning_rate": 3.8591238198123395e-06, "loss": 0.10194388777017593, "step": 2743 }, { "epoch": 0.7123572170301142, "grad_norm": 3.3174240589141846, "learning_rate": 3.8583006356939926e-06, "loss": 0.09004443138837814, "step": 2744 }, { "epoch": 0.7126168224299065, "grad_norm": 2.8242709636688232, "learning_rate": 3.85747725565347e-06, "loss": 0.10959497094154358, "step": 2745 }, { "epoch": 0.7128764278296988, "grad_norm": 1.4388976097106934, "learning_rate": 3.856653679836326e-06, "loss": 0.1215684562921524, "step": 2746 }, { "epoch": 0.7131360332294912, "grad_norm": 1.6859103441238403, "learning_rate": 3.85582990838815e-06, "loss": 0.1201399639248848, "step": 2747 }, { "epoch": 0.7133956386292835, "grad_norm": 3.046604633331299, "learning_rate": 3.855005941454567e-06, "loss": 0.14437006413936615, "step": 2748 }, { "epoch": 0.7136552440290758, "grad_norm": 1.1807026863098145, "learning_rate": 3.854181779181235e-06, "loss": 0.1019837036728859, "step": 2749 }, { "epoch": 0.7139148494288681, "grad_norm": 3.3923192024230957, "learning_rate": 3.853357421713847e-06, "loss": 0.08163154870271683, "step": 2750 }, { "epoch": 0.7141744548286605, "grad_norm": 2.736819267272949, "learning_rate": 3.852532869198129e-06, "loss": 0.11807844042778015, "step": 2751 }, { "epoch": 0.7144340602284528, "grad_norm": 1.817275881767273, "learning_rate": 3.851708121779846e-06, "loss": 0.10869024693965912, "step": 2752 }, { "epoch": 0.7146936656282451, "grad_norm": 1.0973401069641113, "learning_rate": 3.8508831796047925e-06, "loss": 0.11284659802913666, "step": 2753 }, { "epoch": 0.7149532710280374, "grad_norm": 1.0015244483947754, "learning_rate": 3.850058042818799e-06, "loss": 0.11144551634788513, "step": 2754 }, { "epoch": 0.7152128764278297, "grad_norm": 1.2287354469299316, "learning_rate": 3.849232711567732e-06, "loss": 0.10932505130767822, "step": 2755 }, { "epoch": 0.715472481827622, "grad_norm": 2.8737399578094482, "learning_rate": 3.848407185997491e-06, "loss": 0.1306695193052292, "step": 2756 }, { "epoch": 0.7157320872274143, "grad_norm": 2.491034507751465, "learning_rate": 3.847581466254009e-06, "loss": 0.11873014271259308, "step": 2757 }, { "epoch": 0.7159916926272066, "grad_norm": 2.6802451610565186, "learning_rate": 3.846755552483256e-06, "loss": 0.08988969027996063, "step": 2758 }, { "epoch": 0.7162512980269989, "grad_norm": 0.9746320843696594, "learning_rate": 3.845929444831234e-06, "loss": 0.10637657344341278, "step": 2759 }, { "epoch": 0.7165109034267912, "grad_norm": 2.7779266834259033, "learning_rate": 3.845103143443978e-06, "loss": 0.10134578496217728, "step": 2760 }, { "epoch": 0.7167705088265836, "grad_norm": 4.0273027420043945, "learning_rate": 3.844276648467561e-06, "loss": 0.08678871393203735, "step": 2761 }, { "epoch": 0.7170301142263759, "grad_norm": 1.5341030359268188, "learning_rate": 3.843449960048089e-06, "loss": 0.1168866828083992, "step": 2762 }, { "epoch": 0.7172897196261683, "grad_norm": 1.3975979089736938, "learning_rate": 3.8426230783317e-06, "loss": 0.1117420569062233, "step": 2763 }, { "epoch": 0.7175493250259606, "grad_norm": 3.1648313999176025, "learning_rate": 3.841796003464569e-06, "loss": 0.13013166189193726, "step": 2764 }, { "epoch": 0.7178089304257529, "grad_norm": 3.816190004348755, "learning_rate": 3.8409687355929035e-06, "loss": 0.07370639592409134, "step": 2765 }, { "epoch": 0.7180685358255452, "grad_norm": 3.4554946422576904, "learning_rate": 3.840141274862945e-06, "loss": 0.12605905532836914, "step": 2766 }, { "epoch": 0.7183281412253375, "grad_norm": 1.6476225852966309, "learning_rate": 3.83931362142097e-06, "loss": 0.10567348450422287, "step": 2767 }, { "epoch": 0.7185877466251298, "grad_norm": 3.0715572834014893, "learning_rate": 3.83848577541329e-06, "loss": 0.10092608630657196, "step": 2768 }, { "epoch": 0.7188473520249221, "grad_norm": 4.422815322875977, "learning_rate": 3.837657736986246e-06, "loss": 0.11601007729768753, "step": 2769 }, { "epoch": 0.7191069574247144, "grad_norm": 3.174771547317505, "learning_rate": 3.8368295062862196e-06, "loss": 0.11194704473018646, "step": 2770 }, { "epoch": 0.7193665628245067, "grad_norm": 3.4004178047180176, "learning_rate": 3.836001083459621e-06, "loss": 0.12373678386211395, "step": 2771 }, { "epoch": 0.719626168224299, "grad_norm": 3.465616226196289, "learning_rate": 3.835172468652897e-06, "loss": 0.08862891793251038, "step": 2772 }, { "epoch": 0.7198857736240913, "grad_norm": 1.276238203048706, "learning_rate": 3.834343662012528e-06, "loss": 0.11546702682971954, "step": 2773 }, { "epoch": 0.7201453790238836, "grad_norm": 3.1883797645568848, "learning_rate": 3.833514663685027e-06, "loss": 0.1229027509689331, "step": 2774 }, { "epoch": 0.7204049844236761, "grad_norm": 4.407207489013672, "learning_rate": 3.832685473816944e-06, "loss": 0.11835417151451111, "step": 2775 }, { "epoch": 0.7206645898234684, "grad_norm": 3.478095531463623, "learning_rate": 3.831856092554858e-06, "loss": 0.13172046840190887, "step": 2776 }, { "epoch": 0.7209241952232607, "grad_norm": 2.708688735961914, "learning_rate": 3.8310265200453865e-06, "loss": 0.1013835147023201, "step": 2777 }, { "epoch": 0.721183800623053, "grad_norm": 1.0798612833023071, "learning_rate": 3.830196756435179e-06, "loss": 0.11363302171230316, "step": 2778 }, { "epoch": 0.7214434060228453, "grad_norm": 3.1648411750793457, "learning_rate": 3.829366801870917e-06, "loss": 0.09261544793844223, "step": 2779 }, { "epoch": 0.7217030114226376, "grad_norm": 2.500426769256592, "learning_rate": 3.828536656499321e-06, "loss": 0.09349626302719116, "step": 2780 }, { "epoch": 0.7219626168224299, "grad_norm": 3.2033028602600098, "learning_rate": 3.827706320467137e-06, "loss": 0.10992266237735748, "step": 2781 }, { "epoch": 0.7222222222222222, "grad_norm": 1.9920241832733154, "learning_rate": 3.826875793921152e-06, "loss": 0.1074216216802597, "step": 2782 }, { "epoch": 0.7224818276220145, "grad_norm": 1.41390061378479, "learning_rate": 3.826045077008185e-06, "loss": 0.1100463792681694, "step": 2783 }, { "epoch": 0.7227414330218068, "grad_norm": 4.353673458099365, "learning_rate": 3.825214169875085e-06, "loss": 0.11053430289030075, "step": 2784 }, { "epoch": 0.7230010384215991, "grad_norm": 2.895326614379883, "learning_rate": 3.82438307266874e-06, "loss": 0.09492702782154083, "step": 2785 }, { "epoch": 0.7232606438213914, "grad_norm": 1.526610016822815, "learning_rate": 3.8235517855360685e-06, "loss": 0.11383644491434097, "step": 2786 }, { "epoch": 0.7235202492211839, "grad_norm": 1.482895016670227, "learning_rate": 3.822720308624021e-06, "loss": 0.10609569400548935, "step": 2787 }, { "epoch": 0.7237798546209762, "grad_norm": 2.4683265686035156, "learning_rate": 3.821888642079586e-06, "loss": 0.08029334247112274, "step": 2788 }, { "epoch": 0.7240394600207685, "grad_norm": 2.148895740509033, "learning_rate": 3.821056786049781e-06, "loss": 0.10835231840610504, "step": 2789 }, { "epoch": 0.7242990654205608, "grad_norm": 2.7282063961029053, "learning_rate": 3.8202247406816614e-06, "loss": 0.09512601792812347, "step": 2790 }, { "epoch": 0.7245586708203531, "grad_norm": 2.940929412841797, "learning_rate": 3.819392506122312e-06, "loss": 0.06869596242904663, "step": 2791 }, { "epoch": 0.7248182762201454, "grad_norm": 2.4153518676757812, "learning_rate": 3.8185600825188526e-06, "loss": 0.13563494384288788, "step": 2792 }, { "epoch": 0.7250778816199377, "grad_norm": 2.3171539306640625, "learning_rate": 3.817727470018439e-06, "loss": 0.12293761223554611, "step": 2793 }, { "epoch": 0.72533748701973, "grad_norm": 2.204407215118408, "learning_rate": 3.816894668768255e-06, "loss": 0.09957071393728256, "step": 2794 }, { "epoch": 0.7255970924195223, "grad_norm": 1.3047082424163818, "learning_rate": 3.816061678915523e-06, "loss": 0.11029761284589767, "step": 2795 }, { "epoch": 0.7258566978193146, "grad_norm": 2.597583293914795, "learning_rate": 3.815228500607495e-06, "loss": 0.08363259583711624, "step": 2796 }, { "epoch": 0.7261163032191069, "grad_norm": 3.276848316192627, "learning_rate": 3.8143951339914596e-06, "loss": 0.11443699896335602, "step": 2797 }, { "epoch": 0.7263759086188992, "grad_norm": 3.1264302730560303, "learning_rate": 3.813561579214736e-06, "loss": 0.10373024642467499, "step": 2798 }, { "epoch": 0.7266355140186916, "grad_norm": 2.5502514839172363, "learning_rate": 3.8127278364246766e-06, "loss": 0.08339212834835052, "step": 2799 }, { "epoch": 0.726895119418484, "grad_norm": 2.9420907497406006, "learning_rate": 3.811893905768669e-06, "loss": 0.09167426079511642, "step": 2800 }, { "epoch": 0.7271547248182763, "grad_norm": 1.0954324007034302, "learning_rate": 3.811059787394133e-06, "loss": 0.10742047429084778, "step": 2801 }, { "epoch": 0.7274143302180686, "grad_norm": 3.54691219329834, "learning_rate": 3.8102254814485225e-06, "loss": 0.10479333996772766, "step": 2802 }, { "epoch": 0.7276739356178609, "grad_norm": 2.00466251373291, "learning_rate": 3.8093909880793223e-06, "loss": 0.09894717484712601, "step": 2803 }, { "epoch": 0.7279335410176532, "grad_norm": 2.1279854774475098, "learning_rate": 3.8085563074340513e-06, "loss": 0.12046593427658081, "step": 2804 }, { "epoch": 0.7281931464174455, "grad_norm": 2.8970205783843994, "learning_rate": 3.8077214396602642e-06, "loss": 0.1325354129076004, "step": 2805 }, { "epoch": 0.7284527518172378, "grad_norm": 1.7442665100097656, "learning_rate": 3.806886384905545e-06, "loss": 0.11417869478464127, "step": 2806 }, { "epoch": 0.7287123572170301, "grad_norm": 4.422211170196533, "learning_rate": 3.806051143317512e-06, "loss": 0.11238957196474075, "step": 2807 }, { "epoch": 0.7289719626168224, "grad_norm": 2.51019024848938, "learning_rate": 3.805215715043818e-06, "loss": 0.08255362510681152, "step": 2808 }, { "epoch": 0.7292315680166147, "grad_norm": 2.774848461151123, "learning_rate": 3.8043801002321457e-06, "loss": 0.11091409623622894, "step": 2809 }, { "epoch": 0.729491173416407, "grad_norm": 5.274651050567627, "learning_rate": 3.8035442990302145e-06, "loss": 0.11319656670093536, "step": 2810 }, { "epoch": 0.7297507788161994, "grad_norm": 3.1749675273895264, "learning_rate": 3.802708311585774e-06, "loss": 0.10810724645853043, "step": 2811 }, { "epoch": 0.7300103842159917, "grad_norm": 3.1956207752227783, "learning_rate": 3.801872138046608e-06, "loss": 0.11029506474733353, "step": 2812 }, { "epoch": 0.730269989615784, "grad_norm": 2.975314140319824, "learning_rate": 3.801035778560532e-06, "loss": 0.11861775815486908, "step": 2813 }, { "epoch": 0.7305295950155763, "grad_norm": 3.6017727851867676, "learning_rate": 3.800199233275397e-06, "loss": 0.08031661808490753, "step": 2814 }, { "epoch": 0.7307892004153687, "grad_norm": 4.1253838539123535, "learning_rate": 3.7993625023390835e-06, "loss": 0.0831037312746048, "step": 2815 }, { "epoch": 0.731048805815161, "grad_norm": 3.2702462673187256, "learning_rate": 3.7985255858995067e-06, "loss": 0.10235198587179184, "step": 2816 }, { "epoch": 0.7313084112149533, "grad_norm": 2.9076123237609863, "learning_rate": 3.7976884841046135e-06, "loss": 0.12149199098348618, "step": 2817 }, { "epoch": 0.7315680166147456, "grad_norm": 2.7706449031829834, "learning_rate": 3.7968511971023863e-06, "loss": 0.09183628857135773, "step": 2818 }, { "epoch": 0.7318276220145379, "grad_norm": 2.282996654510498, "learning_rate": 3.7960137250408374e-06, "loss": 0.11919976025819778, "step": 2819 }, { "epoch": 0.7320872274143302, "grad_norm": 2.7916531562805176, "learning_rate": 3.795176068068012e-06, "loss": 0.1338341236114502, "step": 2820 }, { "epoch": 0.7323468328141225, "grad_norm": 1.4422352313995361, "learning_rate": 3.794338226331989e-06, "loss": 0.11056434363126755, "step": 2821 }, { "epoch": 0.7326064382139148, "grad_norm": 4.044903755187988, "learning_rate": 3.7935001999808796e-06, "loss": 0.13308623433113098, "step": 2822 }, { "epoch": 0.7328660436137072, "grad_norm": 1.9536778926849365, "learning_rate": 3.792661989162828e-06, "loss": 0.11185299605131149, "step": 2823 }, { "epoch": 0.7331256490134995, "grad_norm": 2.6676950454711914, "learning_rate": 3.7918235940260108e-06, "loss": 0.11592521518468857, "step": 2824 }, { "epoch": 0.7333852544132918, "grad_norm": 2.3236312866210938, "learning_rate": 3.790985014718636e-06, "loss": 0.11407686769962311, "step": 2825 }, { "epoch": 0.7336448598130841, "grad_norm": 1.967716097831726, "learning_rate": 3.790146251388947e-06, "loss": 0.10879065096378326, "step": 2826 }, { "epoch": 0.7339044652128764, "grad_norm": 2.9968066215515137, "learning_rate": 3.789307304185216e-06, "loss": 0.1357310563325882, "step": 2827 }, { "epoch": 0.7341640706126688, "grad_norm": 3.071347713470459, "learning_rate": 3.7884681732557517e-06, "loss": 0.09854000806808472, "step": 2828 }, { "epoch": 0.7344236760124611, "grad_norm": 2.9375252723693848, "learning_rate": 3.7876288587488913e-06, "loss": 0.09066905081272125, "step": 2829 }, { "epoch": 0.7346832814122534, "grad_norm": 3.994521379470825, "learning_rate": 3.786789360813007e-06, "loss": 0.12330921739339828, "step": 2830 }, { "epoch": 0.7349428868120457, "grad_norm": 3.339177370071411, "learning_rate": 3.785949679596504e-06, "loss": 0.097890205681324, "step": 2831 }, { "epoch": 0.735202492211838, "grad_norm": 1.0276217460632324, "learning_rate": 3.785109815247817e-06, "loss": 0.11152671277523041, "step": 2832 }, { "epoch": 0.7354620976116303, "grad_norm": 2.9148216247558594, "learning_rate": 3.784269767915416e-06, "loss": 0.09228479862213135, "step": 2833 }, { "epoch": 0.7357217030114226, "grad_norm": 1.4893500804901123, "learning_rate": 3.7834295377478013e-06, "loss": 0.11087460815906525, "step": 2834 }, { "epoch": 0.735981308411215, "grad_norm": 2.0420150756835938, "learning_rate": 3.7825891248935066e-06, "loss": 0.09664168953895569, "step": 2835 }, { "epoch": 0.7362409138110073, "grad_norm": 1.5990501642227173, "learning_rate": 3.7817485295010982e-06, "loss": 0.09380993247032166, "step": 2836 }, { "epoch": 0.7365005192107996, "grad_norm": 3.084500789642334, "learning_rate": 3.7809077517191733e-06, "loss": 0.11926563084125519, "step": 2837 }, { "epoch": 0.7367601246105919, "grad_norm": 1.0641382932662964, "learning_rate": 3.780066791696363e-06, "loss": 0.10258917510509491, "step": 2838 }, { "epoch": 0.7370197300103842, "grad_norm": 2.070456027984619, "learning_rate": 3.779225649581329e-06, "loss": 0.13596728444099426, "step": 2839 }, { "epoch": 0.7372793354101765, "grad_norm": 2.388727903366089, "learning_rate": 3.778384325522767e-06, "loss": 0.07446841895580292, "step": 2840 }, { "epoch": 0.7375389408099688, "grad_norm": 1.2980884313583374, "learning_rate": 3.7775428196694023e-06, "loss": 0.10668644309043884, "step": 2841 }, { "epoch": 0.7377985462097612, "grad_norm": 3.2464711666107178, "learning_rate": 3.776701132169995e-06, "loss": 0.08340005576610565, "step": 2842 }, { "epoch": 0.7380581516095535, "grad_norm": 2.9294965267181396, "learning_rate": 3.7758592631733354e-06, "loss": 0.12455902248620987, "step": 2843 }, { "epoch": 0.7383177570093458, "grad_norm": 2.05690598487854, "learning_rate": 3.775017212828248e-06, "loss": 0.10681188106536865, "step": 2844 }, { "epoch": 0.7385773624091381, "grad_norm": 2.467395067214966, "learning_rate": 3.774174981283587e-06, "loss": 0.13255785405635834, "step": 2845 }, { "epoch": 0.7388369678089304, "grad_norm": 2.066502571105957, "learning_rate": 3.773332568688239e-06, "loss": 0.10690824687480927, "step": 2846 }, { "epoch": 0.7390965732087228, "grad_norm": 3.7038466930389404, "learning_rate": 3.7724899751911254e-06, "loss": 0.10153557360172272, "step": 2847 }, { "epoch": 0.7393561786085151, "grad_norm": 1.3263121843338013, "learning_rate": 3.7716472009411944e-06, "loss": 0.11261065304279327, "step": 2848 }, { "epoch": 0.7396157840083074, "grad_norm": 1.696021318435669, "learning_rate": 3.7708042460874315e-06, "loss": 0.1125967800617218, "step": 2849 }, { "epoch": 0.7398753894080997, "grad_norm": 1.686103105545044, "learning_rate": 3.769961110778852e-06, "loss": 0.11405329406261444, "step": 2850 }, { "epoch": 0.740134994807892, "grad_norm": 2.858151435852051, "learning_rate": 3.7691177951645018e-06, "loss": 0.10901167243719101, "step": 2851 }, { "epoch": 0.7403946002076843, "grad_norm": 2.232591390609741, "learning_rate": 3.7682742993934595e-06, "loss": 0.11875215172767639, "step": 2852 }, { "epoch": 0.7406542056074766, "grad_norm": 1.156159520149231, "learning_rate": 3.7674306236148374e-06, "loss": 0.11527848988771439, "step": 2853 }, { "epoch": 0.7409138110072689, "grad_norm": 1.6483993530273438, "learning_rate": 3.7665867679777763e-06, "loss": 0.09593701362609863, "step": 2854 }, { "epoch": 0.7411734164070612, "grad_norm": 2.7221462726593018, "learning_rate": 3.7657427326314523e-06, "loss": 0.10399343818426132, "step": 2855 }, { "epoch": 0.7414330218068536, "grad_norm": 2.640930414199829, "learning_rate": 3.7648985177250697e-06, "loss": 0.1120663732290268, "step": 2856 }, { "epoch": 0.7416926272066459, "grad_norm": 3.0941579341888428, "learning_rate": 3.764054123407867e-06, "loss": 0.08988057076931, "step": 2857 }, { "epoch": 0.7419522326064382, "grad_norm": 3.0118801593780518, "learning_rate": 3.763209549829115e-06, "loss": 0.0969114899635315, "step": 2858 }, { "epoch": 0.7422118380062306, "grad_norm": 2.214704990386963, "learning_rate": 3.7623647971381132e-06, "loss": 0.10981233417987823, "step": 2859 }, { "epoch": 0.7424714434060229, "grad_norm": 2.924917459487915, "learning_rate": 3.7615198654841957e-06, "loss": 0.10785075277090073, "step": 2860 }, { "epoch": 0.7427310488058152, "grad_norm": 2.4805829524993896, "learning_rate": 3.760674755016727e-06, "loss": 0.11246054619550705, "step": 2861 }, { "epoch": 0.7429906542056075, "grad_norm": 2.2392325401306152, "learning_rate": 3.7598294658851025e-06, "loss": 0.09801600873470306, "step": 2862 }, { "epoch": 0.7432502596053998, "grad_norm": 3.542355537414551, "learning_rate": 3.758983998238751e-06, "loss": 0.10971163958311081, "step": 2863 }, { "epoch": 0.7435098650051921, "grad_norm": 2.3797926902770996, "learning_rate": 3.758138352227131e-06, "loss": 0.08636285364627838, "step": 2864 }, { "epoch": 0.7437694704049844, "grad_norm": 2.8806309700012207, "learning_rate": 3.757292527999734e-06, "loss": 0.09817864745855331, "step": 2865 }, { "epoch": 0.7440290758047767, "grad_norm": 4.185187816619873, "learning_rate": 3.756446525706081e-06, "loss": 0.09794910997152328, "step": 2866 }, { "epoch": 0.744288681204569, "grad_norm": 2.707728147506714, "learning_rate": 3.7556003454957273e-06, "loss": 0.09590545296669006, "step": 2867 }, { "epoch": 0.7445482866043613, "grad_norm": 1.7570892572402954, "learning_rate": 3.754753987518259e-06, "loss": 0.11092053353786469, "step": 2868 }, { "epoch": 0.7448078920041536, "grad_norm": 3.155379056930542, "learning_rate": 3.75390745192329e-06, "loss": 0.1291487216949463, "step": 2869 }, { "epoch": 0.745067497403946, "grad_norm": 3.102084159851074, "learning_rate": 3.75306073886047e-06, "loss": 0.08646117895841599, "step": 2870 }, { "epoch": 0.7453271028037384, "grad_norm": 2.464350700378418, "learning_rate": 3.752213848479479e-06, "loss": 0.11271552741527557, "step": 2871 }, { "epoch": 0.7455867082035307, "grad_norm": 2.398306131362915, "learning_rate": 3.751366780930027e-06, "loss": 0.09125930815935135, "step": 2872 }, { "epoch": 0.745846313603323, "grad_norm": 3.5555925369262695, "learning_rate": 3.7505195363618563e-06, "loss": 0.12626442313194275, "step": 2873 }, { "epoch": 0.7461059190031153, "grad_norm": 1.780712366104126, "learning_rate": 3.7496721149247407e-06, "loss": 0.10231341421604156, "step": 2874 }, { "epoch": 0.7463655244029076, "grad_norm": 3.123634099960327, "learning_rate": 3.7488245167684846e-06, "loss": 0.09861491620540619, "step": 2875 }, { "epoch": 0.7466251298026999, "grad_norm": 4.428478240966797, "learning_rate": 3.7479767420429235e-06, "loss": 0.11365856230258942, "step": 2876 }, { "epoch": 0.7468847352024922, "grad_norm": 1.4998849630355835, "learning_rate": 3.7471287908979254e-06, "loss": 0.11528154462575912, "step": 2877 }, { "epoch": 0.7471443406022845, "grad_norm": 3.040135145187378, "learning_rate": 3.746280663483389e-06, "loss": 0.08517034351825714, "step": 2878 }, { "epoch": 0.7474039460020768, "grad_norm": 1.8779114484786987, "learning_rate": 3.7454323599492414e-06, "loss": 0.103194460272789, "step": 2879 }, { "epoch": 0.7476635514018691, "grad_norm": 4.75319242477417, "learning_rate": 3.744583880445446e-06, "loss": 0.10632888227701187, "step": 2880 }, { "epoch": 0.7479231568016614, "grad_norm": 3.4828288555145264, "learning_rate": 3.743735225121994e-06, "loss": 0.09467214345932007, "step": 2881 }, { "epoch": 0.7481827622014537, "grad_norm": 2.0686278343200684, "learning_rate": 3.7428863941289067e-06, "loss": 0.08680947124958038, "step": 2882 }, { "epoch": 0.7484423676012462, "grad_norm": 3.125185012817383, "learning_rate": 3.742037387616239e-06, "loss": 0.06448923051357269, "step": 2883 }, { "epoch": 0.7487019730010385, "grad_norm": 3.333513021469116, "learning_rate": 3.7411882057340766e-06, "loss": 0.10190518200397491, "step": 2884 }, { "epoch": 0.7489615784008308, "grad_norm": 2.777353525161743, "learning_rate": 3.7403388486325345e-06, "loss": 0.11520372331142426, "step": 2885 }, { "epoch": 0.7492211838006231, "grad_norm": 4.30525541305542, "learning_rate": 3.739489316461759e-06, "loss": 0.10027921199798584, "step": 2886 }, { "epoch": 0.7494807892004154, "grad_norm": 3.0431759357452393, "learning_rate": 3.7386396093719297e-06, "loss": 0.12037675827741623, "step": 2887 }, { "epoch": 0.7497403946002077, "grad_norm": 2.2445106506347656, "learning_rate": 3.7377897275132534e-06, "loss": 0.110720694065094, "step": 2888 }, { "epoch": 0.75, "grad_norm": 3.546402931213379, "learning_rate": 3.736939671035971e-06, "loss": 0.10935632139444351, "step": 2889 }, { "epoch": 0.7502596053997923, "grad_norm": 3.592240571975708, "learning_rate": 3.736089440090353e-06, "loss": 0.09959754347801208, "step": 2890 }, { "epoch": 0.7505192107995846, "grad_norm": 4.560543060302734, "learning_rate": 3.7352390348266997e-06, "loss": 0.13979506492614746, "step": 2891 }, { "epoch": 0.7507788161993769, "grad_norm": 3.659817934036255, "learning_rate": 3.734388455395344e-06, "loss": 0.10430701076984406, "step": 2892 }, { "epoch": 0.7510384215991692, "grad_norm": 3.3203470706939697, "learning_rate": 3.7335377019466497e-06, "loss": 0.09053434431552887, "step": 2893 }, { "epoch": 0.7512980269989615, "grad_norm": 3.887042999267578, "learning_rate": 3.73268677463101e-06, "loss": 0.08797118067741394, "step": 2894 }, { "epoch": 0.7515576323987538, "grad_norm": 2.206733465194702, "learning_rate": 3.731835673598848e-06, "loss": 0.096535325050354, "step": 2895 }, { "epoch": 0.7518172377985463, "grad_norm": 1.2827610969543457, "learning_rate": 3.7309843990006207e-06, "loss": 0.09712225943803787, "step": 2896 }, { "epoch": 0.7520768431983386, "grad_norm": 4.355175018310547, "learning_rate": 3.7301329509868124e-06, "loss": 0.08971931785345078, "step": 2897 }, { "epoch": 0.7523364485981309, "grad_norm": 2.7454962730407715, "learning_rate": 3.7292813297079416e-06, "loss": 0.11219754815101624, "step": 2898 }, { "epoch": 0.7525960539979232, "grad_norm": 1.2443009614944458, "learning_rate": 3.728429535314554e-06, "loss": 0.10550359636545181, "step": 2899 }, { "epoch": 0.7528556593977155, "grad_norm": 0.9780333638191223, "learning_rate": 3.727577567957227e-06, "loss": 0.10834605991840363, "step": 2900 }, { "epoch": 0.7531152647975078, "grad_norm": 3.023102045059204, "learning_rate": 3.72672542778657e-06, "loss": 0.09757771342992783, "step": 2901 }, { "epoch": 0.7533748701973001, "grad_norm": 3.078062057495117, "learning_rate": 3.7258731149532207e-06, "loss": 0.0803239643573761, "step": 2902 }, { "epoch": 0.7536344755970924, "grad_norm": 2.33957839012146, "learning_rate": 3.72502062960785e-06, "loss": 0.13642515242099762, "step": 2903 }, { "epoch": 0.7538940809968847, "grad_norm": 1.1676210165023804, "learning_rate": 3.724167971901157e-06, "loss": 0.11265534162521362, "step": 2904 }, { "epoch": 0.754153686396677, "grad_norm": 2.5279879570007324, "learning_rate": 3.7233151419838713e-06, "loss": 0.11378364264965057, "step": 2905 }, { "epoch": 0.7544132917964693, "grad_norm": 3.015137195587158, "learning_rate": 3.722462140006755e-06, "loss": 0.09715361893177032, "step": 2906 }, { "epoch": 0.7546728971962616, "grad_norm": 2.40987491607666, "learning_rate": 3.721608966120599e-06, "loss": 0.08206356316804886, "step": 2907 }, { "epoch": 0.754932502596054, "grad_norm": 2.1604273319244385, "learning_rate": 3.7207556204762248e-06, "loss": 0.12271226942539215, "step": 2908 }, { "epoch": 0.7551921079958464, "grad_norm": 1.2631919384002686, "learning_rate": 3.7199021032244843e-06, "loss": 0.10956041514873505, "step": 2909 }, { "epoch": 0.7554517133956387, "grad_norm": 2.4178311824798584, "learning_rate": 3.7190484145162585e-06, "loss": 0.08997152745723724, "step": 2910 }, { "epoch": 0.755711318795431, "grad_norm": 3.159114360809326, "learning_rate": 3.7181945545024622e-06, "loss": 0.07373043894767761, "step": 2911 }, { "epoch": 0.7559709241952233, "grad_norm": 3.077089548110962, "learning_rate": 3.717340523334037e-06, "loss": 0.10505452752113342, "step": 2912 }, { "epoch": 0.7562305295950156, "grad_norm": 1.521982192993164, "learning_rate": 3.716486321161957e-06, "loss": 0.10610099881887436, "step": 2913 }, { "epoch": 0.7564901349948079, "grad_norm": 0.7939885258674622, "learning_rate": 3.7156319481372247e-06, "loss": 0.10942485928535461, "step": 2914 }, { "epoch": 0.7567497403946002, "grad_norm": 1.3107753992080688, "learning_rate": 3.7147774044108743e-06, "loss": 0.1176816076040268, "step": 2915 }, { "epoch": 0.7570093457943925, "grad_norm": 2.679466724395752, "learning_rate": 3.7139226901339685e-06, "loss": 0.11767441779375076, "step": 2916 }, { "epoch": 0.7572689511941848, "grad_norm": 1.6202523708343506, "learning_rate": 3.713067805457603e-06, "loss": 0.11558456718921661, "step": 2917 }, { "epoch": 0.7575285565939771, "grad_norm": 1.555321455001831, "learning_rate": 3.7122127505329003e-06, "loss": 0.09444384276866913, "step": 2918 }, { "epoch": 0.7577881619937694, "grad_norm": 1.9782214164733887, "learning_rate": 3.7113575255110155e-06, "loss": 0.11203914135694504, "step": 2919 }, { "epoch": 0.7580477673935618, "grad_norm": 0.8227517008781433, "learning_rate": 3.7105021305431317e-06, "loss": 0.11785292625427246, "step": 2920 }, { "epoch": 0.7583073727933541, "grad_norm": 1.2231205701828003, "learning_rate": 3.709646565780464e-06, "loss": 0.11940731853246689, "step": 2921 }, { "epoch": 0.7585669781931464, "grad_norm": 3.37502384185791, "learning_rate": 3.708790831374257e-06, "loss": 0.08581811189651489, "step": 2922 }, { "epoch": 0.7588265835929388, "grad_norm": 1.8681460618972778, "learning_rate": 3.7079349274757838e-06, "loss": 0.10009102523326874, "step": 2923 }, { "epoch": 0.7590861889927311, "grad_norm": 2.724271774291992, "learning_rate": 3.7070788542363495e-06, "loss": 0.1383010596036911, "step": 2924 }, { "epoch": 0.7593457943925234, "grad_norm": 3.481766700744629, "learning_rate": 3.706222611807288e-06, "loss": 0.09477891027927399, "step": 2925 }, { "epoch": 0.7596053997923157, "grad_norm": 2.627134084701538, "learning_rate": 3.705366200339964e-06, "loss": 0.13558726012706757, "step": 2926 }, { "epoch": 0.759865005192108, "grad_norm": 2.5125651359558105, "learning_rate": 3.7045096199857694e-06, "loss": 0.08022217452526093, "step": 2927 }, { "epoch": 0.7601246105919003, "grad_norm": 1.5285530090332031, "learning_rate": 3.7036528708961297e-06, "loss": 0.10037972033023834, "step": 2928 }, { "epoch": 0.7603842159916926, "grad_norm": 1.0107578039169312, "learning_rate": 3.702795953222498e-06, "loss": 0.12024834007024765, "step": 2929 }, { "epoch": 0.7606438213914849, "grad_norm": 1.4171563386917114, "learning_rate": 3.7019388671163588e-06, "loss": 0.11157095432281494, "step": 2930 }, { "epoch": 0.7609034267912772, "grad_norm": 1.5373544692993164, "learning_rate": 3.7010816127292245e-06, "loss": 0.10892349481582642, "step": 2931 }, { "epoch": 0.7611630321910696, "grad_norm": 1.5722825527191162, "learning_rate": 3.7002241902126366e-06, "loss": 0.08635362982749939, "step": 2932 }, { "epoch": 0.7614226375908619, "grad_norm": 1.4136059284210205, "learning_rate": 3.6993665997181698e-06, "loss": 0.09811611473560333, "step": 2933 }, { "epoch": 0.7616822429906542, "grad_norm": 1.9224627017974854, "learning_rate": 3.698508841397426e-06, "loss": 0.11215249449014664, "step": 2934 }, { "epoch": 0.7619418483904465, "grad_norm": 1.9792654514312744, "learning_rate": 3.6976509154020375e-06, "loss": 0.09671438485383987, "step": 2935 }, { "epoch": 0.7622014537902388, "grad_norm": 1.0172137022018433, "learning_rate": 3.6967928218836646e-06, "loss": 0.12099442631006241, "step": 2936 }, { "epoch": 0.7624610591900312, "grad_norm": 1.4385548830032349, "learning_rate": 3.695934560993999e-06, "loss": 0.09589669108390808, "step": 2937 }, { "epoch": 0.7627206645898235, "grad_norm": 3.3277533054351807, "learning_rate": 3.6950761328847634e-06, "loss": 0.09589970856904984, "step": 2938 }, { "epoch": 0.7629802699896158, "grad_norm": 3.036402463912964, "learning_rate": 3.694217537707706e-06, "loss": 0.08576485514640808, "step": 2939 }, { "epoch": 0.7632398753894081, "grad_norm": 2.943925142288208, "learning_rate": 3.6933587756146085e-06, "loss": 0.11056594550609589, "step": 2940 }, { "epoch": 0.7634994807892004, "grad_norm": 2.0437674522399902, "learning_rate": 3.6924998467572782e-06, "loss": 0.10560473799705505, "step": 2941 }, { "epoch": 0.7637590861889927, "grad_norm": 1.3596291542053223, "learning_rate": 3.6916407512875567e-06, "loss": 0.1236262172460556, "step": 2942 }, { "epoch": 0.764018691588785, "grad_norm": 3.522167205810547, "learning_rate": 3.6907814893573102e-06, "loss": 0.10227158665657043, "step": 2943 }, { "epoch": 0.7642782969885774, "grad_norm": 3.011026620864868, "learning_rate": 3.6899220611184384e-06, "loss": 0.11388054490089417, "step": 2944 }, { "epoch": 0.7645379023883697, "grad_norm": 2.8795268535614014, "learning_rate": 3.6890624667228664e-06, "loss": 0.07056811451911926, "step": 2945 }, { "epoch": 0.764797507788162, "grad_norm": 2.454580068588257, "learning_rate": 3.6882027063225527e-06, "loss": 0.09546218067407608, "step": 2946 }, { "epoch": 0.7650571131879543, "grad_norm": 3.722407341003418, "learning_rate": 3.687342780069482e-06, "loss": 0.0670834556221962, "step": 2947 }, { "epoch": 0.7653167185877466, "grad_norm": 2.1490278244018555, "learning_rate": 3.68648268811567e-06, "loss": 0.11395300924777985, "step": 2948 }, { "epoch": 0.7655763239875389, "grad_norm": 3.0979256629943848, "learning_rate": 3.6856224306131622e-06, "loss": 0.06522957980632782, "step": 2949 }, { "epoch": 0.7658359293873312, "grad_norm": 2.6589527130126953, "learning_rate": 3.684762007714031e-06, "loss": 0.10127602517604828, "step": 2950 }, { "epoch": 0.7660955347871236, "grad_norm": 2.2869181632995605, "learning_rate": 3.68390141957038e-06, "loss": 0.09843724966049194, "step": 2951 }, { "epoch": 0.7663551401869159, "grad_norm": 1.1487191915512085, "learning_rate": 3.683040666334342e-06, "loss": 0.10789376497268677, "step": 2952 }, { "epoch": 0.7666147455867082, "grad_norm": 1.1946136951446533, "learning_rate": 3.682179748158078e-06, "loss": 0.09951385855674744, "step": 2953 }, { "epoch": 0.7668743509865005, "grad_norm": 3.1858627796173096, "learning_rate": 3.6813186651937775e-06, "loss": 0.08154433965682983, "step": 2954 }, { "epoch": 0.7671339563862928, "grad_norm": 3.52532958984375, "learning_rate": 3.6804574175936623e-06, "loss": 0.08992680162191391, "step": 2955 }, { "epoch": 0.7673935617860852, "grad_norm": 3.2516677379608154, "learning_rate": 3.6795960055099804e-06, "loss": 0.08487334847450256, "step": 2956 }, { "epoch": 0.7676531671858775, "grad_norm": 1.4773688316345215, "learning_rate": 3.6787344290950094e-06, "loss": 0.11160717904567719, "step": 2957 }, { "epoch": 0.7679127725856698, "grad_norm": 3.1126279830932617, "learning_rate": 3.677872688501057e-06, "loss": 0.1057235449552536, "step": 2958 }, { "epoch": 0.7681723779854621, "grad_norm": 3.238727569580078, "learning_rate": 3.6770107838804577e-06, "loss": 0.12220252305269241, "step": 2959 }, { "epoch": 0.7684319833852544, "grad_norm": 1.513532280921936, "learning_rate": 3.676148715385578e-06, "loss": 0.10633565485477448, "step": 2960 }, { "epoch": 0.7686915887850467, "grad_norm": 2.706756591796875, "learning_rate": 3.6752864831688124e-06, "loss": 0.1051703691482544, "step": 2961 }, { "epoch": 0.768951194184839, "grad_norm": 3.6507492065429688, "learning_rate": 3.6744240873825816e-06, "loss": 0.10238061845302582, "step": 2962 }, { "epoch": 0.7692107995846313, "grad_norm": 1.5756382942199707, "learning_rate": 3.673561528179339e-06, "loss": 0.08072027564048767, "step": 2963 }, { "epoch": 0.7694704049844237, "grad_norm": 1.761518955230713, "learning_rate": 3.6726988057115646e-06, "loss": 0.1091669350862503, "step": 2964 }, { "epoch": 0.769730010384216, "grad_norm": 1.9431525468826294, "learning_rate": 3.6718359201317695e-06, "loss": 0.12676158547401428, "step": 2965 }, { "epoch": 0.7699896157840083, "grad_norm": 3.436779499053955, "learning_rate": 3.67097287159249e-06, "loss": 0.09367139637470245, "step": 2966 }, { "epoch": 0.7702492211838006, "grad_norm": 1.5854122638702393, "learning_rate": 3.6701096602462944e-06, "loss": 0.09404732286930084, "step": 2967 }, { "epoch": 0.770508826583593, "grad_norm": 5.060603141784668, "learning_rate": 3.6692462862457784e-06, "loss": 0.15194718539714813, "step": 2968 }, { "epoch": 0.7707684319833853, "grad_norm": 1.913884162902832, "learning_rate": 3.6683827497435678e-06, "loss": 0.10635307431221008, "step": 2969 }, { "epoch": 0.7710280373831776, "grad_norm": 1.0931075811386108, "learning_rate": 3.667519050892315e-06, "loss": 0.10374563932418823, "step": 2970 }, { "epoch": 0.7712876427829699, "grad_norm": 1.4482958316802979, "learning_rate": 3.666655189844701e-06, "loss": 0.10631010681390762, "step": 2971 }, { "epoch": 0.7715472481827622, "grad_norm": 3.257211685180664, "learning_rate": 3.6657911667534397e-06, "loss": 0.09409548342227936, "step": 2972 }, { "epoch": 0.7718068535825545, "grad_norm": 3.6907742023468018, "learning_rate": 3.6649269817712686e-06, "loss": 0.08359424024820328, "step": 2973 }, { "epoch": 0.7720664589823468, "grad_norm": 3.154405355453491, "learning_rate": 3.6640626350509557e-06, "loss": 0.06406259536743164, "step": 2974 }, { "epoch": 0.7723260643821391, "grad_norm": 1.752647042274475, "learning_rate": 3.663198126745299e-06, "loss": 0.09479852020740509, "step": 2975 }, { "epoch": 0.7725856697819314, "grad_norm": 2.894646406173706, "learning_rate": 3.6623334570071216e-06, "loss": 0.07451385259628296, "step": 2976 }, { "epoch": 0.7728452751817237, "grad_norm": 2.203784227371216, "learning_rate": 3.6614686259892794e-06, "loss": 0.09841102361679077, "step": 2977 }, { "epoch": 0.773104880581516, "grad_norm": 2.4990103244781494, "learning_rate": 3.660603633844654e-06, "loss": 0.07571238279342651, "step": 2978 }, { "epoch": 0.7733644859813084, "grad_norm": 2.6113834381103516, "learning_rate": 3.6597384807261564e-06, "loss": 0.12217006832361221, "step": 2979 }, { "epoch": 0.7736240913811008, "grad_norm": 1.7350701093673706, "learning_rate": 3.658873166786726e-06, "loss": 0.10315471887588501, "step": 2980 }, { "epoch": 0.7738836967808931, "grad_norm": 3.1194217205047607, "learning_rate": 3.6580076921793295e-06, "loss": 0.07956518232822418, "step": 2981 }, { "epoch": 0.7741433021806854, "grad_norm": 2.5979466438293457, "learning_rate": 3.6571420570569643e-06, "loss": 0.08734594285488129, "step": 2982 }, { "epoch": 0.7744029075804777, "grad_norm": 1.1706093549728394, "learning_rate": 3.6562762615726538e-06, "loss": 0.11205142736434937, "step": 2983 }, { "epoch": 0.77466251298027, "grad_norm": 2.726717948913574, "learning_rate": 3.6554103058794504e-06, "loss": 0.08034725487232208, "step": 2984 }, { "epoch": 0.7749221183800623, "grad_norm": 3.030496120452881, "learning_rate": 3.6545441901304366e-06, "loss": 0.11450525373220444, "step": 2985 }, { "epoch": 0.7751817237798546, "grad_norm": 1.2212153673171997, "learning_rate": 3.6536779144787223e-06, "loss": 0.10809682309627533, "step": 2986 }, { "epoch": 0.7754413291796469, "grad_norm": 1.0741099119186401, "learning_rate": 3.652811479077443e-06, "loss": 0.09633190929889679, "step": 2987 }, { "epoch": 0.7757009345794392, "grad_norm": 0.834225594997406, "learning_rate": 3.6519448840797667e-06, "loss": 0.10681737959384918, "step": 2988 }, { "epoch": 0.7759605399792315, "grad_norm": 3.659925937652588, "learning_rate": 3.651078129638886e-06, "loss": 0.1049046516418457, "step": 2989 }, { "epoch": 0.7762201453790238, "grad_norm": 4.340736389160156, "learning_rate": 3.6502112159080243e-06, "loss": 0.09412133693695068, "step": 2990 }, { "epoch": 0.7764797507788161, "grad_norm": 1.607904314994812, "learning_rate": 3.6493441430404315e-06, "loss": 0.12353739142417908, "step": 2991 }, { "epoch": 0.7767393561786086, "grad_norm": 4.299932956695557, "learning_rate": 3.648476911189387e-06, "loss": 0.10329504311084747, "step": 2992 }, { "epoch": 0.7769989615784009, "grad_norm": 2.8937158584594727, "learning_rate": 3.6476095205081963e-06, "loss": 0.11183382570743561, "step": 2993 }, { "epoch": 0.7772585669781932, "grad_norm": 1.0096023082733154, "learning_rate": 3.646741971150195e-06, "loss": 0.10666720569133759, "step": 2994 }, { "epoch": 0.7775181723779855, "grad_norm": 2.675485849380493, "learning_rate": 3.6458742632687462e-06, "loss": 0.09965081512928009, "step": 2995 }, { "epoch": 0.7777777777777778, "grad_norm": 1.145786166191101, "learning_rate": 3.64500639701724e-06, "loss": 0.1297408938407898, "step": 2996 }, { "epoch": 0.7780373831775701, "grad_norm": 3.252908706665039, "learning_rate": 3.6441383725490957e-06, "loss": 0.10176564753055573, "step": 2997 }, { "epoch": 0.7782969885773624, "grad_norm": 2.4919211864471436, "learning_rate": 3.643270190017761e-06, "loss": 0.1311972290277481, "step": 2998 }, { "epoch": 0.7785565939771547, "grad_norm": 3.085723876953125, "learning_rate": 3.642401849576709e-06, "loss": 0.1269976943731308, "step": 2999 }, { "epoch": 0.778816199376947, "grad_norm": 2.5596253871917725, "learning_rate": 3.6415333513794442e-06, "loss": 0.08625443279743195, "step": 3000 }, { "epoch": 0.7790758047767393, "grad_norm": 1.824508786201477, "learning_rate": 3.640664695579496e-06, "loss": 0.10798764228820801, "step": 3001 }, { "epoch": 0.7793354101765316, "grad_norm": 1.6225173473358154, "learning_rate": 3.639795882330423e-06, "loss": 0.12220963090658188, "step": 3002 }, { "epoch": 0.7795950155763239, "grad_norm": 1.2686970233917236, "learning_rate": 3.638926911785811e-06, "loss": 0.10413956642150879, "step": 3003 }, { "epoch": 0.7798546209761164, "grad_norm": 3.225878953933716, "learning_rate": 3.6380577840992758e-06, "loss": 0.09163599461317062, "step": 3004 }, { "epoch": 0.7801142263759087, "grad_norm": 1.4708735942840576, "learning_rate": 3.637188499424457e-06, "loss": 0.09331601113080978, "step": 3005 }, { "epoch": 0.780373831775701, "grad_norm": 2.518576145172119, "learning_rate": 3.636319057915027e-06, "loss": 0.12534692883491516, "step": 3006 }, { "epoch": 0.7806334371754933, "grad_norm": 1.46015465259552, "learning_rate": 3.63544945972468e-06, "loss": 0.0970194935798645, "step": 3007 }, { "epoch": 0.7808930425752856, "grad_norm": 2.894796848297119, "learning_rate": 3.634579705007144e-06, "loss": 0.09351807087659836, "step": 3008 }, { "epoch": 0.7811526479750779, "grad_norm": 1.012978434562683, "learning_rate": 3.6337097939161698e-06, "loss": 0.12226735055446625, "step": 3009 }, { "epoch": 0.7814122533748702, "grad_norm": 2.4503681659698486, "learning_rate": 3.632839726605538e-06, "loss": 0.07571861147880554, "step": 3010 }, { "epoch": 0.7816718587746625, "grad_norm": 3.872098684310913, "learning_rate": 3.6319695032290565e-06, "loss": 0.08294646441936493, "step": 3011 }, { "epoch": 0.7819314641744548, "grad_norm": 3.048680067062378, "learning_rate": 3.6310991239405625e-06, "loss": 0.09479150176048279, "step": 3012 }, { "epoch": 0.7821910695742471, "grad_norm": 2.3348512649536133, "learning_rate": 3.630228588893917e-06, "loss": 0.08638118952512741, "step": 3013 }, { "epoch": 0.7824506749740394, "grad_norm": 2.4828171730041504, "learning_rate": 3.629357898243011e-06, "loss": 0.11047562956809998, "step": 3014 }, { "epoch": 0.7827102803738317, "grad_norm": 3.9179484844207764, "learning_rate": 3.6284870521417646e-06, "loss": 0.08764071762561798, "step": 3015 }, { "epoch": 0.7829698857736241, "grad_norm": 2.003004312515259, "learning_rate": 3.627616050744121e-06, "loss": 0.1101437509059906, "step": 3016 }, { "epoch": 0.7832294911734164, "grad_norm": 2.5557072162628174, "learning_rate": 3.6267448942040553e-06, "loss": 0.07654780149459839, "step": 3017 }, { "epoch": 0.7834890965732088, "grad_norm": 2.6837310791015625, "learning_rate": 3.6258735826755663e-06, "loss": 0.04404466226696968, "step": 3018 }, { "epoch": 0.7837487019730011, "grad_norm": 2.9477992057800293, "learning_rate": 3.625002116312683e-06, "loss": 0.07450111210346222, "step": 3019 }, { "epoch": 0.7840083073727934, "grad_norm": 2.5267293453216553, "learning_rate": 3.6241304952694606e-06, "loss": 0.12516885995864868, "step": 3020 }, { "epoch": 0.7842679127725857, "grad_norm": 2.900761604309082, "learning_rate": 3.623258719699982e-06, "loss": 0.09753304719924927, "step": 3021 }, { "epoch": 0.784527518172378, "grad_norm": 1.6292879581451416, "learning_rate": 3.6223867897583555e-06, "loss": 0.11383086442947388, "step": 3022 }, { "epoch": 0.7847871235721703, "grad_norm": 3.2143774032592773, "learning_rate": 3.6215147055987204e-06, "loss": 0.12464810162782669, "step": 3023 }, { "epoch": 0.7850467289719626, "grad_norm": 2.232170343399048, "learning_rate": 3.62064246737524e-06, "loss": 0.06031098961830139, "step": 3024 }, { "epoch": 0.7853063343717549, "grad_norm": 0.9703103303909302, "learning_rate": 3.6197700752421066e-06, "loss": 0.11424864828586578, "step": 3025 }, { "epoch": 0.7855659397715472, "grad_norm": 3.180067539215088, "learning_rate": 3.6188975293535393e-06, "loss": 0.1031109020113945, "step": 3026 }, { "epoch": 0.7858255451713395, "grad_norm": 1.6508538722991943, "learning_rate": 3.618024829863784e-06, "loss": 0.11324851959943771, "step": 3027 }, { "epoch": 0.7860851505711319, "grad_norm": 2.9484968185424805, "learning_rate": 3.617151976927113e-06, "loss": 0.09486891329288483, "step": 3028 }, { "epoch": 0.7863447559709242, "grad_norm": 3.3403666019439697, "learning_rate": 3.6162789706978287e-06, "loss": 0.0806526243686676, "step": 3029 }, { "epoch": 0.7866043613707165, "grad_norm": 2.45332407951355, "learning_rate": 3.6154058113302566e-06, "loss": 0.14598524570465088, "step": 3030 }, { "epoch": 0.7868639667705088, "grad_norm": 1.798910140991211, "learning_rate": 3.614532498978753e-06, "loss": 0.0954815074801445, "step": 3031 }, { "epoch": 0.7871235721703012, "grad_norm": 2.3952460289001465, "learning_rate": 3.6136590337976985e-06, "loss": 0.09580262005329132, "step": 3032 }, { "epoch": 0.7873831775700935, "grad_norm": 1.7038460969924927, "learning_rate": 3.6127854159415014e-06, "loss": 0.10245285928249359, "step": 3033 }, { "epoch": 0.7876427829698858, "grad_norm": 1.493492841720581, "learning_rate": 3.611911645564599e-06, "loss": 0.11477454751729965, "step": 3034 }, { "epoch": 0.7879023883696781, "grad_norm": 4.086459159851074, "learning_rate": 3.611037722821452e-06, "loss": 0.05930968374013901, "step": 3035 }, { "epoch": 0.7881619937694704, "grad_norm": 3.1633079051971436, "learning_rate": 3.610163647866552e-06, "loss": 0.1149190217256546, "step": 3036 }, { "epoch": 0.7884215991692627, "grad_norm": 2.773495674133301, "learning_rate": 3.6092894208544128e-06, "loss": 0.08511839807033539, "step": 3037 }, { "epoch": 0.788681204569055, "grad_norm": 1.2018814086914062, "learning_rate": 3.608415041939579e-06, "loss": 0.10953633487224579, "step": 3038 }, { "epoch": 0.7889408099688473, "grad_norm": 1.606197476387024, "learning_rate": 3.6075405112766213e-06, "loss": 0.11512884497642517, "step": 3039 }, { "epoch": 0.7892004153686397, "grad_norm": 2.7824065685272217, "learning_rate": 3.606665829020137e-06, "loss": 0.10801713168621063, "step": 3040 }, { "epoch": 0.789460020768432, "grad_norm": 2.054959535598755, "learning_rate": 3.605790995324749e-06, "loss": 0.11544696241617203, "step": 3041 }, { "epoch": 0.7897196261682243, "grad_norm": 3.3552656173706055, "learning_rate": 3.604916010345107e-06, "loss": 0.07504107803106308, "step": 3042 }, { "epoch": 0.7899792315680166, "grad_norm": 1.3647633790969849, "learning_rate": 3.60404087423589e-06, "loss": 0.1054520308971405, "step": 3043 }, { "epoch": 0.7902388369678089, "grad_norm": 3.080144166946411, "learning_rate": 3.6031655871518017e-06, "loss": 0.09912461042404175, "step": 3044 }, { "epoch": 0.7904984423676013, "grad_norm": 1.1624826192855835, "learning_rate": 3.602290149247572e-06, "loss": 0.10191594809293747, "step": 3045 }, { "epoch": 0.7907580477673936, "grad_norm": 1.739915132522583, "learning_rate": 3.6014145606779586e-06, "loss": 0.09877707064151764, "step": 3046 }, { "epoch": 0.7910176531671859, "grad_norm": 3.8915042877197266, "learning_rate": 3.600538821597746e-06, "loss": 0.07205112278461456, "step": 3047 }, { "epoch": 0.7912772585669782, "grad_norm": 3.2214434146881104, "learning_rate": 3.599662932161745e-06, "loss": 0.10975973308086395, "step": 3048 }, { "epoch": 0.7915368639667705, "grad_norm": 3.2576584815979004, "learning_rate": 3.5987868925247916e-06, "loss": 0.09203645586967468, "step": 3049 }, { "epoch": 0.7917964693665628, "grad_norm": 2.070086717605591, "learning_rate": 3.597910702841751e-06, "loss": 0.1150725781917572, "step": 3050 }, { "epoch": 0.7920560747663551, "grad_norm": 3.5875439643859863, "learning_rate": 3.5970343632675115e-06, "loss": 0.06834519654512405, "step": 3051 }, { "epoch": 0.7923156801661475, "grad_norm": 2.5657272338867188, "learning_rate": 3.5961578739569926e-06, "loss": 0.1013827919960022, "step": 3052 }, { "epoch": 0.7925752855659398, "grad_norm": 3.9628875255584717, "learning_rate": 3.5952812350651346e-06, "loss": 0.10236360877752304, "step": 3053 }, { "epoch": 0.7928348909657321, "grad_norm": 1.8869860172271729, "learning_rate": 3.5944044467469095e-06, "loss": 0.09874256700277328, "step": 3054 }, { "epoch": 0.7930944963655244, "grad_norm": 2.068758010864258, "learning_rate": 3.5935275091573124e-06, "loss": 0.11209842562675476, "step": 3055 }, { "epoch": 0.7933541017653167, "grad_norm": 1.9463753700256348, "learning_rate": 3.5926504224513655e-06, "loss": 0.11798632144927979, "step": 3056 }, { "epoch": 0.793613707165109, "grad_norm": 1.6274899244308472, "learning_rate": 3.591773186784118e-06, "loss": 0.10931436717510223, "step": 3057 }, { "epoch": 0.7938733125649013, "grad_norm": 3.580794334411621, "learning_rate": 3.590895802310646e-06, "loss": 0.08437328040599823, "step": 3058 }, { "epoch": 0.7941329179646937, "grad_norm": 1.8874400854110718, "learning_rate": 3.590018269186049e-06, "loss": 0.0921873077750206, "step": 3059 }, { "epoch": 0.794392523364486, "grad_norm": 3.2365925312042236, "learning_rate": 3.589140587565456e-06, "loss": 0.06225676089525223, "step": 3060 }, { "epoch": 0.7946521287642783, "grad_norm": 1.6590244770050049, "learning_rate": 3.5882627576040212e-06, "loss": 0.10234035551548004, "step": 3061 }, { "epoch": 0.7949117341640706, "grad_norm": 3.63851261138916, "learning_rate": 3.587384779456925e-06, "loss": 0.11825715005397797, "step": 3062 }, { "epoch": 0.7951713395638629, "grad_norm": 1.0490630865097046, "learning_rate": 3.5865066532793737e-06, "loss": 0.09964064508676529, "step": 3063 }, { "epoch": 0.7954309449636553, "grad_norm": 2.6045544147491455, "learning_rate": 3.5856283792265984e-06, "loss": 0.12331333756446838, "step": 3064 }, { "epoch": 0.7956905503634476, "grad_norm": 3.258615016937256, "learning_rate": 3.5847499574538598e-06, "loss": 0.11909718811511993, "step": 3065 }, { "epoch": 0.7959501557632399, "grad_norm": 3.551915168762207, "learning_rate": 3.5838713881164413e-06, "loss": 0.11298912763595581, "step": 3066 }, { "epoch": 0.7962097611630322, "grad_norm": 0.9122962355613708, "learning_rate": 3.582992671369655e-06, "loss": 0.11960870027542114, "step": 3067 }, { "epoch": 0.7964693665628245, "grad_norm": 2.4174299240112305, "learning_rate": 3.5821138073688373e-06, "loss": 0.07200673222541809, "step": 3068 }, { "epoch": 0.7967289719626168, "grad_norm": 1.2406840324401855, "learning_rate": 3.581234796269351e-06, "loss": 0.0948893204331398, "step": 3069 }, { "epoch": 0.7969885773624091, "grad_norm": 3.241270065307617, "learning_rate": 3.5803556382265854e-06, "loss": 0.07427316904067993, "step": 3070 }, { "epoch": 0.7972481827622014, "grad_norm": 4.049853324890137, "learning_rate": 3.5794763333959564e-06, "loss": 0.10495506972074509, "step": 3071 }, { "epoch": 0.7975077881619937, "grad_norm": 2.568082809448242, "learning_rate": 3.578596881932904e-06, "loss": 0.11802726984024048, "step": 3072 }, { "epoch": 0.797767393561786, "grad_norm": 3.551081418991089, "learning_rate": 3.5777172839928943e-06, "loss": 0.1131201982498169, "step": 3073 }, { "epoch": 0.7980269989615784, "grad_norm": 2.441112518310547, "learning_rate": 3.5768375397314215e-06, "loss": 0.07503484189510345, "step": 3074 }, { "epoch": 0.7982866043613707, "grad_norm": 2.3512816429138184, "learning_rate": 3.5759576493040045e-06, "loss": 0.09006187319755554, "step": 3075 }, { "epoch": 0.7985462097611631, "grad_norm": 1.2638568878173828, "learning_rate": 3.5750776128661865e-06, "loss": 0.11159925162792206, "step": 3076 }, { "epoch": 0.7988058151609554, "grad_norm": 2.700911045074463, "learning_rate": 3.5741974305735377e-06, "loss": 0.1181270107626915, "step": 3077 }, { "epoch": 0.7990654205607477, "grad_norm": 3.42879581451416, "learning_rate": 3.5733171025816547e-06, "loss": 0.09702778607606888, "step": 3078 }, { "epoch": 0.79932502596054, "grad_norm": 1.1741958856582642, "learning_rate": 3.5724366290461604e-06, "loss": 0.09828900545835495, "step": 3079 }, { "epoch": 0.7995846313603323, "grad_norm": 4.61802864074707, "learning_rate": 3.5715560101227013e-06, "loss": 0.08620487153530121, "step": 3080 }, { "epoch": 0.7998442367601246, "grad_norm": 1.775658130645752, "learning_rate": 3.5706752459669506e-06, "loss": 0.11234187334775925, "step": 3081 }, { "epoch": 0.8001038421599169, "grad_norm": 2.527442455291748, "learning_rate": 3.5697943367346066e-06, "loss": 0.12596599757671356, "step": 3082 }, { "epoch": 0.8003634475597092, "grad_norm": 0.9785979390144348, "learning_rate": 3.5689132825813953e-06, "loss": 0.11772428452968597, "step": 3083 }, { "epoch": 0.8006230529595015, "grad_norm": 4.3885416984558105, "learning_rate": 3.5680320836630667e-06, "loss": 0.10190486907958984, "step": 3084 }, { "epoch": 0.8008826583592938, "grad_norm": 2.5228285789489746, "learning_rate": 3.567150740135396e-06, "loss": 0.09684579074382782, "step": 3085 }, { "epoch": 0.8011422637590861, "grad_norm": 1.8575495481491089, "learning_rate": 3.566269252154184e-06, "loss": 0.09510334581136703, "step": 3086 }, { "epoch": 0.8014018691588785, "grad_norm": 2.483375310897827, "learning_rate": 3.565387619875259e-06, "loss": 0.09493915736675262, "step": 3087 }, { "epoch": 0.8016614745586709, "grad_norm": 3.693053960800171, "learning_rate": 3.564505843454472e-06, "loss": 0.14164890348911285, "step": 3088 }, { "epoch": 0.8019210799584632, "grad_norm": 2.8553943634033203, "learning_rate": 3.5636239230477025e-06, "loss": 0.05781259387731552, "step": 3089 }, { "epoch": 0.8021806853582555, "grad_norm": 2.743440628051758, "learning_rate": 3.562741858810852e-06, "loss": 0.07834305614233017, "step": 3090 }, { "epoch": 0.8024402907580478, "grad_norm": 5.2179036140441895, "learning_rate": 3.561859650899851e-06, "loss": 0.08700898289680481, "step": 3091 }, { "epoch": 0.8026998961578401, "grad_norm": 3.1057393550872803, "learning_rate": 3.560977299470653e-06, "loss": 0.05610049143433571, "step": 3092 }, { "epoch": 0.8029595015576324, "grad_norm": 1.635635495185852, "learning_rate": 3.560094804679237e-06, "loss": 0.10319702327251434, "step": 3093 }, { "epoch": 0.8032191069574247, "grad_norm": 1.7203636169433594, "learning_rate": 3.5592121666816083e-06, "loss": 0.1108507513999939, "step": 3094 }, { "epoch": 0.803478712357217, "grad_norm": 1.8606107234954834, "learning_rate": 3.5583293856337963e-06, "loss": 0.09971112012863159, "step": 3095 }, { "epoch": 0.8037383177570093, "grad_norm": 3.2434744834899902, "learning_rate": 3.557446461691858e-06, "loss": 0.06787510216236115, "step": 3096 }, { "epoch": 0.8039979231568016, "grad_norm": 2.7350971698760986, "learning_rate": 3.5565633950118727e-06, "loss": 0.08428919315338135, "step": 3097 }, { "epoch": 0.8042575285565939, "grad_norm": 2.805492877960205, "learning_rate": 3.5556801857499474e-06, "loss": 0.10034428536891937, "step": 3098 }, { "epoch": 0.8045171339563862, "grad_norm": 1.4847712516784668, "learning_rate": 3.554796834062213e-06, "loss": 0.11695072054862976, "step": 3099 }, { "epoch": 0.8047767393561787, "grad_norm": 1.612877368927002, "learning_rate": 3.553913340104826e-06, "loss": 0.1032075360417366, "step": 3100 }, { "epoch": 0.805036344755971, "grad_norm": 4.164183616638184, "learning_rate": 3.5530297040339667e-06, "loss": 0.09927793592214584, "step": 3101 }, { "epoch": 0.8052959501557633, "grad_norm": 2.703950881958008, "learning_rate": 3.5521459260058435e-06, "loss": 0.06686782836914062, "step": 3102 }, { "epoch": 0.8055555555555556, "grad_norm": 1.53022301197052, "learning_rate": 3.5512620061766866e-06, "loss": 0.10778675973415375, "step": 3103 }, { "epoch": 0.8058151609553479, "grad_norm": 2.135795831680298, "learning_rate": 3.550377944702755e-06, "loss": 0.10016871988773346, "step": 3104 }, { "epoch": 0.8060747663551402, "grad_norm": 1.8811590671539307, "learning_rate": 3.549493741740327e-06, "loss": 0.10368974506855011, "step": 3105 }, { "epoch": 0.8063343717549325, "grad_norm": 1.9932022094726562, "learning_rate": 3.5486093974457133e-06, "loss": 0.10117236524820328, "step": 3106 }, { "epoch": 0.8065939771547248, "grad_norm": 2.7162957191467285, "learning_rate": 3.5477249119752426e-06, "loss": 0.10524043440818787, "step": 3107 }, { "epoch": 0.8068535825545171, "grad_norm": 2.551501989364624, "learning_rate": 3.546840285485273e-06, "loss": 0.1033266931772232, "step": 3108 }, { "epoch": 0.8071131879543094, "grad_norm": 2.521026849746704, "learning_rate": 3.5459555181321856e-06, "loss": 0.09798362851142883, "step": 3109 }, { "epoch": 0.8073727933541017, "grad_norm": 1.5618469715118408, "learning_rate": 3.545070610072389e-06, "loss": 0.09912095963954926, "step": 3110 }, { "epoch": 0.807632398753894, "grad_norm": 1.3894364833831787, "learning_rate": 3.5441855614623126e-06, "loss": 0.12128438800573349, "step": 3111 }, { "epoch": 0.8078920041536864, "grad_norm": 3.7238075733184814, "learning_rate": 3.5433003724584143e-06, "loss": 0.11319156736135483, "step": 3112 }, { "epoch": 0.8081516095534788, "grad_norm": 3.6596500873565674, "learning_rate": 3.5424150432171727e-06, "loss": 0.09841364622116089, "step": 3113 }, { "epoch": 0.8084112149532711, "grad_norm": 4.912769794464111, "learning_rate": 3.5415295738950965e-06, "loss": 0.09851647168397903, "step": 3114 }, { "epoch": 0.8086708203530634, "grad_norm": 2.5801920890808105, "learning_rate": 3.5406439646487144e-06, "loss": 0.09086352586746216, "step": 3115 }, { "epoch": 0.8089304257528557, "grad_norm": 3.404467821121216, "learning_rate": 3.539758215634584e-06, "loss": 0.07660047709941864, "step": 3116 }, { "epoch": 0.809190031152648, "grad_norm": 1.472694993019104, "learning_rate": 3.5388723270092824e-06, "loss": 0.10702046751976013, "step": 3117 }, { "epoch": 0.8094496365524403, "grad_norm": 3.8047232627868652, "learning_rate": 3.5379862989294174e-06, "loss": 0.09421013295650482, "step": 3118 }, { "epoch": 0.8097092419522326, "grad_norm": 2.8869364261627197, "learning_rate": 3.537100131551617e-06, "loss": 0.10835807025432587, "step": 3119 }, { "epoch": 0.8099688473520249, "grad_norm": 3.041673421859741, "learning_rate": 3.536213825032535e-06, "loss": 0.0829821452498436, "step": 3120 }, { "epoch": 0.8102284527518172, "grad_norm": 2.6580870151519775, "learning_rate": 3.535327379528851e-06, "loss": 0.09785380959510803, "step": 3121 }, { "epoch": 0.8104880581516095, "grad_norm": 1.7556310892105103, "learning_rate": 3.5344407951972687e-06, "loss": 0.08377068489789963, "step": 3122 }, { "epoch": 0.8107476635514018, "grad_norm": 2.7886738777160645, "learning_rate": 3.5335540721945145e-06, "loss": 0.13305892050266266, "step": 3123 }, { "epoch": 0.8110072689511942, "grad_norm": 3.227242946624756, "learning_rate": 3.532667210677341e-06, "loss": 0.1051730215549469, "step": 3124 }, { "epoch": 0.8112668743509865, "grad_norm": 3.510571241378784, "learning_rate": 3.531780210802526e-06, "loss": 0.0867043063044548, "step": 3125 }, { "epoch": 0.8115264797507789, "grad_norm": 2.8208634853363037, "learning_rate": 3.5308930727268693e-06, "loss": 0.11225098371505737, "step": 3126 }, { "epoch": 0.8117860851505712, "grad_norm": 3.4610936641693115, "learning_rate": 3.530005796607198e-06, "loss": 0.11076037585735321, "step": 3127 }, { "epoch": 0.8120456905503635, "grad_norm": 5.820727825164795, "learning_rate": 3.529118382600361e-06, "loss": 0.12931384146213531, "step": 3128 }, { "epoch": 0.8123052959501558, "grad_norm": 1.0435541868209839, "learning_rate": 3.5282308308632348e-06, "loss": 0.10874348133802414, "step": 3129 }, { "epoch": 0.8125649013499481, "grad_norm": 3.858350992202759, "learning_rate": 3.527343141552715e-06, "loss": 0.08135294169187546, "step": 3130 }, { "epoch": 0.8128245067497404, "grad_norm": 1.4569789171218872, "learning_rate": 3.526455314825728e-06, "loss": 0.11244998872280121, "step": 3131 }, { "epoch": 0.8130841121495327, "grad_norm": 1.3324227333068848, "learning_rate": 3.525567350839219e-06, "loss": 0.10069162398576736, "step": 3132 }, { "epoch": 0.813343717549325, "grad_norm": 2.7823965549468994, "learning_rate": 3.5246792497501615e-06, "loss": 0.11852237582206726, "step": 3133 }, { "epoch": 0.8136033229491173, "grad_norm": 2.832019329071045, "learning_rate": 3.5237910117155494e-06, "loss": 0.08846433460712433, "step": 3134 }, { "epoch": 0.8138629283489096, "grad_norm": 3.2163193225860596, "learning_rate": 3.5229026368924043e-06, "loss": 0.09344912320375443, "step": 3135 }, { "epoch": 0.814122533748702, "grad_norm": 2.8125667572021484, "learning_rate": 3.5220141254377697e-06, "loss": 0.044414322823286057, "step": 3136 }, { "epoch": 0.8143821391484943, "grad_norm": 2.2272019386291504, "learning_rate": 3.521125477508715e-06, "loss": 0.11180514842271805, "step": 3137 }, { "epoch": 0.8146417445482866, "grad_norm": 2.0429744720458984, "learning_rate": 3.5202366932623327e-06, "loss": 0.10835283994674683, "step": 3138 }, { "epoch": 0.814901349948079, "grad_norm": 0.9017481207847595, "learning_rate": 3.519347772855738e-06, "loss": 0.1045888140797615, "step": 3139 }, { "epoch": 0.8151609553478713, "grad_norm": 3.2215843200683594, "learning_rate": 3.518458716446074e-06, "loss": 0.08997808396816254, "step": 3140 }, { "epoch": 0.8154205607476636, "grad_norm": 1.8477412462234497, "learning_rate": 3.517569524190503e-06, "loss": 0.10052371770143509, "step": 3141 }, { "epoch": 0.8156801661474559, "grad_norm": 2.7297115325927734, "learning_rate": 3.516680196246216e-06, "loss": 0.1111760139465332, "step": 3142 }, { "epoch": 0.8159397715472482, "grad_norm": 2.1537280082702637, "learning_rate": 3.5157907327704244e-06, "loss": 0.11232058703899384, "step": 3143 }, { "epoch": 0.8161993769470405, "grad_norm": 4.923956394195557, "learning_rate": 3.514901133920366e-06, "loss": 0.10723499953746796, "step": 3144 }, { "epoch": 0.8164589823468328, "grad_norm": 1.1775063276290894, "learning_rate": 3.5140113998533015e-06, "loss": 0.11605291068553925, "step": 3145 }, { "epoch": 0.8167185877466251, "grad_norm": 1.5453096628189087, "learning_rate": 3.513121530726515e-06, "loss": 0.10879356414079666, "step": 3146 }, { "epoch": 0.8169781931464174, "grad_norm": 1.4733694791793823, "learning_rate": 3.5122315266973146e-06, "loss": 0.11641211062669754, "step": 3147 }, { "epoch": 0.8172377985462098, "grad_norm": 3.462418794631958, "learning_rate": 3.511341387923032e-06, "loss": 0.06970974057912827, "step": 3148 }, { "epoch": 0.8174974039460021, "grad_norm": 4.035385608673096, "learning_rate": 3.5104511145610255e-06, "loss": 0.12066319584846497, "step": 3149 }, { "epoch": 0.8177570093457944, "grad_norm": 3.8617124557495117, "learning_rate": 3.509560706768674e-06, "loss": 0.10655723512172699, "step": 3150 }, { "epoch": 0.8180166147455867, "grad_norm": 1.4256528615951538, "learning_rate": 3.508670164703382e-06, "loss": 0.11228343844413757, "step": 3151 }, { "epoch": 0.818276220145379, "grad_norm": 2.2769997119903564, "learning_rate": 3.5077794885225747e-06, "loss": 0.09835906326770782, "step": 3152 }, { "epoch": 0.8185358255451713, "grad_norm": 2.2660300731658936, "learning_rate": 3.5068886783837047e-06, "loss": 0.10084262490272522, "step": 3153 }, { "epoch": 0.8187954309449637, "grad_norm": 3.6687402725219727, "learning_rate": 3.505997734444248e-06, "loss": 0.062310464680194855, "step": 3154 }, { "epoch": 0.819055036344756, "grad_norm": 4.483529090881348, "learning_rate": 3.505106656861701e-06, "loss": 0.09336303174495697, "step": 3155 }, { "epoch": 0.8193146417445483, "grad_norm": 2.5951552391052246, "learning_rate": 3.5042154457935867e-06, "loss": 0.10164127498865128, "step": 3156 }, { "epoch": 0.8195742471443406, "grad_norm": 2.9802074432373047, "learning_rate": 3.5033241013974512e-06, "loss": 0.06361529231071472, "step": 3157 }, { "epoch": 0.8198338525441329, "grad_norm": 2.5135021209716797, "learning_rate": 3.502432623830863e-06, "loss": 0.11073184013366699, "step": 3158 }, { "epoch": 0.8200934579439252, "grad_norm": 2.839215040206909, "learning_rate": 3.5015410132514153e-06, "loss": 0.07471852004528046, "step": 3159 }, { "epoch": 0.8203530633437176, "grad_norm": 1.153519868850708, "learning_rate": 3.5006492698167256e-06, "loss": 0.11111534386873245, "step": 3160 }, { "epoch": 0.8206126687435099, "grad_norm": 1.881235122680664, "learning_rate": 3.499757393684431e-06, "loss": 0.08745106309652328, "step": 3161 }, { "epoch": 0.8208722741433022, "grad_norm": 1.3821781873703003, "learning_rate": 3.4988653850121975e-06, "loss": 0.11050866544246674, "step": 3162 }, { "epoch": 0.8211318795430945, "grad_norm": 1.1894410848617554, "learning_rate": 3.4979732439577103e-06, "loss": 0.11606267839670181, "step": 3163 }, { "epoch": 0.8213914849428868, "grad_norm": 1.8060494661331177, "learning_rate": 3.4970809706786802e-06, "loss": 0.1280202567577362, "step": 3164 }, { "epoch": 0.8216510903426791, "grad_norm": 2.0025014877319336, "learning_rate": 3.49618856533284e-06, "loss": 0.08936047554016113, "step": 3165 }, { "epoch": 0.8219106957424714, "grad_norm": 3.5182321071624756, "learning_rate": 3.4952960280779463e-06, "loss": 0.06107208877801895, "step": 3166 }, { "epoch": 0.8221703011422637, "grad_norm": 1.9516398906707764, "learning_rate": 3.494403359071781e-06, "loss": 0.10835623741149902, "step": 3167 }, { "epoch": 0.822429906542056, "grad_norm": 4.723561763763428, "learning_rate": 3.4935105584721455e-06, "loss": 0.12387724965810776, "step": 3168 }, { "epoch": 0.8226895119418484, "grad_norm": 2.9264492988586426, "learning_rate": 3.4926176264368683e-06, "loss": 0.09562981128692627, "step": 3169 }, { "epoch": 0.8229491173416407, "grad_norm": 2.927402973175049, "learning_rate": 3.491724563123798e-06, "loss": 0.06004648655653, "step": 3170 }, { "epoch": 0.823208722741433, "grad_norm": 1.074096441268921, "learning_rate": 3.4908313686908074e-06, "loss": 0.1097528487443924, "step": 3171 }, { "epoch": 0.8234683281412254, "grad_norm": 1.1532106399536133, "learning_rate": 3.4899380432957948e-06, "loss": 0.1063658595085144, "step": 3172 }, { "epoch": 0.8237279335410177, "grad_norm": 1.8446227312088013, "learning_rate": 3.4890445870966783e-06, "loss": 0.10849609971046448, "step": 3173 }, { "epoch": 0.82398753894081, "grad_norm": 4.938416004180908, "learning_rate": 3.4881510002514e-06, "loss": 0.0721956267952919, "step": 3174 }, { "epoch": 0.8242471443406023, "grad_norm": 1.0959088802337646, "learning_rate": 3.487257282917927e-06, "loss": 0.10568992793560028, "step": 3175 }, { "epoch": 0.8245067497403946, "grad_norm": 1.127467155456543, "learning_rate": 3.4863634352542465e-06, "loss": 0.1067514643073082, "step": 3176 }, { "epoch": 0.8247663551401869, "grad_norm": 2.4168217182159424, "learning_rate": 3.4854694574183716e-06, "loss": 0.07940662652254105, "step": 3177 }, { "epoch": 0.8250259605399792, "grad_norm": 2.046720504760742, "learning_rate": 3.4845753495683366e-06, "loss": 0.09474382549524307, "step": 3178 }, { "epoch": 0.8252855659397715, "grad_norm": 1.7865831851959229, "learning_rate": 3.4836811118621995e-06, "loss": 0.11396799236536026, "step": 3179 }, { "epoch": 0.8255451713395638, "grad_norm": 2.8241896629333496, "learning_rate": 3.482786744458041e-06, "loss": 0.07334466278553009, "step": 3180 }, { "epoch": 0.8258047767393562, "grad_norm": 1.89207124710083, "learning_rate": 3.4818922475139643e-06, "loss": 0.09700317680835724, "step": 3181 }, { "epoch": 0.8260643821391485, "grad_norm": 3.400149345397949, "learning_rate": 3.4809976211880963e-06, "loss": 0.10612291842699051, "step": 3182 }, { "epoch": 0.8263239875389408, "grad_norm": 2.0546412467956543, "learning_rate": 3.4801028656385867e-06, "loss": 0.10519750416278839, "step": 3183 }, { "epoch": 0.8265835929387332, "grad_norm": 3.3138928413391113, "learning_rate": 3.4792079810236078e-06, "loss": 0.07538927346467972, "step": 3184 }, { "epoch": 0.8268431983385255, "grad_norm": 2.8639450073242188, "learning_rate": 3.4783129675013545e-06, "loss": 0.10413146018981934, "step": 3185 }, { "epoch": 0.8271028037383178, "grad_norm": 2.782681465148926, "learning_rate": 3.4774178252300445e-06, "loss": 0.12337732315063477, "step": 3186 }, { "epoch": 0.8273624091381101, "grad_norm": 1.331648349761963, "learning_rate": 3.476522554367918e-06, "loss": 0.10756801068782806, "step": 3187 }, { "epoch": 0.8276220145379024, "grad_norm": 2.6579039096832275, "learning_rate": 3.4756271550732397e-06, "loss": 0.09304298460483551, "step": 3188 }, { "epoch": 0.8278816199376947, "grad_norm": 1.7008917331695557, "learning_rate": 3.474731627504296e-06, "loss": 0.10484129935503006, "step": 3189 }, { "epoch": 0.828141225337487, "grad_norm": 3.0118682384490967, "learning_rate": 3.4738359718193932e-06, "loss": 0.07210870087146759, "step": 3190 }, { "epoch": 0.8284008307372793, "grad_norm": 1.4500380754470825, "learning_rate": 3.4729401881768655e-06, "loss": 0.10181548446416855, "step": 3191 }, { "epoch": 0.8286604361370716, "grad_norm": 1.7196515798568726, "learning_rate": 3.4720442767350637e-06, "loss": 0.10044166445732117, "step": 3192 }, { "epoch": 0.8289200415368639, "grad_norm": 3.470391035079956, "learning_rate": 3.4711482376523687e-06, "loss": 0.09362506866455078, "step": 3193 }, { "epoch": 0.8291796469366562, "grad_norm": 2.069002866744995, "learning_rate": 3.4702520710871757e-06, "loss": 0.09558701515197754, "step": 3194 }, { "epoch": 0.8294392523364486, "grad_norm": 2.616335391998291, "learning_rate": 3.469355777197909e-06, "loss": 0.10880221426486969, "step": 3195 }, { "epoch": 0.829698857736241, "grad_norm": 3.137407064437866, "learning_rate": 3.468459356143012e-06, "loss": 0.10462237149477005, "step": 3196 }, { "epoch": 0.8299584631360333, "grad_norm": 1.2559493780136108, "learning_rate": 3.4675628080809508e-06, "loss": 0.11454613506793976, "step": 3197 }, { "epoch": 0.8302180685358256, "grad_norm": 2.5996692180633545, "learning_rate": 3.466666133170215e-06, "loss": 0.1001986414194107, "step": 3198 }, { "epoch": 0.8304776739356179, "grad_norm": 2.9334025382995605, "learning_rate": 3.4657693315693177e-06, "loss": 0.07563033699989319, "step": 3199 }, { "epoch": 0.8307372793354102, "grad_norm": 3.4418141841888428, "learning_rate": 3.4648724034367903e-06, "loss": 0.07411079108715057, "step": 3200 }, { "epoch": 0.8309968847352025, "grad_norm": 4.230822563171387, "learning_rate": 3.4639753489311905e-06, "loss": 0.06261239945888519, "step": 3201 }, { "epoch": 0.8312564901349948, "grad_norm": 3.4382688999176025, "learning_rate": 3.4630781682110975e-06, "loss": 0.04889526218175888, "step": 3202 }, { "epoch": 0.8315160955347871, "grad_norm": 3.4057700634002686, "learning_rate": 3.462180861435111e-06, "loss": 0.07225097715854645, "step": 3203 }, { "epoch": 0.8317757009345794, "grad_norm": 3.1474571228027344, "learning_rate": 3.4612834287618556e-06, "loss": 0.10154461860656738, "step": 3204 }, { "epoch": 0.8320353063343717, "grad_norm": 2.971553325653076, "learning_rate": 3.4603858703499747e-06, "loss": 0.11578083038330078, "step": 3205 }, { "epoch": 0.832294911734164, "grad_norm": 1.2304573059082031, "learning_rate": 3.4594881863581397e-06, "loss": 0.11766916513442993, "step": 3206 }, { "epoch": 0.8325545171339563, "grad_norm": 1.2971638441085815, "learning_rate": 3.4585903769450372e-06, "loss": 0.11348866671323776, "step": 3207 }, { "epoch": 0.8328141225337488, "grad_norm": 4.24677848815918, "learning_rate": 3.4576924422693813e-06, "loss": 0.1084866151213646, "step": 3208 }, { "epoch": 0.8330737279335411, "grad_norm": 6.017414093017578, "learning_rate": 3.4567943824899055e-06, "loss": 0.09127038717269897, "step": 3209 }, { "epoch": 0.8333333333333334, "grad_norm": 1.1653039455413818, "learning_rate": 3.455896197765366e-06, "loss": 0.09598015248775482, "step": 3210 }, { "epoch": 0.8335929387331257, "grad_norm": 4.147595405578613, "learning_rate": 3.454997888254541e-06, "loss": 0.10821082442998886, "step": 3211 }, { "epoch": 0.833852544132918, "grad_norm": 3.681870460510254, "learning_rate": 3.4540994541162333e-06, "loss": 0.06187322735786438, "step": 3212 }, { "epoch": 0.8341121495327103, "grad_norm": 3.598620653152466, "learning_rate": 3.4532008955092627e-06, "loss": 0.05544571951031685, "step": 3213 }, { "epoch": 0.8343717549325026, "grad_norm": 5.067573547363281, "learning_rate": 3.452302212592475e-06, "loss": 0.11966492235660553, "step": 3214 }, { "epoch": 0.8346313603322949, "grad_norm": 4.22280740737915, "learning_rate": 3.4514034055247366e-06, "loss": 0.10154921561479568, "step": 3215 }, { "epoch": 0.8348909657320872, "grad_norm": 2.988894462585449, "learning_rate": 3.450504474464937e-06, "loss": 0.06244547665119171, "step": 3216 }, { "epoch": 0.8351505711318795, "grad_norm": 2.3161141872406006, "learning_rate": 3.4496054195719847e-06, "loss": 0.1151372417807579, "step": 3217 }, { "epoch": 0.8354101765316718, "grad_norm": 3.187459707260132, "learning_rate": 3.4487062410048135e-06, "loss": 0.10828912258148193, "step": 3218 }, { "epoch": 0.8356697819314641, "grad_norm": 2.6390066146850586, "learning_rate": 3.447806938922377e-06, "loss": 0.09027349948883057, "step": 3219 }, { "epoch": 0.8359293873312564, "grad_norm": 4.3210859298706055, "learning_rate": 3.4469075134836516e-06, "loss": 0.13967251777648926, "step": 3220 }, { "epoch": 0.8361889927310489, "grad_norm": 3.8049802780151367, "learning_rate": 3.4460079648476354e-06, "loss": 0.11388306319713593, "step": 3221 }, { "epoch": 0.8364485981308412, "grad_norm": 2.5633931159973145, "learning_rate": 3.4451082931733463e-06, "loss": 0.06507068872451782, "step": 3222 }, { "epoch": 0.8367082035306335, "grad_norm": 2.5947301387786865, "learning_rate": 3.444208498619827e-06, "loss": 0.09883345663547516, "step": 3223 }, { "epoch": 0.8369678089304258, "grad_norm": 1.2568707466125488, "learning_rate": 3.4433085813461407e-06, "loss": 0.10665461421012878, "step": 3224 }, { "epoch": 0.8372274143302181, "grad_norm": 1.7386674880981445, "learning_rate": 3.4424085415113722e-06, "loss": 0.09594065696001053, "step": 3225 }, { "epoch": 0.8374870197300104, "grad_norm": 1.1013284921646118, "learning_rate": 3.4415083792746284e-06, "loss": 0.10449261218309402, "step": 3226 }, { "epoch": 0.8377466251298027, "grad_norm": 1.4860293865203857, "learning_rate": 3.440608094795035e-06, "loss": 0.09953641891479492, "step": 3227 }, { "epoch": 0.838006230529595, "grad_norm": 2.652709722518921, "learning_rate": 3.4397076882317438e-06, "loss": 0.07636306434869766, "step": 3228 }, { "epoch": 0.8382658359293873, "grad_norm": 1.7976821660995483, "learning_rate": 3.438807159743926e-06, "loss": 0.08361149579286575, "step": 3229 }, { "epoch": 0.8385254413291796, "grad_norm": 1.0960252285003662, "learning_rate": 3.4379065094907734e-06, "loss": 0.10233230888843536, "step": 3230 }, { "epoch": 0.8387850467289719, "grad_norm": 1.852165937423706, "learning_rate": 3.4370057376315e-06, "loss": 0.12435039132833481, "step": 3231 }, { "epoch": 0.8390446521287642, "grad_norm": 3.0784430503845215, "learning_rate": 3.4361048443253443e-06, "loss": 0.08322489261627197, "step": 3232 }, { "epoch": 0.8393042575285566, "grad_norm": 2.6712570190429688, "learning_rate": 3.4352038297315616e-06, "loss": 0.09489075094461441, "step": 3233 }, { "epoch": 0.839563862928349, "grad_norm": 2.0175416469573975, "learning_rate": 3.4343026940094297e-06, "loss": 0.07408430427312851, "step": 3234 }, { "epoch": 0.8398234683281413, "grad_norm": 3.643214464187622, "learning_rate": 3.433401437318251e-06, "loss": 0.1291552484035492, "step": 3235 }, { "epoch": 0.8400830737279336, "grad_norm": 1.9298104047775269, "learning_rate": 3.4325000598173454e-06, "loss": 0.09809579700231552, "step": 3236 }, { "epoch": 0.8403426791277259, "grad_norm": 1.1826763153076172, "learning_rate": 3.4315985616660572e-06, "loss": 0.11056619137525558, "step": 3237 }, { "epoch": 0.8406022845275182, "grad_norm": 3.1782777309417725, "learning_rate": 3.430696943023749e-06, "loss": 0.08508563041687012, "step": 3238 }, { "epoch": 0.8408618899273105, "grad_norm": 3.6980440616607666, "learning_rate": 3.429795204049808e-06, "loss": 0.10593272745609283, "step": 3239 }, { "epoch": 0.8411214953271028, "grad_norm": 2.690756320953369, "learning_rate": 3.428893344903639e-06, "loss": 0.07956257462501526, "step": 3240 }, { "epoch": 0.8413811007268951, "grad_norm": 4.5212202072143555, "learning_rate": 3.427991365744672e-06, "loss": 0.07006420195102692, "step": 3241 }, { "epoch": 0.8416407061266874, "grad_norm": 2.462413787841797, "learning_rate": 3.4270892667323553e-06, "loss": 0.10556433349847794, "step": 3242 }, { "epoch": 0.8419003115264797, "grad_norm": 2.8630130290985107, "learning_rate": 3.42618704802616e-06, "loss": 0.10686361789703369, "step": 3243 }, { "epoch": 0.842159916926272, "grad_norm": 2.4886679649353027, "learning_rate": 3.4252847097855764e-06, "loss": 0.10949264466762543, "step": 3244 }, { "epoch": 0.8424195223260644, "grad_norm": 3.596330165863037, "learning_rate": 3.4243822521701183e-06, "loss": 0.07692268490791321, "step": 3245 }, { "epoch": 0.8426791277258567, "grad_norm": 2.1271743774414062, "learning_rate": 3.4234796753393197e-06, "loss": 0.09919354319572449, "step": 3246 }, { "epoch": 0.842938733125649, "grad_norm": 1.98509681224823, "learning_rate": 3.4225769794527354e-06, "loss": 0.09600389003753662, "step": 3247 }, { "epoch": 0.8431983385254413, "grad_norm": 1.9130356311798096, "learning_rate": 3.4216741646699396e-06, "loss": 0.09236568212509155, "step": 3248 }, { "epoch": 0.8434579439252337, "grad_norm": 3.542762041091919, "learning_rate": 3.420771231150532e-06, "loss": 0.049964144825935364, "step": 3249 }, { "epoch": 0.843717549325026, "grad_norm": 3.20415997505188, "learning_rate": 3.4198681790541296e-06, "loss": 0.08319489657878876, "step": 3250 }, { "epoch": 0.8439771547248183, "grad_norm": 3.6096744537353516, "learning_rate": 3.418965008540371e-06, "loss": 0.09132485091686249, "step": 3251 }, { "epoch": 0.8442367601246106, "grad_norm": 2.2919962406158447, "learning_rate": 3.418061719768916e-06, "loss": 0.1021500900387764, "step": 3252 }, { "epoch": 0.8444963655244029, "grad_norm": 2.6772983074188232, "learning_rate": 3.4171583128994453e-06, "loss": 0.08372238278388977, "step": 3253 }, { "epoch": 0.8447559709241952, "grad_norm": 2.344167709350586, "learning_rate": 3.41625478809166e-06, "loss": 0.11768631637096405, "step": 3254 }, { "epoch": 0.8450155763239875, "grad_norm": 3.7808969020843506, "learning_rate": 3.4153511455052844e-06, "loss": 0.11714616417884827, "step": 3255 }, { "epoch": 0.8452751817237798, "grad_norm": 2.1015560626983643, "learning_rate": 3.414447385300061e-06, "loss": 0.10412711650133133, "step": 3256 }, { "epoch": 0.8455347871235722, "grad_norm": 3.7245731353759766, "learning_rate": 3.4135435076357537e-06, "loss": 0.09050905704498291, "step": 3257 }, { "epoch": 0.8457943925233645, "grad_norm": 2.967721939086914, "learning_rate": 3.412639512672147e-06, "loss": 0.059012219309806824, "step": 3258 }, { "epoch": 0.8460539979231568, "grad_norm": 2.2360453605651855, "learning_rate": 3.411735400569046e-06, "loss": 0.11365523189306259, "step": 3259 }, { "epoch": 0.8463136033229491, "grad_norm": 3.371739387512207, "learning_rate": 3.41083117148628e-06, "loss": 0.11882872879505157, "step": 3260 }, { "epoch": 0.8465732087227414, "grad_norm": 3.51747989654541, "learning_rate": 3.409926825583692e-06, "loss": 0.09240303188562393, "step": 3261 }, { "epoch": 0.8468328141225338, "grad_norm": 3.2642323970794678, "learning_rate": 3.4090223630211523e-06, "loss": 0.0708475261926651, "step": 3262 }, { "epoch": 0.8470924195223261, "grad_norm": 2.1870603561401367, "learning_rate": 3.4081177839585485e-06, "loss": 0.11289942264556885, "step": 3263 }, { "epoch": 0.8473520249221184, "grad_norm": 1.335564374923706, "learning_rate": 3.407213088555789e-06, "loss": 0.09744168072938919, "step": 3264 }, { "epoch": 0.8476116303219107, "grad_norm": 2.7074718475341797, "learning_rate": 3.4063082769728045e-06, "loss": 0.0781979039311409, "step": 3265 }, { "epoch": 0.847871235721703, "grad_norm": 4.22532844543457, "learning_rate": 3.405403349369544e-06, "loss": 0.07093390077352524, "step": 3266 }, { "epoch": 0.8481308411214953, "grad_norm": 3.833172559738159, "learning_rate": 3.4044983059059766e-06, "loss": 0.07448619604110718, "step": 3267 }, { "epoch": 0.8483904465212876, "grad_norm": 2.882520914077759, "learning_rate": 3.4035931467420964e-06, "loss": 0.12260347604751587, "step": 3268 }, { "epoch": 0.84865005192108, "grad_norm": 2.612701177597046, "learning_rate": 3.402687872037912e-06, "loss": 0.08725111186504364, "step": 3269 }, { "epoch": 0.8489096573208723, "grad_norm": 3.6873011589050293, "learning_rate": 3.401782481953457e-06, "loss": 0.04081885516643524, "step": 3270 }, { "epoch": 0.8491692627206646, "grad_norm": 2.262361526489258, "learning_rate": 3.4008769766487826e-06, "loss": 0.10765595734119415, "step": 3271 }, { "epoch": 0.8494288681204569, "grad_norm": 1.7101101875305176, "learning_rate": 3.399971356283963e-06, "loss": 0.11014223098754883, "step": 3272 }, { "epoch": 0.8496884735202492, "grad_norm": 3.3358030319213867, "learning_rate": 3.399065621019089e-06, "loss": 0.0497620552778244, "step": 3273 }, { "epoch": 0.8499480789200415, "grad_norm": 2.579113721847534, "learning_rate": 3.3981597710142757e-06, "loss": 0.08986924588680267, "step": 3274 }, { "epoch": 0.8502076843198338, "grad_norm": 2.553670883178711, "learning_rate": 3.3972538064296545e-06, "loss": 0.10735590755939484, "step": 3275 }, { "epoch": 0.8504672897196262, "grad_norm": 1.6120482683181763, "learning_rate": 3.396347727425381e-06, "loss": 0.10650384426116943, "step": 3276 }, { "epoch": 0.8507268951194185, "grad_norm": 1.8498172760009766, "learning_rate": 3.3954415341616285e-06, "loss": 0.11324034631252289, "step": 3277 }, { "epoch": 0.8509865005192108, "grad_norm": 3.9378113746643066, "learning_rate": 3.394535226798592e-06, "loss": 0.05961160734295845, "step": 3278 }, { "epoch": 0.8512461059190031, "grad_norm": 2.99786114692688, "learning_rate": 3.393628805496485e-06, "loss": 0.08840987086296082, "step": 3279 }, { "epoch": 0.8515057113187954, "grad_norm": 2.0733048915863037, "learning_rate": 3.3927222704155415e-06, "loss": 0.11487150192260742, "step": 3280 }, { "epoch": 0.8517653167185878, "grad_norm": 2.356922149658203, "learning_rate": 3.3918156217160182e-06, "loss": 0.09628163278102875, "step": 3281 }, { "epoch": 0.8520249221183801, "grad_norm": 2.0906026363372803, "learning_rate": 3.3909088595581884e-06, "loss": 0.10686805844306946, "step": 3282 }, { "epoch": 0.8522845275181724, "grad_norm": 1.8042415380477905, "learning_rate": 3.3900019841023476e-06, "loss": 0.09156176447868347, "step": 3283 }, { "epoch": 0.8525441329179647, "grad_norm": 2.0881340503692627, "learning_rate": 3.389094995508809e-06, "loss": 0.08996759355068207, "step": 3284 }, { "epoch": 0.852803738317757, "grad_norm": 4.79214334487915, "learning_rate": 3.3881878939379086e-06, "loss": 0.09361228346824646, "step": 3285 }, { "epoch": 0.8530633437175493, "grad_norm": 3.5045785903930664, "learning_rate": 3.3872806795500014e-06, "loss": 0.070243239402771, "step": 3286 }, { "epoch": 0.8533229491173416, "grad_norm": 2.467745065689087, "learning_rate": 3.386373352505462e-06, "loss": 0.08650142699480057, "step": 3287 }, { "epoch": 0.8535825545171339, "grad_norm": 1.1167913675308228, "learning_rate": 3.385465912964685e-06, "loss": 0.11998043209314346, "step": 3288 }, { "epoch": 0.8538421599169262, "grad_norm": 3.6735470294952393, "learning_rate": 3.3845583610880844e-06, "loss": 0.0820886492729187, "step": 3289 }, { "epoch": 0.8541017653167186, "grad_norm": 3.305781364440918, "learning_rate": 3.383650697036095e-06, "loss": 0.069145068526268, "step": 3290 }, { "epoch": 0.8543613707165109, "grad_norm": 1.8486928939819336, "learning_rate": 3.3827429209691722e-06, "loss": 0.1057630181312561, "step": 3291 }, { "epoch": 0.8546209761163032, "grad_norm": 2.9778738021850586, "learning_rate": 3.3818350330477874e-06, "loss": 0.11087633669376373, "step": 3292 }, { "epoch": 0.8548805815160956, "grad_norm": 2.596025228500366, "learning_rate": 3.3809270334324364e-06, "loss": 0.11956492066383362, "step": 3293 }, { "epoch": 0.8551401869158879, "grad_norm": 1.9650355577468872, "learning_rate": 3.3800189222836316e-06, "loss": 0.09371155500411987, "step": 3294 }, { "epoch": 0.8553997923156802, "grad_norm": 2.135634660720825, "learning_rate": 3.3791106997619078e-06, "loss": 0.08428966253995895, "step": 3295 }, { "epoch": 0.8556593977154725, "grad_norm": 2.1900675296783447, "learning_rate": 3.378202366027817e-06, "loss": 0.1086149662733078, "step": 3296 }, { "epoch": 0.8559190031152648, "grad_norm": 3.2220637798309326, "learning_rate": 3.3772939212419314e-06, "loss": 0.12388548254966736, "step": 3297 }, { "epoch": 0.8561786085150571, "grad_norm": 2.5803282260894775, "learning_rate": 3.3763853655648436e-06, "loss": 0.06991089880466461, "step": 3298 }, { "epoch": 0.8564382139148494, "grad_norm": 2.8968117237091064, "learning_rate": 3.3754766991571657e-06, "loss": 0.050346437841653824, "step": 3299 }, { "epoch": 0.8566978193146417, "grad_norm": 3.156750202178955, "learning_rate": 3.3745679221795284e-06, "loss": 0.08305796980857849, "step": 3300 }, { "epoch": 0.856957424714434, "grad_norm": 1.8762311935424805, "learning_rate": 3.3736590347925842e-06, "loss": 0.0997258722782135, "step": 3301 }, { "epoch": 0.8572170301142263, "grad_norm": 1.5964587926864624, "learning_rate": 3.3727500371570014e-06, "loss": 0.0926867350935936, "step": 3302 }, { "epoch": 0.8574766355140186, "grad_norm": 1.3308483362197876, "learning_rate": 3.371840929433472e-06, "loss": 0.13423886895179749, "step": 3303 }, { "epoch": 0.857736240913811, "grad_norm": 1.9488080739974976, "learning_rate": 3.370931711782704e-06, "loss": 0.08706329762935638, "step": 3304 }, { "epoch": 0.8579958463136034, "grad_norm": 2.7813186645507812, "learning_rate": 3.370022384365427e-06, "loss": 0.10325184464454651, "step": 3305 }, { "epoch": 0.8582554517133957, "grad_norm": 1.5304410457611084, "learning_rate": 3.3691129473423895e-06, "loss": 0.09780575335025787, "step": 3306 }, { "epoch": 0.858515057113188, "grad_norm": 4.3861188888549805, "learning_rate": 3.368203400874358e-06, "loss": 0.09929993748664856, "step": 3307 }, { "epoch": 0.8587746625129803, "grad_norm": 1.4085694551467896, "learning_rate": 3.3672937451221206e-06, "loss": 0.09328845143318176, "step": 3308 }, { "epoch": 0.8590342679127726, "grad_norm": 4.026661396026611, "learning_rate": 3.366383980246483e-06, "loss": 0.07267071306705475, "step": 3309 }, { "epoch": 0.8592938733125649, "grad_norm": 3.940964937210083, "learning_rate": 3.3654741064082712e-06, "loss": 0.07721388339996338, "step": 3310 }, { "epoch": 0.8595534787123572, "grad_norm": 3.1676366329193115, "learning_rate": 3.3645641237683295e-06, "loss": 0.10643068701028824, "step": 3311 }, { "epoch": 0.8598130841121495, "grad_norm": 2.8873603343963623, "learning_rate": 3.363654032487523e-06, "loss": 0.09465242177248001, "step": 3312 }, { "epoch": 0.8600726895119418, "grad_norm": 1.2975306510925293, "learning_rate": 3.362743832726735e-06, "loss": 0.09749461710453033, "step": 3313 }, { "epoch": 0.8603322949117341, "grad_norm": 3.9013943672180176, "learning_rate": 3.3618335246468657e-06, "loss": 0.05910643935203552, "step": 3314 }, { "epoch": 0.8605919003115264, "grad_norm": 2.8054144382476807, "learning_rate": 3.360923108408839e-06, "loss": 0.07618468999862671, "step": 3315 }, { "epoch": 0.8608515057113187, "grad_norm": 1.0063657760620117, "learning_rate": 3.3600125841735955e-06, "loss": 0.11660123616456985, "step": 3316 }, { "epoch": 0.8611111111111112, "grad_norm": 1.7061681747436523, "learning_rate": 3.359101952102094e-06, "loss": 0.10471618920564651, "step": 3317 }, { "epoch": 0.8613707165109035, "grad_norm": 4.118768692016602, "learning_rate": 3.3581912123553144e-06, "loss": 0.062274254858493805, "step": 3318 }, { "epoch": 0.8616303219106958, "grad_norm": 3.0237324237823486, "learning_rate": 3.357280365094254e-06, "loss": 0.11273296177387238, "step": 3319 }, { "epoch": 0.8618899273104881, "grad_norm": 1.665008783340454, "learning_rate": 3.3563694104799292e-06, "loss": 0.10912498831748962, "step": 3320 }, { "epoch": 0.8621495327102804, "grad_norm": 3.1336159706115723, "learning_rate": 3.3554583486733776e-06, "loss": 0.09333860129117966, "step": 3321 }, { "epoch": 0.8624091381100727, "grad_norm": 1.7760605812072754, "learning_rate": 3.3545471798356526e-06, "loss": 0.08051146566867828, "step": 3322 }, { "epoch": 0.862668743509865, "grad_norm": 3.1579489707946777, "learning_rate": 3.353635904127829e-06, "loss": 0.07630165666341782, "step": 3323 }, { "epoch": 0.8629283489096573, "grad_norm": 2.0590779781341553, "learning_rate": 3.3527245217109986e-06, "loss": 0.0919349193572998, "step": 3324 }, { "epoch": 0.8631879543094496, "grad_norm": 1.788443684577942, "learning_rate": 3.3518130327462734e-06, "loss": 0.10477029532194138, "step": 3325 }, { "epoch": 0.8634475597092419, "grad_norm": 2.1995975971221924, "learning_rate": 3.3509014373947834e-06, "loss": 0.10771869122982025, "step": 3326 }, { "epoch": 0.8637071651090342, "grad_norm": 1.6228153705596924, "learning_rate": 3.3499897358176793e-06, "loss": 0.10467325150966644, "step": 3327 }, { "epoch": 0.8639667705088265, "grad_norm": 2.6832525730133057, "learning_rate": 3.349077928176126e-06, "loss": 0.09690730273723602, "step": 3328 }, { "epoch": 0.864226375908619, "grad_norm": 2.987459897994995, "learning_rate": 3.3481660146313133e-06, "loss": 0.06540210545063019, "step": 3329 }, { "epoch": 0.8644859813084113, "grad_norm": 2.9703924655914307, "learning_rate": 3.3472539953444453e-06, "loss": 0.07259806990623474, "step": 3330 }, { "epoch": 0.8647455867082036, "grad_norm": 2.786783218383789, "learning_rate": 3.346341870476746e-06, "loss": 0.052033681422472, "step": 3331 }, { "epoch": 0.8650051921079959, "grad_norm": 3.0266144275665283, "learning_rate": 3.3454296401894585e-06, "loss": 0.08120229095220566, "step": 3332 }, { "epoch": 0.8652647975077882, "grad_norm": 3.5849530696868896, "learning_rate": 3.344517304643844e-06, "loss": 0.09768601506948471, "step": 3333 }, { "epoch": 0.8655244029075805, "grad_norm": 4.575412750244141, "learning_rate": 3.3436048640011832e-06, "loss": 0.09512767940759659, "step": 3334 }, { "epoch": 0.8657840083073728, "grad_norm": 2.320138692855835, "learning_rate": 3.342692318422773e-06, "loss": 0.10113190114498138, "step": 3335 }, { "epoch": 0.8660436137071651, "grad_norm": 1.5873031616210938, "learning_rate": 3.3417796680699334e-06, "loss": 0.10938134044408798, "step": 3336 }, { "epoch": 0.8663032191069574, "grad_norm": 3.541621208190918, "learning_rate": 3.3408669131039972e-06, "loss": 0.07145588845014572, "step": 3337 }, { "epoch": 0.8665628245067497, "grad_norm": 3.756950616836548, "learning_rate": 3.33995405368632e-06, "loss": 0.10264040529727936, "step": 3338 }, { "epoch": 0.866822429906542, "grad_norm": 2.7081360816955566, "learning_rate": 3.3390410899782748e-06, "loss": 0.10994812846183777, "step": 3339 }, { "epoch": 0.8670820353063343, "grad_norm": 2.3437981605529785, "learning_rate": 3.3381280221412517e-06, "loss": 0.1140141636133194, "step": 3340 }, { "epoch": 0.8673416407061267, "grad_norm": 3.1985528469085693, "learning_rate": 3.3372148503366607e-06, "loss": 0.04194235801696777, "step": 3341 }, { "epoch": 0.867601246105919, "grad_norm": 1.8097463846206665, "learning_rate": 3.3363015747259294e-06, "loss": 0.10340937227010727, "step": 3342 }, { "epoch": 0.8678608515057114, "grad_norm": 2.1703100204467773, "learning_rate": 3.3353881954705047e-06, "loss": 0.0900895893573761, "step": 3343 }, { "epoch": 0.8681204569055037, "grad_norm": 2.262378215789795, "learning_rate": 3.3344747127318506e-06, "loss": 0.10389894247055054, "step": 3344 }, { "epoch": 0.868380062305296, "grad_norm": 5.362878799438477, "learning_rate": 3.33356112667145e-06, "loss": 0.09438082575798035, "step": 3345 }, { "epoch": 0.8686396677050883, "grad_norm": 2.5976414680480957, "learning_rate": 3.332647437450803e-06, "loss": 0.07378719002008438, "step": 3346 }, { "epoch": 0.8688992731048806, "grad_norm": 3.6777398586273193, "learning_rate": 3.3317336452314308e-06, "loss": 0.0917738750576973, "step": 3347 }, { "epoch": 0.8691588785046729, "grad_norm": 1.5369070768356323, "learning_rate": 3.33081975017487e-06, "loss": 0.10917273163795471, "step": 3348 }, { "epoch": 0.8694184839044652, "grad_norm": 1.1556483507156372, "learning_rate": 3.329905752442677e-06, "loss": 0.10429979860782623, "step": 3349 }, { "epoch": 0.8696780893042575, "grad_norm": 2.270705223083496, "learning_rate": 3.328991652196424e-06, "loss": 0.10683336853981018, "step": 3350 }, { "epoch": 0.8699376947040498, "grad_norm": 2.0787625312805176, "learning_rate": 3.3280774495977048e-06, "loss": 0.09241293370723724, "step": 3351 }, { "epoch": 0.8701973001038421, "grad_norm": 3.0600762367248535, "learning_rate": 3.3271631448081277e-06, "loss": 0.052597321569919586, "step": 3352 }, { "epoch": 0.8704569055036345, "grad_norm": 2.0309908390045166, "learning_rate": 3.3262487379893227e-06, "loss": 0.09047000855207443, "step": 3353 }, { "epoch": 0.8707165109034268, "grad_norm": 1.5179184675216675, "learning_rate": 3.325334229302935e-06, "loss": 0.10463333129882812, "step": 3354 }, { "epoch": 0.8709761163032191, "grad_norm": 2.0766749382019043, "learning_rate": 3.3244196189106287e-06, "loss": 0.10043435543775558, "step": 3355 }, { "epoch": 0.8712357217030114, "grad_norm": 3.3984627723693848, "learning_rate": 3.3235049069740864e-06, "loss": 0.0783149003982544, "step": 3356 }, { "epoch": 0.8714953271028038, "grad_norm": 2.820598840713501, "learning_rate": 3.3225900936550094e-06, "loss": 0.07382917404174805, "step": 3357 }, { "epoch": 0.8717549325025961, "grad_norm": 1.9271620512008667, "learning_rate": 3.3216751791151124e-06, "loss": 0.10422137379646301, "step": 3358 }, { "epoch": 0.8720145379023884, "grad_norm": 4.647981643676758, "learning_rate": 3.320760163516134e-06, "loss": 0.06888523697853088, "step": 3359 }, { "epoch": 0.8722741433021807, "grad_norm": 1.6285640001296997, "learning_rate": 3.3198450470198273e-06, "loss": 0.11220492422580719, "step": 3360 }, { "epoch": 0.872533748701973, "grad_norm": 1.844852089881897, "learning_rate": 3.3189298297879653e-06, "loss": 0.08849607408046722, "step": 3361 }, { "epoch": 0.8727933541017653, "grad_norm": 2.305673122406006, "learning_rate": 3.3180145119823347e-06, "loss": 0.08799023181200027, "step": 3362 }, { "epoch": 0.8730529595015576, "grad_norm": 3.829368829727173, "learning_rate": 3.3170990937647445e-06, "loss": 0.07611213624477386, "step": 3363 }, { "epoch": 0.8733125649013499, "grad_norm": 1.3836959600448608, "learning_rate": 3.316183575297019e-06, "loss": 0.10029268264770508, "step": 3364 }, { "epoch": 0.8735721703011423, "grad_norm": 1.007566213607788, "learning_rate": 3.3152679567410023e-06, "loss": 0.10984668135643005, "step": 3365 }, { "epoch": 0.8738317757009346, "grad_norm": 2.9664621353149414, "learning_rate": 3.3143522382585536e-06, "loss": 0.0845390185713768, "step": 3366 }, { "epoch": 0.8740913811007269, "grad_norm": 2.8034827709198, "learning_rate": 3.3134364200115504e-06, "loss": 0.10913977026939392, "step": 3367 }, { "epoch": 0.8743509865005192, "grad_norm": 1.9915424585342407, "learning_rate": 3.3125205021618893e-06, "loss": 0.10543590784072876, "step": 3368 }, { "epoch": 0.8746105919003115, "grad_norm": 2.699167013168335, "learning_rate": 3.3116044848714833e-06, "loss": 0.061869896948337555, "step": 3369 }, { "epoch": 0.8748701973001038, "grad_norm": 3.421632766723633, "learning_rate": 3.3106883683022646e-06, "loss": 0.0790172666311264, "step": 3370 }, { "epoch": 0.8751298026998962, "grad_norm": 1.2015289068222046, "learning_rate": 3.309772152616179e-06, "loss": 0.10676752030849457, "step": 3371 }, { "epoch": 0.8753894080996885, "grad_norm": 0.9017152190208435, "learning_rate": 3.3088558379751945e-06, "loss": 0.10887964069843292, "step": 3372 }, { "epoch": 0.8756490134994808, "grad_norm": 1.4661153554916382, "learning_rate": 3.307939424541294e-06, "loss": 0.10570351779460907, "step": 3373 }, { "epoch": 0.8759086188992731, "grad_norm": 4.178006172180176, "learning_rate": 3.3070229124764784e-06, "loss": 0.07682529091835022, "step": 3374 }, { "epoch": 0.8761682242990654, "grad_norm": 3.5319080352783203, "learning_rate": 3.3061063019427653e-06, "loss": 0.11385034024715424, "step": 3375 }, { "epoch": 0.8764278296988577, "grad_norm": 1.2918800115585327, "learning_rate": 3.305189593102192e-06, "loss": 0.10759054124355316, "step": 3376 }, { "epoch": 0.8766874350986501, "grad_norm": 1.51865816116333, "learning_rate": 3.30427278611681e-06, "loss": 0.09719176590442657, "step": 3377 }, { "epoch": 0.8769470404984424, "grad_norm": 2.4901115894317627, "learning_rate": 3.3033558811486915e-06, "loss": 0.1008400171995163, "step": 3378 }, { "epoch": 0.8772066458982347, "grad_norm": 3.582440137863159, "learning_rate": 3.302438878359923e-06, "loss": 0.09280642867088318, "step": 3379 }, { "epoch": 0.877466251298027, "grad_norm": 2.2978856563568115, "learning_rate": 3.30152177791261e-06, "loss": 0.08765435218811035, "step": 3380 }, { "epoch": 0.8777258566978193, "grad_norm": 1.0986675024032593, "learning_rate": 3.300604579968875e-06, "loss": 0.10271956026554108, "step": 3381 }, { "epoch": 0.8779854620976116, "grad_norm": 0.9388576149940491, "learning_rate": 3.2996872846908574e-06, "loss": 0.10340006649494171, "step": 3382 }, { "epoch": 0.8782450674974039, "grad_norm": 4.126948833465576, "learning_rate": 3.298769892240713e-06, "loss": 0.06499031186103821, "step": 3383 }, { "epoch": 0.8785046728971962, "grad_norm": 4.082151412963867, "learning_rate": 3.2978524027806185e-06, "loss": 0.07987446337938309, "step": 3384 }, { "epoch": 0.8787642782969886, "grad_norm": 1.6163806915283203, "learning_rate": 3.2969348164727624e-06, "loss": 0.09774161130189896, "step": 3385 }, { "epoch": 0.8790238836967809, "grad_norm": 2.5447487831115723, "learning_rate": 3.296017133479354e-06, "loss": 0.049160778522491455, "step": 3386 }, { "epoch": 0.8792834890965732, "grad_norm": 3.1760990619659424, "learning_rate": 3.295099353962618e-06, "loss": 0.10449287295341492, "step": 3387 }, { "epoch": 0.8795430944963655, "grad_norm": 1.9684492349624634, "learning_rate": 3.2941814780847987e-06, "loss": 0.09901832789182663, "step": 3388 }, { "epoch": 0.8798026998961579, "grad_norm": 3.758225202560425, "learning_rate": 3.293263506008153e-06, "loss": 0.05627552419900894, "step": 3389 }, { "epoch": 0.8800623052959502, "grad_norm": 4.1428608894348145, "learning_rate": 3.2923454378949587e-06, "loss": 0.06769782304763794, "step": 3390 }, { "epoch": 0.8803219106957425, "grad_norm": 2.282472610473633, "learning_rate": 3.291427273907509e-06, "loss": 0.08706668019294739, "step": 3391 }, { "epoch": 0.8805815160955348, "grad_norm": 3.6344361305236816, "learning_rate": 3.2905090142081147e-06, "loss": 0.10017132759094238, "step": 3392 }, { "epoch": 0.8808411214953271, "grad_norm": 1.751694917678833, "learning_rate": 3.289590658959103e-06, "loss": 0.09245622158050537, "step": 3393 }, { "epoch": 0.8811007268951194, "grad_norm": 3.8516030311584473, "learning_rate": 3.2886722083228177e-06, "loss": 0.06690734624862671, "step": 3394 }, { "epoch": 0.8813603322949117, "grad_norm": 3.463491439819336, "learning_rate": 3.2877536624616198e-06, "loss": 0.09796062111854553, "step": 3395 }, { "epoch": 0.881619937694704, "grad_norm": 3.823514699935913, "learning_rate": 3.286835021537886e-06, "loss": 0.05087421461939812, "step": 3396 }, { "epoch": 0.8818795430944963, "grad_norm": 2.321967840194702, "learning_rate": 3.2859162857140143e-06, "loss": 0.11789585649967194, "step": 3397 }, { "epoch": 0.8821391484942886, "grad_norm": 3.544651985168457, "learning_rate": 3.2849974551524135e-06, "loss": 0.059721723198890686, "step": 3398 }, { "epoch": 0.882398753894081, "grad_norm": 2.1073131561279297, "learning_rate": 3.284078530015512e-06, "loss": 0.10313821583986282, "step": 3399 }, { "epoch": 0.8826583592938733, "grad_norm": 2.347104787826538, "learning_rate": 3.2831595104657547e-06, "loss": 0.07998717576265335, "step": 3400 }, { "epoch": 0.8829179646936657, "grad_norm": 4.943865776062012, "learning_rate": 3.2822403966656045e-06, "loss": 0.068760946393013, "step": 3401 }, { "epoch": 0.883177570093458, "grad_norm": 1.1198137998580933, "learning_rate": 3.2813211887775386e-06, "loss": 0.10691170394420624, "step": 3402 }, { "epoch": 0.8834371754932503, "grad_norm": 2.4345109462738037, "learning_rate": 3.2804018869640525e-06, "loss": 0.10178033262491226, "step": 3403 }, { "epoch": 0.8836967808930426, "grad_norm": 2.4529428482055664, "learning_rate": 3.279482491387657e-06, "loss": 0.10965871065855026, "step": 3404 }, { "epoch": 0.8839563862928349, "grad_norm": 3.251282215118408, "learning_rate": 3.2785630022108804e-06, "loss": 0.12587058544158936, "step": 3405 }, { "epoch": 0.8842159916926272, "grad_norm": 2.9050261974334717, "learning_rate": 3.2776434195962682e-06, "loss": 0.06368359923362732, "step": 3406 }, { "epoch": 0.8844755970924195, "grad_norm": 1.6138484477996826, "learning_rate": 3.2767237437063803e-06, "loss": 0.11406600475311279, "step": 3407 }, { "epoch": 0.8847352024922118, "grad_norm": 2.9036805629730225, "learning_rate": 3.2758039747037944e-06, "loss": 0.08110520243644714, "step": 3408 }, { "epoch": 0.8849948078920041, "grad_norm": 1.173680305480957, "learning_rate": 3.274884112751106e-06, "loss": 0.10675381124019623, "step": 3409 }, { "epoch": 0.8852544132917964, "grad_norm": 3.445234775543213, "learning_rate": 3.2739641580109243e-06, "loss": 0.09479954838752747, "step": 3410 }, { "epoch": 0.8855140186915887, "grad_norm": 2.995326280593872, "learning_rate": 3.273044110645878e-06, "loss": 0.11673299968242645, "step": 3411 }, { "epoch": 0.885773624091381, "grad_norm": 2.656395435333252, "learning_rate": 3.2721239708186077e-06, "loss": 0.06722040474414825, "step": 3412 }, { "epoch": 0.8860332294911735, "grad_norm": 1.8202356100082397, "learning_rate": 3.271203738691774e-06, "loss": 0.09290148317813873, "step": 3413 }, { "epoch": 0.8862928348909658, "grad_norm": 1.2213587760925293, "learning_rate": 3.2702834144280536e-06, "loss": 0.10981513559818268, "step": 3414 }, { "epoch": 0.8865524402907581, "grad_norm": 1.731101393699646, "learning_rate": 3.2693629981901392e-06, "loss": 0.10252199321985245, "step": 3415 }, { "epoch": 0.8868120456905504, "grad_norm": 1.230921983718872, "learning_rate": 3.268442490140738e-06, "loss": 0.09504536539316177, "step": 3416 }, { "epoch": 0.8870716510903427, "grad_norm": 0.9416756629943848, "learning_rate": 3.2675218904425748e-06, "loss": 0.09895117580890656, "step": 3417 }, { "epoch": 0.887331256490135, "grad_norm": 2.173450469970703, "learning_rate": 3.266601199258391e-06, "loss": 0.0991683080792427, "step": 3418 }, { "epoch": 0.8875908618899273, "grad_norm": 2.0125443935394287, "learning_rate": 3.2656804167509456e-06, "loss": 0.07791497558355331, "step": 3419 }, { "epoch": 0.8878504672897196, "grad_norm": 1.1586095094680786, "learning_rate": 3.264759543083009e-06, "loss": 0.09923776239156723, "step": 3420 }, { "epoch": 0.8881100726895119, "grad_norm": 2.664475917816162, "learning_rate": 3.2638385784173704e-06, "loss": 0.10289636254310608, "step": 3421 }, { "epoch": 0.8883696780893042, "grad_norm": 1.7391725778579712, "learning_rate": 3.262917522916838e-06, "loss": 0.11800447106361389, "step": 3422 }, { "epoch": 0.8886292834890965, "grad_norm": 3.6189780235290527, "learning_rate": 3.2619963767442314e-06, "loss": 0.08043959736824036, "step": 3423 }, { "epoch": 0.8888888888888888, "grad_norm": 4.4290900230407715, "learning_rate": 3.2610751400623887e-06, "loss": 0.07610565423965454, "step": 3424 }, { "epoch": 0.8891484942886813, "grad_norm": 5.355878829956055, "learning_rate": 3.2601538130341627e-06, "loss": 0.07178694009780884, "step": 3425 }, { "epoch": 0.8894080996884736, "grad_norm": 2.1991629600524902, "learning_rate": 3.2592323958224236e-06, "loss": 0.08144809305667877, "step": 3426 }, { "epoch": 0.8896677050882659, "grad_norm": 4.420435905456543, "learning_rate": 3.258310888590058e-06, "loss": 0.06446611881256104, "step": 3427 }, { "epoch": 0.8899273104880582, "grad_norm": 2.3978335857391357, "learning_rate": 3.257389291499965e-06, "loss": 0.11547042429447174, "step": 3428 }, { "epoch": 0.8901869158878505, "grad_norm": 2.830754041671753, "learning_rate": 3.256467604715062e-06, "loss": 0.12178684771060944, "step": 3429 }, { "epoch": 0.8904465212876428, "grad_norm": 1.9269102811813354, "learning_rate": 3.255545828398285e-06, "loss": 0.09863042831420898, "step": 3430 }, { "epoch": 0.8907061266874351, "grad_norm": 3.2185678482055664, "learning_rate": 3.2546239627125796e-06, "loss": 0.11158527433872223, "step": 3431 }, { "epoch": 0.8909657320872274, "grad_norm": 4.081255912780762, "learning_rate": 3.2537020078209126e-06, "loss": 0.06318625062704086, "step": 3432 }, { "epoch": 0.8912253374870197, "grad_norm": 2.725797176361084, "learning_rate": 3.252779963886264e-06, "loss": 0.10001571476459503, "step": 3433 }, { "epoch": 0.891484942886812, "grad_norm": 2.344642162322998, "learning_rate": 3.25185783107163e-06, "loss": 0.06497841328382492, "step": 3434 }, { "epoch": 0.8917445482866043, "grad_norm": 3.587358236312866, "learning_rate": 3.250935609540021e-06, "loss": 0.11537284404039383, "step": 3435 }, { "epoch": 0.8920041536863966, "grad_norm": 4.251770496368408, "learning_rate": 3.250013299454468e-06, "loss": 0.07921083271503448, "step": 3436 }, { "epoch": 0.892263759086189, "grad_norm": 1.0783756971359253, "learning_rate": 3.2490909009780117e-06, "loss": 0.10525444149971008, "step": 3437 }, { "epoch": 0.8925233644859814, "grad_norm": 1.4292547702789307, "learning_rate": 3.248168414273712e-06, "loss": 0.0978788360953331, "step": 3438 }, { "epoch": 0.8927829698857737, "grad_norm": 4.0458879470825195, "learning_rate": 3.2472458395046423e-06, "loss": 0.07977134734392166, "step": 3439 }, { "epoch": 0.893042575285566, "grad_norm": 3.371062994003296, "learning_rate": 3.2463231768338954e-06, "loss": 0.11743417382240295, "step": 3440 }, { "epoch": 0.8933021806853583, "grad_norm": 1.9280084371566772, "learning_rate": 3.2454004264245735e-06, "loss": 0.09630699455738068, "step": 3441 }, { "epoch": 0.8935617860851506, "grad_norm": 2.7671570777893066, "learning_rate": 3.2444775884398e-06, "loss": 0.07712680101394653, "step": 3442 }, { "epoch": 0.8938213914849429, "grad_norm": 1.4685622453689575, "learning_rate": 3.2435546630427105e-06, "loss": 0.09384092688560486, "step": 3443 }, { "epoch": 0.8940809968847352, "grad_norm": 2.147320032119751, "learning_rate": 3.2426316503964583e-06, "loss": 0.07882310450077057, "step": 3444 }, { "epoch": 0.8943406022845275, "grad_norm": 1.8371983766555786, "learning_rate": 3.2417085506642095e-06, "loss": 0.10728079080581665, "step": 3445 }, { "epoch": 0.8946002076843198, "grad_norm": 2.916015386581421, "learning_rate": 3.240785364009148e-06, "loss": 0.09710890054702759, "step": 3446 }, { "epoch": 0.8948598130841121, "grad_norm": 1.8574448823928833, "learning_rate": 3.239862090594471e-06, "loss": 0.09546913951635361, "step": 3447 }, { "epoch": 0.8951194184839044, "grad_norm": 1.8666731119155884, "learning_rate": 3.2389387305833932e-06, "loss": 0.10615326464176178, "step": 3448 }, { "epoch": 0.8953790238836968, "grad_norm": 2.295621633529663, "learning_rate": 3.238015284139143e-06, "loss": 0.07037478685379028, "step": 3449 }, { "epoch": 0.8956386292834891, "grad_norm": 2.194178581237793, "learning_rate": 3.2370917514249646e-06, "loss": 0.10557761043310165, "step": 3450 }, { "epoch": 0.8958982346832814, "grad_norm": 1.2766932249069214, "learning_rate": 3.2361681326041167e-06, "loss": 0.08933389186859131, "step": 3451 }, { "epoch": 0.8961578400830738, "grad_norm": 2.6940152645111084, "learning_rate": 3.2352444278398754e-06, "loss": 0.08356189727783203, "step": 3452 }, { "epoch": 0.8964174454828661, "grad_norm": 3.1885573863983154, "learning_rate": 3.2343206372955305e-06, "loss": 0.07446250319480896, "step": 3453 }, { "epoch": 0.8966770508826584, "grad_norm": 2.988466739654541, "learning_rate": 3.2333967611343854e-06, "loss": 0.1103920042514801, "step": 3454 }, { "epoch": 0.8969366562824507, "grad_norm": 2.5293490886688232, "learning_rate": 3.232472799519762e-06, "loss": 0.11896616220474243, "step": 3455 }, { "epoch": 0.897196261682243, "grad_norm": 3.303678512573242, "learning_rate": 3.231548752614994e-06, "loss": 0.09280577301979065, "step": 3456 }, { "epoch": 0.8974558670820353, "grad_norm": 3.4276123046875, "learning_rate": 3.2306246205834333e-06, "loss": 0.08565492928028107, "step": 3457 }, { "epoch": 0.8977154724818276, "grad_norm": 2.039881467819214, "learning_rate": 3.229700403588444e-06, "loss": 0.10863137245178223, "step": 3458 }, { "epoch": 0.8979750778816199, "grad_norm": 3.8591151237487793, "learning_rate": 3.2287761017934076e-06, "loss": 0.10353905707597733, "step": 3459 }, { "epoch": 0.8982346832814122, "grad_norm": 4.729366302490234, "learning_rate": 3.2278517153617186e-06, "loss": 0.08646658062934875, "step": 3460 }, { "epoch": 0.8984942886812046, "grad_norm": 2.765111207962036, "learning_rate": 3.226927244456788e-06, "loss": 0.06314390152692795, "step": 3461 }, { "epoch": 0.8987538940809969, "grad_norm": 1.5069429874420166, "learning_rate": 3.2260026892420397e-06, "loss": 0.09974793344736099, "step": 3462 }, { "epoch": 0.8990134994807892, "grad_norm": 3.3815925121307373, "learning_rate": 3.2250780498809157e-06, "loss": 0.0614091157913208, "step": 3463 }, { "epoch": 0.8992731048805815, "grad_norm": 2.0596511363983154, "learning_rate": 3.22415332653687e-06, "loss": 0.10076969861984253, "step": 3464 }, { "epoch": 0.8995327102803738, "grad_norm": 1.2474595308303833, "learning_rate": 3.2232285193733735e-06, "loss": 0.10657304525375366, "step": 3465 }, { "epoch": 0.8997923156801662, "grad_norm": 2.9027583599090576, "learning_rate": 3.2223036285539092e-06, "loss": 0.09155216813087463, "step": 3466 }, { "epoch": 0.9000519210799585, "grad_norm": 3.5956671237945557, "learning_rate": 3.2213786542419796e-06, "loss": 0.07637512683868408, "step": 3467 }, { "epoch": 0.9003115264797508, "grad_norm": 2.0336103439331055, "learning_rate": 3.220453596601096e-06, "loss": 0.09760017693042755, "step": 3468 }, { "epoch": 0.9005711318795431, "grad_norm": 1.9069572687149048, "learning_rate": 3.219528455794788e-06, "loss": 0.10498269647359848, "step": 3469 }, { "epoch": 0.9008307372793354, "grad_norm": 3.2853355407714844, "learning_rate": 3.2186032319866e-06, "loss": 0.0784992128610611, "step": 3470 }, { "epoch": 0.9010903426791277, "grad_norm": 3.77716326713562, "learning_rate": 3.217677925340091e-06, "loss": 0.06390920281410217, "step": 3471 }, { "epoch": 0.90134994807892, "grad_norm": 1.489317774772644, "learning_rate": 3.2167525360188317e-06, "loss": 0.09998571872711182, "step": 3472 }, { "epoch": 0.9016095534787124, "grad_norm": 1.9700888395309448, "learning_rate": 3.2158270641864126e-06, "loss": 0.08668561279773712, "step": 3473 }, { "epoch": 0.9018691588785047, "grad_norm": 2.764577865600586, "learning_rate": 3.2149015100064333e-06, "loss": 0.09943072497844696, "step": 3474 }, { "epoch": 0.902128764278297, "grad_norm": 3.220576286315918, "learning_rate": 3.213975873642513e-06, "loss": 0.10069432109594345, "step": 3475 }, { "epoch": 0.9023883696780893, "grad_norm": 3.6775670051574707, "learning_rate": 3.2130501552582823e-06, "loss": 0.06847278773784637, "step": 3476 }, { "epoch": 0.9026479750778816, "grad_norm": 3.6445510387420654, "learning_rate": 3.2121243550173854e-06, "loss": 0.12289843708276749, "step": 3477 }, { "epoch": 0.9029075804776739, "grad_norm": 1.8173902034759521, "learning_rate": 3.2111984730834837e-06, "loss": 0.0916096568107605, "step": 3478 }, { "epoch": 0.9031671858774662, "grad_norm": 1.3718395233154297, "learning_rate": 3.2102725096202527e-06, "loss": 0.10248346626758575, "step": 3479 }, { "epoch": 0.9034267912772586, "grad_norm": 2.539567470550537, "learning_rate": 3.2093464647913815e-06, "loss": 0.08837020397186279, "step": 3480 }, { "epoch": 0.9036863966770509, "grad_norm": 1.1781985759735107, "learning_rate": 3.2084203387605716e-06, "loss": 0.11575949192047119, "step": 3481 }, { "epoch": 0.9039460020768432, "grad_norm": 2.5146267414093018, "learning_rate": 3.207494131691542e-06, "loss": 0.09862785786390305, "step": 3482 }, { "epoch": 0.9042056074766355, "grad_norm": 3.8409929275512695, "learning_rate": 3.2065678437480265e-06, "loss": 0.10107126832008362, "step": 3483 }, { "epoch": 0.9044652128764278, "grad_norm": 3.4219000339508057, "learning_rate": 3.2056414750937696e-06, "loss": 0.07201497256755829, "step": 3484 }, { "epoch": 0.9047248182762202, "grad_norm": 3.6310372352600098, "learning_rate": 3.2047150258925324e-06, "loss": 0.0843178927898407, "step": 3485 }, { "epoch": 0.9049844236760125, "grad_norm": 2.3372962474823, "learning_rate": 3.203788496308091e-06, "loss": 0.08423762023448944, "step": 3486 }, { "epoch": 0.9052440290758048, "grad_norm": 2.206798791885376, "learning_rate": 3.2028618865042327e-06, "loss": 0.10687319934368134, "step": 3487 }, { "epoch": 0.9055036344755971, "grad_norm": 2.9429779052734375, "learning_rate": 3.201935196644763e-06, "loss": 0.06612204015254974, "step": 3488 }, { "epoch": 0.9057632398753894, "grad_norm": 1.649469256401062, "learning_rate": 3.2010084268934984e-06, "loss": 0.10623440146446228, "step": 3489 }, { "epoch": 0.9060228452751817, "grad_norm": 2.3660120964050293, "learning_rate": 3.2000815774142703e-06, "loss": 0.1035357415676117, "step": 3490 }, { "epoch": 0.906282450674974, "grad_norm": 4.28488302230835, "learning_rate": 3.199154648370925e-06, "loss": 0.07125988602638245, "step": 3491 }, { "epoch": 0.9065420560747663, "grad_norm": 3.4388411045074463, "learning_rate": 3.198227639927322e-06, "loss": 0.05692137032747269, "step": 3492 }, { "epoch": 0.9068016614745587, "grad_norm": 2.5871634483337402, "learning_rate": 3.1973005522473355e-06, "loss": 0.047336678951978683, "step": 3493 }, { "epoch": 0.907061266874351, "grad_norm": 3.996628761291504, "learning_rate": 3.196373385494853e-06, "loss": 0.08222590386867523, "step": 3494 }, { "epoch": 0.9073208722741433, "grad_norm": 3.263329029083252, "learning_rate": 3.1954461398337766e-06, "loss": 0.11899209767580032, "step": 3495 }, { "epoch": 0.9075804776739356, "grad_norm": 3.2619435787200928, "learning_rate": 3.194518815428022e-06, "loss": 0.09378582239151001, "step": 3496 }, { "epoch": 0.907840083073728, "grad_norm": 2.0873231887817383, "learning_rate": 3.193591412441519e-06, "loss": 0.11977951228618622, "step": 3497 }, { "epoch": 0.9080996884735203, "grad_norm": 0.9963669180870056, "learning_rate": 3.1926639310382114e-06, "loss": 0.09747263789176941, "step": 3498 }, { "epoch": 0.9083592938733126, "grad_norm": 3.9037561416625977, "learning_rate": 3.1917363713820565e-06, "loss": 0.07065834105014801, "step": 3499 }, { "epoch": 0.9086188992731049, "grad_norm": 2.685457944869995, "learning_rate": 3.1908087336370245e-06, "loss": 0.11329935491085052, "step": 3500 }, { "epoch": 0.9088785046728972, "grad_norm": 3.4184398651123047, "learning_rate": 3.189881017967103e-06, "loss": 0.08853499591350555, "step": 3501 }, { "epoch": 0.9091381100726895, "grad_norm": 3.521406412124634, "learning_rate": 3.188953224536289e-06, "loss": 0.10433415323495865, "step": 3502 }, { "epoch": 0.9093977154724818, "grad_norm": 2.8399505615234375, "learning_rate": 3.188025353508595e-06, "loss": 0.0689791738986969, "step": 3503 }, { "epoch": 0.9096573208722741, "grad_norm": 3.4226765632629395, "learning_rate": 3.187097405048049e-06, "loss": 0.09485507756471634, "step": 3504 }, { "epoch": 0.9099169262720664, "grad_norm": 3.4639217853546143, "learning_rate": 3.18616937931869e-06, "loss": 0.06869897991418839, "step": 3505 }, { "epoch": 0.9101765316718587, "grad_norm": 1.2196186780929565, "learning_rate": 3.1852412764845704e-06, "loss": 0.11410767585039139, "step": 3506 }, { "epoch": 0.910436137071651, "grad_norm": 2.3926053047180176, "learning_rate": 3.1843130967097596e-06, "loss": 0.06538031250238419, "step": 3507 }, { "epoch": 0.9106957424714434, "grad_norm": 2.170056104660034, "learning_rate": 3.183384840158338e-06, "loss": 0.08518895506858826, "step": 3508 }, { "epoch": 0.9109553478712358, "grad_norm": 3.1314985752105713, "learning_rate": 3.1824565069943993e-06, "loss": 0.07764025032520294, "step": 3509 }, { "epoch": 0.9112149532710281, "grad_norm": 2.0227913856506348, "learning_rate": 3.181528097382052e-06, "loss": 0.10345545411109924, "step": 3510 }, { "epoch": 0.9114745586708204, "grad_norm": 2.577392101287842, "learning_rate": 3.180599611485418e-06, "loss": 0.06115550547838211, "step": 3511 }, { "epoch": 0.9117341640706127, "grad_norm": 2.3188817501068115, "learning_rate": 3.179671049468632e-06, "loss": 0.09873498231172562, "step": 3512 }, { "epoch": 0.911993769470405, "grad_norm": 2.8343288898468018, "learning_rate": 3.1787424114958433e-06, "loss": 0.08545972406864166, "step": 3513 }, { "epoch": 0.9122533748701973, "grad_norm": 2.7621476650238037, "learning_rate": 3.1778136977312125e-06, "loss": 0.11177079379558563, "step": 3514 }, { "epoch": 0.9125129802699896, "grad_norm": 2.0280063152313232, "learning_rate": 3.1768849083389164e-06, "loss": 0.09868116676807404, "step": 3515 }, { "epoch": 0.9127725856697819, "grad_norm": 2.75993275642395, "learning_rate": 3.175956043483142e-06, "loss": 0.08920136094093323, "step": 3516 }, { "epoch": 0.9130321910695742, "grad_norm": 2.6606407165527344, "learning_rate": 3.1750271033280925e-06, "loss": 0.10496070235967636, "step": 3517 }, { "epoch": 0.9132917964693665, "grad_norm": 2.2926766872406006, "learning_rate": 3.1740980880379823e-06, "loss": 0.08586503565311432, "step": 3518 }, { "epoch": 0.9135514018691588, "grad_norm": 2.6169676780700684, "learning_rate": 3.1731689977770414e-06, "loss": 0.07916779816150665, "step": 3519 }, { "epoch": 0.9138110072689511, "grad_norm": 1.7630535364151, "learning_rate": 3.172239832709511e-06, "loss": 0.10720044374465942, "step": 3520 }, { "epoch": 0.9140706126687436, "grad_norm": 2.111301898956299, "learning_rate": 3.171310592999646e-06, "loss": 0.09419569373130798, "step": 3521 }, { "epoch": 0.9143302180685359, "grad_norm": 3.8772761821746826, "learning_rate": 3.1703812788117157e-06, "loss": 0.12031093239784241, "step": 3522 }, { "epoch": 0.9145898234683282, "grad_norm": 4.096920013427734, "learning_rate": 3.1694518903099993e-06, "loss": 0.0661889910697937, "step": 3523 }, { "epoch": 0.9148494288681205, "grad_norm": 3.848659038543701, "learning_rate": 3.168522427658794e-06, "loss": 0.06796815246343613, "step": 3524 }, { "epoch": 0.9151090342679128, "grad_norm": 1.323261022567749, "learning_rate": 3.1675928910224063e-06, "loss": 0.10278507322072983, "step": 3525 }, { "epoch": 0.9153686396677051, "grad_norm": 3.627566337585449, "learning_rate": 3.1666632805651564e-06, "loss": 0.08146144449710846, "step": 3526 }, { "epoch": 0.9156282450674974, "grad_norm": 3.2962210178375244, "learning_rate": 3.165733596451379e-06, "loss": 0.11543820798397064, "step": 3527 }, { "epoch": 0.9158878504672897, "grad_norm": 2.332174062728882, "learning_rate": 3.164803838845421e-06, "loss": 0.10067121684551239, "step": 3528 }, { "epoch": 0.916147455867082, "grad_norm": 1.843226671218872, "learning_rate": 3.1638740079116427e-06, "loss": 0.11069148033857346, "step": 3529 }, { "epoch": 0.9164070612668743, "grad_norm": 2.2078475952148438, "learning_rate": 3.1629441038144152e-06, "loss": 0.10162486135959625, "step": 3530 }, { "epoch": 0.9166666666666666, "grad_norm": 3.541583776473999, "learning_rate": 3.1620141267181253e-06, "loss": 0.09704533219337463, "step": 3531 }, { "epoch": 0.9169262720664589, "grad_norm": 3.352815866470337, "learning_rate": 3.1610840767871726e-06, "loss": 0.08314324915409088, "step": 3532 }, { "epoch": 0.9171858774662512, "grad_norm": 1.9753936529159546, "learning_rate": 3.160153954185967e-06, "loss": 0.09824618697166443, "step": 3533 }, { "epoch": 0.9174454828660437, "grad_norm": 2.4203743934631348, "learning_rate": 3.159223759078934e-06, "loss": 0.10338559746742249, "step": 3534 }, { "epoch": 0.917705088265836, "grad_norm": 2.3511908054351807, "learning_rate": 3.15829349163051e-06, "loss": 0.052875734865665436, "step": 3535 }, { "epoch": 0.9179646936656283, "grad_norm": 2.731170654296875, "learning_rate": 3.157363152005145e-06, "loss": 0.09451096504926682, "step": 3536 }, { "epoch": 0.9182242990654206, "grad_norm": 3.6958813667297363, "learning_rate": 3.1564327403673024e-06, "loss": 0.09708530455827713, "step": 3537 }, { "epoch": 0.9184839044652129, "grad_norm": 3.8556535243988037, "learning_rate": 3.155502256881458e-06, "loss": 0.07847501337528229, "step": 3538 }, { "epoch": 0.9187435098650052, "grad_norm": 1.2295475006103516, "learning_rate": 3.1545717017120993e-06, "loss": 0.09662891924381256, "step": 3539 }, { "epoch": 0.9190031152647975, "grad_norm": 3.530275344848633, "learning_rate": 3.153641075023726e-06, "loss": 0.07424813508987427, "step": 3540 }, { "epoch": 0.9192627206645898, "grad_norm": 4.129073143005371, "learning_rate": 3.152710376980854e-06, "loss": 0.06240028887987137, "step": 3541 }, { "epoch": 0.9195223260643821, "grad_norm": 2.463906764984131, "learning_rate": 3.151779607748008e-06, "loss": 0.1011628806591034, "step": 3542 }, { "epoch": 0.9197819314641744, "grad_norm": 1.8215316534042358, "learning_rate": 3.150848767489726e-06, "loss": 0.09826289117336273, "step": 3543 }, { "epoch": 0.9200415368639667, "grad_norm": 1.6334617137908936, "learning_rate": 3.1499178563705605e-06, "loss": 0.10334564745426178, "step": 3544 }, { "epoch": 0.920301142263759, "grad_norm": 2.8949482440948486, "learning_rate": 3.1489868745550754e-06, "loss": 0.099741131067276, "step": 3545 }, { "epoch": 0.9205607476635514, "grad_norm": 2.3669936656951904, "learning_rate": 3.1480558222078467e-06, "loss": 0.1140979677438736, "step": 3546 }, { "epoch": 0.9208203530633438, "grad_norm": 2.37319016456604, "learning_rate": 3.147124699493462e-06, "loss": 0.10991272330284119, "step": 3547 }, { "epoch": 0.9210799584631361, "grad_norm": 1.634840488433838, "learning_rate": 3.146193506576524e-06, "loss": 0.10106093436479568, "step": 3548 }, { "epoch": 0.9213395638629284, "grad_norm": 2.899932861328125, "learning_rate": 3.1452622436216452e-06, "loss": 0.07449059933423996, "step": 3549 }, { "epoch": 0.9215991692627207, "grad_norm": 1.3300598859786987, "learning_rate": 3.144330910793453e-06, "loss": 0.11245125532150269, "step": 3550 }, { "epoch": 0.921858774662513, "grad_norm": 3.307194232940674, "learning_rate": 3.143399508256584e-06, "loss": 0.056044526398181915, "step": 3551 }, { "epoch": 0.9221183800623053, "grad_norm": 2.630854606628418, "learning_rate": 3.14246803617569e-06, "loss": 0.08776434510946274, "step": 3552 }, { "epoch": 0.9223779854620976, "grad_norm": 2.594311475753784, "learning_rate": 3.1415364947154323e-06, "loss": 0.11022783815860748, "step": 3553 }, { "epoch": 0.9226375908618899, "grad_norm": 2.6780953407287598, "learning_rate": 3.1406048840404886e-06, "loss": 0.09507179260253906, "step": 3554 }, { "epoch": 0.9228971962616822, "grad_norm": 3.824110746383667, "learning_rate": 3.1396732043155447e-06, "loss": 0.0536641925573349, "step": 3555 }, { "epoch": 0.9231568016614745, "grad_norm": 3.5952043533325195, "learning_rate": 3.1387414557053013e-06, "loss": 0.07153306156396866, "step": 3556 }, { "epoch": 0.9234164070612668, "grad_norm": 1.2807670831680298, "learning_rate": 3.1378096383744685e-06, "loss": 0.10569284111261368, "step": 3557 }, { "epoch": 0.9236760124610592, "grad_norm": 3.8666796684265137, "learning_rate": 3.1368777524877713e-06, "loss": 0.09261107444763184, "step": 3558 }, { "epoch": 0.9239356178608515, "grad_norm": 3.5056562423706055, "learning_rate": 3.1359457982099467e-06, "loss": 0.11477115750312805, "step": 3559 }, { "epoch": 0.9241952232606438, "grad_norm": 2.459590435028076, "learning_rate": 3.1350137757057412e-06, "loss": 0.08340436220169067, "step": 3560 }, { "epoch": 0.9244548286604362, "grad_norm": 1.3207225799560547, "learning_rate": 3.1340816851399164e-06, "loss": 0.10088369995355606, "step": 3561 }, { "epoch": 0.9247144340602285, "grad_norm": 2.6630687713623047, "learning_rate": 3.133149526677243e-06, "loss": 0.10041570663452148, "step": 3562 }, { "epoch": 0.9249740394600208, "grad_norm": 2.369821310043335, "learning_rate": 3.1322173004825076e-06, "loss": 0.11280803382396698, "step": 3563 }, { "epoch": 0.9252336448598131, "grad_norm": 4.071226119995117, "learning_rate": 3.1312850067205036e-06, "loss": 0.1090482622385025, "step": 3564 }, { "epoch": 0.9254932502596054, "grad_norm": 3.420064687728882, "learning_rate": 3.130352645556042e-06, "loss": 0.05828237533569336, "step": 3565 }, { "epoch": 0.9257528556593977, "grad_norm": 1.3190752267837524, "learning_rate": 3.1294202171539406e-06, "loss": 0.08876402676105499, "step": 3566 }, { "epoch": 0.92601246105919, "grad_norm": 1.3287774324417114, "learning_rate": 3.128487721679033e-06, "loss": 0.10938721895217896, "step": 3567 }, { "epoch": 0.9262720664589823, "grad_norm": 1.5731830596923828, "learning_rate": 3.1275551592961617e-06, "loss": 0.09032082557678223, "step": 3568 }, { "epoch": 0.9265316718587746, "grad_norm": 2.2007699012756348, "learning_rate": 3.1266225301701842e-06, "loss": 0.09005336463451385, "step": 3569 }, { "epoch": 0.926791277258567, "grad_norm": 6.019087791442871, "learning_rate": 3.1256898344659657e-06, "loss": 0.10151579976081848, "step": 3570 }, { "epoch": 0.9270508826583593, "grad_norm": 2.9151768684387207, "learning_rate": 3.124757072348388e-06, "loss": 0.075886070728302, "step": 3571 }, { "epoch": 0.9273104880581516, "grad_norm": 1.935316801071167, "learning_rate": 3.1238242439823393e-06, "loss": 0.10652238130569458, "step": 3572 }, { "epoch": 0.927570093457944, "grad_norm": 2.953238010406494, "learning_rate": 3.1228913495327245e-06, "loss": 0.05515282228589058, "step": 3573 }, { "epoch": 0.9278296988577363, "grad_norm": 2.5823187828063965, "learning_rate": 3.1219583891644574e-06, "loss": 0.06600020825862885, "step": 3574 }, { "epoch": 0.9280893042575286, "grad_norm": 1.8108105659484863, "learning_rate": 3.121025363042463e-06, "loss": 0.08815757185220718, "step": 3575 }, { "epoch": 0.9283489096573209, "grad_norm": 3.482081890106201, "learning_rate": 3.12009227133168e-06, "loss": 0.09817135334014893, "step": 3576 }, { "epoch": 0.9286085150571132, "grad_norm": 5.27702522277832, "learning_rate": 3.119159114197058e-06, "loss": 0.06697569787502289, "step": 3577 }, { "epoch": 0.9288681204569055, "grad_norm": 4.793358325958252, "learning_rate": 3.118225891803556e-06, "loss": 0.07540681958198547, "step": 3578 }, { "epoch": 0.9291277258566978, "grad_norm": 1.688765525817871, "learning_rate": 3.1172926043161496e-06, "loss": 0.09100484848022461, "step": 3579 }, { "epoch": 0.9293873312564901, "grad_norm": 3.646355628967285, "learning_rate": 3.116359251899819e-06, "loss": 0.06546974927186966, "step": 3580 }, { "epoch": 0.9296469366562824, "grad_norm": 1.990419864654541, "learning_rate": 3.1154258347195614e-06, "loss": 0.10714177787303925, "step": 3581 }, { "epoch": 0.9299065420560748, "grad_norm": 2.062875270843506, "learning_rate": 3.114492352940384e-06, "loss": 0.10733528435230255, "step": 3582 }, { "epoch": 0.9301661474558671, "grad_norm": 4.586723327636719, "learning_rate": 3.1135588067273044e-06, "loss": 0.08923974633216858, "step": 3583 }, { "epoch": 0.9304257528556594, "grad_norm": 1.9867161512374878, "learning_rate": 3.1126251962453507e-06, "loss": 0.09610868990421295, "step": 3584 }, { "epoch": 0.9306853582554517, "grad_norm": 1.9713680744171143, "learning_rate": 3.111691521659567e-06, "loss": 0.09094630926847458, "step": 3585 }, { "epoch": 0.930944963655244, "grad_norm": 1.6762821674346924, "learning_rate": 3.1107577831350033e-06, "loss": 0.10664359480142593, "step": 3586 }, { "epoch": 0.9312045690550363, "grad_norm": 4.067721366882324, "learning_rate": 3.1098239808367237e-06, "loss": 0.06644079834222794, "step": 3587 }, { "epoch": 0.9314641744548287, "grad_norm": 3.5372958183288574, "learning_rate": 3.108890114929803e-06, "loss": 0.0691530704498291, "step": 3588 }, { "epoch": 0.931723779854621, "grad_norm": 4.558583736419678, "learning_rate": 3.1079561855793277e-06, "loss": 0.13485975563526154, "step": 3589 }, { "epoch": 0.9319833852544133, "grad_norm": 4.58333683013916, "learning_rate": 3.107022192950395e-06, "loss": 0.0474349707365036, "step": 3590 }, { "epoch": 0.9322429906542056, "grad_norm": 3.092576026916504, "learning_rate": 3.106088137208113e-06, "loss": 0.08605216443538666, "step": 3591 }, { "epoch": 0.9325025960539979, "grad_norm": 3.6044118404388428, "learning_rate": 3.105154018517602e-06, "loss": 0.10664473474025726, "step": 3592 }, { "epoch": 0.9327622014537902, "grad_norm": 3.383073568344116, "learning_rate": 3.1042198370439925e-06, "loss": 0.09714318811893463, "step": 3593 }, { "epoch": 0.9330218068535826, "grad_norm": 1.4393717050552368, "learning_rate": 3.103285592952427e-06, "loss": 0.10270993411540985, "step": 3594 }, { "epoch": 0.9332814122533749, "grad_norm": 1.8748183250427246, "learning_rate": 3.1023512864080563e-06, "loss": 0.07633957266807556, "step": 3595 }, { "epoch": 0.9335410176531672, "grad_norm": 2.9284701347351074, "learning_rate": 3.101416917576047e-06, "loss": 0.07927748560905457, "step": 3596 }, { "epoch": 0.9338006230529595, "grad_norm": 2.0847666263580322, "learning_rate": 3.1004824866215723e-06, "loss": 0.1052212119102478, "step": 3597 }, { "epoch": 0.9340602284527518, "grad_norm": 2.245163917541504, "learning_rate": 3.09954799370982e-06, "loss": 0.07536093890666962, "step": 3598 }, { "epoch": 0.9343198338525441, "grad_norm": 3.7811367511749268, "learning_rate": 3.0986134390059847e-06, "loss": 0.07230962812900543, "step": 3599 }, { "epoch": 0.9345794392523364, "grad_norm": 1.4620462656021118, "learning_rate": 3.0976788226752767e-06, "loss": 0.10290399193763733, "step": 3600 }, { "epoch": 0.9348390446521287, "grad_norm": 1.6140049695968628, "learning_rate": 3.0967441448829132e-06, "loss": 0.10502134263515472, "step": 3601 }, { "epoch": 0.935098650051921, "grad_norm": 2.8910129070281982, "learning_rate": 3.0958094057941245e-06, "loss": 0.10444015264511108, "step": 3602 }, { "epoch": 0.9353582554517134, "grad_norm": 1.3213433027267456, "learning_rate": 3.09487460557415e-06, "loss": 0.0919160470366478, "step": 3603 }, { "epoch": 0.9356178608515057, "grad_norm": 3.6871607303619385, "learning_rate": 3.093939744388243e-06, "loss": 0.06211552023887634, "step": 3604 }, { "epoch": 0.935877466251298, "grad_norm": 3.2293784618377686, "learning_rate": 3.093004822401664e-06, "loss": 0.07394708693027496, "step": 3605 }, { "epoch": 0.9361370716510904, "grad_norm": 1.9978735446929932, "learning_rate": 3.092069839779686e-06, "loss": 0.07907595485448837, "step": 3606 }, { "epoch": 0.9363966770508827, "grad_norm": 3.8571293354034424, "learning_rate": 3.0911347966875925e-06, "loss": 0.07445584237575531, "step": 3607 }, { "epoch": 0.936656282450675, "grad_norm": 3.409294605255127, "learning_rate": 3.090199693290678e-06, "loss": 0.07172717154026031, "step": 3608 }, { "epoch": 0.9369158878504673, "grad_norm": 4.2520976066589355, "learning_rate": 3.0892645297542475e-06, "loss": 0.07040563225746155, "step": 3609 }, { "epoch": 0.9371754932502596, "grad_norm": 2.7870535850524902, "learning_rate": 3.088329306243616e-06, "loss": 0.09092546254396439, "step": 3610 }, { "epoch": 0.9374350986500519, "grad_norm": 3.1169581413269043, "learning_rate": 3.0873940229241095e-06, "loss": 0.08159227669239044, "step": 3611 }, { "epoch": 0.9376947040498442, "grad_norm": 3.7074697017669678, "learning_rate": 3.086458679961066e-06, "loss": 0.03281460329890251, "step": 3612 }, { "epoch": 0.9379543094496365, "grad_norm": 4.149311065673828, "learning_rate": 3.0855232775198317e-06, "loss": 0.04749062657356262, "step": 3613 }, { "epoch": 0.9382139148494288, "grad_norm": 2.7662506103515625, "learning_rate": 3.084587815765764e-06, "loss": 0.07675160467624664, "step": 3614 }, { "epoch": 0.9384735202492211, "grad_norm": 3.694411039352417, "learning_rate": 3.0836522948642317e-06, "loss": 0.07135970890522003, "step": 3615 }, { "epoch": 0.9387331256490135, "grad_norm": 2.30017352104187, "learning_rate": 3.0827167149806126e-06, "loss": 0.09809732437133789, "step": 3616 }, { "epoch": 0.9389927310488058, "grad_norm": 1.9571093320846558, "learning_rate": 3.0817810762802974e-06, "loss": 0.10150012373924255, "step": 3617 }, { "epoch": 0.9392523364485982, "grad_norm": 1.919835090637207, "learning_rate": 3.0808453789286842e-06, "loss": 0.12387322634458542, "step": 3618 }, { "epoch": 0.9395119418483905, "grad_norm": 2.3333208560943604, "learning_rate": 3.079909623091183e-06, "loss": 0.09177359938621521, "step": 3619 }, { "epoch": 0.9397715472481828, "grad_norm": 4.158233165740967, "learning_rate": 3.078973808933215e-06, "loss": 0.08808566629886627, "step": 3620 }, { "epoch": 0.9400311526479751, "grad_norm": 3.3563528060913086, "learning_rate": 3.0780379366202106e-06, "loss": 0.07156821340322495, "step": 3621 }, { "epoch": 0.9402907580477674, "grad_norm": 2.7772724628448486, "learning_rate": 3.077102006317609e-06, "loss": 0.06445442885160446, "step": 3622 }, { "epoch": 0.9405503634475597, "grad_norm": 3.324150323867798, "learning_rate": 3.0761660181908624e-06, "loss": 0.10120414197444916, "step": 3623 }, { "epoch": 0.940809968847352, "grad_norm": 3.852102756500244, "learning_rate": 3.0752299724054324e-06, "loss": 0.07142437994480133, "step": 3624 }, { "epoch": 0.9410695742471443, "grad_norm": 2.7443771362304688, "learning_rate": 3.0742938691267903e-06, "loss": 0.087434321641922, "step": 3625 }, { "epoch": 0.9413291796469366, "grad_norm": 3.209564208984375, "learning_rate": 3.073357708520417e-06, "loss": 0.062038518488407135, "step": 3626 }, { "epoch": 0.9415887850467289, "grad_norm": 2.3241665363311768, "learning_rate": 3.072421490751805e-06, "loss": 0.097371906042099, "step": 3627 }, { "epoch": 0.9418483904465212, "grad_norm": 1.8297531604766846, "learning_rate": 3.071485215986455e-06, "loss": 0.08035165071487427, "step": 3628 }, { "epoch": 0.9421079958463136, "grad_norm": 3.473917007446289, "learning_rate": 3.0705488843898816e-06, "loss": 0.0784931406378746, "step": 3629 }, { "epoch": 0.942367601246106, "grad_norm": 2.852487564086914, "learning_rate": 3.0696124961276036e-06, "loss": 0.08102560043334961, "step": 3630 }, { "epoch": 0.9426272066458983, "grad_norm": 3.850759506225586, "learning_rate": 3.0686760513651555e-06, "loss": 0.05814626067876816, "step": 3631 }, { "epoch": 0.9428868120456906, "grad_norm": 3.2162091732025146, "learning_rate": 3.0677395502680774e-06, "loss": 0.05533629655838013, "step": 3632 }, { "epoch": 0.9431464174454829, "grad_norm": 1.4104835987091064, "learning_rate": 3.066802993001922e-06, "loss": 0.08501909673213959, "step": 3633 }, { "epoch": 0.9434060228452752, "grad_norm": 3.141735553741455, "learning_rate": 3.0658663797322518e-06, "loss": 0.10441616177558899, "step": 3634 }, { "epoch": 0.9436656282450675, "grad_norm": 2.9011282920837402, "learning_rate": 3.0649297106246386e-06, "loss": 0.09729303419589996, "step": 3635 }, { "epoch": 0.9439252336448598, "grad_norm": 1.6933470964431763, "learning_rate": 3.0639929858446626e-06, "loss": 0.08684133738279343, "step": 3636 }, { "epoch": 0.9441848390446521, "grad_norm": 1.5805567502975464, "learning_rate": 3.0630562055579154e-06, "loss": 0.09455917030572891, "step": 3637 }, { "epoch": 0.9444444444444444, "grad_norm": 1.792495608329773, "learning_rate": 3.0621193699300004e-06, "loss": 0.09498351067304611, "step": 3638 }, { "epoch": 0.9447040498442367, "grad_norm": 2.491041421890259, "learning_rate": 3.0611824791265278e-06, "loss": 0.09655852615833282, "step": 3639 }, { "epoch": 0.944963655244029, "grad_norm": 2.8179407119750977, "learning_rate": 3.060245533313117e-06, "loss": 0.10104959458112717, "step": 3640 }, { "epoch": 0.9452232606438213, "grad_norm": 3.438504934310913, "learning_rate": 3.0593085326553995e-06, "loss": 0.08276398479938507, "step": 3641 }, { "epoch": 0.9454828660436138, "grad_norm": 2.2822787761688232, "learning_rate": 3.058371477319016e-06, "loss": 0.09945939481258392, "step": 3642 }, { "epoch": 0.9457424714434061, "grad_norm": 4.808245658874512, "learning_rate": 3.0574343674696162e-06, "loss": 0.08351193368434906, "step": 3643 }, { "epoch": 0.9460020768431984, "grad_norm": 3.785538911819458, "learning_rate": 3.05649720327286e-06, "loss": 0.06864461302757263, "step": 3644 }, { "epoch": 0.9462616822429907, "grad_norm": 4.03400993347168, "learning_rate": 3.0555599848944152e-06, "loss": 0.12419643998146057, "step": 3645 }, { "epoch": 0.946521287642783, "grad_norm": 2.6110289096832275, "learning_rate": 3.054622712499962e-06, "loss": 0.09626883268356323, "step": 3646 }, { "epoch": 0.9467808930425753, "grad_norm": 2.036942720413208, "learning_rate": 3.0536853862551875e-06, "loss": 0.10204406082630157, "step": 3647 }, { "epoch": 0.9470404984423676, "grad_norm": 3.2886767387390137, "learning_rate": 3.0527480063257908e-06, "loss": 0.06466753780841827, "step": 3648 }, { "epoch": 0.9473001038421599, "grad_norm": 2.2754151821136475, "learning_rate": 3.0518105728774782e-06, "loss": 0.08396749943494797, "step": 3649 }, { "epoch": 0.9475597092419522, "grad_norm": 3.4507968425750732, "learning_rate": 3.0508730860759673e-06, "loss": 0.060655541718006134, "step": 3650 }, { "epoch": 0.9478193146417445, "grad_norm": 3.109053611755371, "learning_rate": 3.0499355460869833e-06, "loss": 0.0938805490732193, "step": 3651 }, { "epoch": 0.9480789200415368, "grad_norm": 3.649104595184326, "learning_rate": 3.048997953076263e-06, "loss": 0.049153752624988556, "step": 3652 }, { "epoch": 0.9483385254413291, "grad_norm": 4.0764641761779785, "learning_rate": 3.0480603072095507e-06, "loss": 0.0805991068482399, "step": 3653 }, { "epoch": 0.9485981308411215, "grad_norm": 1.7393970489501953, "learning_rate": 3.0471226086526e-06, "loss": 0.10276515781879425, "step": 3654 }, { "epoch": 0.9488577362409139, "grad_norm": 3.0460331439971924, "learning_rate": 3.046184857571176e-06, "loss": 0.04332907497882843, "step": 3655 }, { "epoch": 0.9491173416407062, "grad_norm": 1.559369444847107, "learning_rate": 3.0452470541310507e-06, "loss": 0.09227748215198517, "step": 3656 }, { "epoch": 0.9493769470404985, "grad_norm": 3.2399637699127197, "learning_rate": 3.0443091984980066e-06, "loss": 0.06501825898885727, "step": 3657 }, { "epoch": 0.9496365524402908, "grad_norm": 2.5927867889404297, "learning_rate": 3.0433712908378356e-06, "loss": 0.0863061398267746, "step": 3658 }, { "epoch": 0.9498961578400831, "grad_norm": 3.5219874382019043, "learning_rate": 3.042433331316336e-06, "loss": 0.06372235715389252, "step": 3659 }, { "epoch": 0.9501557632398754, "grad_norm": 3.3524186611175537, "learning_rate": 3.04149532009932e-06, "loss": 0.0947607010602951, "step": 3660 }, { "epoch": 0.9504153686396677, "grad_norm": 3.904222011566162, "learning_rate": 3.0405572573526067e-06, "loss": 0.07106590270996094, "step": 3661 }, { "epoch": 0.95067497403946, "grad_norm": 3.5873162746429443, "learning_rate": 3.0396191432420225e-06, "loss": 0.0850900262594223, "step": 3662 }, { "epoch": 0.9509345794392523, "grad_norm": 2.1031129360198975, "learning_rate": 3.0386809779334055e-06, "loss": 0.09163197875022888, "step": 3663 }, { "epoch": 0.9511941848390446, "grad_norm": 2.6164331436157227, "learning_rate": 3.037742761592601e-06, "loss": 0.09848974645137787, "step": 3664 }, { "epoch": 0.9514537902388369, "grad_norm": 3.83526873588562, "learning_rate": 3.0368044943854647e-06, "loss": 0.07136912643909454, "step": 3665 }, { "epoch": 0.9517133956386293, "grad_norm": 3.555373430252075, "learning_rate": 3.0358661764778614e-06, "loss": 0.10415089130401611, "step": 3666 }, { "epoch": 0.9519730010384216, "grad_norm": 1.6597917079925537, "learning_rate": 3.0349278080356626e-06, "loss": 0.09696243703365326, "step": 3667 }, { "epoch": 0.952232606438214, "grad_norm": 3.1503801345825195, "learning_rate": 3.0339893892247518e-06, "loss": 0.07111958414316177, "step": 3668 }, { "epoch": 0.9524922118380063, "grad_norm": 2.8843271732330322, "learning_rate": 3.0330509202110197e-06, "loss": 0.08994778990745544, "step": 3669 }, { "epoch": 0.9527518172377986, "grad_norm": 2.9423534870147705, "learning_rate": 3.032112401160366e-06, "loss": 0.09677101671695709, "step": 3670 }, { "epoch": 0.9530114226375909, "grad_norm": 4.22330904006958, "learning_rate": 3.0311738322386996e-06, "loss": 0.08588729798793793, "step": 3671 }, { "epoch": 0.9532710280373832, "grad_norm": 1.9754998683929443, "learning_rate": 3.0302352136119364e-06, "loss": 0.11954523622989655, "step": 3672 }, { "epoch": 0.9535306334371755, "grad_norm": 2.147373914718628, "learning_rate": 3.029296545446005e-06, "loss": 0.08694875985383987, "step": 3673 }, { "epoch": 0.9537902388369678, "grad_norm": 2.086873769760132, "learning_rate": 3.0283578279068387e-06, "loss": 0.06516259908676147, "step": 3674 }, { "epoch": 0.9540498442367601, "grad_norm": 3.4975273609161377, "learning_rate": 3.027419061160383e-06, "loss": 0.0975157767534256, "step": 3675 }, { "epoch": 0.9543094496365524, "grad_norm": 3.5002589225769043, "learning_rate": 3.026480245372588e-06, "loss": 0.07654890418052673, "step": 3676 }, { "epoch": 0.9545690550363447, "grad_norm": 4.0864152908325195, "learning_rate": 3.025541380709417e-06, "loss": 0.07856874167919159, "step": 3677 }, { "epoch": 0.9548286604361371, "grad_norm": 3.9259371757507324, "learning_rate": 3.024602467336839e-06, "loss": 0.09297142922878265, "step": 3678 }, { "epoch": 0.9550882658359294, "grad_norm": 2.15194034576416, "learning_rate": 3.0236635054208322e-06, "loss": 0.0999661386013031, "step": 3679 }, { "epoch": 0.9553478712357217, "grad_norm": 5.21980094909668, "learning_rate": 3.022724495127384e-06, "loss": 0.08062779903411865, "step": 3680 }, { "epoch": 0.955607476635514, "grad_norm": 1.9303585290908813, "learning_rate": 3.0217854366224885e-06, "loss": 0.10652735829353333, "step": 3681 }, { "epoch": 0.9558670820353063, "grad_norm": 2.478130578994751, "learning_rate": 3.020846330072152e-06, "loss": 0.07057280093431473, "step": 3682 }, { "epoch": 0.9561266874350987, "grad_norm": 3.2954025268554688, "learning_rate": 3.019907175642386e-06, "loss": 0.05934841185808182, "step": 3683 }, { "epoch": 0.956386292834891, "grad_norm": 4.434176445007324, "learning_rate": 3.0189679734992107e-06, "loss": 0.06901536136865616, "step": 3684 }, { "epoch": 0.9566458982346833, "grad_norm": 3.4923267364501953, "learning_rate": 3.0180287238086563e-06, "loss": 0.0730733722448349, "step": 3685 }, { "epoch": 0.9569055036344756, "grad_norm": 1.3238036632537842, "learning_rate": 3.0170894267367605e-06, "loss": 0.10792563110589981, "step": 3686 }, { "epoch": 0.9571651090342679, "grad_norm": 5.077177047729492, "learning_rate": 3.0161500824495704e-06, "loss": 0.09220407158136368, "step": 3687 }, { "epoch": 0.9574247144340602, "grad_norm": 3.5096757411956787, "learning_rate": 3.015210691113139e-06, "loss": 0.05164702609181404, "step": 3688 }, { "epoch": 0.9576843198338525, "grad_norm": 1.6894973516464233, "learning_rate": 3.014271252893531e-06, "loss": 0.09786063432693481, "step": 3689 }, { "epoch": 0.9579439252336449, "grad_norm": 3.192328691482544, "learning_rate": 3.0133317679568145e-06, "loss": 0.06264142692089081, "step": 3690 }, { "epoch": 0.9582035306334372, "grad_norm": 1.7270042896270752, "learning_rate": 3.012392236469073e-06, "loss": 0.09139712154865265, "step": 3691 }, { "epoch": 0.9584631360332295, "grad_norm": 1.6761152744293213, "learning_rate": 3.0114526585963903e-06, "loss": 0.08249901235103607, "step": 3692 }, { "epoch": 0.9587227414330218, "grad_norm": 2.524346351623535, "learning_rate": 3.0105130345048654e-06, "loss": 0.08801355212926865, "step": 3693 }, { "epoch": 0.9589823468328141, "grad_norm": 1.4325950145721436, "learning_rate": 3.0095733643606005e-06, "loss": 0.09265615046024323, "step": 3694 }, { "epoch": 0.9592419522326064, "grad_norm": 2.2226979732513428, "learning_rate": 3.0086336483297084e-06, "loss": 0.08518834412097931, "step": 3695 }, { "epoch": 0.9595015576323987, "grad_norm": 1.4803756475448608, "learning_rate": 3.007693886578309e-06, "loss": 0.10476462543010712, "step": 3696 }, { "epoch": 0.9597611630321911, "grad_norm": 3.035676956176758, "learning_rate": 3.0067540792725304e-06, "loss": 0.10327483713626862, "step": 3697 }, { "epoch": 0.9600207684319834, "grad_norm": 4.590161323547363, "learning_rate": 3.00581422657851e-06, "loss": 0.08616869151592255, "step": 3698 }, { "epoch": 0.9602803738317757, "grad_norm": 2.579085350036621, "learning_rate": 3.004874328662391e-06, "loss": 0.11474411934614182, "step": 3699 }, { "epoch": 0.960539979231568, "grad_norm": 3.4116766452789307, "learning_rate": 3.003934385690328e-06, "loss": 0.0446806401014328, "step": 3700 }, { "epoch": 0.9607995846313603, "grad_norm": 4.846246719360352, "learning_rate": 3.002994397828479e-06, "loss": 0.04490792751312256, "step": 3701 }, { "epoch": 0.9610591900311527, "grad_norm": 3.1138710975646973, "learning_rate": 3.002054365243014e-06, "loss": 0.04757894575595856, "step": 3702 }, { "epoch": 0.961318795430945, "grad_norm": 2.2233052253723145, "learning_rate": 3.001114288100107e-06, "loss": 0.09221889078617096, "step": 3703 }, { "epoch": 0.9615784008307373, "grad_norm": 3.7113802433013916, "learning_rate": 3.0001741665659444e-06, "loss": 0.05142088979482651, "step": 3704 }, { "epoch": 0.9618380062305296, "grad_norm": 2.293431520462036, "learning_rate": 2.9992340008067177e-06, "loss": 0.0780731588602066, "step": 3705 }, { "epoch": 0.9620976116303219, "grad_norm": 2.4052655696868896, "learning_rate": 2.9982937909886267e-06, "loss": 0.08302688598632812, "step": 3706 }, { "epoch": 0.9623572170301142, "grad_norm": 3.9964373111724854, "learning_rate": 2.997353537277878e-06, "loss": 0.07777144014835358, "step": 3707 }, { "epoch": 0.9626168224299065, "grad_norm": 2.871767997741699, "learning_rate": 2.996413239840687e-06, "loss": 0.09979604184627533, "step": 3708 }, { "epoch": 0.9628764278296988, "grad_norm": 1.9852237701416016, "learning_rate": 2.995472898843278e-06, "loss": 0.08937931060791016, "step": 3709 }, { "epoch": 0.9631360332294912, "grad_norm": 3.4854965209960938, "learning_rate": 2.994532514451881e-06, "loss": 0.07387816160917282, "step": 3710 }, { "epoch": 0.9633956386292835, "grad_norm": 3.1634857654571533, "learning_rate": 2.9935920868327345e-06, "loss": 0.0629347562789917, "step": 3711 }, { "epoch": 0.9636552440290758, "grad_norm": 3.2425789833068848, "learning_rate": 2.9926516161520848e-06, "loss": 0.09172509610652924, "step": 3712 }, { "epoch": 0.9639148494288681, "grad_norm": 1.9876742362976074, "learning_rate": 2.9917111025761848e-06, "loss": 0.10603801906108856, "step": 3713 }, { "epoch": 0.9641744548286605, "grad_norm": 3.0556092262268066, "learning_rate": 2.9907705462712974e-06, "loss": 0.07466690242290497, "step": 3714 }, { "epoch": 0.9644340602284528, "grad_norm": 4.011180877685547, "learning_rate": 2.989829947403689e-06, "loss": 0.05445224791765213, "step": 3715 }, { "epoch": 0.9646936656282451, "grad_norm": 1.5820177793502808, "learning_rate": 2.9888893061396375e-06, "loss": 0.10634486377239227, "step": 3716 }, { "epoch": 0.9649532710280374, "grad_norm": 4.193573951721191, "learning_rate": 2.987948622645427e-06, "loss": 0.0435626283288002, "step": 3717 }, { "epoch": 0.9652128764278297, "grad_norm": 1.4295320510864258, "learning_rate": 2.9870078970873483e-06, "loss": 0.09915536642074585, "step": 3718 }, { "epoch": 0.965472481827622, "grad_norm": 3.415971517562866, "learning_rate": 2.9860671296317e-06, "loss": 0.09220358729362488, "step": 3719 }, { "epoch": 0.9657320872274143, "grad_norm": 1.4427517652511597, "learning_rate": 2.985126320444789e-06, "loss": 0.09866921603679657, "step": 3720 }, { "epoch": 0.9659916926272066, "grad_norm": 1.8539936542510986, "learning_rate": 2.9841854696929266e-06, "loss": 0.10184159129858017, "step": 3721 }, { "epoch": 0.9662512980269989, "grad_norm": 3.5466597080230713, "learning_rate": 2.983244577542437e-06, "loss": 0.07947169989347458, "step": 3722 }, { "epoch": 0.9665109034267912, "grad_norm": 3.985182046890259, "learning_rate": 2.9823036441596465e-06, "loss": 0.051976513117551804, "step": 3723 }, { "epoch": 0.9667705088265836, "grad_norm": 1.6992570161819458, "learning_rate": 2.98136266971089e-06, "loss": 0.09476928412914276, "step": 3724 }, { "epoch": 0.9670301142263759, "grad_norm": 1.6180031299591064, "learning_rate": 2.980421654362511e-06, "loss": 0.12050864100456238, "step": 3725 }, { "epoch": 0.9672897196261683, "grad_norm": 3.2203938961029053, "learning_rate": 2.97948059828086e-06, "loss": 0.08575300872325897, "step": 3726 }, { "epoch": 0.9675493250259606, "grad_norm": 2.926093816757202, "learning_rate": 2.978539501632293e-06, "loss": 0.07690463960170746, "step": 3727 }, { "epoch": 0.9678089304257529, "grad_norm": 4.335335731506348, "learning_rate": 2.977598364583176e-06, "loss": 0.06276843696832657, "step": 3728 }, { "epoch": 0.9680685358255452, "grad_norm": 2.462658405303955, "learning_rate": 2.9766571872998777e-06, "loss": 0.08779865503311157, "step": 3729 }, { "epoch": 0.9683281412253375, "grad_norm": 3.327953815460205, "learning_rate": 2.9757159699487798e-06, "loss": 0.0783466026186943, "step": 3730 }, { "epoch": 0.9685877466251298, "grad_norm": 3.169152021408081, "learning_rate": 2.9747747126962666e-06, "loss": 0.06906729936599731, "step": 3731 }, { "epoch": 0.9688473520249221, "grad_norm": 2.4702348709106445, "learning_rate": 2.9738334157087307e-06, "loss": 0.1074284166097641, "step": 3732 }, { "epoch": 0.9691069574247144, "grad_norm": 3.061734914779663, "learning_rate": 2.9728920791525724e-06, "loss": 0.09584398567676544, "step": 3733 }, { "epoch": 0.9693665628245067, "grad_norm": 3.083012104034424, "learning_rate": 2.971950703194197e-06, "loss": 0.07649669051170349, "step": 3734 }, { "epoch": 0.969626168224299, "grad_norm": 3.909170389175415, "learning_rate": 2.97100928800002e-06, "loss": 0.04407615214586258, "step": 3735 }, { "epoch": 0.9698857736240913, "grad_norm": 3.2984261512756348, "learning_rate": 2.9700678337364615e-06, "loss": 0.09736901521682739, "step": 3736 }, { "epoch": 0.9701453790238836, "grad_norm": 3.506584882736206, "learning_rate": 2.9691263405699485e-06, "loss": 0.06430939584970474, "step": 3737 }, { "epoch": 0.9704049844236761, "grad_norm": 3.37536358833313, "learning_rate": 2.968184808666916e-06, "loss": 0.07377149164676666, "step": 3738 }, { "epoch": 0.9706645898234684, "grad_norm": 3.5671441555023193, "learning_rate": 2.967243238193805e-06, "loss": 0.0704464465379715, "step": 3739 }, { "epoch": 0.9709241952232607, "grad_norm": 2.6039252281188965, "learning_rate": 2.9663016293170632e-06, "loss": 0.0953335389494896, "step": 3740 }, { "epoch": 0.971183800623053, "grad_norm": 3.324192762374878, "learning_rate": 2.965359982203147e-06, "loss": 0.04852358251810074, "step": 3741 }, { "epoch": 0.9714434060228453, "grad_norm": 2.7928667068481445, "learning_rate": 2.9644182970185163e-06, "loss": 0.036407358944416046, "step": 3742 }, { "epoch": 0.9717030114226376, "grad_norm": 2.5585992336273193, "learning_rate": 2.96347657392964e-06, "loss": 0.0498909056186676, "step": 3743 }, { "epoch": 0.9719626168224299, "grad_norm": 1.3509206771850586, "learning_rate": 2.962534813102994e-06, "loss": 0.125161275267601, "step": 3744 }, { "epoch": 0.9722222222222222, "grad_norm": 2.4331867694854736, "learning_rate": 2.9615930147050597e-06, "loss": 0.10183379054069519, "step": 3745 }, { "epoch": 0.9724818276220145, "grad_norm": 2.2444562911987305, "learning_rate": 2.9606511789023247e-06, "loss": 0.08055628836154938, "step": 3746 }, { "epoch": 0.9727414330218068, "grad_norm": 3.893171548843384, "learning_rate": 2.9597093058612843e-06, "loss": 0.05955097824335098, "step": 3747 }, { "epoch": 0.9730010384215991, "grad_norm": 2.9648499488830566, "learning_rate": 2.958767395748441e-06, "loss": 0.08589915186166763, "step": 3748 }, { "epoch": 0.9732606438213914, "grad_norm": 3.061478853225708, "learning_rate": 2.957825448730302e-06, "loss": 0.09218522906303406, "step": 3749 }, { "epoch": 0.9735202492211839, "grad_norm": 2.062434434890747, "learning_rate": 2.9568834649733825e-06, "loss": 0.09892693161964417, "step": 3750 }, { "epoch": 0.9737798546209762, "grad_norm": 3.526191234588623, "learning_rate": 2.9559414446442024e-06, "loss": 0.06243719160556793, "step": 3751 }, { "epoch": 0.9740394600207685, "grad_norm": 3.280308485031128, "learning_rate": 2.9549993879092912e-06, "loss": 0.04777250438928604, "step": 3752 }, { "epoch": 0.9742990654205608, "grad_norm": 1.977278709411621, "learning_rate": 2.954057294935182e-06, "loss": 0.08786669373512268, "step": 3753 }, { "epoch": 0.9745586708203531, "grad_norm": 2.942662239074707, "learning_rate": 2.9531151658884156e-06, "loss": 0.08520466089248657, "step": 3754 }, { "epoch": 0.9748182762201454, "grad_norm": 3.741246461868286, "learning_rate": 2.9521730009355383e-06, "loss": 0.06840794533491135, "step": 3755 }, { "epoch": 0.9750778816199377, "grad_norm": 4.004042148590088, "learning_rate": 2.951230800243103e-06, "loss": 0.07868140190839767, "step": 3756 }, { "epoch": 0.97533748701973, "grad_norm": 3.9458999633789062, "learning_rate": 2.95028856397767e-06, "loss": 0.06736665964126587, "step": 3757 }, { "epoch": 0.9755970924195223, "grad_norm": 4.222728252410889, "learning_rate": 2.9493462923058057e-06, "loss": 0.05304492264986038, "step": 3758 }, { "epoch": 0.9758566978193146, "grad_norm": 2.76507568359375, "learning_rate": 2.948403985394081e-06, "loss": 0.08746056258678436, "step": 3759 }, { "epoch": 0.9761163032191069, "grad_norm": 2.4647514820098877, "learning_rate": 2.9474616434090737e-06, "loss": 0.09291726350784302, "step": 3760 }, { "epoch": 0.9763759086188992, "grad_norm": 2.452462673187256, "learning_rate": 2.9465192665173697e-06, "loss": 0.11098437011241913, "step": 3761 }, { "epoch": 0.9766355140186916, "grad_norm": 3.1244137287139893, "learning_rate": 2.9455768548855597e-06, "loss": 0.07193635404109955, "step": 3762 }, { "epoch": 0.976895119418484, "grad_norm": 1.4568109512329102, "learning_rate": 2.944634408680239e-06, "loss": 0.10758285969495773, "step": 3763 }, { "epoch": 0.9771547248182763, "grad_norm": 2.081653118133545, "learning_rate": 2.9436919280680115e-06, "loss": 0.10985273122787476, "step": 3764 }, { "epoch": 0.9774143302180686, "grad_norm": 1.9905556440353394, "learning_rate": 2.942749413215486e-06, "loss": 0.10191197693347931, "step": 3765 }, { "epoch": 0.9776739356178609, "grad_norm": 1.1725423336029053, "learning_rate": 2.941806864289279e-06, "loss": 0.10802099108695984, "step": 3766 }, { "epoch": 0.9779335410176532, "grad_norm": 2.1450886726379395, "learning_rate": 2.9408642814560082e-06, "loss": 0.08798013627529144, "step": 3767 }, { "epoch": 0.9781931464174455, "grad_norm": 3.0806078910827637, "learning_rate": 2.939921664882304e-06, "loss": 0.056140609085559845, "step": 3768 }, { "epoch": 0.9784527518172378, "grad_norm": 4.105889320373535, "learning_rate": 2.9389790147347973e-06, "loss": 0.08531048148870468, "step": 3769 }, { "epoch": 0.9787123572170301, "grad_norm": 3.8876986503601074, "learning_rate": 2.9380363311801286e-06, "loss": 0.06665592640638351, "step": 3770 }, { "epoch": 0.9789719626168224, "grad_norm": 3.009997606277466, "learning_rate": 2.9370936143849415e-06, "loss": 0.07043245434761047, "step": 3771 }, { "epoch": 0.9792315680166147, "grad_norm": 1.670413851737976, "learning_rate": 2.9361508645158875e-06, "loss": 0.09495474398136139, "step": 3772 }, { "epoch": 0.979491173416407, "grad_norm": 3.182772636413574, "learning_rate": 2.9352080817396227e-06, "loss": 0.08953788131475449, "step": 3773 }, { "epoch": 0.9797507788161994, "grad_norm": 3.12426495552063, "learning_rate": 2.93426526622281e-06, "loss": 0.05081429332494736, "step": 3774 }, { "epoch": 0.9800103842159917, "grad_norm": 1.2466167211532593, "learning_rate": 2.933322418132117e-06, "loss": 0.11008072644472122, "step": 3775 }, { "epoch": 0.980269989615784, "grad_norm": 1.7667672634124756, "learning_rate": 2.932379537634218e-06, "loss": 0.06767766177654266, "step": 3776 }, { "epoch": 0.9805295950155763, "grad_norm": 3.3567590713500977, "learning_rate": 2.9314366248957927e-06, "loss": 0.06775026023387909, "step": 3777 }, { "epoch": 0.9807892004153687, "grad_norm": 2.695427656173706, "learning_rate": 2.9304936800835264e-06, "loss": 0.07222101837396622, "step": 3778 }, { "epoch": 0.981048805815161, "grad_norm": 2.858752727508545, "learning_rate": 2.9295507033641105e-06, "loss": 0.05979746952652931, "step": 3779 }, { "epoch": 0.9813084112149533, "grad_norm": 1.2893275022506714, "learning_rate": 2.9286076949042415e-06, "loss": 0.10386250913143158, "step": 3780 }, { "epoch": 0.9815680166147456, "grad_norm": 1.5968986749649048, "learning_rate": 2.9276646548706222e-06, "loss": 0.10970909893512726, "step": 3781 }, { "epoch": 0.9818276220145379, "grad_norm": 1.1487246751785278, "learning_rate": 2.9267215834299585e-06, "loss": 0.1018112301826477, "step": 3782 }, { "epoch": 0.9820872274143302, "grad_norm": 4.1533355712890625, "learning_rate": 2.9257784807489663e-06, "loss": 0.04547521844506264, "step": 3783 }, { "epoch": 0.9823468328141225, "grad_norm": 4.833178520202637, "learning_rate": 2.924835346994363e-06, "loss": 0.04166565090417862, "step": 3784 }, { "epoch": 0.9826064382139148, "grad_norm": 1.4472782611846924, "learning_rate": 2.9238921823328737e-06, "loss": 0.10360710322856903, "step": 3785 }, { "epoch": 0.9828660436137072, "grad_norm": 1.6010929346084595, "learning_rate": 2.9229489869312276e-06, "loss": 0.09803107380867004, "step": 3786 }, { "epoch": 0.9831256490134995, "grad_norm": 1.8327300548553467, "learning_rate": 2.9220057609561616e-06, "loss": 0.07521498203277588, "step": 3787 }, { "epoch": 0.9833852544132918, "grad_norm": 3.3864636421203613, "learning_rate": 2.9210625045744143e-06, "loss": 0.08821599185466766, "step": 3788 }, { "epoch": 0.9836448598130841, "grad_norm": 1.5069103240966797, "learning_rate": 2.920119217952735e-06, "loss": 0.09881038963794708, "step": 3789 }, { "epoch": 0.9839044652128764, "grad_norm": 4.091759204864502, "learning_rate": 2.9191759012578712e-06, "loss": 0.05472932010889053, "step": 3790 }, { "epoch": 0.9841640706126688, "grad_norm": 1.399294137954712, "learning_rate": 2.9182325546565822e-06, "loss": 0.10376382619142532, "step": 3791 }, { "epoch": 0.9844236760124611, "grad_norm": 1.52107834815979, "learning_rate": 2.91728917831563e-06, "loss": 0.10856848210096359, "step": 3792 }, { "epoch": 0.9846832814122534, "grad_norm": 2.9445078372955322, "learning_rate": 2.916345772401781e-06, "loss": 0.09065871685743332, "step": 3793 }, { "epoch": 0.9849428868120457, "grad_norm": 1.0215723514556885, "learning_rate": 2.9154023370818085e-06, "loss": 0.11458351463079453, "step": 3794 }, { "epoch": 0.985202492211838, "grad_norm": 2.2398810386657715, "learning_rate": 2.91445887252249e-06, "loss": 0.08003892749547958, "step": 3795 }, { "epoch": 0.9854620976116303, "grad_norm": 2.7007882595062256, "learning_rate": 2.9135153788906084e-06, "loss": 0.10366986691951752, "step": 3796 }, { "epoch": 0.9857217030114226, "grad_norm": 1.5618597269058228, "learning_rate": 2.912571856352952e-06, "loss": 0.09336338937282562, "step": 3797 }, { "epoch": 0.985981308411215, "grad_norm": 3.29398250579834, "learning_rate": 2.9116283050763137e-06, "loss": 0.0816405862569809, "step": 3798 }, { "epoch": 0.9862409138110073, "grad_norm": 2.448446273803711, "learning_rate": 2.910684725227492e-06, "loss": 0.09362006187438965, "step": 3799 }, { "epoch": 0.9865005192107996, "grad_norm": 2.0245590209960938, "learning_rate": 2.909741116973289e-06, "loss": 0.08376061916351318, "step": 3800 }, { "epoch": 0.9867601246105919, "grad_norm": 3.3590621948242188, "learning_rate": 2.908797480480515e-06, "loss": 0.07700704038143158, "step": 3801 }, { "epoch": 0.9870197300103842, "grad_norm": 3.7882883548736572, "learning_rate": 2.9078538159159823e-06, "loss": 0.07170895487070084, "step": 3802 }, { "epoch": 0.9872793354101765, "grad_norm": 3.7251322269439697, "learning_rate": 2.90691012344651e-06, "loss": 0.04322303459048271, "step": 3803 }, { "epoch": 0.9875389408099688, "grad_norm": 3.7260594367980957, "learning_rate": 2.90596640323892e-06, "loss": 0.09856033325195312, "step": 3804 }, { "epoch": 0.9877985462097612, "grad_norm": 2.252643585205078, "learning_rate": 2.905022655460041e-06, "loss": 0.09319594502449036, "step": 3805 }, { "epoch": 0.9880581516095535, "grad_norm": 1.6973110437393188, "learning_rate": 2.9040788802767065e-06, "loss": 0.0936354249715805, "step": 3806 }, { "epoch": 0.9883177570093458, "grad_norm": 1.5243748426437378, "learning_rate": 2.903135077855754e-06, "loss": 0.10813826322555542, "step": 3807 }, { "epoch": 0.9885773624091381, "grad_norm": 3.2546443939208984, "learning_rate": 2.902191248364025e-06, "loss": 0.06690040230751038, "step": 3808 }, { "epoch": 0.9888369678089304, "grad_norm": 2.944709300994873, "learning_rate": 2.9012473919683686e-06, "loss": 0.10308218002319336, "step": 3809 }, { "epoch": 0.9890965732087228, "grad_norm": 1.7229604721069336, "learning_rate": 2.900303508835637e-06, "loss": 0.10637867450714111, "step": 3810 }, { "epoch": 0.9893561786085151, "grad_norm": 2.7517292499542236, "learning_rate": 2.8993595991326854e-06, "loss": 0.042876310646533966, "step": 3811 }, { "epoch": 0.9896157840083074, "grad_norm": 3.993664264678955, "learning_rate": 2.8984156630263777e-06, "loss": 0.05761632323265076, "step": 3812 }, { "epoch": 0.9898753894080997, "grad_norm": 3.3270621299743652, "learning_rate": 2.8974717006835785e-06, "loss": 0.08131992816925049, "step": 3813 }, { "epoch": 0.990134994807892, "grad_norm": 3.358372211456299, "learning_rate": 2.8965277122711593e-06, "loss": 0.0830269604921341, "step": 3814 }, { "epoch": 0.9903946002076843, "grad_norm": 3.4792490005493164, "learning_rate": 2.895583697955995e-06, "loss": 0.09690818190574646, "step": 3815 }, { "epoch": 0.9906542056074766, "grad_norm": 2.3108737468719482, "learning_rate": 2.8946396579049674e-06, "loss": 0.09592828154563904, "step": 3816 }, { "epoch": 0.9909138110072689, "grad_norm": 1.6545164585113525, "learning_rate": 2.8936955922849597e-06, "loss": 0.09630692005157471, "step": 3817 }, { "epoch": 0.9911734164070612, "grad_norm": 3.739079713821411, "learning_rate": 2.8927515012628616e-06, "loss": 0.0684753805398941, "step": 3818 }, { "epoch": 0.9914330218068536, "grad_norm": 2.689544677734375, "learning_rate": 2.8918073850055663e-06, "loss": 0.08591587841510773, "step": 3819 }, { "epoch": 0.9916926272066459, "grad_norm": 1.9300663471221924, "learning_rate": 2.8908632436799725e-06, "loss": 0.10644059628248215, "step": 3820 }, { "epoch": 0.9919522326064382, "grad_norm": 4.077075481414795, "learning_rate": 2.8899190774529823e-06, "loss": 0.08032970130443573, "step": 3821 }, { "epoch": 0.9922118380062306, "grad_norm": 4.151669979095459, "learning_rate": 2.888974886491503e-06, "loss": 0.07340988516807556, "step": 3822 }, { "epoch": 0.9924714434060229, "grad_norm": 3.1289732456207275, "learning_rate": 2.888030670962446e-06, "loss": 0.04899340867996216, "step": 3823 }, { "epoch": 0.9927310488058152, "grad_norm": 2.919896125793457, "learning_rate": 2.8870864310327267e-06, "loss": 0.04613886773586273, "step": 3824 }, { "epoch": 0.9929906542056075, "grad_norm": 1.931528925895691, "learning_rate": 2.886142166869265e-06, "loss": 0.10775148868560791, "step": 3825 }, { "epoch": 0.9932502596053998, "grad_norm": 2.737668752670288, "learning_rate": 2.885197878638986e-06, "loss": 0.08356829732656479, "step": 3826 }, { "epoch": 0.9935098650051921, "grad_norm": 3.6735165119171143, "learning_rate": 2.8842535665088177e-06, "loss": 0.039730098098516464, "step": 3827 }, { "epoch": 0.9937694704049844, "grad_norm": 3.367408275604248, "learning_rate": 2.8833092306456934e-06, "loss": 0.06326065957546234, "step": 3828 }, { "epoch": 0.9940290758047767, "grad_norm": 4.293264865875244, "learning_rate": 2.882364871216548e-06, "loss": 0.07341136038303375, "step": 3829 }, { "epoch": 0.994288681204569, "grad_norm": 2.580933094024658, "learning_rate": 2.881420488388325e-06, "loss": 0.10738139599561691, "step": 3830 }, { "epoch": 0.9945482866043613, "grad_norm": 1.732674479484558, "learning_rate": 2.8804760823279687e-06, "loss": 0.10052920132875443, "step": 3831 }, { "epoch": 0.9948078920041536, "grad_norm": 4.753111362457275, "learning_rate": 2.8795316532024287e-06, "loss": 0.08734848350286484, "step": 3832 }, { "epoch": 0.995067497403946, "grad_norm": 3.1519367694854736, "learning_rate": 2.878587201178658e-06, "loss": 0.10483953356742859, "step": 3833 }, { "epoch": 0.9953271028037384, "grad_norm": 3.4728996753692627, "learning_rate": 2.8776427264236147e-06, "loss": 0.08082173764705658, "step": 3834 }, { "epoch": 0.9955867082035307, "grad_norm": 4.086656093597412, "learning_rate": 2.8766982291042593e-06, "loss": 0.07104691863059998, "step": 3835 }, { "epoch": 0.995846313603323, "grad_norm": 2.4220738410949707, "learning_rate": 2.875753709387559e-06, "loss": 0.0907975286245346, "step": 3836 }, { "epoch": 0.9961059190031153, "grad_norm": 3.564396858215332, "learning_rate": 2.8748091674404816e-06, "loss": 0.06687536090612411, "step": 3837 }, { "epoch": 0.9963655244029076, "grad_norm": 4.964117050170898, "learning_rate": 2.8738646034300015e-06, "loss": 0.0630548894405365, "step": 3838 }, { "epoch": 0.9966251298026999, "grad_norm": 3.6025493144989014, "learning_rate": 2.8729200175230955e-06, "loss": 0.09373842179775238, "step": 3839 }, { "epoch": 0.9968847352024922, "grad_norm": 5.050782680511475, "learning_rate": 2.8719754098867446e-06, "loss": 0.05710114538669586, "step": 3840 }, { "epoch": 0.9971443406022845, "grad_norm": 2.8958818912506104, "learning_rate": 2.8710307806879343e-06, "loss": 0.08186859637498856, "step": 3841 }, { "epoch": 0.9974039460020768, "grad_norm": 4.340834617614746, "learning_rate": 2.870086130093653e-06, "loss": 0.0547458678483963, "step": 3842 }, { "epoch": 0.9976635514018691, "grad_norm": 4.009760856628418, "learning_rate": 2.8691414582708943e-06, "loss": 0.060613393783569336, "step": 3843 }, { "epoch": 0.9979231568016614, "grad_norm": 4.712045192718506, "learning_rate": 2.8681967653866527e-06, "loss": 0.034244637936353683, "step": 3844 }, { "epoch": 0.9981827622014537, "grad_norm": 2.4105358123779297, "learning_rate": 2.86725205160793e-06, "loss": 0.10404165089130402, "step": 3845 }, { "epoch": 0.9984423676012462, "grad_norm": 2.678699493408203, "learning_rate": 2.8663073171017284e-06, "loss": 0.061971768736839294, "step": 3846 }, { "epoch": 0.9987019730010385, "grad_norm": 3.452995777130127, "learning_rate": 2.865362562035058e-06, "loss": 0.06541416049003601, "step": 3847 }, { "epoch": 0.9989615784008308, "grad_norm": 2.95284366607666, "learning_rate": 2.8644177865749266e-06, "loss": 0.07216471433639526, "step": 3848 }, { "epoch": 0.9992211838006231, "grad_norm": 3.837205171585083, "learning_rate": 2.8634729908883506e-06, "loss": 0.08157125115394592, "step": 3849 }, { "epoch": 0.9994807892004154, "grad_norm": 3.829353094100952, "learning_rate": 2.8625281751423485e-06, "loss": 0.09287507832050323, "step": 3850 }, { "epoch": 0.9997403946002077, "grad_norm": 2.552730083465576, "learning_rate": 2.861583339503941e-06, "loss": 0.07623539119958878, "step": 3851 }, { "epoch": 1.0, "grad_norm": 3.0614569187164307, "learning_rate": 2.8606384841401548e-06, "loss": 0.06908125430345535, "step": 3852 }, { "epoch": 1.0002596053997923, "grad_norm": 2.7439146041870117, "learning_rate": 2.8596936092180173e-06, "loss": 0.0821584090590477, "step": 3853 }, { "epoch": 1.0005192107995846, "grad_norm": 2.9751148223876953, "learning_rate": 2.858748714904561e-06, "loss": 0.057912662625312805, "step": 3854 }, { "epoch": 1.000778816199377, "grad_norm": 1.7403590679168701, "learning_rate": 2.857803801366823e-06, "loss": 0.08801832795143127, "step": 3855 }, { "epoch": 1.0010384215991692, "grad_norm": 3.1241466999053955, "learning_rate": 2.85685886877184e-06, "loss": 0.037718646228313446, "step": 3856 }, { "epoch": 1.0012980269989615, "grad_norm": 3.117330551147461, "learning_rate": 2.8559139172866563e-06, "loss": 0.024086985737085342, "step": 3857 }, { "epoch": 1.0015576323987538, "grad_norm": 2.3678231239318848, "learning_rate": 2.854968947078317e-06, "loss": 0.09018373489379883, "step": 3858 }, { "epoch": 1.0018172377985461, "grad_norm": 2.956388473510742, "learning_rate": 2.854023958313872e-06, "loss": 0.09736403822898865, "step": 3859 }, { "epoch": 1.0020768431983385, "grad_norm": 2.8233156204223633, "learning_rate": 2.853078951160373e-06, "loss": 0.01583479717373848, "step": 3860 }, { "epoch": 1.0023364485981308, "grad_norm": 1.6313724517822266, "learning_rate": 2.852133925784875e-06, "loss": 0.08958549052476883, "step": 3861 }, { "epoch": 1.002596053997923, "grad_norm": 1.8397681713104248, "learning_rate": 2.8511888823544376e-06, "loss": 0.09160225093364716, "step": 3862 }, { "epoch": 1.0028556593977154, "grad_norm": 4.438880920410156, "learning_rate": 2.850243821036123e-06, "loss": 0.03866400569677353, "step": 3863 }, { "epoch": 1.0031152647975077, "grad_norm": 2.534456729888916, "learning_rate": 2.849298741996996e-06, "loss": 0.09086019545793533, "step": 3864 }, { "epoch": 1.0033748701973002, "grad_norm": 5.155349254608154, "learning_rate": 2.8483536454041245e-06, "loss": 0.03551696613430977, "step": 3865 }, { "epoch": 1.0036344755970925, "grad_norm": 1.4293770790100098, "learning_rate": 2.847408531424581e-06, "loss": 0.08136923611164093, "step": 3866 }, { "epoch": 1.0038940809968848, "grad_norm": 3.1569859981536865, "learning_rate": 2.8464634002254387e-06, "loss": 0.028140071779489517, "step": 3867 }, { "epoch": 1.0041536863966771, "grad_norm": 3.168240547180176, "learning_rate": 2.8455182519737774e-06, "loss": 0.06799830496311188, "step": 3868 }, { "epoch": 1.0044132917964694, "grad_norm": 3.680553436279297, "learning_rate": 2.844573086836675e-06, "loss": 0.030363963916897774, "step": 3869 }, { "epoch": 1.0046728971962617, "grad_norm": 2.827592372894287, "learning_rate": 2.8436279049812167e-06, "loss": 0.029755622148513794, "step": 3870 }, { "epoch": 1.004932502596054, "grad_norm": 4.4003190994262695, "learning_rate": 2.8426827065744887e-06, "loss": 0.04505976289510727, "step": 3871 }, { "epoch": 1.0051921079958464, "grad_norm": 2.51214861869812, "learning_rate": 2.8417374917835804e-06, "loss": 0.06864924728870392, "step": 3872 }, { "epoch": 1.0054517133956387, "grad_norm": 1.3747435808181763, "learning_rate": 2.8407922607755843e-06, "loss": 0.09255468100309372, "step": 3873 }, { "epoch": 1.005711318795431, "grad_norm": 2.2452046871185303, "learning_rate": 2.839847013717595e-06, "loss": 0.08827945590019226, "step": 3874 }, { "epoch": 1.0059709241952233, "grad_norm": 3.398263692855835, "learning_rate": 2.838901750776711e-06, "loss": 0.02888551726937294, "step": 3875 }, { "epoch": 1.0062305295950156, "grad_norm": 3.6826565265655518, "learning_rate": 2.8379564721200333e-06, "loss": 0.04702798277139664, "step": 3876 }, { "epoch": 1.0064901349948079, "grad_norm": 1.8600789308547974, "learning_rate": 2.8370111779146646e-06, "loss": 0.11257217079401016, "step": 3877 }, { "epoch": 1.0067497403946002, "grad_norm": 1.824422001838684, "learning_rate": 2.8360658683277125e-06, "loss": 0.08739437907934189, "step": 3878 }, { "epoch": 1.0070093457943925, "grad_norm": 2.632884979248047, "learning_rate": 2.835120543526285e-06, "loss": 0.05447334051132202, "step": 3879 }, { "epoch": 1.0072689511941848, "grad_norm": 4.053382873535156, "learning_rate": 2.8341752036774945e-06, "loss": 0.04529604688286781, "step": 3880 }, { "epoch": 1.007528556593977, "grad_norm": 2.842876434326172, "learning_rate": 2.8332298489484555e-06, "loss": 0.049735039472579956, "step": 3881 }, { "epoch": 1.0077881619937694, "grad_norm": 2.515897274017334, "learning_rate": 2.832284479506286e-06, "loss": 0.06593941152095795, "step": 3882 }, { "epoch": 1.0080477673935617, "grad_norm": 3.377268075942993, "learning_rate": 2.8313390955181025e-06, "loss": 0.047624971717596054, "step": 3883 }, { "epoch": 1.008307372793354, "grad_norm": 3.633416175842285, "learning_rate": 2.830393697151031e-06, "loss": 0.1155373752117157, "step": 3884 }, { "epoch": 1.0085669781931463, "grad_norm": 2.4512743949890137, "learning_rate": 2.8294482845721934e-06, "loss": 0.0690910741686821, "step": 3885 }, { "epoch": 1.0088265835929386, "grad_norm": 2.210526704788208, "learning_rate": 2.828502857948719e-06, "loss": 0.07545661181211472, "step": 3886 }, { "epoch": 1.009086188992731, "grad_norm": 2.566495418548584, "learning_rate": 2.8275574174477372e-06, "loss": 0.06500071287155151, "step": 3887 }, { "epoch": 1.0093457943925233, "grad_norm": 2.480783224105835, "learning_rate": 2.8266119632363786e-06, "loss": 0.08104701340198517, "step": 3888 }, { "epoch": 1.0096053997923158, "grad_norm": 2.1972739696502686, "learning_rate": 2.8256664954817805e-06, "loss": 0.08489517122507095, "step": 3889 }, { "epoch": 1.009865005192108, "grad_norm": 1.8923053741455078, "learning_rate": 2.8247210143510774e-06, "loss": 0.08622998744249344, "step": 3890 }, { "epoch": 1.0101246105919004, "grad_norm": 4.16513729095459, "learning_rate": 2.8237755200114105e-06, "loss": 0.05584811419248581, "step": 3891 }, { "epoch": 1.0103842159916927, "grad_norm": 4.069782733917236, "learning_rate": 2.82283001262992e-06, "loss": 0.032432056963443756, "step": 3892 }, { "epoch": 1.010643821391485, "grad_norm": 2.6284966468811035, "learning_rate": 2.8218844923737517e-06, "loss": 0.06262995302677155, "step": 3893 }, { "epoch": 1.0109034267912773, "grad_norm": 1.453033447265625, "learning_rate": 2.8209389594100507e-06, "loss": 0.09289449453353882, "step": 3894 }, { "epoch": 1.0111630321910696, "grad_norm": 2.688459873199463, "learning_rate": 2.8199934139059664e-06, "loss": 0.0665975958108902, "step": 3895 }, { "epoch": 1.011422637590862, "grad_norm": 3.3629674911499023, "learning_rate": 2.8190478560286488e-06, "loss": 0.04926665872335434, "step": 3896 }, { "epoch": 1.0116822429906542, "grad_norm": 3.51141357421875, "learning_rate": 2.818102285945251e-06, "loss": 0.039052993059158325, "step": 3897 }, { "epoch": 1.0119418483904465, "grad_norm": 2.540029764175415, "learning_rate": 2.8171567038229286e-06, "loss": 0.07106764614582062, "step": 3898 }, { "epoch": 1.0122014537902388, "grad_norm": 3.086373805999756, "learning_rate": 2.8162111098288393e-06, "loss": 0.021100036799907684, "step": 3899 }, { "epoch": 1.0124610591900312, "grad_norm": 4.145785808563232, "learning_rate": 2.8152655041301413e-06, "loss": 0.096808061003685, "step": 3900 }, { "epoch": 1.0127206645898235, "grad_norm": 3.397812604904175, "learning_rate": 2.8143198868939957e-06, "loss": 0.04936348646879196, "step": 3901 }, { "epoch": 1.0129802699896158, "grad_norm": 3.4401793479919434, "learning_rate": 2.8133742582875677e-06, "loss": 0.045203737914562225, "step": 3902 }, { "epoch": 1.013239875389408, "grad_norm": 3.0262515544891357, "learning_rate": 2.8124286184780224e-06, "loss": 0.03168366849422455, "step": 3903 }, { "epoch": 1.0134994807892004, "grad_norm": 1.7772440910339355, "learning_rate": 2.8114829676325263e-06, "loss": 0.08794403821229935, "step": 3904 }, { "epoch": 1.0137590861889927, "grad_norm": 2.8097732067108154, "learning_rate": 2.810537305918249e-06, "loss": 0.06478134542703629, "step": 3905 }, { "epoch": 1.014018691588785, "grad_norm": 2.5284435749053955, "learning_rate": 2.809591633502362e-06, "loss": 0.02305014617741108, "step": 3906 }, { "epoch": 1.0142782969885773, "grad_norm": 2.008112907409668, "learning_rate": 2.808645950552039e-06, "loss": 0.06978386640548706, "step": 3907 }, { "epoch": 1.0145379023883696, "grad_norm": 4.603976726531982, "learning_rate": 2.807700257234455e-06, "loss": 0.05117189884185791, "step": 3908 }, { "epoch": 1.014797507788162, "grad_norm": 3.7616329193115234, "learning_rate": 2.8067545537167874e-06, "loss": 0.04437851905822754, "step": 3909 }, { "epoch": 1.0150571131879542, "grad_norm": 4.137739181518555, "learning_rate": 2.805808840166212e-06, "loss": 0.053877852857112885, "step": 3910 }, { "epoch": 1.0153167185877465, "grad_norm": 2.093984365463257, "learning_rate": 2.804863116749913e-06, "loss": 0.08901694416999817, "step": 3911 }, { "epoch": 1.0155763239875388, "grad_norm": 2.424070119857788, "learning_rate": 2.8039173836350714e-06, "loss": 0.03368374705314636, "step": 3912 }, { "epoch": 1.0158359293873314, "grad_norm": 1.6240442991256714, "learning_rate": 2.80297164098887e-06, "loss": 0.08390182256698608, "step": 3913 }, { "epoch": 1.0160955347871237, "grad_norm": 4.0733184814453125, "learning_rate": 2.802025888978496e-06, "loss": 0.05504004657268524, "step": 3914 }, { "epoch": 1.016355140186916, "grad_norm": 3.0028328895568848, "learning_rate": 2.801080127771136e-06, "loss": 0.028692159801721573, "step": 3915 }, { "epoch": 1.0166147455867083, "grad_norm": 3.085200071334839, "learning_rate": 2.800134357533979e-06, "loss": 0.02359565533697605, "step": 3916 }, { "epoch": 1.0168743509865006, "grad_norm": 3.089411973953247, "learning_rate": 2.7991885784342154e-06, "loss": 0.024532940238714218, "step": 3917 }, { "epoch": 1.017133956386293, "grad_norm": 4.0335283279418945, "learning_rate": 2.7982427906390374e-06, "loss": 0.0288909412920475, "step": 3918 }, { "epoch": 1.0173935617860852, "grad_norm": 2.3545637130737305, "learning_rate": 2.797296994315638e-06, "loss": 0.07317750155925751, "step": 3919 }, { "epoch": 1.0176531671858775, "grad_norm": 3.5695371627807617, "learning_rate": 2.796351189631214e-06, "loss": 0.04398123174905777, "step": 3920 }, { "epoch": 1.0179127725856698, "grad_norm": 2.940814256668091, "learning_rate": 2.795405376752961e-06, "loss": 0.06045273318886757, "step": 3921 }, { "epoch": 1.0181723779854621, "grad_norm": 1.8928711414337158, "learning_rate": 2.7944595558480765e-06, "loss": 0.0712544322013855, "step": 3922 }, { "epoch": 1.0184319833852544, "grad_norm": 3.854687213897705, "learning_rate": 2.7935137270837605e-06, "loss": 0.05936316400766373, "step": 3923 }, { "epoch": 1.0186915887850467, "grad_norm": 2.174966812133789, "learning_rate": 2.7925678906272143e-06, "loss": 0.0810035914182663, "step": 3924 }, { "epoch": 1.018951194184839, "grad_norm": 2.8906190395355225, "learning_rate": 2.7916220466456384e-06, "loss": 0.08084101974964142, "step": 3925 }, { "epoch": 1.0192107995846313, "grad_norm": 3.473522424697876, "learning_rate": 2.7906761953062396e-06, "loss": 0.07507621496915817, "step": 3926 }, { "epoch": 1.0194704049844237, "grad_norm": 3.8653311729431152, "learning_rate": 2.78973033677622e-06, "loss": 0.049873895943164825, "step": 3927 }, { "epoch": 1.019730010384216, "grad_norm": 3.9908607006073, "learning_rate": 2.788784471222786e-06, "loss": 0.029896780848503113, "step": 3928 }, { "epoch": 1.0199896157840083, "grad_norm": 1.6528176069259644, "learning_rate": 2.7878385988131455e-06, "loss": 0.08684729784727097, "step": 3929 }, { "epoch": 1.0202492211838006, "grad_norm": 2.3459842205047607, "learning_rate": 2.7868927197145086e-06, "loss": 0.08467750251293182, "step": 3930 }, { "epoch": 1.0205088265835929, "grad_norm": 3.2477402687072754, "learning_rate": 2.785946834094082e-06, "loss": 0.011677428148686886, "step": 3931 }, { "epoch": 1.0207684319833852, "grad_norm": 2.229412794113159, "learning_rate": 2.7850009421190787e-06, "loss": 0.08626257628202438, "step": 3932 }, { "epoch": 1.0210280373831775, "grad_norm": 3.7638113498687744, "learning_rate": 2.7840550439567104e-06, "loss": 0.04219447448849678, "step": 3933 }, { "epoch": 1.0212876427829698, "grad_norm": 3.5162057876586914, "learning_rate": 2.7831091397741905e-06, "loss": 0.053571056574583054, "step": 3934 }, { "epoch": 1.021547248182762, "grad_norm": 4.139199256896973, "learning_rate": 2.782163229738732e-06, "loss": 0.044286176562309265, "step": 3935 }, { "epoch": 1.0218068535825544, "grad_norm": 1.572709083557129, "learning_rate": 2.7812173140175506e-06, "loss": 0.10284431278705597, "step": 3936 }, { "epoch": 1.022066458982347, "grad_norm": 4.175275802612305, "learning_rate": 2.780271392777864e-06, "loss": 0.018436651676893234, "step": 3937 }, { "epoch": 1.0223260643821392, "grad_norm": 3.0032382011413574, "learning_rate": 2.7793254661868877e-06, "loss": 0.05983491986989975, "step": 3938 }, { "epoch": 1.0225856697819315, "grad_norm": 3.3440933227539062, "learning_rate": 2.778379534411841e-06, "loss": 0.0641261413693428, "step": 3939 }, { "epoch": 1.0228452751817239, "grad_norm": 2.320543050765991, "learning_rate": 2.7774335976199422e-06, "loss": 0.087073415517807, "step": 3940 }, { "epoch": 1.0231048805815162, "grad_norm": 3.039219379425049, "learning_rate": 2.7764876559784105e-06, "loss": 0.019788989797234535, "step": 3941 }, { "epoch": 1.0233644859813085, "grad_norm": 4.190389633178711, "learning_rate": 2.775541709654469e-06, "loss": 0.06837867200374603, "step": 3942 }, { "epoch": 1.0236240913811008, "grad_norm": 4.889593601226807, "learning_rate": 2.7745957588153376e-06, "loss": 0.03978399932384491, "step": 3943 }, { "epoch": 1.023883696780893, "grad_norm": 2.504082441329956, "learning_rate": 2.7736498036282388e-06, "loss": 0.08112341165542603, "step": 3944 }, { "epoch": 1.0241433021806854, "grad_norm": 3.0524258613586426, "learning_rate": 2.7727038442603967e-06, "loss": 0.0641295537352562, "step": 3945 }, { "epoch": 1.0244029075804777, "grad_norm": 1.3348926305770874, "learning_rate": 2.7717578808790342e-06, "loss": 0.09703397750854492, "step": 3946 }, { "epoch": 1.02466251298027, "grad_norm": 1.9785901308059692, "learning_rate": 2.7708119136513778e-06, "loss": 0.0955120399594307, "step": 3947 }, { "epoch": 1.0249221183800623, "grad_norm": 2.8497841358184814, "learning_rate": 2.7698659427446507e-06, "loss": 0.03770347312092781, "step": 3948 }, { "epoch": 1.0251817237798546, "grad_norm": 2.7761895656585693, "learning_rate": 2.768919968326079e-06, "loss": 0.09526559710502625, "step": 3949 }, { "epoch": 1.025441329179647, "grad_norm": 2.5931806564331055, "learning_rate": 2.767973990562891e-06, "loss": 0.06812390685081482, "step": 3950 }, { "epoch": 1.0257009345794392, "grad_norm": 3.938485622406006, "learning_rate": 2.7670280096223136e-06, "loss": 0.03966488689184189, "step": 3951 }, { "epoch": 1.0259605399792315, "grad_norm": 3.006927490234375, "learning_rate": 2.766082025671573e-06, "loss": 0.03754935786128044, "step": 3952 }, { "epoch": 1.0262201453790238, "grad_norm": 3.2878222465515137, "learning_rate": 2.7651360388778993e-06, "loss": 0.03896132856607437, "step": 3953 }, { "epoch": 1.0264797507788161, "grad_norm": 2.139326333999634, "learning_rate": 2.7641900494085193e-06, "loss": 0.10176488757133484, "step": 3954 }, { "epoch": 1.0267393561786085, "grad_norm": 2.506192922592163, "learning_rate": 2.763244057430665e-06, "loss": 0.016956429928541183, "step": 3955 }, { "epoch": 1.0269989615784008, "grad_norm": 2.637286901473999, "learning_rate": 2.762298063111563e-06, "loss": 0.025671277195215225, "step": 3956 }, { "epoch": 1.027258566978193, "grad_norm": 4.4823527336120605, "learning_rate": 2.7613520666184466e-06, "loss": 0.04011429101228714, "step": 3957 }, { "epoch": 1.0275181723779854, "grad_norm": 3.638679265975952, "learning_rate": 2.7604060681185442e-06, "loss": 0.027441199868917465, "step": 3958 }, { "epoch": 1.0277777777777777, "grad_norm": 3.9311389923095703, "learning_rate": 2.7594600677790866e-06, "loss": 0.040281083434820175, "step": 3959 }, { "epoch": 1.02803738317757, "grad_norm": 1.577609896659851, "learning_rate": 2.758514065767306e-06, "loss": 0.006718968041241169, "step": 3960 }, { "epoch": 1.0282969885773625, "grad_norm": 1.6149389743804932, "learning_rate": 2.7575680622504342e-06, "loss": 0.09144975244998932, "step": 3961 }, { "epoch": 1.0285565939771548, "grad_norm": 2.7003867626190186, "learning_rate": 2.7566220573957015e-06, "loss": 0.0584682896733284, "step": 3962 }, { "epoch": 1.0288161993769471, "grad_norm": 4.062076568603516, "learning_rate": 2.755676051370341e-06, "loss": 0.048331934958696365, "step": 3963 }, { "epoch": 1.0290758047767394, "grad_norm": 2.0782408714294434, "learning_rate": 2.7547300443415836e-06, "loss": 0.07219375669956207, "step": 3964 }, { "epoch": 1.0293354101765317, "grad_norm": 1.5268653631210327, "learning_rate": 2.753784036476664e-06, "loss": 0.09120845794677734, "step": 3965 }, { "epoch": 1.029595015576324, "grad_norm": 3.0634849071502686, "learning_rate": 2.752838027942813e-06, "loss": 0.029507078230381012, "step": 3966 }, { "epoch": 1.0298546209761164, "grad_norm": 1.3790923357009888, "learning_rate": 2.751892018907263e-06, "loss": 0.09669916331768036, "step": 3967 }, { "epoch": 1.0301142263759087, "grad_norm": 2.0366570949554443, "learning_rate": 2.7509460095372477e-06, "loss": 0.09541483968496323, "step": 3968 }, { "epoch": 1.030373831775701, "grad_norm": 1.356662392616272, "learning_rate": 2.7500000000000004e-06, "loss": 0.09329788386821747, "step": 3969 }, { "epoch": 1.0306334371754933, "grad_norm": 1.9294637441635132, "learning_rate": 2.749053990462752e-06, "loss": 0.0760013535618782, "step": 3970 }, { "epoch": 1.0308930425752856, "grad_norm": 1.5242077112197876, "learning_rate": 2.7481079810927374e-06, "loss": 0.09433136880397797, "step": 3971 }, { "epoch": 1.0311526479750779, "grad_norm": 1.813827395439148, "learning_rate": 2.747161972057188e-06, "loss": 0.08820557594299316, "step": 3972 }, { "epoch": 1.0314122533748702, "grad_norm": 3.5673515796661377, "learning_rate": 2.7462159635233372e-06, "loss": 0.03829487785696983, "step": 3973 }, { "epoch": 1.0316718587746625, "grad_norm": 3.9640915393829346, "learning_rate": 2.7452699556584167e-06, "loss": 0.043137893080711365, "step": 3974 }, { "epoch": 1.0319314641744548, "grad_norm": 2.8416008949279785, "learning_rate": 2.7443239486296603e-06, "loss": 0.04989931732416153, "step": 3975 }, { "epoch": 1.032191069574247, "grad_norm": 3.392849922180176, "learning_rate": 2.7433779426042997e-06, "loss": 0.06625586748123169, "step": 3976 }, { "epoch": 1.0324506749740394, "grad_norm": 1.6053979396820068, "learning_rate": 2.742431937749567e-06, "loss": 0.08397123962640762, "step": 3977 }, { "epoch": 1.0327102803738317, "grad_norm": 3.2950503826141357, "learning_rate": 2.7414859342326943e-06, "loss": 0.05522729083895683, "step": 3978 }, { "epoch": 1.032969885773624, "grad_norm": 1.466151475906372, "learning_rate": 2.740539932220915e-06, "loss": 0.10040596127510071, "step": 3979 }, { "epoch": 1.0332294911734163, "grad_norm": 3.0148112773895264, "learning_rate": 2.739593931881457e-06, "loss": 0.03378322720527649, "step": 3980 }, { "epoch": 1.0334890965732086, "grad_norm": 2.0857787132263184, "learning_rate": 2.738647933381554e-06, "loss": 0.08599846065044403, "step": 3981 }, { "epoch": 1.033748701973001, "grad_norm": 3.6174046993255615, "learning_rate": 2.7377019368884367e-06, "loss": 0.04707502946257591, "step": 3982 }, { "epoch": 1.0340083073727933, "grad_norm": 1.7080374956130981, "learning_rate": 2.7367559425693364e-06, "loss": 0.090047687292099, "step": 3983 }, { "epoch": 1.0342679127725856, "grad_norm": 1.6451809406280518, "learning_rate": 2.7358099505914814e-06, "loss": 0.09308139979839325, "step": 3984 }, { "epoch": 1.034527518172378, "grad_norm": 3.0345675945281982, "learning_rate": 2.7348639611221014e-06, "loss": 0.02550310268998146, "step": 3985 }, { "epoch": 1.0347871235721704, "grad_norm": 3.937973976135254, "learning_rate": 2.7339179743284273e-06, "loss": 0.04381372779607773, "step": 3986 }, { "epoch": 1.0350467289719627, "grad_norm": 4.048559188842773, "learning_rate": 2.732971990377688e-06, "loss": 0.04065912216901779, "step": 3987 }, { "epoch": 1.035306334371755, "grad_norm": 3.2888717651367188, "learning_rate": 2.7320260094371097e-06, "loss": 0.027598656713962555, "step": 3988 }, { "epoch": 1.0355659397715473, "grad_norm": 2.6655964851379395, "learning_rate": 2.7310800316739215e-06, "loss": 0.039265226572752, "step": 3989 }, { "epoch": 1.0358255451713396, "grad_norm": 3.544816255569458, "learning_rate": 2.7301340572553505e-06, "loss": 0.052528589963912964, "step": 3990 }, { "epoch": 1.036085150571132, "grad_norm": 3.6720998287200928, "learning_rate": 2.729188086348624e-06, "loss": 0.06252562254667282, "step": 3991 }, { "epoch": 1.0363447559709242, "grad_norm": 3.190408706665039, "learning_rate": 2.728242119120967e-06, "loss": 0.05642625316977501, "step": 3992 }, { "epoch": 1.0366043613707165, "grad_norm": 2.171942710876465, "learning_rate": 2.7272961557396044e-06, "loss": 0.08242321759462357, "step": 3993 }, { "epoch": 1.0368639667705088, "grad_norm": 4.600465774536133, "learning_rate": 2.726350196371762e-06, "loss": 0.032257623970508575, "step": 3994 }, { "epoch": 1.0371235721703012, "grad_norm": 2.516791820526123, "learning_rate": 2.7254042411846636e-06, "loss": 0.01869191974401474, "step": 3995 }, { "epoch": 1.0373831775700935, "grad_norm": 2.368593692779541, "learning_rate": 2.724458290345532e-06, "loss": 0.06692339479923248, "step": 3996 }, { "epoch": 1.0376427829698858, "grad_norm": 3.9954404830932617, "learning_rate": 2.7235123440215898e-06, "loss": 0.04482244327664375, "step": 3997 }, { "epoch": 1.037902388369678, "grad_norm": 2.468771457672119, "learning_rate": 2.7225664023800597e-06, "loss": 0.08396980166435242, "step": 3998 }, { "epoch": 1.0381619937694704, "grad_norm": 2.848997116088867, "learning_rate": 2.72162046558816e-06, "loss": 0.024717744439840317, "step": 3999 }, { "epoch": 1.0384215991692627, "grad_norm": 1.7482929229736328, "learning_rate": 2.720674533813113e-06, "loss": 0.09570075571537018, "step": 4000 }, { "epoch": 1.038681204569055, "grad_norm": 3.7631638050079346, "learning_rate": 2.719728607222136e-06, "loss": 0.03635646030306816, "step": 4001 }, { "epoch": 1.0389408099688473, "grad_norm": 1.493350625038147, "learning_rate": 2.7187826859824497e-06, "loss": 0.08552553504705429, "step": 4002 }, { "epoch": 1.0392004153686396, "grad_norm": 2.9481348991394043, "learning_rate": 2.717836770261269e-06, "loss": 0.02740093693137169, "step": 4003 }, { "epoch": 1.039460020768432, "grad_norm": 2.8448238372802734, "learning_rate": 2.7168908602258106e-06, "loss": 0.038349054753780365, "step": 4004 }, { "epoch": 1.0397196261682242, "grad_norm": 2.7656917572021484, "learning_rate": 2.7159449560432903e-06, "loss": 0.051474831998348236, "step": 4005 }, { "epoch": 1.0399792315680165, "grad_norm": 3.9072155952453613, "learning_rate": 2.714999057880922e-06, "loss": 0.03222525864839554, "step": 4006 }, { "epoch": 1.0402388369678088, "grad_norm": 2.7897493839263916, "learning_rate": 2.714053165905919e-06, "loss": 0.07154998183250427, "step": 4007 }, { "epoch": 1.0404984423676011, "grad_norm": 1.5471274852752686, "learning_rate": 2.7131072802854925e-06, "loss": 0.09953596442937851, "step": 4008 }, { "epoch": 1.0407580477673937, "grad_norm": 2.4412550926208496, "learning_rate": 2.7121614011868547e-06, "loss": 0.044319383800029755, "step": 4009 }, { "epoch": 1.041017653167186, "grad_norm": 2.7776739597320557, "learning_rate": 2.711215528777216e-06, "loss": 0.05222005397081375, "step": 4010 }, { "epoch": 1.0412772585669783, "grad_norm": 2.4688711166381836, "learning_rate": 2.7102696632237813e-06, "loss": 0.08536887913942337, "step": 4011 }, { "epoch": 1.0415368639667706, "grad_norm": 3.585768938064575, "learning_rate": 2.7093238046937615e-06, "loss": 0.03417842462658882, "step": 4012 }, { "epoch": 1.041796469366563, "grad_norm": 4.707372188568115, "learning_rate": 2.708377953354362e-06, "loss": 0.05575156584382057, "step": 4013 }, { "epoch": 1.0420560747663552, "grad_norm": 4.110029220581055, "learning_rate": 2.7074321093727877e-06, "loss": 0.023141970857977867, "step": 4014 }, { "epoch": 1.0423156801661475, "grad_norm": 2.6947364807128906, "learning_rate": 2.706486272916241e-06, "loss": 0.09709420800209045, "step": 4015 }, { "epoch": 1.0425752855659398, "grad_norm": 1.600274682044983, "learning_rate": 2.7055404441519246e-06, "loss": 0.0989212617278099, "step": 4016 }, { "epoch": 1.0428348909657321, "grad_norm": 3.245651960372925, "learning_rate": 2.70459462324704e-06, "loss": 0.06181341037154198, "step": 4017 }, { "epoch": 1.0430944963655244, "grad_norm": 2.834902763366699, "learning_rate": 2.7036488103687868e-06, "loss": 0.08381226658821106, "step": 4018 }, { "epoch": 1.0433541017653167, "grad_norm": 1.9912270307540894, "learning_rate": 2.702703005684363e-06, "loss": 0.07480700314044952, "step": 4019 }, { "epoch": 1.043613707165109, "grad_norm": 2.118088483810425, "learning_rate": 2.7017572093609633e-06, "loss": 0.09297561645507812, "step": 4020 }, { "epoch": 1.0438733125649013, "grad_norm": 1.447504997253418, "learning_rate": 2.700811421565785e-06, "loss": 0.09996850788593292, "step": 4021 }, { "epoch": 1.0441329179646937, "grad_norm": 2.2271249294281006, "learning_rate": 2.6998656424660226e-06, "loss": 0.08612946420907974, "step": 4022 }, { "epoch": 1.044392523364486, "grad_norm": 2.5600264072418213, "learning_rate": 2.698919872228865e-06, "loss": 0.07486042380332947, "step": 4023 }, { "epoch": 1.0446521287642783, "grad_norm": 2.674595832824707, "learning_rate": 2.6979741110215045e-06, "loss": 0.0466763973236084, "step": 4024 }, { "epoch": 1.0449117341640706, "grad_norm": 2.8126392364501953, "learning_rate": 2.6970283590111307e-06, "loss": 0.023769378662109375, "step": 4025 }, { "epoch": 1.0451713395638629, "grad_norm": 2.1751604080200195, "learning_rate": 2.6960826163649297e-06, "loss": 0.09192458540201187, "step": 4026 }, { "epoch": 1.0454309449636552, "grad_norm": 3.368004560470581, "learning_rate": 2.695136883250088e-06, "loss": 0.03276674449443817, "step": 4027 }, { "epoch": 1.0456905503634475, "grad_norm": 3.096442461013794, "learning_rate": 2.6941911598337885e-06, "loss": 0.02544855885207653, "step": 4028 }, { "epoch": 1.0459501557632398, "grad_norm": 1.7553772926330566, "learning_rate": 2.6932454462832146e-06, "loss": 0.09723693132400513, "step": 4029 }, { "epoch": 1.046209761163032, "grad_norm": 2.921396255493164, "learning_rate": 2.6922997427655456e-06, "loss": 0.028456319123506546, "step": 4030 }, { "epoch": 1.0464693665628244, "grad_norm": 2.389998197555542, "learning_rate": 2.6913540494479613e-06, "loss": 0.06526590883731842, "step": 4031 }, { "epoch": 1.0467289719626167, "grad_norm": 3.690566062927246, "learning_rate": 2.6904083664976384e-06, "loss": 0.05225120112299919, "step": 4032 }, { "epoch": 1.0469885773624092, "grad_norm": 3.21195650100708, "learning_rate": 2.689462694081752e-06, "loss": 0.07025641202926636, "step": 4033 }, { "epoch": 1.0472481827622016, "grad_norm": 2.649714946746826, "learning_rate": 2.6885170323674748e-06, "loss": 0.07118889689445496, "step": 4034 }, { "epoch": 1.0475077881619939, "grad_norm": 3.1698262691497803, "learning_rate": 2.6875713815219783e-06, "loss": 0.030820205807685852, "step": 4035 }, { "epoch": 1.0477673935617862, "grad_norm": 1.9878491163253784, "learning_rate": 2.6866257417124326e-06, "loss": 0.08192302286624908, "step": 4036 }, { "epoch": 1.0480269989615785, "grad_norm": 2.1541922092437744, "learning_rate": 2.6856801131060046e-06, "loss": 0.06637127697467804, "step": 4037 }, { "epoch": 1.0482866043613708, "grad_norm": 2.979593276977539, "learning_rate": 2.6847344958698594e-06, "loss": 0.04644976556301117, "step": 4038 }, { "epoch": 1.048546209761163, "grad_norm": 2.6115338802337646, "learning_rate": 2.683788890171162e-06, "loss": 0.02788042090833187, "step": 4039 }, { "epoch": 1.0488058151609554, "grad_norm": 2.433579206466675, "learning_rate": 2.6828432961770717e-06, "loss": 0.06793923676013947, "step": 4040 }, { "epoch": 1.0490654205607477, "grad_norm": 3.384035348892212, "learning_rate": 2.68189771405475e-06, "loss": 0.0332099050283432, "step": 4041 }, { "epoch": 1.04932502596054, "grad_norm": 1.944540023803711, "learning_rate": 2.680952143971352e-06, "loss": 0.08157407492399216, "step": 4042 }, { "epoch": 1.0495846313603323, "grad_norm": 2.2567191123962402, "learning_rate": 2.6800065860940343e-06, "loss": 0.07759096473455429, "step": 4043 }, { "epoch": 1.0498442367601246, "grad_norm": 2.322727918624878, "learning_rate": 2.6790610405899496e-06, "loss": 0.013983847573399544, "step": 4044 }, { "epoch": 1.050103842159917, "grad_norm": 3.430367946624756, "learning_rate": 2.6781155076262494e-06, "loss": 0.05069349333643913, "step": 4045 }, { "epoch": 1.0503634475597092, "grad_norm": 2.7190232276916504, "learning_rate": 2.6771699873700806e-06, "loss": 0.054574400186538696, "step": 4046 }, { "epoch": 1.0506230529595015, "grad_norm": 2.3924853801727295, "learning_rate": 2.6762244799885906e-06, "loss": 0.07253812253475189, "step": 4047 }, { "epoch": 1.0508826583592938, "grad_norm": 3.0575430393218994, "learning_rate": 2.675278985648923e-06, "loss": 0.05577671155333519, "step": 4048 }, { "epoch": 1.0511422637590861, "grad_norm": 1.6060243844985962, "learning_rate": 2.674333504518221e-06, "loss": 0.09102092683315277, "step": 4049 }, { "epoch": 1.0514018691588785, "grad_norm": 3.0683600902557373, "learning_rate": 2.673388036763622e-06, "loss": 0.08076492697000504, "step": 4050 }, { "epoch": 1.0516614745586708, "grad_norm": 3.2997539043426514, "learning_rate": 2.672442582552264e-06, "loss": 0.019480954855680466, "step": 4051 }, { "epoch": 1.051921079958463, "grad_norm": 3.074047565460205, "learning_rate": 2.671497142051282e-06, "loss": 0.032373834401369095, "step": 4052 }, { "epoch": 1.0521806853582554, "grad_norm": 2.3276684284210205, "learning_rate": 2.6705517154278073e-06, "loss": 0.061611540615558624, "step": 4053 }, { "epoch": 1.0524402907580477, "grad_norm": 4.068581581115723, "learning_rate": 2.66960630284897e-06, "loss": 0.02804327756166458, "step": 4054 }, { "epoch": 1.05269989615784, "grad_norm": 2.7821707725524902, "learning_rate": 2.668660904481898e-06, "loss": 0.05365408957004547, "step": 4055 }, { "epoch": 1.0529595015576323, "grad_norm": 2.1781344413757324, "learning_rate": 2.667715520493716e-06, "loss": 0.08002577722072601, "step": 4056 }, { "epoch": 1.0532191069574246, "grad_norm": 1.766291618347168, "learning_rate": 2.666770151051545e-06, "loss": 0.10908893495798111, "step": 4057 }, { "epoch": 1.0534787123572171, "grad_norm": 4.186858654022217, "learning_rate": 2.665824796322506e-06, "loss": 0.051531944423913956, "step": 4058 }, { "epoch": 1.0537383177570094, "grad_norm": 3.769212484359741, "learning_rate": 2.6648794564737156e-06, "loss": 0.03133492171764374, "step": 4059 }, { "epoch": 1.0539979231568017, "grad_norm": 3.144989013671875, "learning_rate": 2.663934131672289e-06, "loss": 0.058954354375600815, "step": 4060 }, { "epoch": 1.054257528556594, "grad_norm": 2.325281858444214, "learning_rate": 2.6629888220853365e-06, "loss": 0.05852513015270233, "step": 4061 }, { "epoch": 1.0545171339563864, "grad_norm": 2.33276104927063, "learning_rate": 2.662043527879968e-06, "loss": 0.051188595592975616, "step": 4062 }, { "epoch": 1.0547767393561787, "grad_norm": 2.7250559329986572, "learning_rate": 2.66109824922329e-06, "loss": 0.07454364001750946, "step": 4063 }, { "epoch": 1.055036344755971, "grad_norm": 2.047347068786621, "learning_rate": 2.660152986282406e-06, "loss": 0.08857395499944687, "step": 4064 }, { "epoch": 1.0552959501557633, "grad_norm": 2.3472671508789062, "learning_rate": 2.659207739224417e-06, "loss": 0.06462789326906204, "step": 4065 }, { "epoch": 1.0555555555555556, "grad_norm": 2.64546537399292, "learning_rate": 2.65826250821642e-06, "loss": 0.08237029612064362, "step": 4066 }, { "epoch": 1.0558151609553479, "grad_norm": 2.8957278728485107, "learning_rate": 2.6573172934255116e-06, "loss": 0.02088775858283043, "step": 4067 }, { "epoch": 1.0560747663551402, "grad_norm": 2.3724021911621094, "learning_rate": 2.656372095018784e-06, "loss": 0.0777602344751358, "step": 4068 }, { "epoch": 1.0563343717549325, "grad_norm": 3.6290290355682373, "learning_rate": 2.6554269131633254e-06, "loss": 0.03071206621825695, "step": 4069 }, { "epoch": 1.0565939771547248, "grad_norm": 4.016747951507568, "learning_rate": 2.6544817480262237e-06, "loss": 0.04657578468322754, "step": 4070 }, { "epoch": 1.0568535825545171, "grad_norm": 1.773616075515747, "learning_rate": 2.6535365997745616e-06, "loss": 0.0851423591375351, "step": 4071 }, { "epoch": 1.0571131879543094, "grad_norm": 2.058293342590332, "learning_rate": 2.65259146857542e-06, "loss": 0.09665881097316742, "step": 4072 }, { "epoch": 1.0573727933541017, "grad_norm": 1.8778573274612427, "learning_rate": 2.6516463545958766e-06, "loss": 0.07107338309288025, "step": 4073 }, { "epoch": 1.057632398753894, "grad_norm": 4.625175476074219, "learning_rate": 2.650701258003005e-06, "loss": 0.0620405375957489, "step": 4074 }, { "epoch": 1.0578920041536863, "grad_norm": 2.1994729042053223, "learning_rate": 2.6497561789638783e-06, "loss": 0.07918239384889603, "step": 4075 }, { "epoch": 1.0581516095534786, "grad_norm": 1.8920437097549438, "learning_rate": 2.648811117645564e-06, "loss": 0.10092388838529587, "step": 4076 }, { "epoch": 1.058411214953271, "grad_norm": 3.0711302757263184, "learning_rate": 2.6478660742151257e-06, "loss": 0.04649528115987778, "step": 4077 }, { "epoch": 1.0586708203530633, "grad_norm": 2.5979654788970947, "learning_rate": 2.646921048839628e-06, "loss": 0.04630441218614578, "step": 4078 }, { "epoch": 1.0589304257528556, "grad_norm": 2.431591272354126, "learning_rate": 2.6459760416861293e-06, "loss": 0.06313815712928772, "step": 4079 }, { "epoch": 1.0591900311526479, "grad_norm": 3.4282443523406982, "learning_rate": 2.6450310529216837e-06, "loss": 0.05509592965245247, "step": 4080 }, { "epoch": 1.0594496365524404, "grad_norm": 2.4812209606170654, "learning_rate": 2.644086082713345e-06, "loss": 0.06957733631134033, "step": 4081 }, { "epoch": 1.0597092419522327, "grad_norm": 3.138118267059326, "learning_rate": 2.64314113122816e-06, "loss": 0.0392177551984787, "step": 4082 }, { "epoch": 1.059968847352025, "grad_norm": 1.550652027130127, "learning_rate": 2.6421961986331786e-06, "loss": 0.1003250926733017, "step": 4083 }, { "epoch": 1.0602284527518173, "grad_norm": 4.140512466430664, "learning_rate": 2.6412512850954397e-06, "loss": 0.040859125554561615, "step": 4084 }, { "epoch": 1.0604880581516096, "grad_norm": 2.9083943367004395, "learning_rate": 2.640306390781984e-06, "loss": 0.03137354180216789, "step": 4085 }, { "epoch": 1.060747663551402, "grad_norm": 2.336470603942871, "learning_rate": 2.6393615158598464e-06, "loss": 0.030945315957069397, "step": 4086 }, { "epoch": 1.0610072689511942, "grad_norm": 3.673325777053833, "learning_rate": 2.6384166604960596e-06, "loss": 0.015696678310632706, "step": 4087 }, { "epoch": 1.0612668743509865, "grad_norm": 2.832279920578003, "learning_rate": 2.6374718248576526e-06, "loss": 0.0832982063293457, "step": 4088 }, { "epoch": 1.0615264797507789, "grad_norm": 3.479811429977417, "learning_rate": 2.6365270091116497e-06, "loss": 0.010625971481204033, "step": 4089 }, { "epoch": 1.0617860851505712, "grad_norm": 1.7402461767196655, "learning_rate": 2.635582213425074e-06, "loss": 0.10046546906232834, "step": 4090 }, { "epoch": 1.0620456905503635, "grad_norm": 2.1549534797668457, "learning_rate": 2.6346374379649432e-06, "loss": 0.08990339934825897, "step": 4091 }, { "epoch": 1.0623052959501558, "grad_norm": 1.7616523504257202, "learning_rate": 2.633692682898272e-06, "loss": 0.0917896032333374, "step": 4092 }, { "epoch": 1.062564901349948, "grad_norm": 3.612013578414917, "learning_rate": 2.632747948392071e-06, "loss": 0.02186252363026142, "step": 4093 }, { "epoch": 1.0628245067497404, "grad_norm": 4.342433929443359, "learning_rate": 2.6318032346133476e-06, "loss": 0.02583080530166626, "step": 4094 }, { "epoch": 1.0630841121495327, "grad_norm": 2.0781900882720947, "learning_rate": 2.6308585417291073e-06, "loss": 0.09400241076946259, "step": 4095 }, { "epoch": 1.063343717549325, "grad_norm": 4.185186386108398, "learning_rate": 2.6299138699063475e-06, "loss": 0.04244356229901314, "step": 4096 }, { "epoch": 1.0636033229491173, "grad_norm": 2.212658643722534, "learning_rate": 2.6289692193120664e-06, "loss": 0.10347354412078857, "step": 4097 }, { "epoch": 1.0638629283489096, "grad_norm": 2.061983585357666, "learning_rate": 2.628024590113256e-06, "loss": 0.08330054581165314, "step": 4098 }, { "epoch": 1.064122533748702, "grad_norm": 3.4343583583831787, "learning_rate": 2.6270799824769056e-06, "loss": 0.015814006328582764, "step": 4099 }, { "epoch": 1.0643821391484942, "grad_norm": 2.9444921016693115, "learning_rate": 2.626135396569999e-06, "loss": 0.07110892236232758, "step": 4100 }, { "epoch": 1.0646417445482865, "grad_norm": 2.493931531906128, "learning_rate": 2.6251908325595187e-06, "loss": 0.08127719163894653, "step": 4101 }, { "epoch": 1.0649013499480788, "grad_norm": 4.704492568969727, "learning_rate": 2.6242462906124415e-06, "loss": 0.030227910727262497, "step": 4102 }, { "epoch": 1.0651609553478711, "grad_norm": 2.923896074295044, "learning_rate": 2.623301770895741e-06, "loss": 0.01576124131679535, "step": 4103 }, { "epoch": 1.0654205607476634, "grad_norm": 3.2193005084991455, "learning_rate": 2.6223572735763864e-06, "loss": 0.06697461009025574, "step": 4104 }, { "epoch": 1.0656801661474558, "grad_norm": 5.301481246948242, "learning_rate": 2.621412798821343e-06, "loss": 0.02468080446124077, "step": 4105 }, { "epoch": 1.0659397715472483, "grad_norm": 4.952271461486816, "learning_rate": 2.6204683467975716e-06, "loss": 0.030161939561367035, "step": 4106 }, { "epoch": 1.0661993769470406, "grad_norm": 2.051990509033203, "learning_rate": 2.619523917672033e-06, "loss": 0.08948693424463272, "step": 4107 }, { "epoch": 1.066458982346833, "grad_norm": 4.324248313903809, "learning_rate": 2.6185795116116756e-06, "loss": 0.06165216118097305, "step": 4108 }, { "epoch": 1.0667185877466252, "grad_norm": 4.159121990203857, "learning_rate": 2.617635128783452e-06, "loss": 0.023330893367528915, "step": 4109 }, { "epoch": 1.0669781931464175, "grad_norm": 5.601207733154297, "learning_rate": 2.6166907693543085e-06, "loss": 0.013735822401940823, "step": 4110 }, { "epoch": 1.0672377985462098, "grad_norm": 3.6971609592437744, "learning_rate": 2.615746433491183e-06, "loss": 0.03844009339809418, "step": 4111 }, { "epoch": 1.0674974039460021, "grad_norm": 2.820391893386841, "learning_rate": 2.6148021213610147e-06, "loss": 0.038584694266319275, "step": 4112 }, { "epoch": 1.0677570093457944, "grad_norm": 1.5139353275299072, "learning_rate": 2.613857833130735e-06, "loss": 0.10524242371320724, "step": 4113 }, { "epoch": 1.0680166147455867, "grad_norm": 1.9117552042007446, "learning_rate": 2.6129135689672745e-06, "loss": 0.09874506294727325, "step": 4114 }, { "epoch": 1.068276220145379, "grad_norm": 1.2890706062316895, "learning_rate": 2.6119693290375555e-06, "loss": 0.09181550145149231, "step": 4115 }, { "epoch": 1.0685358255451713, "grad_norm": 2.077787399291992, "learning_rate": 2.611025113508498e-06, "loss": 0.09290182590484619, "step": 4116 }, { "epoch": 1.0687954309449637, "grad_norm": 1.901715874671936, "learning_rate": 2.610080922547019e-06, "loss": 0.07169216126203537, "step": 4117 }, { "epoch": 1.069055036344756, "grad_norm": 3.354509115219116, "learning_rate": 2.6091367563200294e-06, "loss": 0.06130516156554222, "step": 4118 }, { "epoch": 1.0693146417445483, "grad_norm": 1.7237602472305298, "learning_rate": 2.608192614994435e-06, "loss": 0.10476624965667725, "step": 4119 }, { "epoch": 1.0695742471443406, "grad_norm": 1.431862711906433, "learning_rate": 2.6072484987371395e-06, "loss": 0.10379011183977127, "step": 4120 }, { "epoch": 1.0698338525441329, "grad_norm": 3.2784101963043213, "learning_rate": 2.6063044077150414e-06, "loss": 0.04655951261520386, "step": 4121 }, { "epoch": 1.0700934579439252, "grad_norm": 3.4592440128326416, "learning_rate": 2.6053603420950337e-06, "loss": 0.040031515061855316, "step": 4122 }, { "epoch": 1.0703530633437175, "grad_norm": 3.1983423233032227, "learning_rate": 2.6044163020440055e-06, "loss": 0.058932580053806305, "step": 4123 }, { "epoch": 1.0706126687435098, "grad_norm": 2.324713945388794, "learning_rate": 2.6034722877288415e-06, "loss": 0.009404126554727554, "step": 4124 }, { "epoch": 1.070872274143302, "grad_norm": 3.3153371810913086, "learning_rate": 2.6025282993164226e-06, "loss": 0.0258273147046566, "step": 4125 }, { "epoch": 1.0711318795430944, "grad_norm": 2.0428237915039062, "learning_rate": 2.6015843369736234e-06, "loss": 0.08260546624660492, "step": 4126 }, { "epoch": 1.0713914849428867, "grad_norm": 3.088172674179077, "learning_rate": 2.6006404008673153e-06, "loss": 0.05646265298128128, "step": 4127 }, { "epoch": 1.071651090342679, "grad_norm": 4.772261142730713, "learning_rate": 2.599696491164364e-06, "loss": 0.04360925406217575, "step": 4128 }, { "epoch": 1.0719106957424716, "grad_norm": 4.407276153564453, "learning_rate": 2.5987526080316316e-06, "loss": 0.032720260322093964, "step": 4129 }, { "epoch": 1.0721703011422639, "grad_norm": 3.062180280685425, "learning_rate": 2.5978087516359756e-06, "loss": 0.08647119998931885, "step": 4130 }, { "epoch": 1.0724299065420562, "grad_norm": 4.216299533843994, "learning_rate": 2.5968649221442473e-06, "loss": 0.03850763291120529, "step": 4131 }, { "epoch": 1.0726895119418485, "grad_norm": 3.274062156677246, "learning_rate": 2.595921119723294e-06, "loss": 0.04949379339814186, "step": 4132 }, { "epoch": 1.0729491173416408, "grad_norm": 3.1963305473327637, "learning_rate": 2.5949773445399595e-06, "loss": 0.024690646678209305, "step": 4133 }, { "epoch": 1.073208722741433, "grad_norm": 2.4225258827209473, "learning_rate": 2.5940335967610807e-06, "loss": 0.0779728889465332, "step": 4134 }, { "epoch": 1.0734683281412254, "grad_norm": 4.257212162017822, "learning_rate": 2.5930898765534906e-06, "loss": 0.055456630885601044, "step": 4135 }, { "epoch": 1.0737279335410177, "grad_norm": 3.4830729961395264, "learning_rate": 2.592146184084018e-06, "loss": 0.06164859980344772, "step": 4136 }, { "epoch": 1.07398753894081, "grad_norm": 3.5573527812957764, "learning_rate": 2.5912025195194855e-06, "loss": 0.07836706191301346, "step": 4137 }, { "epoch": 1.0742471443406023, "grad_norm": 1.8759008646011353, "learning_rate": 2.5902588830267116e-06, "loss": 0.08186164498329163, "step": 4138 }, { "epoch": 1.0745067497403946, "grad_norm": 3.3772082328796387, "learning_rate": 2.5893152747725094e-06, "loss": 0.050400905311107635, "step": 4139 }, { "epoch": 1.074766355140187, "grad_norm": 3.5407583713531494, "learning_rate": 2.588371694923687e-06, "loss": 0.03311251103878021, "step": 4140 }, { "epoch": 1.0750259605399792, "grad_norm": 3.2026989459991455, "learning_rate": 2.5874281436470494e-06, "loss": 0.0295974463224411, "step": 4141 }, { "epoch": 1.0752855659397715, "grad_norm": 2.910358190536499, "learning_rate": 2.586484621109393e-06, "loss": 0.08116450905799866, "step": 4142 }, { "epoch": 1.0755451713395638, "grad_norm": 2.7002997398376465, "learning_rate": 2.585541127477511e-06, "loss": 0.0615374818444252, "step": 4143 }, { "epoch": 1.0758047767393562, "grad_norm": 1.8987804651260376, "learning_rate": 2.5845976629181917e-06, "loss": 0.0901188999414444, "step": 4144 }, { "epoch": 1.0760643821391485, "grad_norm": 3.891918897628784, "learning_rate": 2.58365422759822e-06, "loss": 0.03131946921348572, "step": 4145 }, { "epoch": 1.0763239875389408, "grad_norm": 1.9959802627563477, "learning_rate": 2.582710821684371e-06, "loss": 0.08450855314731598, "step": 4146 }, { "epoch": 1.076583592938733, "grad_norm": 3.5367648601531982, "learning_rate": 2.5817674453434185e-06, "loss": 0.024320099502801895, "step": 4147 }, { "epoch": 1.0768431983385254, "grad_norm": 3.4217841625213623, "learning_rate": 2.5808240987421295e-06, "loss": 0.0729474127292633, "step": 4148 }, { "epoch": 1.0771028037383177, "grad_norm": 3.1962640285491943, "learning_rate": 2.579880782047267e-06, "loss": 0.04654790461063385, "step": 4149 }, { "epoch": 1.07736240913811, "grad_norm": 5.082945346832275, "learning_rate": 2.5789374954255864e-06, "loss": 0.07401392608880997, "step": 4150 }, { "epoch": 1.0776220145379023, "grad_norm": 3.1947340965270996, "learning_rate": 2.5779942390438396e-06, "loss": 0.06343073397874832, "step": 4151 }, { "epoch": 1.0778816199376946, "grad_norm": 2.4925637245178223, "learning_rate": 2.5770510130687726e-06, "loss": 0.018097393214702606, "step": 4152 }, { "epoch": 1.078141225337487, "grad_norm": 2.8923165798187256, "learning_rate": 2.5761078176671275e-06, "loss": 0.0158468559384346, "step": 4153 }, { "epoch": 1.0784008307372794, "grad_norm": 3.4513790607452393, "learning_rate": 2.5751646530056378e-06, "loss": 0.045674458146095276, "step": 4154 }, { "epoch": 1.0786604361370717, "grad_norm": 3.205904722213745, "learning_rate": 2.574221519251035e-06, "loss": 0.05219442769885063, "step": 4155 }, { "epoch": 1.078920041536864, "grad_norm": 3.9109787940979004, "learning_rate": 2.5732784165700418e-06, "loss": 0.02759280987083912, "step": 4156 }, { "epoch": 1.0791796469366564, "grad_norm": 3.401820421218872, "learning_rate": 2.5723353451293793e-06, "loss": 0.05777042359113693, "step": 4157 }, { "epoch": 1.0794392523364487, "grad_norm": 2.6133179664611816, "learning_rate": 2.571392305095759e-06, "loss": 0.02928595058619976, "step": 4158 }, { "epoch": 1.079698857736241, "grad_norm": 1.9735596179962158, "learning_rate": 2.5704492966358897e-06, "loss": 0.09748577326536179, "step": 4159 }, { "epoch": 1.0799584631360333, "grad_norm": 4.029669761657715, "learning_rate": 2.569506319916474e-06, "loss": 0.028011314570903778, "step": 4160 }, { "epoch": 1.0802180685358256, "grad_norm": 3.555030107498169, "learning_rate": 2.5685633751042084e-06, "loss": 0.0379687175154686, "step": 4161 }, { "epoch": 1.0804776739356179, "grad_norm": 2.3187947273254395, "learning_rate": 2.5676204623657824e-06, "loss": 0.09077322483062744, "step": 4162 }, { "epoch": 1.0807372793354102, "grad_norm": 2.869283437728882, "learning_rate": 2.566677581867884e-06, "loss": 0.08908284455537796, "step": 4163 }, { "epoch": 1.0809968847352025, "grad_norm": 4.189727306365967, "learning_rate": 2.565734733777192e-06, "loss": 0.05830421298742294, "step": 4164 }, { "epoch": 1.0812564901349948, "grad_norm": 3.1482815742492676, "learning_rate": 2.564791918260378e-06, "loss": 0.058758076280355453, "step": 4165 }, { "epoch": 1.0815160955347871, "grad_norm": 2.614305257797241, "learning_rate": 2.563849135484113e-06, "loss": 0.054633334279060364, "step": 4166 }, { "epoch": 1.0817757009345794, "grad_norm": 3.429266929626465, "learning_rate": 2.562906385615059e-06, "loss": 0.06064419075846672, "step": 4167 }, { "epoch": 1.0820353063343717, "grad_norm": 3.1591081619262695, "learning_rate": 2.561963668819873e-06, "loss": 0.04199685528874397, "step": 4168 }, { "epoch": 1.082294911734164, "grad_norm": 3.8386056423187256, "learning_rate": 2.5610209852652034e-06, "loss": 0.02585495635867119, "step": 4169 }, { "epoch": 1.0825545171339563, "grad_norm": 2.196687698364258, "learning_rate": 2.5600783351176966e-06, "loss": 0.06999862194061279, "step": 4170 }, { "epoch": 1.0828141225337486, "grad_norm": 2.7574758529663086, "learning_rate": 2.559135718543992e-06, "loss": 0.055462971329689026, "step": 4171 }, { "epoch": 1.083073727933541, "grad_norm": 3.092477321624756, "learning_rate": 2.5581931357107226e-06, "loss": 0.037210430949926376, "step": 4172 }, { "epoch": 1.0833333333333333, "grad_norm": 1.654099702835083, "learning_rate": 2.5572505867845145e-06, "loss": 0.09362807124853134, "step": 4173 }, { "epoch": 1.0835929387331256, "grad_norm": 2.004789352416992, "learning_rate": 2.5563080719319892e-06, "loss": 0.0868455171585083, "step": 4174 }, { "epoch": 1.0838525441329179, "grad_norm": 2.203688859939575, "learning_rate": 2.5553655913197612e-06, "loss": 0.06610292196273804, "step": 4175 }, { "epoch": 1.0841121495327102, "grad_norm": 2.269199848175049, "learning_rate": 2.554423145114442e-06, "loss": 0.08050164580345154, "step": 4176 }, { "epoch": 1.0843717549325027, "grad_norm": 2.2758140563964844, "learning_rate": 2.553480733482631e-06, "loss": 0.07595323026180267, "step": 4177 }, { "epoch": 1.084631360332295, "grad_norm": 4.201223373413086, "learning_rate": 2.5525383565909274e-06, "loss": 0.030788371339440346, "step": 4178 }, { "epoch": 1.0848909657320873, "grad_norm": 2.8541066646575928, "learning_rate": 2.5515960146059204e-06, "loss": 0.028981853276491165, "step": 4179 }, { "epoch": 1.0851505711318796, "grad_norm": 3.8346288204193115, "learning_rate": 2.5506537076941955e-06, "loss": 0.03447093442082405, "step": 4180 }, { "epoch": 1.085410176531672, "grad_norm": 3.8453452587127686, "learning_rate": 2.5497114360223306e-06, "loss": 0.029667694121599197, "step": 4181 }, { "epoch": 1.0856697819314642, "grad_norm": 1.7260948419570923, "learning_rate": 2.548769199756898e-06, "loss": 0.08918649703264236, "step": 4182 }, { "epoch": 1.0859293873312565, "grad_norm": 1.920691967010498, "learning_rate": 2.547826999064463e-06, "loss": 0.0828423947095871, "step": 4183 }, { "epoch": 1.0861889927310489, "grad_norm": 3.2547028064727783, "learning_rate": 2.546884834111586e-06, "loss": 0.06588059663772583, "step": 4184 }, { "epoch": 1.0864485981308412, "grad_norm": 3.3557119369506836, "learning_rate": 2.545942705064819e-06, "loss": 0.055067941546440125, "step": 4185 }, { "epoch": 1.0867082035306335, "grad_norm": 2.6064469814300537, "learning_rate": 2.5450006120907095e-06, "loss": 0.07318414002656937, "step": 4186 }, { "epoch": 1.0869678089304258, "grad_norm": 3.007657051086426, "learning_rate": 2.544058555355798e-06, "loss": 0.030916623771190643, "step": 4187 }, { "epoch": 1.087227414330218, "grad_norm": 3.7842917442321777, "learning_rate": 2.543116535026619e-06, "loss": 0.03445238620042801, "step": 4188 }, { "epoch": 1.0874870197300104, "grad_norm": 1.6876580715179443, "learning_rate": 2.5421745512696988e-06, "loss": 0.08978071808815002, "step": 4189 }, { "epoch": 1.0877466251298027, "grad_norm": 2.389127731323242, "learning_rate": 2.54123260425156e-06, "loss": 0.08524592965841293, "step": 4190 }, { "epoch": 1.088006230529595, "grad_norm": 2.5778093338012695, "learning_rate": 2.540290694138717e-06, "loss": 0.0545169897377491, "step": 4191 }, { "epoch": 1.0882658359293873, "grad_norm": 2.261699914932251, "learning_rate": 2.5393488210976764e-06, "loss": 0.06913450360298157, "step": 4192 }, { "epoch": 1.0885254413291796, "grad_norm": 2.3642256259918213, "learning_rate": 2.5384069852949415e-06, "loss": 0.019315969198942184, "step": 4193 }, { "epoch": 1.088785046728972, "grad_norm": 3.556007146835327, "learning_rate": 2.537465186897006e-06, "loss": 0.05670822411775589, "step": 4194 }, { "epoch": 1.0890446521287642, "grad_norm": 3.094174861907959, "learning_rate": 2.5365234260703607e-06, "loss": 0.06270401179790497, "step": 4195 }, { "epoch": 1.0893042575285565, "grad_norm": 3.8436126708984375, "learning_rate": 2.5355817029814844e-06, "loss": 0.04092046618461609, "step": 4196 }, { "epoch": 1.0895638629283488, "grad_norm": 3.182455539703369, "learning_rate": 2.534640017796854e-06, "loss": 0.07331593334674835, "step": 4197 }, { "epoch": 1.0898234683281411, "grad_norm": 8.258829116821289, "learning_rate": 2.533698370682937e-06, "loss": 0.027112498879432678, "step": 4198 }, { "epoch": 1.0900830737279334, "grad_norm": 2.855506658554077, "learning_rate": 2.5327567618061965e-06, "loss": 0.07427683472633362, "step": 4199 }, { "epoch": 1.0903426791277258, "grad_norm": 1.6589627265930176, "learning_rate": 2.5318151913330857e-06, "loss": 0.10314308106899261, "step": 4200 }, { "epoch": 1.090602284527518, "grad_norm": 2.9990248680114746, "learning_rate": 2.5308736594300522e-06, "loss": 0.043254751712083817, "step": 4201 }, { "epoch": 1.0908618899273106, "grad_norm": 4.037078380584717, "learning_rate": 2.529932166263539e-06, "loss": 0.0539054200053215, "step": 4202 }, { "epoch": 1.091121495327103, "grad_norm": 2.524402141571045, "learning_rate": 2.5289907119999814e-06, "loss": 0.07060201466083527, "step": 4203 }, { "epoch": 1.0913811007268952, "grad_norm": 3.5414652824401855, "learning_rate": 2.528049296805804e-06, "loss": 0.052718229591846466, "step": 4204 }, { "epoch": 1.0916407061266875, "grad_norm": 3.674147844314575, "learning_rate": 2.5271079208474288e-06, "loss": 0.06275558471679688, "step": 4205 }, { "epoch": 1.0919003115264798, "grad_norm": 3.2556254863739014, "learning_rate": 2.5261665842912696e-06, "loss": 0.02425164170563221, "step": 4206 }, { "epoch": 1.0921599169262721, "grad_norm": 3.906675100326538, "learning_rate": 2.525225287303735e-06, "loss": 0.022858401760458946, "step": 4207 }, { "epoch": 1.0924195223260644, "grad_norm": 3.0658152103424072, "learning_rate": 2.5242840300512205e-06, "loss": 0.03229320049285889, "step": 4208 }, { "epoch": 1.0926791277258567, "grad_norm": 2.4377126693725586, "learning_rate": 2.5233428127001226e-06, "loss": 0.0463896244764328, "step": 4209 }, { "epoch": 1.092938733125649, "grad_norm": 2.8878588676452637, "learning_rate": 2.522401635416825e-06, "loss": 0.035569410771131516, "step": 4210 }, { "epoch": 1.0931983385254413, "grad_norm": 2.4809954166412354, "learning_rate": 2.5214604983677084e-06, "loss": 0.06603089720010757, "step": 4211 }, { "epoch": 1.0934579439252337, "grad_norm": 3.474226951599121, "learning_rate": 2.5205194017191414e-06, "loss": 0.06928106397390366, "step": 4212 }, { "epoch": 1.093717549325026, "grad_norm": 3.9154868125915527, "learning_rate": 2.5195783456374902e-06, "loss": 0.037373293191194534, "step": 4213 }, { "epoch": 1.0939771547248183, "grad_norm": 3.8423404693603516, "learning_rate": 2.518637330289111e-06, "loss": 0.046135492622852325, "step": 4214 }, { "epoch": 1.0942367601246106, "grad_norm": 3.524576425552368, "learning_rate": 2.517696355840355e-06, "loss": 0.04242807999253273, "step": 4215 }, { "epoch": 1.0944963655244029, "grad_norm": 1.7290689945220947, "learning_rate": 2.5167554224575643e-06, "loss": 0.08709125220775604, "step": 4216 }, { "epoch": 1.0947559709241952, "grad_norm": 2.5796473026275635, "learning_rate": 2.5158145303070737e-06, "loss": 0.06673219054937363, "step": 4217 }, { "epoch": 1.0950155763239875, "grad_norm": 5.4406232833862305, "learning_rate": 2.5148736795552125e-06, "loss": 0.027400020509958267, "step": 4218 }, { "epoch": 1.0952751817237798, "grad_norm": 3.969048023223877, "learning_rate": 2.5139328703683007e-06, "loss": 0.07083737850189209, "step": 4219 }, { "epoch": 1.095534787123572, "grad_norm": 2.4903461933135986, "learning_rate": 2.5129921029126524e-06, "loss": 0.08036099374294281, "step": 4220 }, { "epoch": 1.0957943925233644, "grad_norm": 4.1460652351379395, "learning_rate": 2.5120513773545737e-06, "loss": 0.03023618459701538, "step": 4221 }, { "epoch": 1.0960539979231567, "grad_norm": 2.593364953994751, "learning_rate": 2.511110693860363e-06, "loss": 0.0430821031332016, "step": 4222 }, { "epoch": 1.096313603322949, "grad_norm": 3.030496120452881, "learning_rate": 2.510170052596312e-06, "loss": 0.02533452957868576, "step": 4223 }, { "epoch": 1.0965732087227413, "grad_norm": 2.550734758377075, "learning_rate": 2.5092294537287038e-06, "loss": 0.07607066631317139, "step": 4224 }, { "epoch": 1.0968328141225339, "grad_norm": 2.0401759147644043, "learning_rate": 2.5082888974238155e-06, "loss": 0.07621943950653076, "step": 4225 }, { "epoch": 1.0970924195223262, "grad_norm": 2.794407606124878, "learning_rate": 2.5073483838479164e-06, "loss": 0.035684533417224884, "step": 4226 }, { "epoch": 1.0973520249221185, "grad_norm": 3.0834896564483643, "learning_rate": 2.5064079131672658e-06, "loss": 0.02229895256459713, "step": 4227 }, { "epoch": 1.0976116303219108, "grad_norm": 2.536264657974243, "learning_rate": 2.5054674855481198e-06, "loss": 0.015090042725205421, "step": 4228 }, { "epoch": 1.097871235721703, "grad_norm": 3.3633837699890137, "learning_rate": 2.5045271011567225e-06, "loss": 0.03353634476661682, "step": 4229 }, { "epoch": 1.0981308411214954, "grad_norm": 3.4734437465667725, "learning_rate": 2.5035867601593145e-06, "loss": 0.0674486830830574, "step": 4230 }, { "epoch": 1.0983904465212877, "grad_norm": 3.961317777633667, "learning_rate": 2.5026464627221233e-06, "loss": 0.036779507994651794, "step": 4231 }, { "epoch": 1.09865005192108, "grad_norm": 5.2935004234313965, "learning_rate": 2.501706209011374e-06, "loss": 0.05176430940628052, "step": 4232 }, { "epoch": 1.0989096573208723, "grad_norm": 3.7870664596557617, "learning_rate": 2.5007659991932826e-06, "loss": 0.029548197984695435, "step": 4233 }, { "epoch": 1.0991692627206646, "grad_norm": 3.4388296604156494, "learning_rate": 2.4998258334340563e-06, "loss": 0.034707196056842804, "step": 4234 }, { "epoch": 1.099428868120457, "grad_norm": 2.2769968509674072, "learning_rate": 2.498885711899894e-06, "loss": 0.0358012355864048, "step": 4235 }, { "epoch": 1.0996884735202492, "grad_norm": 2.4919281005859375, "learning_rate": 2.497945634756987e-06, "loss": 0.07248447835445404, "step": 4236 }, { "epoch": 1.0999480789200415, "grad_norm": 1.8114632368087769, "learning_rate": 2.497005602171522e-06, "loss": 0.09291020035743713, "step": 4237 }, { "epoch": 1.1002076843198338, "grad_norm": 4.37881326675415, "learning_rate": 2.4960656143096732e-06, "loss": 0.036322902888059616, "step": 4238 }, { "epoch": 1.1004672897196262, "grad_norm": 2.8531386852264404, "learning_rate": 2.4951256713376092e-06, "loss": 0.029626918956637383, "step": 4239 }, { "epoch": 1.1007268951194185, "grad_norm": 1.8465299606323242, "learning_rate": 2.494185773421491e-06, "loss": 0.10635048896074295, "step": 4240 }, { "epoch": 1.1009865005192108, "grad_norm": 3.034996271133423, "learning_rate": 2.4932459207274694e-06, "loss": 0.019033968448638916, "step": 4241 }, { "epoch": 1.101246105919003, "grad_norm": 2.276562452316284, "learning_rate": 2.4923061134216927e-06, "loss": 0.06905971467494965, "step": 4242 }, { "epoch": 1.1015057113187954, "grad_norm": 3.7382097244262695, "learning_rate": 2.4913663516702927e-06, "loss": 0.03365258127450943, "step": 4243 }, { "epoch": 1.1017653167185877, "grad_norm": 3.116299629211426, "learning_rate": 2.4904266356394003e-06, "loss": 0.03284734860062599, "step": 4244 }, { "epoch": 1.10202492211838, "grad_norm": 2.3390843868255615, "learning_rate": 2.4894869654951353e-06, "loss": 0.05328363925218582, "step": 4245 }, { "epoch": 1.1022845275181723, "grad_norm": 3.137934923171997, "learning_rate": 2.48854734140361e-06, "loss": 0.029984721913933754, "step": 4246 }, { "epoch": 1.1025441329179646, "grad_norm": 2.4917495250701904, "learning_rate": 2.4876077635309283e-06, "loss": 0.04865095391869545, "step": 4247 }, { "epoch": 1.102803738317757, "grad_norm": 3.4995229244232178, "learning_rate": 2.4866682320431858e-06, "loss": 0.05035588890314102, "step": 4248 }, { "epoch": 1.1030633437175492, "grad_norm": 2.0117578506469727, "learning_rate": 2.4857287471064707e-06, "loss": 0.09270529448986053, "step": 4249 }, { "epoch": 1.1033229491173417, "grad_norm": 2.573829174041748, "learning_rate": 2.484789308886862e-06, "loss": 0.012749850749969482, "step": 4250 }, { "epoch": 1.103582554517134, "grad_norm": 8.334599494934082, "learning_rate": 2.4838499175504303e-06, "loss": 0.04122734069824219, "step": 4251 }, { "epoch": 1.1038421599169264, "grad_norm": 3.1508920192718506, "learning_rate": 2.4829105732632397e-06, "loss": 0.06805526465177536, "step": 4252 }, { "epoch": 1.1041017653167187, "grad_norm": 3.1751208305358887, "learning_rate": 2.4819712761913444e-06, "loss": 0.048867367208004, "step": 4253 }, { "epoch": 1.104361370716511, "grad_norm": 3.55332612991333, "learning_rate": 2.48103202650079e-06, "loss": 0.041252221912145615, "step": 4254 }, { "epoch": 1.1046209761163033, "grad_norm": 3.152250051498413, "learning_rate": 2.480092824357615e-06, "loss": 0.03421918302774429, "step": 4255 }, { "epoch": 1.1048805815160956, "grad_norm": 3.547065019607544, "learning_rate": 2.4791536699278487e-06, "loss": 0.031451933085918427, "step": 4256 }, { "epoch": 1.105140186915888, "grad_norm": 3.248905658721924, "learning_rate": 2.478214563377512e-06, "loss": 0.04038873314857483, "step": 4257 }, { "epoch": 1.1053997923156802, "grad_norm": 1.870849370956421, "learning_rate": 2.4772755048726172e-06, "loss": 0.0923592746257782, "step": 4258 }, { "epoch": 1.1056593977154725, "grad_norm": 2.7804853916168213, "learning_rate": 2.4763364945791685e-06, "loss": 0.06537124514579773, "step": 4259 }, { "epoch": 1.1059190031152648, "grad_norm": 2.6495325565338135, "learning_rate": 2.475397532663161e-06, "loss": 0.04711098223924637, "step": 4260 }, { "epoch": 1.1061786085150571, "grad_norm": 3.0786914825439453, "learning_rate": 2.474458619290584e-06, "loss": 0.04321243241429329, "step": 4261 }, { "epoch": 1.1064382139148494, "grad_norm": 1.9949735403060913, "learning_rate": 2.4735197546274125e-06, "loss": 0.014474743977189064, "step": 4262 }, { "epoch": 1.1066978193146417, "grad_norm": 2.3775556087493896, "learning_rate": 2.4725809388396175e-06, "loss": 0.010234012268483639, "step": 4263 }, { "epoch": 1.106957424714434, "grad_norm": 3.7792646884918213, "learning_rate": 2.4716421720931616e-06, "loss": 0.03050573542714119, "step": 4264 }, { "epoch": 1.1072170301142263, "grad_norm": 1.7015506029129028, "learning_rate": 2.4707034545539963e-06, "loss": 0.08536569029092789, "step": 4265 }, { "epoch": 1.1074766355140186, "grad_norm": 2.1432785987854004, "learning_rate": 2.4697647863880648e-06, "loss": 0.07180243730545044, "step": 4266 }, { "epoch": 1.107736240913811, "grad_norm": 1.9383387565612793, "learning_rate": 2.468826167761301e-06, "loss": 0.09516294300556183, "step": 4267 }, { "epoch": 1.1079958463136033, "grad_norm": 1.7483532428741455, "learning_rate": 2.4678875988396344e-06, "loss": 0.08351202309131622, "step": 4268 }, { "epoch": 1.1082554517133956, "grad_norm": 1.7108334302902222, "learning_rate": 2.466949079788981e-06, "loss": 0.08959706872701645, "step": 4269 }, { "epoch": 1.1085150571131879, "grad_norm": 2.834111452102661, "learning_rate": 2.4660106107752485e-06, "loss": 0.04816553369164467, "step": 4270 }, { "epoch": 1.1087746625129802, "grad_norm": 6.759758472442627, "learning_rate": 2.4650721919643376e-06, "loss": 0.058697640895843506, "step": 4271 }, { "epoch": 1.1090342679127725, "grad_norm": 2.4563441276550293, "learning_rate": 2.46413382352214e-06, "loss": 0.027214497327804565, "step": 4272 }, { "epoch": 1.109293873312565, "grad_norm": 3.3016979694366455, "learning_rate": 2.463195505614536e-06, "loss": 0.049344684928655624, "step": 4273 }, { "epoch": 1.1095534787123573, "grad_norm": 3.120934009552002, "learning_rate": 2.4622572384074e-06, "loss": 0.03646431863307953, "step": 4274 }, { "epoch": 1.1098130841121496, "grad_norm": 1.7136847972869873, "learning_rate": 2.4613190220665957e-06, "loss": 0.09607680886983871, "step": 4275 }, { "epoch": 1.110072689511942, "grad_norm": 1.379296064376831, "learning_rate": 2.4603808567579787e-06, "loss": 0.08902357518672943, "step": 4276 }, { "epoch": 1.1103322949117342, "grad_norm": 2.1836140155792236, "learning_rate": 2.4594427426473944e-06, "loss": 0.05073413997888565, "step": 4277 }, { "epoch": 1.1105919003115265, "grad_norm": 2.7285361289978027, "learning_rate": 2.45850467990068e-06, "loss": 0.06717626750469208, "step": 4278 }, { "epoch": 1.1108515057113189, "grad_norm": 2.098693370819092, "learning_rate": 2.457566668683664e-06, "loss": 0.09698792546987534, "step": 4279 }, { "epoch": 1.1111111111111112, "grad_norm": 4.729648113250732, "learning_rate": 2.456628709162166e-06, "loss": 0.04080312326550484, "step": 4280 }, { "epoch": 1.1113707165109035, "grad_norm": 2.055534601211548, "learning_rate": 2.455690801501994e-06, "loss": 0.09722219407558441, "step": 4281 }, { "epoch": 1.1116303219106958, "grad_norm": 2.687061071395874, "learning_rate": 2.4547529458689496e-06, "loss": 0.03284420073032379, "step": 4282 }, { "epoch": 1.111889927310488, "grad_norm": 2.9485366344451904, "learning_rate": 2.4538151424288246e-06, "loss": 0.05214754864573479, "step": 4283 }, { "epoch": 1.1121495327102804, "grad_norm": 3.7567355632781982, "learning_rate": 2.452877391347401e-06, "loss": 0.02970290742814541, "step": 4284 }, { "epoch": 1.1124091381100727, "grad_norm": 2.3208022117614746, "learning_rate": 2.4519396927904505e-06, "loss": 0.07460277527570724, "step": 4285 }, { "epoch": 1.112668743509865, "grad_norm": 3.029618978500366, "learning_rate": 2.4510020469237376e-06, "loss": 0.04212760925292969, "step": 4286 }, { "epoch": 1.1129283489096573, "grad_norm": 2.3300321102142334, "learning_rate": 2.450064453913017e-06, "loss": 0.08085070550441742, "step": 4287 }, { "epoch": 1.1131879543094496, "grad_norm": 2.3613295555114746, "learning_rate": 2.449126913924034e-06, "loss": 0.07311539351940155, "step": 4288 }, { "epoch": 1.113447559709242, "grad_norm": 3.78210711479187, "learning_rate": 2.4481894271225225e-06, "loss": 0.05835399404168129, "step": 4289 }, { "epoch": 1.1137071651090342, "grad_norm": 3.858863115310669, "learning_rate": 2.44725199367421e-06, "loss": 0.049335576593875885, "step": 4290 }, { "epoch": 1.1139667705088265, "grad_norm": 2.5862600803375244, "learning_rate": 2.4463146137448127e-06, "loss": 0.07633822411298752, "step": 4291 }, { "epoch": 1.1142263759086188, "grad_norm": 3.237941265106201, "learning_rate": 2.4453772875000398e-06, "loss": 0.06913082301616669, "step": 4292 }, { "epoch": 1.1144859813084111, "grad_norm": 2.6918559074401855, "learning_rate": 2.444440015105586e-06, "loss": 0.014900344423949718, "step": 4293 }, { "epoch": 1.1147455867082035, "grad_norm": 2.2676303386688232, "learning_rate": 2.4435027967271413e-06, "loss": 0.08540085703134537, "step": 4294 }, { "epoch": 1.1150051921079958, "grad_norm": 3.197388172149658, "learning_rate": 2.442565632530384e-06, "loss": 0.04086950793862343, "step": 4295 }, { "epoch": 1.115264797507788, "grad_norm": 2.5956623554229736, "learning_rate": 2.441628522680985e-06, "loss": 0.01568928174674511, "step": 4296 }, { "epoch": 1.1155244029075804, "grad_norm": 2.5083072185516357, "learning_rate": 2.4406914673446012e-06, "loss": 0.06605704128742218, "step": 4297 }, { "epoch": 1.1157840083073727, "grad_norm": 1.6355764865875244, "learning_rate": 2.4397544666868837e-06, "loss": 0.09132993221282959, "step": 4298 }, { "epoch": 1.1160436137071652, "grad_norm": 2.5103282928466797, "learning_rate": 2.438817520873474e-06, "loss": 0.028999052941799164, "step": 4299 }, { "epoch": 1.1163032191069575, "grad_norm": 3.5017404556274414, "learning_rate": 2.4378806300700003e-06, "loss": 0.0573747456073761, "step": 4300 }, { "epoch": 1.1165628245067498, "grad_norm": 2.6176042556762695, "learning_rate": 2.436943794442085e-06, "loss": 0.060067348182201385, "step": 4301 }, { "epoch": 1.1168224299065421, "grad_norm": 2.681525230407715, "learning_rate": 2.436007014155339e-06, "loss": 0.07253782451152802, "step": 4302 }, { "epoch": 1.1170820353063344, "grad_norm": 4.523741245269775, "learning_rate": 2.435070289375363e-06, "loss": 0.038430165499448776, "step": 4303 }, { "epoch": 1.1173416407061267, "grad_norm": 2.5928289890289307, "learning_rate": 2.434133620267749e-06, "loss": 0.07457245141267776, "step": 4304 }, { "epoch": 1.117601246105919, "grad_norm": 2.6150476932525635, "learning_rate": 2.433197006998078e-06, "loss": 0.09089282155036926, "step": 4305 }, { "epoch": 1.1178608515057114, "grad_norm": 3.3907275199890137, "learning_rate": 2.432260449731923e-06, "loss": 0.03186897188425064, "step": 4306 }, { "epoch": 1.1181204569055037, "grad_norm": 2.872286558151245, "learning_rate": 2.431323948634846e-06, "loss": 0.05578925833106041, "step": 4307 }, { "epoch": 1.118380062305296, "grad_norm": 3.249728202819824, "learning_rate": 2.4303875038723975e-06, "loss": 0.014649536460638046, "step": 4308 }, { "epoch": 1.1186396677050883, "grad_norm": 2.3074207305908203, "learning_rate": 2.429451115610119e-06, "loss": 0.029574943706393242, "step": 4309 }, { "epoch": 1.1188992731048806, "grad_norm": 2.2195425033569336, "learning_rate": 2.428514784013545e-06, "loss": 0.0864180326461792, "step": 4310 }, { "epoch": 1.1191588785046729, "grad_norm": 3.991398572921753, "learning_rate": 2.4275785092481965e-06, "loss": 0.027047431096434593, "step": 4311 }, { "epoch": 1.1194184839044652, "grad_norm": 2.739335775375366, "learning_rate": 2.426642291479584e-06, "loss": 0.050833798944950104, "step": 4312 }, { "epoch": 1.1196780893042575, "grad_norm": 4.268163204193115, "learning_rate": 2.4257061308732104e-06, "loss": 0.0431484691798687, "step": 4313 }, { "epoch": 1.1199376947040498, "grad_norm": 2.3660693168640137, "learning_rate": 2.424770027594568e-06, "loss": 0.04897264391183853, "step": 4314 }, { "epoch": 1.120197300103842, "grad_norm": 2.872368335723877, "learning_rate": 2.4238339818091383e-06, "loss": 0.0327291339635849, "step": 4315 }, { "epoch": 1.1204569055036344, "grad_norm": 2.582491636276245, "learning_rate": 2.422897993682392e-06, "loss": 0.08392935246229172, "step": 4316 }, { "epoch": 1.1207165109034267, "grad_norm": 3.5830070972442627, "learning_rate": 2.421962063379791e-06, "loss": 0.07824160158634186, "step": 4317 }, { "epoch": 1.120976116303219, "grad_norm": 3.750608205795288, "learning_rate": 2.421026191066785e-06, "loss": 0.04927720129489899, "step": 4318 }, { "epoch": 1.1212357217030113, "grad_norm": 3.2960400581359863, "learning_rate": 2.4200903769088174e-06, "loss": 0.04096376895904541, "step": 4319 }, { "epoch": 1.1214953271028036, "grad_norm": 3.076082706451416, "learning_rate": 2.419154621071317e-06, "loss": 0.05012643337249756, "step": 4320 }, { "epoch": 1.1217549325025962, "grad_norm": 2.8113796710968018, "learning_rate": 2.4182189237197033e-06, "loss": 0.04445149004459381, "step": 4321 }, { "epoch": 1.1220145379023885, "grad_norm": 2.10410213470459, "learning_rate": 2.4172832850193877e-06, "loss": 0.09205010533332825, "step": 4322 }, { "epoch": 1.1222741433021808, "grad_norm": 2.130810499191284, "learning_rate": 2.41634770513577e-06, "loss": 0.0644829198718071, "step": 4323 }, { "epoch": 1.122533748701973, "grad_norm": 1.576909065246582, "learning_rate": 2.4154121842342363e-06, "loss": 0.08819034695625305, "step": 4324 }, { "epoch": 1.1227933541017654, "grad_norm": 3.0635926723480225, "learning_rate": 2.414476722480169e-06, "loss": 0.03610574081540108, "step": 4325 }, { "epoch": 1.1230529595015577, "grad_norm": 3.7828640937805176, "learning_rate": 2.413541320038934e-06, "loss": 0.037322454154491425, "step": 4326 }, { "epoch": 1.12331256490135, "grad_norm": 2.7240641117095947, "learning_rate": 2.4126059770758907e-06, "loss": 0.055171214044094086, "step": 4327 }, { "epoch": 1.1235721703011423, "grad_norm": 1.518164873123169, "learning_rate": 2.4116706937563848e-06, "loss": 0.10337449610233307, "step": 4328 }, { "epoch": 1.1238317757009346, "grad_norm": 3.0971312522888184, "learning_rate": 2.4107354702457524e-06, "loss": 0.04918487370014191, "step": 4329 }, { "epoch": 1.124091381100727, "grad_norm": 3.71771240234375, "learning_rate": 2.409800306709323e-06, "loss": 0.0662480816245079, "step": 4330 }, { "epoch": 1.1243509865005192, "grad_norm": 2.534728765487671, "learning_rate": 2.4088652033124086e-06, "loss": 0.051925767213106155, "step": 4331 }, { "epoch": 1.1246105919003115, "grad_norm": 1.7949581146240234, "learning_rate": 2.407930160220315e-06, "loss": 0.10456417500972748, "step": 4332 }, { "epoch": 1.1248701973001038, "grad_norm": 2.8469221591949463, "learning_rate": 2.406995177598337e-06, "loss": 0.019070778042078018, "step": 4333 }, { "epoch": 1.1251298026998962, "grad_norm": 1.8021135330200195, "learning_rate": 2.4060602556117578e-06, "loss": 0.08660170435905457, "step": 4334 }, { "epoch": 1.1253894080996885, "grad_norm": 2.302227020263672, "learning_rate": 2.40512539442585e-06, "loss": 0.011319989338517189, "step": 4335 }, { "epoch": 1.1256490134994808, "grad_norm": 2.264055013656616, "learning_rate": 2.404190594205876e-06, "loss": 0.0770774856209755, "step": 4336 }, { "epoch": 1.125908618899273, "grad_norm": 2.190784454345703, "learning_rate": 2.4032558551170875e-06, "loss": 0.0955021008849144, "step": 4337 }, { "epoch": 1.1261682242990654, "grad_norm": 3.626385450363159, "learning_rate": 2.402321177324724e-06, "loss": 0.010523397475481033, "step": 4338 }, { "epoch": 1.1264278296988577, "grad_norm": 3.6259522438049316, "learning_rate": 2.4013865609940156e-06, "loss": 0.06314434111118317, "step": 4339 }, { "epoch": 1.12668743509865, "grad_norm": 2.9313910007476807, "learning_rate": 2.400452006290181e-06, "loss": 0.04722636193037033, "step": 4340 }, { "epoch": 1.1269470404984423, "grad_norm": 2.7129147052764893, "learning_rate": 2.399517513378428e-06, "loss": 0.030992714688181877, "step": 4341 }, { "epoch": 1.1272066458982346, "grad_norm": 2.6577770709991455, "learning_rate": 2.398583082423954e-06, "loss": 0.04949129372835159, "step": 4342 }, { "epoch": 1.127466251298027, "grad_norm": 2.7430877685546875, "learning_rate": 2.3976487135919444e-06, "loss": 0.060867808759212494, "step": 4343 }, { "epoch": 1.1277258566978192, "grad_norm": 2.9080636501312256, "learning_rate": 2.3967144070475746e-06, "loss": 0.03689063340425491, "step": 4344 }, { "epoch": 1.1279854620976115, "grad_norm": 3.3435254096984863, "learning_rate": 2.395780162956008e-06, "loss": 0.03889165073633194, "step": 4345 }, { "epoch": 1.1282450674974038, "grad_norm": 2.9955172538757324, "learning_rate": 2.3948459814823986e-06, "loss": 0.01778462715446949, "step": 4346 }, { "epoch": 1.1285046728971961, "grad_norm": 2.7976551055908203, "learning_rate": 2.3939118627918874e-06, "loss": 0.04475903883576393, "step": 4347 }, { "epoch": 1.1287642782969887, "grad_norm": 2.8353915214538574, "learning_rate": 2.3929778070496055e-06, "loss": 0.01993386074900627, "step": 4348 }, { "epoch": 1.129023883696781, "grad_norm": 3.070665121078491, "learning_rate": 2.3920438144206726e-06, "loss": 0.029060354456305504, "step": 4349 }, { "epoch": 1.1292834890965733, "grad_norm": 2.105898857116699, "learning_rate": 2.3911098850701975e-06, "loss": 0.0816187858581543, "step": 4350 }, { "epoch": 1.1295430944963656, "grad_norm": 3.335293769836426, "learning_rate": 2.3901760191632774e-06, "loss": 0.02294009178876877, "step": 4351 }, { "epoch": 1.129802699896158, "grad_norm": 3.0795466899871826, "learning_rate": 2.389242216864997e-06, "loss": 0.01617066189646721, "step": 4352 }, { "epoch": 1.1300623052959502, "grad_norm": 2.4619972705841064, "learning_rate": 2.3883084783404337e-06, "loss": 0.10092340409755707, "step": 4353 }, { "epoch": 1.1303219106957425, "grad_norm": 1.7102686166763306, "learning_rate": 2.38737480375465e-06, "loss": 0.10316724330186844, "step": 4354 }, { "epoch": 1.1305815160955348, "grad_norm": 4.127305507659912, "learning_rate": 2.3864411932726967e-06, "loss": 0.047792982310056686, "step": 4355 }, { "epoch": 1.1308411214953271, "grad_norm": 3.743783473968506, "learning_rate": 2.3855076470596165e-06, "loss": 0.018102038651704788, "step": 4356 }, { "epoch": 1.1311007268951194, "grad_norm": 4.115577220916748, "learning_rate": 2.3845741652804393e-06, "loss": 0.02843625098466873, "step": 4357 }, { "epoch": 1.1313603322949117, "grad_norm": 2.2706644535064697, "learning_rate": 2.3836407481001823e-06, "loss": 0.08072276413440704, "step": 4358 }, { "epoch": 1.131619937694704, "grad_norm": 4.90341329574585, "learning_rate": 2.3827073956838515e-06, "loss": 0.020424554124474525, "step": 4359 }, { "epoch": 1.1318795430944963, "grad_norm": 1.9931150674819946, "learning_rate": 2.3817741081964433e-06, "loss": 0.08275728672742844, "step": 4360 }, { "epoch": 1.1321391484942886, "grad_norm": 2.5116240978240967, "learning_rate": 2.3808408858029437e-06, "loss": 0.0645834431052208, "step": 4361 }, { "epoch": 1.132398753894081, "grad_norm": 2.897813320159912, "learning_rate": 2.379907728668321e-06, "loss": 0.013225406408309937, "step": 4362 }, { "epoch": 1.1326583592938733, "grad_norm": 3.1194252967834473, "learning_rate": 2.378974636957538e-06, "loss": 0.027853313833475113, "step": 4363 }, { "epoch": 1.1329179646936656, "grad_norm": 2.3218307495117188, "learning_rate": 2.3780416108355433e-06, "loss": 0.07050076872110367, "step": 4364 }, { "epoch": 1.1331775700934579, "grad_norm": 2.585477352142334, "learning_rate": 2.3771086504672762e-06, "loss": 0.0513433963060379, "step": 4365 }, { "epoch": 1.1334371754932502, "grad_norm": 3.951512098312378, "learning_rate": 2.3761757560176614e-06, "loss": 0.030840575695037842, "step": 4366 }, { "epoch": 1.1336967808930425, "grad_norm": 2.7506978511810303, "learning_rate": 2.3752429276516132e-06, "loss": 0.061605993658304214, "step": 4367 }, { "epoch": 1.1339563862928348, "grad_norm": 2.004042148590088, "learning_rate": 2.3743101655340346e-06, "loss": 0.08792445063591003, "step": 4368 }, { "epoch": 1.1342159916926273, "grad_norm": 1.8011810779571533, "learning_rate": 2.3733774698298173e-06, "loss": 0.09880810976028442, "step": 4369 }, { "epoch": 1.1344755970924196, "grad_norm": 4.2979936599731445, "learning_rate": 2.3724448407038386e-06, "loss": 0.028526373207569122, "step": 4370 }, { "epoch": 1.134735202492212, "grad_norm": 2.7790746688842773, "learning_rate": 2.371512278320968e-06, "loss": 0.029069801792502403, "step": 4371 }, { "epoch": 1.1349948078920042, "grad_norm": 6.006535053253174, "learning_rate": 2.3705797828460596e-06, "loss": 0.04343223571777344, "step": 4372 }, { "epoch": 1.1352544132917965, "grad_norm": 3.804785966873169, "learning_rate": 2.369647354443959e-06, "loss": 0.056201107800006866, "step": 4373 }, { "epoch": 1.1355140186915889, "grad_norm": 3.1598517894744873, "learning_rate": 2.3687149932794967e-06, "loss": 0.04461398720741272, "step": 4374 }, { "epoch": 1.1357736240913812, "grad_norm": 2.6509454250335693, "learning_rate": 2.3677826995174935e-06, "loss": 0.0772663950920105, "step": 4375 }, { "epoch": 1.1360332294911735, "grad_norm": 3.241834878921509, "learning_rate": 2.366850473322757e-06, "loss": 0.019077204167842865, "step": 4376 }, { "epoch": 1.1362928348909658, "grad_norm": 2.5240354537963867, "learning_rate": 2.3659183148600847e-06, "loss": 0.06237257272005081, "step": 4377 }, { "epoch": 1.136552440290758, "grad_norm": 3.453449010848999, "learning_rate": 2.3649862242942595e-06, "loss": 0.09978923201560974, "step": 4378 }, { "epoch": 1.1368120456905504, "grad_norm": 1.987369179725647, "learning_rate": 2.364054201790054e-06, "loss": 0.08071377873420715, "step": 4379 }, { "epoch": 1.1370716510903427, "grad_norm": 2.142186164855957, "learning_rate": 2.363122247512229e-06, "loss": 0.09436348080635071, "step": 4380 }, { "epoch": 1.137331256490135, "grad_norm": 3.7092506885528564, "learning_rate": 2.3621903616255322e-06, "loss": 0.028189804404973984, "step": 4381 }, { "epoch": 1.1375908618899273, "grad_norm": 3.6490490436553955, "learning_rate": 2.3612585442946994e-06, "loss": 0.016968440264463425, "step": 4382 }, { "epoch": 1.1378504672897196, "grad_norm": 3.098134756088257, "learning_rate": 2.3603267956844555e-06, "loss": 0.06910091638565063, "step": 4383 }, { "epoch": 1.138110072689512, "grad_norm": 2.5220019817352295, "learning_rate": 2.3593951159595125e-06, "loss": 0.07395213842391968, "step": 4384 }, { "epoch": 1.1383696780893042, "grad_norm": 2.6738147735595703, "learning_rate": 2.358463505284568e-06, "loss": 0.07706598192453384, "step": 4385 }, { "epoch": 1.1386292834890965, "grad_norm": 3.084510326385498, "learning_rate": 2.3575319638243103e-06, "loss": 0.06706855446100235, "step": 4386 }, { "epoch": 1.1388888888888888, "grad_norm": 3.5895278453826904, "learning_rate": 2.3566004917434165e-06, "loss": 0.02539140358567238, "step": 4387 }, { "epoch": 1.1391484942886811, "grad_norm": 5.952500820159912, "learning_rate": 2.3556690892065487e-06, "loss": 0.018921226263046265, "step": 4388 }, { "epoch": 1.1394080996884735, "grad_norm": 1.8990405797958374, "learning_rate": 2.354737756378355e-06, "loss": 0.09130971133708954, "step": 4389 }, { "epoch": 1.1396677050882658, "grad_norm": 1.3813124895095825, "learning_rate": 2.353806493423476e-06, "loss": 0.09073250740766525, "step": 4390 }, { "epoch": 1.139927310488058, "grad_norm": 3.403393030166626, "learning_rate": 2.352875300506538e-06, "loss": 0.04251682013273239, "step": 4391 }, { "epoch": 1.1401869158878504, "grad_norm": 2.694044828414917, "learning_rate": 2.351944177792155e-06, "loss": 0.07274764776229858, "step": 4392 }, { "epoch": 1.1404465212876427, "grad_norm": 1.6308026313781738, "learning_rate": 2.3510131254449253e-06, "loss": 0.09421554207801819, "step": 4393 }, { "epoch": 1.140706126687435, "grad_norm": 1.6771286725997925, "learning_rate": 2.35008214362944e-06, "loss": 0.08275527507066727, "step": 4394 }, { "epoch": 1.1409657320872273, "grad_norm": 2.719599962234497, "learning_rate": 2.349151232510274e-06, "loss": 0.019776925444602966, "step": 4395 }, { "epoch": 1.1412253374870198, "grad_norm": 3.0154290199279785, "learning_rate": 2.3482203922519937e-06, "loss": 0.06727102398872375, "step": 4396 }, { "epoch": 1.1414849428868121, "grad_norm": 1.9727606773376465, "learning_rate": 2.3472896230191473e-06, "loss": 0.06707480549812317, "step": 4397 }, { "epoch": 1.1417445482866044, "grad_norm": 2.451082468032837, "learning_rate": 2.3463589249762747e-06, "loss": 0.019919127225875854, "step": 4398 }, { "epoch": 1.1420041536863967, "grad_norm": 1.523006796836853, "learning_rate": 2.345428298287902e-06, "loss": 0.09308203309774399, "step": 4399 }, { "epoch": 1.142263759086189, "grad_norm": 2.201063632965088, "learning_rate": 2.3444977431185432e-06, "loss": 0.05590534210205078, "step": 4400 }, { "epoch": 1.1425233644859814, "grad_norm": 2.792686700820923, "learning_rate": 2.3435672596326987e-06, "loss": 0.0575837716460228, "step": 4401 }, { "epoch": 1.1427829698857737, "grad_norm": 2.700396776199341, "learning_rate": 2.3426368479948556e-06, "loss": 0.05899468809366226, "step": 4402 }, { "epoch": 1.143042575285566, "grad_norm": 2.933868646621704, "learning_rate": 2.3417065083694913e-06, "loss": 0.010825038887560368, "step": 4403 }, { "epoch": 1.1433021806853583, "grad_norm": 3.0356533527374268, "learning_rate": 2.3407762409210677e-06, "loss": 0.04440826177597046, "step": 4404 }, { "epoch": 1.1435617860851506, "grad_norm": 3.4220073223114014, "learning_rate": 2.339846045814034e-06, "loss": 0.02921368181705475, "step": 4405 }, { "epoch": 1.1438213914849429, "grad_norm": 3.3318800926208496, "learning_rate": 2.3389159232128285e-06, "loss": 0.05442798137664795, "step": 4406 }, { "epoch": 1.1440809968847352, "grad_norm": 3.7411575317382812, "learning_rate": 2.337985873281875e-06, "loss": 0.04887353628873825, "step": 4407 }, { "epoch": 1.1443406022845275, "grad_norm": 1.9353922605514526, "learning_rate": 2.337055896185586e-06, "loss": 0.08594200015068054, "step": 4408 }, { "epoch": 1.1446002076843198, "grad_norm": 3.756042242050171, "learning_rate": 2.336125992088359e-06, "loss": 0.02567238360643387, "step": 4409 }, { "epoch": 1.144859813084112, "grad_norm": 2.198913335800171, "learning_rate": 2.3351961611545798e-06, "loss": 0.06296449899673462, "step": 4410 }, { "epoch": 1.1451194184839044, "grad_norm": 2.619852304458618, "learning_rate": 2.3342664035486216e-06, "loss": 0.06017139554023743, "step": 4411 }, { "epoch": 1.1453790238836967, "grad_norm": 2.0230367183685303, "learning_rate": 2.3333367194348443e-06, "loss": 0.08422243595123291, "step": 4412 }, { "epoch": 1.145638629283489, "grad_norm": 1.9364391565322876, "learning_rate": 2.3324071089775953e-06, "loss": 0.0961238294839859, "step": 4413 }, { "epoch": 1.1458982346832813, "grad_norm": 2.7953133583068848, "learning_rate": 2.3314775723412065e-06, "loss": 0.07204920053482056, "step": 4414 }, { "epoch": 1.1461578400830736, "grad_norm": 3.5600903034210205, "learning_rate": 2.330548109690002e-06, "loss": 0.06188317388296127, "step": 4415 }, { "epoch": 1.146417445482866, "grad_norm": 2.540569543838501, "learning_rate": 2.3296187211882854e-06, "loss": 0.06069548428058624, "step": 4416 }, { "epoch": 1.1466770508826585, "grad_norm": 3.0923118591308594, "learning_rate": 2.328689407000354e-06, "loss": 0.05851517617702484, "step": 4417 }, { "epoch": 1.1469366562824508, "grad_norm": 3.744206428527832, "learning_rate": 2.327760167290489e-06, "loss": 0.018435079604387283, "step": 4418 }, { "epoch": 1.147196261682243, "grad_norm": 4.539536476135254, "learning_rate": 2.3268310022229597e-06, "loss": 0.030325882136821747, "step": 4419 }, { "epoch": 1.1474558670820354, "grad_norm": 4.140817642211914, "learning_rate": 2.3259019119620184e-06, "loss": 0.05096834525465965, "step": 4420 }, { "epoch": 1.1477154724818277, "grad_norm": 2.0262346267700195, "learning_rate": 2.324972896671908e-06, "loss": 0.07651692628860474, "step": 4421 }, { "epoch": 1.14797507788162, "grad_norm": 3.837942123413086, "learning_rate": 2.324043956516859e-06, "loss": 0.0261167511343956, "step": 4422 }, { "epoch": 1.1482346832814123, "grad_norm": 2.936257839202881, "learning_rate": 2.323115091661085e-06, "loss": 0.06979265064001083, "step": 4423 }, { "epoch": 1.1484942886812046, "grad_norm": 1.9606668949127197, "learning_rate": 2.3221863022687877e-06, "loss": 0.0857066810131073, "step": 4424 }, { "epoch": 1.148753894080997, "grad_norm": 3.4357872009277344, "learning_rate": 2.321257588504158e-06, "loss": 0.04244130477309227, "step": 4425 }, { "epoch": 1.1490134994807892, "grad_norm": 2.5309417247772217, "learning_rate": 2.3203289505313677e-06, "loss": 0.02488076314330101, "step": 4426 }, { "epoch": 1.1492731048805815, "grad_norm": 3.748934507369995, "learning_rate": 2.3194003885145826e-06, "loss": 0.03914038836956024, "step": 4427 }, { "epoch": 1.1495327102803738, "grad_norm": 2.6024370193481445, "learning_rate": 2.3184719026179485e-06, "loss": 0.08283543586730957, "step": 4428 }, { "epoch": 1.1497923156801662, "grad_norm": 3.0695505142211914, "learning_rate": 2.317543493005602e-06, "loss": 0.050791412591934204, "step": 4429 }, { "epoch": 1.1500519210799585, "grad_norm": 3.2256217002868652, "learning_rate": 2.316615159841663e-06, "loss": 0.04305500537157059, "step": 4430 }, { "epoch": 1.1503115264797508, "grad_norm": 4.73587703704834, "learning_rate": 2.315686903290242e-06, "loss": 0.03833843395113945, "step": 4431 }, { "epoch": 1.150571131879543, "grad_norm": 2.675269365310669, "learning_rate": 2.3147587235154307e-06, "loss": 0.05409738048911095, "step": 4432 }, { "epoch": 1.1508307372793354, "grad_norm": 3.3404741287231445, "learning_rate": 2.3138306206813117e-06, "loss": 0.03107641637325287, "step": 4433 }, { "epoch": 1.1510903426791277, "grad_norm": 2.965667247772217, "learning_rate": 2.3129025949519517e-06, "loss": 0.02689177729189396, "step": 4434 }, { "epoch": 1.15134994807892, "grad_norm": 2.1459548473358154, "learning_rate": 2.3119746464914055e-06, "loss": 0.0720396339893341, "step": 4435 }, { "epoch": 1.1516095534787123, "grad_norm": 3.9420647621154785, "learning_rate": 2.3110467754637115e-06, "loss": 0.02538750320672989, "step": 4436 }, { "epoch": 1.1518691588785046, "grad_norm": 2.2790143489837646, "learning_rate": 2.3101189820328977e-06, "loss": 0.08861587941646576, "step": 4437 }, { "epoch": 1.152128764278297, "grad_norm": 3.0518062114715576, "learning_rate": 2.3091912663629758e-06, "loss": 0.016440406441688538, "step": 4438 }, { "epoch": 1.1523883696780892, "grad_norm": 3.7667930126190186, "learning_rate": 2.3082636286179446e-06, "loss": 0.030529819428920746, "step": 4439 }, { "epoch": 1.1526479750778815, "grad_norm": 2.894895076751709, "learning_rate": 2.3073360689617893e-06, "loss": 0.05083082988858223, "step": 4440 }, { "epoch": 1.1529075804776738, "grad_norm": 3.3557002544403076, "learning_rate": 2.3064085875584814e-06, "loss": 0.018399104475975037, "step": 4441 }, { "epoch": 1.1531671858774661, "grad_norm": 3.139683723449707, "learning_rate": 2.3054811845719786e-06, "loss": 0.013060636818408966, "step": 4442 }, { "epoch": 1.1534267912772584, "grad_norm": 4.06471586227417, "learning_rate": 2.304553860166224e-06, "loss": 0.04121175408363342, "step": 4443 }, { "epoch": 1.153686396677051, "grad_norm": 2.5914554595947266, "learning_rate": 2.303626614505147e-06, "loss": 0.012088527902960777, "step": 4444 }, { "epoch": 1.1539460020768433, "grad_norm": 1.9699496030807495, "learning_rate": 2.302699447752665e-06, "loss": 0.08539823442697525, "step": 4445 }, { "epoch": 1.1542056074766356, "grad_norm": 2.105401039123535, "learning_rate": 2.3017723600726795e-06, "loss": 0.07465869188308716, "step": 4446 }, { "epoch": 1.154465212876428, "grad_norm": 2.0861775875091553, "learning_rate": 2.3008453516290756e-06, "loss": 0.0944344699382782, "step": 4447 }, { "epoch": 1.1547248182762202, "grad_norm": 2.0949501991271973, "learning_rate": 2.29991842258573e-06, "loss": 0.07004192471504211, "step": 4448 }, { "epoch": 1.1549844236760125, "grad_norm": 2.3483614921569824, "learning_rate": 2.2989915731065023e-06, "loss": 0.0661735087633133, "step": 4449 }, { "epoch": 1.1552440290758048, "grad_norm": 2.2748892307281494, "learning_rate": 2.2980648033552384e-06, "loss": 0.07977946847677231, "step": 4450 }, { "epoch": 1.1555036344755971, "grad_norm": 3.978501081466675, "learning_rate": 2.297138113495768e-06, "loss": 0.024108827114105225, "step": 4451 }, { "epoch": 1.1557632398753894, "grad_norm": 3.5796873569488525, "learning_rate": 2.2962115036919095e-06, "loss": 0.03087722137570381, "step": 4452 }, { "epoch": 1.1560228452751817, "grad_norm": 4.029059886932373, "learning_rate": 2.295284974107468e-06, "loss": 0.04312227666378021, "step": 4453 }, { "epoch": 1.156282450674974, "grad_norm": 2.911226987838745, "learning_rate": 2.2943585249062316e-06, "loss": 0.03389168530702591, "step": 4454 }, { "epoch": 1.1565420560747663, "grad_norm": 1.4865012168884277, "learning_rate": 2.2934321562519746e-06, "loss": 0.08316183090209961, "step": 4455 }, { "epoch": 1.1568016614745587, "grad_norm": 3.5482048988342285, "learning_rate": 2.2925058683084582e-06, "loss": 0.06858739256858826, "step": 4456 }, { "epoch": 1.157061266874351, "grad_norm": 3.977023124694824, "learning_rate": 2.291579661239429e-06, "loss": 0.030756942927837372, "step": 4457 }, { "epoch": 1.1573208722741433, "grad_norm": 2.6064224243164062, "learning_rate": 2.2906535352086205e-06, "loss": 0.0455467514693737, "step": 4458 }, { "epoch": 1.1575804776739356, "grad_norm": 3.4514079093933105, "learning_rate": 2.2897274903797484e-06, "loss": 0.04660193622112274, "step": 4459 }, { "epoch": 1.1578400830737279, "grad_norm": 1.5151723623275757, "learning_rate": 2.2888015269165166e-06, "loss": 0.09161611646413803, "step": 4460 }, { "epoch": 1.1580996884735202, "grad_norm": 1.7618815898895264, "learning_rate": 2.2878756449826157e-06, "loss": 0.09335772693157196, "step": 4461 }, { "epoch": 1.1583592938733125, "grad_norm": 3.2426235675811768, "learning_rate": 2.2869498447417193e-06, "loss": 0.02328638732433319, "step": 4462 }, { "epoch": 1.1586188992731048, "grad_norm": 2.991039514541626, "learning_rate": 2.286024126357488e-06, "loss": 0.02468833699822426, "step": 4463 }, { "epoch": 1.158878504672897, "grad_norm": 4.623355865478516, "learning_rate": 2.285098489993567e-06, "loss": 0.04254288971424103, "step": 4464 }, { "epoch": 1.1591381100726896, "grad_norm": 3.388747453689575, "learning_rate": 2.2841729358135885e-06, "loss": 0.05225363001227379, "step": 4465 }, { "epoch": 1.159397715472482, "grad_norm": 1.9823204278945923, "learning_rate": 2.2832474639811686e-06, "loss": 0.0885801762342453, "step": 4466 }, { "epoch": 1.1596573208722742, "grad_norm": 2.522836685180664, "learning_rate": 2.28232207465991e-06, "loss": 0.05283397436141968, "step": 4467 }, { "epoch": 1.1599169262720666, "grad_norm": 2.905766487121582, "learning_rate": 2.281396768013401e-06, "loss": 0.012481451965868473, "step": 4468 }, { "epoch": 1.1601765316718589, "grad_norm": 1.8553816080093384, "learning_rate": 2.280471544205213e-06, "loss": 0.08383597433567047, "step": 4469 }, { "epoch": 1.1604361370716512, "grad_norm": 2.1880502700805664, "learning_rate": 2.279546403398905e-06, "loss": 0.05976169556379318, "step": 4470 }, { "epoch": 1.1606957424714435, "grad_norm": 2.477053642272949, "learning_rate": 2.2786213457580216e-06, "loss": 0.060785964131355286, "step": 4471 }, { "epoch": 1.1609553478712358, "grad_norm": 2.688016891479492, "learning_rate": 2.27769637144609e-06, "loss": 0.09140866249799728, "step": 4472 }, { "epoch": 1.161214953271028, "grad_norm": 3.1737008094787598, "learning_rate": 2.276771480626627e-06, "loss": 0.05358892306685448, "step": 4473 }, { "epoch": 1.1614745586708204, "grad_norm": 2.3906452655792236, "learning_rate": 2.27584667346313e-06, "loss": 0.07307767868041992, "step": 4474 }, { "epoch": 1.1617341640706127, "grad_norm": 3.2988972663879395, "learning_rate": 2.2749219501190846e-06, "loss": 0.04367180913686752, "step": 4475 }, { "epoch": 1.161993769470405, "grad_norm": 2.3066086769104004, "learning_rate": 2.2739973107579606e-06, "loss": 0.08045897632837296, "step": 4476 }, { "epoch": 1.1622533748701973, "grad_norm": 3.4575917720794678, "learning_rate": 2.2730727555432135e-06, "loss": 0.05460960790514946, "step": 4477 }, { "epoch": 1.1625129802699896, "grad_norm": 2.653064012527466, "learning_rate": 2.272148284638282e-06, "loss": 0.0908137708902359, "step": 4478 }, { "epoch": 1.162772585669782, "grad_norm": 2.29707670211792, "learning_rate": 2.271223898206593e-06, "loss": 0.061852723360061646, "step": 4479 }, { "epoch": 1.1630321910695742, "grad_norm": 1.7967246770858765, "learning_rate": 2.270299596411556e-06, "loss": 0.08557597547769547, "step": 4480 }, { "epoch": 1.1632917964693665, "grad_norm": 3.3202290534973145, "learning_rate": 2.269375379416568e-06, "loss": 0.04606924206018448, "step": 4481 }, { "epoch": 1.1635514018691588, "grad_norm": 1.9755884408950806, "learning_rate": 2.268451247385007e-06, "loss": 0.08443823456764221, "step": 4482 }, { "epoch": 1.1638110072689511, "grad_norm": 2.2015819549560547, "learning_rate": 2.2675272004802383e-06, "loss": 0.08026101440191269, "step": 4483 }, { "epoch": 1.1640706126687435, "grad_norm": 4.0056681632995605, "learning_rate": 2.2666032388656144e-06, "loss": 0.023683153092861176, "step": 4484 }, { "epoch": 1.1643302180685358, "grad_norm": 2.4637725353240967, "learning_rate": 2.265679362704471e-06, "loss": 0.023648854345083237, "step": 4485 }, { "epoch": 1.164589823468328, "grad_norm": 3.175926923751831, "learning_rate": 2.264755572160125e-06, "loss": 0.030553873628377914, "step": 4486 }, { "epoch": 1.1648494288681204, "grad_norm": 2.3237178325653076, "learning_rate": 2.2638318673958836e-06, "loss": 0.006837660912424326, "step": 4487 }, { "epoch": 1.1651090342679127, "grad_norm": 3.6748476028442383, "learning_rate": 2.262908248575036e-06, "loss": 0.03589365631341934, "step": 4488 }, { "epoch": 1.165368639667705, "grad_norm": 5.605793476104736, "learning_rate": 2.261984715860858e-06, "loss": 0.04057224094867706, "step": 4489 }, { "epoch": 1.1656282450674973, "grad_norm": 2.5520222187042236, "learning_rate": 2.261061269416608e-06, "loss": 0.09057323634624481, "step": 4490 }, { "epoch": 1.1658878504672896, "grad_norm": 5.260068416595459, "learning_rate": 2.2601379094055297e-06, "loss": 0.028400812298059464, "step": 4491 }, { "epoch": 1.1661474558670821, "grad_norm": 2.0932178497314453, "learning_rate": 2.2592146359908534e-06, "loss": 0.06694462150335312, "step": 4492 }, { "epoch": 1.1664070612668744, "grad_norm": 3.055926561355591, "learning_rate": 2.2582914493357912e-06, "loss": 0.07140643894672394, "step": 4493 }, { "epoch": 1.1666666666666667, "grad_norm": 2.9977355003356934, "learning_rate": 2.2573683496035424e-06, "loss": 0.040010347962379456, "step": 4494 }, { "epoch": 1.166926272066459, "grad_norm": 1.8435711860656738, "learning_rate": 2.2564453369572898e-06, "loss": 0.08749943226575851, "step": 4495 }, { "epoch": 1.1671858774662514, "grad_norm": 2.350431442260742, "learning_rate": 2.255522411560201e-06, "loss": 0.07531620562076569, "step": 4496 }, { "epoch": 1.1674454828660437, "grad_norm": 1.6829075813293457, "learning_rate": 2.254599573575427e-06, "loss": 0.011224732734262943, "step": 4497 }, { "epoch": 1.167705088265836, "grad_norm": 2.4455485343933105, "learning_rate": 2.2536768231661053e-06, "loss": 0.030881090089678764, "step": 4498 }, { "epoch": 1.1679646936656283, "grad_norm": 2.8958749771118164, "learning_rate": 2.252754160495357e-06, "loss": 0.023100245743989944, "step": 4499 }, { "epoch": 1.1682242990654206, "grad_norm": 3.1303110122680664, "learning_rate": 2.251831585726289e-06, "loss": 0.035749148577451706, "step": 4500 }, { "epoch": 1.1684839044652129, "grad_norm": 2.0102927684783936, "learning_rate": 2.250909099021989e-06, "loss": 0.06766438484191895, "step": 4501 }, { "epoch": 1.1687435098650052, "grad_norm": 2.8108651638031006, "learning_rate": 2.249986700545533e-06, "loss": 0.08353219926357269, "step": 4502 }, { "epoch": 1.1690031152647975, "grad_norm": 3.3227150440216064, "learning_rate": 2.2490643904599787e-06, "loss": 0.05597027391195297, "step": 4503 }, { "epoch": 1.1692627206645898, "grad_norm": 2.8608837127685547, "learning_rate": 2.248142168928371e-06, "loss": 0.06895322352647781, "step": 4504 }, { "epoch": 1.1695223260643821, "grad_norm": 3.556090831756592, "learning_rate": 2.247220036113737e-06, "loss": 0.05615787208080292, "step": 4505 }, { "epoch": 1.1697819314641744, "grad_norm": 3.166835069656372, "learning_rate": 2.2462979921790877e-06, "loss": 0.05799485743045807, "step": 4506 }, { "epoch": 1.1700415368639667, "grad_norm": 2.6739392280578613, "learning_rate": 2.2453760372874206e-06, "loss": 0.04582826793193817, "step": 4507 }, { "epoch": 1.170301142263759, "grad_norm": 3.3428947925567627, "learning_rate": 2.2444541716017167e-06, "loss": 0.02469577267765999, "step": 4508 }, { "epoch": 1.1705607476635513, "grad_norm": 1.7799333333969116, "learning_rate": 2.243532395284938e-06, "loss": 0.0804646760225296, "step": 4509 }, { "epoch": 1.1708203530633436, "grad_norm": 1.8507789373397827, "learning_rate": 2.242610708500036e-06, "loss": 0.01412976160645485, "step": 4510 }, { "epoch": 1.171079958463136, "grad_norm": 3.372169017791748, "learning_rate": 2.2416891114099433e-06, "loss": 0.050019703805446625, "step": 4511 }, { "epoch": 1.1713395638629283, "grad_norm": 2.6556591987609863, "learning_rate": 2.240767604177577e-06, "loss": 0.06321568042039871, "step": 4512 }, { "epoch": 1.1715991692627208, "grad_norm": 1.9303078651428223, "learning_rate": 2.2398461869658384e-06, "loss": 0.07445034384727478, "step": 4513 }, { "epoch": 1.171858774662513, "grad_norm": 2.9051434993743896, "learning_rate": 2.2389248599376125e-06, "loss": 0.03389062359929085, "step": 4514 }, { "epoch": 1.1721183800623054, "grad_norm": 3.5149548053741455, "learning_rate": 2.238003623255769e-06, "loss": 0.03197956085205078, "step": 4515 }, { "epoch": 1.1723779854620977, "grad_norm": 3.1639657020568848, "learning_rate": 2.2370824770831634e-06, "loss": 0.026663580909371376, "step": 4516 }, { "epoch": 1.17263759086189, "grad_norm": 4.032985687255859, "learning_rate": 2.23616142158263e-06, "loss": 0.04501304030418396, "step": 4517 }, { "epoch": 1.1728971962616823, "grad_norm": 2.251138210296631, "learning_rate": 2.2352404569169924e-06, "loss": 0.09420903027057648, "step": 4518 }, { "epoch": 1.1731568016614746, "grad_norm": 2.6604745388031006, "learning_rate": 2.234319583249056e-06, "loss": 0.05768046900629997, "step": 4519 }, { "epoch": 1.173416407061267, "grad_norm": 3.1183173656463623, "learning_rate": 2.233398800741609e-06, "loss": 0.026514101773500443, "step": 4520 }, { "epoch": 1.1736760124610592, "grad_norm": 3.4112420082092285, "learning_rate": 2.2324781095574255e-06, "loss": 0.029935281723737717, "step": 4521 }, { "epoch": 1.1739356178608515, "grad_norm": 3.359501361846924, "learning_rate": 2.2315575098592627e-06, "loss": 0.05386754125356674, "step": 4522 }, { "epoch": 1.1741952232606438, "grad_norm": 3.6638803482055664, "learning_rate": 2.2306370018098623e-06, "loss": 0.05880118906497955, "step": 4523 }, { "epoch": 1.1744548286604362, "grad_norm": 3.32902455329895, "learning_rate": 2.229716585571947e-06, "loss": 0.027574900537729263, "step": 4524 }, { "epoch": 1.1747144340602285, "grad_norm": 3.26275634765625, "learning_rate": 2.228796261308227e-06, "loss": 0.02283944934606552, "step": 4525 }, { "epoch": 1.1749740394600208, "grad_norm": 3.6061623096466064, "learning_rate": 2.2278760291813934e-06, "loss": 0.041200295090675354, "step": 4526 }, { "epoch": 1.175233644859813, "grad_norm": 3.216191530227661, "learning_rate": 2.2269558893541237e-06, "loss": 0.028913216665387154, "step": 4527 }, { "epoch": 1.1754932502596054, "grad_norm": 3.1479647159576416, "learning_rate": 2.226035841989076e-06, "loss": 0.0162649005651474, "step": 4528 }, { "epoch": 1.1757528556593977, "grad_norm": 3.2505381107330322, "learning_rate": 2.225115887248894e-06, "loss": 0.05607185512781143, "step": 4529 }, { "epoch": 1.17601246105919, "grad_norm": 2.748478412628174, "learning_rate": 2.2241960252962054e-06, "loss": 0.06912711262702942, "step": 4530 }, { "epoch": 1.1762720664589823, "grad_norm": 2.54534912109375, "learning_rate": 2.2232762562936204e-06, "loss": 0.035893552005290985, "step": 4531 }, { "epoch": 1.1765316718587746, "grad_norm": 1.671777367591858, "learning_rate": 2.222356580403733e-06, "loss": 0.08481024205684662, "step": 4532 }, { "epoch": 1.176791277258567, "grad_norm": 1.832305908203125, "learning_rate": 2.22143699778912e-06, "loss": 0.08789783716201782, "step": 4533 }, { "epoch": 1.1770508826583592, "grad_norm": 2.7828195095062256, "learning_rate": 2.2205175086123436e-06, "loss": 0.06132884323596954, "step": 4534 }, { "epoch": 1.1773104880581515, "grad_norm": 3.348508596420288, "learning_rate": 2.2195981130359482e-06, "loss": 0.0492081455886364, "step": 4535 }, { "epoch": 1.1775700934579438, "grad_norm": 4.422159671783447, "learning_rate": 2.218678811222462e-06, "loss": 0.026354821398854256, "step": 4536 }, { "epoch": 1.1778296988577361, "grad_norm": 3.1021530628204346, "learning_rate": 2.217759603334396e-06, "loss": 0.03398439660668373, "step": 4537 }, { "epoch": 1.1780893042575284, "grad_norm": 2.834196090698242, "learning_rate": 2.2168404895342456e-06, "loss": 0.053061336278915405, "step": 4538 }, { "epoch": 1.1783489096573208, "grad_norm": 2.902085304260254, "learning_rate": 2.2159214699844895e-06, "loss": 0.013065742328763008, "step": 4539 }, { "epoch": 1.1786085150571133, "grad_norm": 2.8928306102752686, "learning_rate": 2.2150025448475876e-06, "loss": 0.016827475279569626, "step": 4540 }, { "epoch": 1.1788681204569056, "grad_norm": 2.5945680141448975, "learning_rate": 2.214083714285987e-06, "loss": 0.06017640233039856, "step": 4541 }, { "epoch": 1.179127725856698, "grad_norm": 3.215191602706909, "learning_rate": 2.213164978462114e-06, "loss": 0.06501281261444092, "step": 4542 }, { "epoch": 1.1793873312564902, "grad_norm": 2.544243335723877, "learning_rate": 2.212246337538382e-06, "loss": 0.033964235335588455, "step": 4543 }, { "epoch": 1.1796469366562825, "grad_norm": 2.6944003105163574, "learning_rate": 2.211327791677183e-06, "loss": 0.05062190443277359, "step": 4544 }, { "epoch": 1.1799065420560748, "grad_norm": 3.904104471206665, "learning_rate": 2.2104093410408972e-06, "loss": 0.05494134873151779, "step": 4545 }, { "epoch": 1.1801661474558671, "grad_norm": 2.786912679672241, "learning_rate": 2.209490985791886e-06, "loss": 0.06257513165473938, "step": 4546 }, { "epoch": 1.1804257528556594, "grad_norm": 2.9075753688812256, "learning_rate": 2.2085727260924916e-06, "loss": 0.06565488874912262, "step": 4547 }, { "epoch": 1.1806853582554517, "grad_norm": 3.263012647628784, "learning_rate": 2.2076545621050416e-06, "loss": 0.04130887985229492, "step": 4548 }, { "epoch": 1.180944963655244, "grad_norm": 3.174534559249878, "learning_rate": 2.2067364939918473e-06, "loss": 0.0440085344016552, "step": 4549 }, { "epoch": 1.1812045690550363, "grad_norm": 2.2153160572052, "learning_rate": 2.205818521915203e-06, "loss": 0.09233090281486511, "step": 4550 }, { "epoch": 1.1814641744548287, "grad_norm": 2.3146300315856934, "learning_rate": 2.2049006460373824e-06, "loss": 0.09876270592212677, "step": 4551 }, { "epoch": 1.181723779854621, "grad_norm": 3.821823835372925, "learning_rate": 2.203982866520647e-06, "loss": 0.041579850018024445, "step": 4552 }, { "epoch": 1.1819833852544133, "grad_norm": 2.3476717472076416, "learning_rate": 2.2030651835272383e-06, "loss": 0.077156662940979, "step": 4553 }, { "epoch": 1.1822429906542056, "grad_norm": 3.2943077087402344, "learning_rate": 2.202147597219383e-06, "loss": 0.048991650342941284, "step": 4554 }, { "epoch": 1.1825025960539979, "grad_norm": 2.5220842361450195, "learning_rate": 2.2012301077592874e-06, "loss": 0.015778731554746628, "step": 4555 }, { "epoch": 1.1827622014537902, "grad_norm": 3.4539833068847656, "learning_rate": 2.2003127153091437e-06, "loss": 0.037894293665885925, "step": 4556 }, { "epoch": 1.1830218068535825, "grad_norm": 2.966153860092163, "learning_rate": 2.199395420031126e-06, "loss": 0.015590347349643707, "step": 4557 }, { "epoch": 1.1832814122533748, "grad_norm": 2.403823137283325, "learning_rate": 2.198478222087391e-06, "loss": 0.0727045014500618, "step": 4558 }, { "epoch": 1.183541017653167, "grad_norm": 4.786617279052734, "learning_rate": 2.1975611216400776e-06, "loss": 0.03229960799217224, "step": 4559 }, { "epoch": 1.1838006230529594, "grad_norm": 2.7562801837921143, "learning_rate": 2.196644118851309e-06, "loss": 0.008421776816248894, "step": 4560 }, { "epoch": 1.184060228452752, "grad_norm": 2.7632927894592285, "learning_rate": 2.19572721388319e-06, "loss": 0.09862959384918213, "step": 4561 }, { "epoch": 1.1843198338525442, "grad_norm": 2.6191954612731934, "learning_rate": 2.194810406897809e-06, "loss": 0.0283890962600708, "step": 4562 }, { "epoch": 1.1845794392523366, "grad_norm": 3.0053551197052, "learning_rate": 2.193893698057235e-06, "loss": 0.034515440464019775, "step": 4563 }, { "epoch": 1.1848390446521289, "grad_norm": 3.4683101177215576, "learning_rate": 2.1929770875235227e-06, "loss": 0.06126849353313446, "step": 4564 }, { "epoch": 1.1850986500519212, "grad_norm": 3.916301965713501, "learning_rate": 2.1920605754587073e-06, "loss": 0.02937385067343712, "step": 4565 }, { "epoch": 1.1853582554517135, "grad_norm": 2.854243755340576, "learning_rate": 2.1911441620248066e-06, "loss": 0.028992613777518272, "step": 4566 }, { "epoch": 1.1856178608515058, "grad_norm": 5.338257312774658, "learning_rate": 2.1902278473838217e-06, "loss": 0.03100827895104885, "step": 4567 }, { "epoch": 1.185877466251298, "grad_norm": 4.657471179962158, "learning_rate": 2.1893116316977365e-06, "loss": 0.018013454973697662, "step": 4568 }, { "epoch": 1.1861370716510904, "grad_norm": 1.902245044708252, "learning_rate": 2.188395515128517e-06, "loss": 0.08273535966873169, "step": 4569 }, { "epoch": 1.1863966770508827, "grad_norm": 2.5574140548706055, "learning_rate": 2.1874794978381114e-06, "loss": 0.01328904740512371, "step": 4570 }, { "epoch": 1.186656282450675, "grad_norm": 1.8715218305587769, "learning_rate": 2.1865635799884507e-06, "loss": 0.07603281736373901, "step": 4571 }, { "epoch": 1.1869158878504673, "grad_norm": 4.3610405921936035, "learning_rate": 2.185647761741447e-06, "loss": 0.024798795580863953, "step": 4572 }, { "epoch": 1.1871754932502596, "grad_norm": 2.9648680686950684, "learning_rate": 2.184732043258998e-06, "loss": 0.03562396764755249, "step": 4573 }, { "epoch": 1.187435098650052, "grad_norm": 4.385250568389893, "learning_rate": 2.183816424702982e-06, "loss": 0.03183583542704582, "step": 4574 }, { "epoch": 1.1876947040498442, "grad_norm": 2.7792677879333496, "learning_rate": 2.1829009062352558e-06, "loss": 0.024035388603806496, "step": 4575 }, { "epoch": 1.1879543094496365, "grad_norm": 2.343782901763916, "learning_rate": 2.181985488017666e-06, "loss": 0.018370695412158966, "step": 4576 }, { "epoch": 1.1882139148494288, "grad_norm": 4.003039360046387, "learning_rate": 2.1810701702120367e-06, "loss": 0.029869385063648224, "step": 4577 }, { "epoch": 1.1884735202492211, "grad_norm": 3.384565591812134, "learning_rate": 2.180154952980173e-06, "loss": 0.0377768911421299, "step": 4578 }, { "epoch": 1.1887331256490135, "grad_norm": 3.4657187461853027, "learning_rate": 2.179239836483866e-06, "loss": 0.05337199568748474, "step": 4579 }, { "epoch": 1.1889927310488058, "grad_norm": 3.8240909576416016, "learning_rate": 2.178324820884888e-06, "loss": 0.028864286839962006, "step": 4580 }, { "epoch": 1.189252336448598, "grad_norm": 3.9989445209503174, "learning_rate": 2.177409906344992e-06, "loss": 0.022635452449321747, "step": 4581 }, { "epoch": 1.1895119418483904, "grad_norm": 2.3035926818847656, "learning_rate": 2.1764950930259143e-06, "loss": 0.07243141531944275, "step": 4582 }, { "epoch": 1.1897715472481827, "grad_norm": 4.473597526550293, "learning_rate": 2.1755803810893716e-06, "loss": 0.03256186097860336, "step": 4583 }, { "epoch": 1.190031152647975, "grad_norm": 3.353803873062134, "learning_rate": 2.174665770697066e-06, "loss": 0.06094156205654144, "step": 4584 }, { "epoch": 1.1902907580477673, "grad_norm": 4.1448259353637695, "learning_rate": 2.1737512620106784e-06, "loss": 0.07281498610973358, "step": 4585 }, { "epoch": 1.1905503634475596, "grad_norm": 2.4856622219085693, "learning_rate": 2.1728368551918726e-06, "loss": 0.07051970064640045, "step": 4586 }, { "epoch": 1.190809968847352, "grad_norm": 4.324007511138916, "learning_rate": 2.1719225504022968e-06, "loss": 0.034985028207302094, "step": 4587 }, { "epoch": 1.1910695742471444, "grad_norm": 4.032293796539307, "learning_rate": 2.1710083478035763e-06, "loss": 0.05295789986848831, "step": 4588 }, { "epoch": 1.1913291796469367, "grad_norm": 3.048546314239502, "learning_rate": 2.1700942475573246e-06, "loss": 0.03803468868136406, "step": 4589 }, { "epoch": 1.191588785046729, "grad_norm": 3.0845863819122314, "learning_rate": 2.169180249825131e-06, "loss": 0.06209590286016464, "step": 4590 }, { "epoch": 1.1918483904465214, "grad_norm": 3.0228447914123535, "learning_rate": 2.1682663547685695e-06, "loss": 0.02910313382744789, "step": 4591 }, { "epoch": 1.1921079958463137, "grad_norm": 2.7584214210510254, "learning_rate": 2.167352562549197e-06, "loss": 0.05040159821510315, "step": 4592 }, { "epoch": 1.192367601246106, "grad_norm": 3.381852626800537, "learning_rate": 2.1664388733285514e-06, "loss": 0.040309760719537735, "step": 4593 }, { "epoch": 1.1926272066458983, "grad_norm": 3.7260820865631104, "learning_rate": 2.1655252872681505e-06, "loss": 0.035708293318748474, "step": 4594 }, { "epoch": 1.1928868120456906, "grad_norm": 2.061501979827881, "learning_rate": 2.164611804529496e-06, "loss": 0.09165188670158386, "step": 4595 }, { "epoch": 1.1931464174454829, "grad_norm": 4.338311195373535, "learning_rate": 2.1636984252740705e-06, "loss": 0.02293953113257885, "step": 4596 }, { "epoch": 1.1934060228452752, "grad_norm": 3.312566041946411, "learning_rate": 2.16278514966334e-06, "loss": 0.021700087934732437, "step": 4597 }, { "epoch": 1.1936656282450675, "grad_norm": 2.4341111183166504, "learning_rate": 2.161871977858749e-06, "loss": 0.07148830592632294, "step": 4598 }, { "epoch": 1.1939252336448598, "grad_norm": 2.9223973751068115, "learning_rate": 2.160958910021726e-06, "loss": 0.06676674634218216, "step": 4599 }, { "epoch": 1.1941848390446521, "grad_norm": 1.8612176179885864, "learning_rate": 2.16004594631368e-06, "loss": 0.08207927644252777, "step": 4600 }, { "epoch": 1.1944444444444444, "grad_norm": 3.5741612911224365, "learning_rate": 2.1591330868960035e-06, "loss": 0.04670602083206177, "step": 4601 }, { "epoch": 1.1947040498442367, "grad_norm": 2.9752609729766846, "learning_rate": 2.1582203319300677e-06, "loss": 0.07645025849342346, "step": 4602 }, { "epoch": 1.194963655244029, "grad_norm": 3.640841007232666, "learning_rate": 2.157307681577227e-06, "loss": 0.02533440664410591, "step": 4603 }, { "epoch": 1.1952232606438213, "grad_norm": 2.459333658218384, "learning_rate": 2.1563951359988183e-06, "loss": 0.09316779673099518, "step": 4604 }, { "epoch": 1.1954828660436136, "grad_norm": 1.940468430519104, "learning_rate": 2.155482695356157e-06, "loss": 0.08938229084014893, "step": 4605 }, { "epoch": 1.195742471443406, "grad_norm": 2.9874534606933594, "learning_rate": 2.154570359810542e-06, "loss": 0.026931986212730408, "step": 4606 }, { "epoch": 1.1960020768431983, "grad_norm": 4.724596977233887, "learning_rate": 2.1536581295232545e-06, "loss": 0.03003878891468048, "step": 4607 }, { "epoch": 1.1962616822429906, "grad_norm": 2.698326826095581, "learning_rate": 2.152746004655556e-06, "loss": 0.07194890081882477, "step": 4608 }, { "epoch": 1.196521287642783, "grad_norm": 4.546741008758545, "learning_rate": 2.1518339853686874e-06, "loss": 0.04759805649518967, "step": 4609 }, { "epoch": 1.1967808930425754, "grad_norm": 2.27023983001709, "learning_rate": 2.1509220718238744e-06, "loss": 0.09212622046470642, "step": 4610 }, { "epoch": 1.1970404984423677, "grad_norm": 4.528189659118652, "learning_rate": 2.150010264182322e-06, "loss": 0.04078741371631622, "step": 4611 }, { "epoch": 1.19730010384216, "grad_norm": 1.688067078590393, "learning_rate": 2.1490985626052173e-06, "loss": 0.10097283124923706, "step": 4612 }, { "epoch": 1.1975597092419523, "grad_norm": 2.568796396255493, "learning_rate": 2.1481869672537277e-06, "loss": 0.01849876344203949, "step": 4613 }, { "epoch": 1.1978193146417446, "grad_norm": 3.2852511405944824, "learning_rate": 2.147275478289002e-06, "loss": 0.031106434762477875, "step": 4614 }, { "epoch": 1.198078920041537, "grad_norm": 3.8341188430786133, "learning_rate": 2.1463640958721714e-06, "loss": 0.03238809481263161, "step": 4615 }, { "epoch": 1.1983385254413292, "grad_norm": 3.7457215785980225, "learning_rate": 2.145452820164348e-06, "loss": 0.03131033480167389, "step": 4616 }, { "epoch": 1.1985981308411215, "grad_norm": 3.0199031829833984, "learning_rate": 2.144541651326623e-06, "loss": 0.048535458743572235, "step": 4617 }, { "epoch": 1.1988577362409139, "grad_norm": 1.9181166887283325, "learning_rate": 2.1436305895200715e-06, "loss": 0.07100682705640793, "step": 4618 }, { "epoch": 1.1991173416407062, "grad_norm": 3.093785047531128, "learning_rate": 2.142719634905747e-06, "loss": 0.045445069670677185, "step": 4619 }, { "epoch": 1.1993769470404985, "grad_norm": 1.8926572799682617, "learning_rate": 2.1418087876446867e-06, "loss": 0.10325121879577637, "step": 4620 }, { "epoch": 1.1996365524402908, "grad_norm": 2.731064558029175, "learning_rate": 2.140898047897907e-06, "loss": 0.03767282888293266, "step": 4621 }, { "epoch": 1.199896157840083, "grad_norm": 1.5746889114379883, "learning_rate": 2.139987415826405e-06, "loss": 0.0813111811876297, "step": 4622 }, { "epoch": 1.2001557632398754, "grad_norm": 2.25423526763916, "learning_rate": 2.1390768915911614e-06, "loss": 0.04831499978899956, "step": 4623 }, { "epoch": 1.2004153686396677, "grad_norm": 2.903899669647217, "learning_rate": 2.138166475353135e-06, "loss": 0.03633582219481468, "step": 4624 }, { "epoch": 1.20067497403946, "grad_norm": 2.232896327972412, "learning_rate": 2.137256167273266e-06, "loss": 0.0819544643163681, "step": 4625 }, { "epoch": 1.2009345794392523, "grad_norm": 2.458757162094116, "learning_rate": 2.136345967512477e-06, "loss": 0.07731559872627258, "step": 4626 }, { "epoch": 1.2011941848390446, "grad_norm": 2.315314769744873, "learning_rate": 2.1354358762316703e-06, "loss": 0.06915505975484848, "step": 4627 }, { "epoch": 1.201453790238837, "grad_norm": 2.9375600814819336, "learning_rate": 2.134525893591729e-06, "loss": 0.0442766472697258, "step": 4628 }, { "epoch": 1.2017133956386292, "grad_norm": 2.2751200199127197, "learning_rate": 2.133616019753517e-06, "loss": 0.07490640878677368, "step": 4629 }, { "epoch": 1.2019730010384215, "grad_norm": 1.8650187253952026, "learning_rate": 2.1327062548778797e-06, "loss": 0.07082365453243256, "step": 4630 }, { "epoch": 1.2022326064382138, "grad_norm": 3.0565032958984375, "learning_rate": 2.131796599125643e-06, "loss": 0.034341298043727875, "step": 4631 }, { "epoch": 1.2024922118380061, "grad_norm": 3.2280781269073486, "learning_rate": 2.1308870526576112e-06, "loss": 0.04793146997690201, "step": 4632 }, { "epoch": 1.2027518172377984, "grad_norm": 2.607015609741211, "learning_rate": 2.129977615634573e-06, "loss": 0.0798201858997345, "step": 4633 }, { "epoch": 1.2030114226375908, "grad_norm": 5.069211483001709, "learning_rate": 2.129068288217296e-06, "loss": 0.03257276862859726, "step": 4634 }, { "epoch": 1.203271028037383, "grad_norm": 3.153198003768921, "learning_rate": 2.128159070566529e-06, "loss": 0.030115513131022453, "step": 4635 }, { "epoch": 1.2035306334371756, "grad_norm": 4.355334281921387, "learning_rate": 2.1272499628429993e-06, "loss": 0.03882531821727753, "step": 4636 }, { "epoch": 1.203790238836968, "grad_norm": 1.9958441257476807, "learning_rate": 2.1263409652074165e-06, "loss": 0.08155709505081177, "step": 4637 }, { "epoch": 1.2040498442367602, "grad_norm": 5.097908973693848, "learning_rate": 2.125432077820471e-06, "loss": 0.041363947093486786, "step": 4638 }, { "epoch": 1.2043094496365525, "grad_norm": 2.38321852684021, "learning_rate": 2.1245233008428355e-06, "loss": 0.013210300356149673, "step": 4639 }, { "epoch": 1.2045690550363448, "grad_norm": 2.2098658084869385, "learning_rate": 2.123614634435158e-06, "loss": 0.08695819973945618, "step": 4640 }, { "epoch": 1.2048286604361371, "grad_norm": 3.0267422199249268, "learning_rate": 2.1227060787580693e-06, "loss": 0.05989180505275726, "step": 4641 }, { "epoch": 1.2050882658359294, "grad_norm": 3.2653872966766357, "learning_rate": 2.1217976339721837e-06, "loss": 0.050081007182598114, "step": 4642 }, { "epoch": 1.2053478712357217, "grad_norm": 5.1559834480285645, "learning_rate": 2.120889300238093e-06, "loss": 0.028562048450112343, "step": 4643 }, { "epoch": 1.205607476635514, "grad_norm": 2.504169225692749, "learning_rate": 2.1199810777163687e-06, "loss": 0.052486252039670944, "step": 4644 }, { "epoch": 1.2058670820353063, "grad_norm": 3.4193527698516846, "learning_rate": 2.1190729665675647e-06, "loss": 0.026583848521113396, "step": 4645 }, { "epoch": 1.2061266874350987, "grad_norm": 2.2663118839263916, "learning_rate": 2.1181649669522138e-06, "loss": 0.0933675467967987, "step": 4646 }, { "epoch": 1.206386292834891, "grad_norm": 3.9388740062713623, "learning_rate": 2.1172570790308293e-06, "loss": 0.05433795601129532, "step": 4647 }, { "epoch": 1.2066458982346833, "grad_norm": 4.507951259613037, "learning_rate": 2.1163493029639056e-06, "loss": 0.03758959472179413, "step": 4648 }, { "epoch": 1.2069055036344756, "grad_norm": 3.3271024227142334, "learning_rate": 2.1154416389119163e-06, "loss": 0.0328688770532608, "step": 4649 }, { "epoch": 1.2071651090342679, "grad_norm": 3.4780235290527344, "learning_rate": 2.114534087035316e-06, "loss": 0.03598114848136902, "step": 4650 }, { "epoch": 1.2074247144340602, "grad_norm": 2.273317337036133, "learning_rate": 2.1136266474945387e-06, "loss": 0.088612399995327, "step": 4651 }, { "epoch": 1.2076843198338525, "grad_norm": 3.2471718788146973, "learning_rate": 2.112719320449999e-06, "loss": 0.017776068300008774, "step": 4652 }, { "epoch": 1.2079439252336448, "grad_norm": 3.1192545890808105, "learning_rate": 2.111812106062092e-06, "loss": 0.07071022689342499, "step": 4653 }, { "epoch": 1.208203530633437, "grad_norm": 3.452873706817627, "learning_rate": 2.1109050044911922e-06, "loss": 0.04924429953098297, "step": 4654 }, { "epoch": 1.2084631360332294, "grad_norm": 4.1817708015441895, "learning_rate": 2.109998015897654e-06, "loss": 0.02962648496031761, "step": 4655 }, { "epoch": 1.2087227414330217, "grad_norm": 3.2224996089935303, "learning_rate": 2.1090911404418123e-06, "loss": 0.030060913413763046, "step": 4656 }, { "epoch": 1.208982346832814, "grad_norm": 3.4029226303100586, "learning_rate": 2.108184378283982e-06, "loss": 0.017864195629954338, "step": 4657 }, { "epoch": 1.2092419522326066, "grad_norm": 2.7196953296661377, "learning_rate": 2.107277729584459e-06, "loss": 0.024115854874253273, "step": 4658 }, { "epoch": 1.2095015576323989, "grad_norm": 1.6206599473953247, "learning_rate": 2.106371194503516e-06, "loss": 0.0941636711359024, "step": 4659 }, { "epoch": 1.2097611630321912, "grad_norm": 3.5306589603424072, "learning_rate": 2.1054647732014088e-06, "loss": 0.046169064939022064, "step": 4660 }, { "epoch": 1.2100207684319835, "grad_norm": 3.0590946674346924, "learning_rate": 2.1045584658383718e-06, "loss": 0.05838320776820183, "step": 4661 }, { "epoch": 1.2102803738317758, "grad_norm": 2.4549362659454346, "learning_rate": 2.1036522725746207e-06, "loss": 0.049729883670806885, "step": 4662 }, { "epoch": 1.210539979231568, "grad_norm": 1.7145438194274902, "learning_rate": 2.102746193570346e-06, "loss": 0.07545654475688934, "step": 4663 }, { "epoch": 1.2107995846313604, "grad_norm": 2.9802052974700928, "learning_rate": 2.1018402289857254e-06, "loss": 0.028662214055657387, "step": 4664 }, { "epoch": 1.2110591900311527, "grad_norm": 2.6306629180908203, "learning_rate": 2.1009343789809116e-06, "loss": 0.03716248273849487, "step": 4665 }, { "epoch": 1.211318795430945, "grad_norm": 2.4895057678222656, "learning_rate": 2.1000286437160385e-06, "loss": 0.03361469507217407, "step": 4666 }, { "epoch": 1.2115784008307373, "grad_norm": 1.8301670551300049, "learning_rate": 2.0991230233512177e-06, "loss": 0.07030385732650757, "step": 4667 }, { "epoch": 1.2118380062305296, "grad_norm": 1.9788613319396973, "learning_rate": 2.098217518046543e-06, "loss": 0.016232198104262352, "step": 4668 }, { "epoch": 1.212097611630322, "grad_norm": 2.462325096130371, "learning_rate": 2.097312127962088e-06, "loss": 0.06577920913696289, "step": 4669 }, { "epoch": 1.2123572170301142, "grad_norm": 3.6180386543273926, "learning_rate": 2.0964068532579047e-06, "loss": 0.01469483133405447, "step": 4670 }, { "epoch": 1.2126168224299065, "grad_norm": 1.8246525526046753, "learning_rate": 2.095501694094024e-06, "loss": 0.08910806477069855, "step": 4671 }, { "epoch": 1.2128764278296988, "grad_norm": 2.8592824935913086, "learning_rate": 2.094596650630457e-06, "loss": 0.05986659973859787, "step": 4672 }, { "epoch": 1.2131360332294912, "grad_norm": 3.063538074493408, "learning_rate": 2.093691723027196e-06, "loss": 0.047573547810316086, "step": 4673 }, { "epoch": 1.2133956386292835, "grad_norm": 3.1735944747924805, "learning_rate": 2.0927869114442118e-06, "loss": 0.054422527551651, "step": 4674 }, { "epoch": 1.2136552440290758, "grad_norm": 4.11083984375, "learning_rate": 2.091882216041452e-06, "loss": 0.02619342878460884, "step": 4675 }, { "epoch": 1.213914849428868, "grad_norm": 2.041865110397339, "learning_rate": 2.090977636978848e-06, "loss": 0.10139475762844086, "step": 4676 }, { "epoch": 1.2141744548286604, "grad_norm": 3.1057918071746826, "learning_rate": 2.090073174416308e-06, "loss": 0.04536714032292366, "step": 4677 }, { "epoch": 1.2144340602284527, "grad_norm": 2.588770866394043, "learning_rate": 2.089168828513722e-06, "loss": 0.06445186585187912, "step": 4678 }, { "epoch": 1.214693665628245, "grad_norm": 2.4182162284851074, "learning_rate": 2.0882645994309543e-06, "loss": 0.08301053941249847, "step": 4679 }, { "epoch": 1.2149532710280373, "grad_norm": 3.925271987915039, "learning_rate": 2.0873604873278543e-06, "loss": 0.026685431599617004, "step": 4680 }, { "epoch": 1.2152128764278296, "grad_norm": 2.9304676055908203, "learning_rate": 2.086456492364247e-06, "loss": 0.055765584111213684, "step": 4681 }, { "epoch": 1.215472481827622, "grad_norm": 2.6537766456604004, "learning_rate": 2.08555261469994e-06, "loss": 0.01783869042992592, "step": 4682 }, { "epoch": 1.2157320872274142, "grad_norm": 3.054157257080078, "learning_rate": 2.084648854494716e-06, "loss": 0.011101140640676022, "step": 4683 }, { "epoch": 1.2159916926272065, "grad_norm": 2.2631542682647705, "learning_rate": 2.08374521190834e-06, "loss": 0.08965957164764404, "step": 4684 }, { "epoch": 1.216251298026999, "grad_norm": 2.853895902633667, "learning_rate": 2.082841687100556e-06, "loss": 0.03278084844350815, "step": 4685 }, { "epoch": 1.2165109034267914, "grad_norm": 1.7317466735839844, "learning_rate": 2.0819382802310857e-06, "loss": 0.09289457648992538, "step": 4686 }, { "epoch": 1.2167705088265837, "grad_norm": 2.2986037731170654, "learning_rate": 2.08103499145963e-06, "loss": 0.07719562947750092, "step": 4687 }, { "epoch": 1.217030114226376, "grad_norm": 2.614619255065918, "learning_rate": 2.080131820945871e-06, "loss": 0.019366100430488586, "step": 4688 }, { "epoch": 1.2172897196261683, "grad_norm": 3.0427019596099854, "learning_rate": 2.0792287688494683e-06, "loss": 0.026073506101965904, "step": 4689 }, { "epoch": 1.2175493250259606, "grad_norm": 2.5346593856811523, "learning_rate": 2.0783258353300607e-06, "loss": 0.024268142879009247, "step": 4690 }, { "epoch": 1.217808930425753, "grad_norm": 2.1072802543640137, "learning_rate": 2.0774230205472657e-06, "loss": 0.05791354179382324, "step": 4691 }, { "epoch": 1.2180685358255452, "grad_norm": 1.4724730253219604, "learning_rate": 2.0765203246606806e-06, "loss": 0.008757305331528187, "step": 4692 }, { "epoch": 1.2183281412253375, "grad_norm": 3.077885866165161, "learning_rate": 2.0756177478298824e-06, "loss": 0.06086632236838341, "step": 4693 }, { "epoch": 1.2185877466251298, "grad_norm": 2.973871946334839, "learning_rate": 2.0747152902144243e-06, "loss": 0.027632908895611763, "step": 4694 }, { "epoch": 1.2188473520249221, "grad_norm": 2.588984966278076, "learning_rate": 2.073812951973841e-06, "loss": 0.05668577551841736, "step": 4695 }, { "epoch": 1.2191069574247144, "grad_norm": 3.322890043258667, "learning_rate": 2.072910733267645e-06, "loss": 0.016124092042446136, "step": 4696 }, { "epoch": 1.2193665628245067, "grad_norm": 3.6381406784057617, "learning_rate": 2.072008634255329e-06, "loss": 0.028633704409003258, "step": 4697 }, { "epoch": 1.219626168224299, "grad_norm": 2.9553399085998535, "learning_rate": 2.0711066550963615e-06, "loss": 0.03791643679141998, "step": 4698 }, { "epoch": 1.2198857736240913, "grad_norm": 3.4332549571990967, "learning_rate": 2.070204795950193e-06, "loss": 0.04797163978219032, "step": 4699 }, { "epoch": 1.2201453790238836, "grad_norm": 2.7408976554870605, "learning_rate": 2.069303056976251e-06, "loss": 0.09416963160037994, "step": 4700 }, { "epoch": 1.220404984423676, "grad_norm": 3.337307929992676, "learning_rate": 2.0684014383339447e-06, "loss": 0.02461859956383705, "step": 4701 }, { "epoch": 1.2206645898234683, "grad_norm": 3.2768208980560303, "learning_rate": 2.0674999401826558e-06, "loss": 0.03658812865614891, "step": 4702 }, { "epoch": 1.2209241952232606, "grad_norm": 1.9224886894226074, "learning_rate": 2.0665985626817496e-06, "loss": 0.08996451646089554, "step": 4703 }, { "epoch": 1.2211838006230529, "grad_norm": 3.2606122493743896, "learning_rate": 2.06569730599057e-06, "loss": 0.019187357276678085, "step": 4704 }, { "epoch": 1.2214434060228452, "grad_norm": 2.1760292053222656, "learning_rate": 2.06479617026844e-06, "loss": 0.08977136015892029, "step": 4705 }, { "epoch": 1.2217030114226377, "grad_norm": 2.4139115810394287, "learning_rate": 2.063895155674657e-06, "loss": 0.06323473900556564, "step": 4706 }, { "epoch": 1.22196261682243, "grad_norm": 3.207484006881714, "learning_rate": 2.0629942623685e-06, "loss": 0.028994733467698097, "step": 4707 }, { "epoch": 1.2222222222222223, "grad_norm": 5.510268688201904, "learning_rate": 2.0620934905092273e-06, "loss": 0.029240477830171585, "step": 4708 }, { "epoch": 1.2224818276220146, "grad_norm": 4.261717796325684, "learning_rate": 2.0611928402560756e-06, "loss": 0.0526605062186718, "step": 4709 }, { "epoch": 1.222741433021807, "grad_norm": 3.0452632904052734, "learning_rate": 2.0602923117682573e-06, "loss": 0.05929543823003769, "step": 4710 }, { "epoch": 1.2230010384215992, "grad_norm": 4.036485195159912, "learning_rate": 2.059391905204966e-06, "loss": 0.0439729318022728, "step": 4711 }, { "epoch": 1.2232606438213915, "grad_norm": 2.9653351306915283, "learning_rate": 2.0584916207253736e-06, "loss": 0.08420787751674652, "step": 4712 }, { "epoch": 1.2235202492211839, "grad_norm": 3.2479312419891357, "learning_rate": 2.0575914584886285e-06, "loss": 0.08057516813278198, "step": 4713 }, { "epoch": 1.2237798546209762, "grad_norm": 3.841543436050415, "learning_rate": 2.0566914186538596e-06, "loss": 0.06123737990856171, "step": 4714 }, { "epoch": 1.2240394600207685, "grad_norm": 3.699324369430542, "learning_rate": 2.0557915013801732e-06, "loss": 0.025459760800004005, "step": 4715 }, { "epoch": 1.2242990654205608, "grad_norm": 2.706258535385132, "learning_rate": 2.054891706826655e-06, "loss": 0.0729423314332962, "step": 4716 }, { "epoch": 1.224558670820353, "grad_norm": 2.3936777114868164, "learning_rate": 2.053992035152366e-06, "loss": 0.07573989778757095, "step": 4717 }, { "epoch": 1.2248182762201454, "grad_norm": 2.4095816612243652, "learning_rate": 2.053092486516349e-06, "loss": 0.07136072218418121, "step": 4718 }, { "epoch": 1.2250778816199377, "grad_norm": 2.9471521377563477, "learning_rate": 2.0521930610776235e-06, "loss": 0.061849478632211685, "step": 4719 }, { "epoch": 1.22533748701973, "grad_norm": 3.5461645126342773, "learning_rate": 2.0512937589951877e-06, "loss": 0.06122208014130592, "step": 4720 }, { "epoch": 1.2255970924195223, "grad_norm": 4.692230701446533, "learning_rate": 2.050394580428016e-06, "loss": 0.021720584481954575, "step": 4721 }, { "epoch": 1.2258566978193146, "grad_norm": 3.4961633682250977, "learning_rate": 2.049495525535064e-06, "loss": 0.07672297954559326, "step": 4722 }, { "epoch": 1.226116303219107, "grad_norm": 2.932926893234253, "learning_rate": 2.048596594475264e-06, "loss": 0.04317757114768028, "step": 4723 }, { "epoch": 1.2263759086188992, "grad_norm": 1.9555124044418335, "learning_rate": 2.047697787407526e-06, "loss": 0.08129993826150894, "step": 4724 }, { "epoch": 1.2266355140186915, "grad_norm": 4.670706748962402, "learning_rate": 2.046799104490738e-06, "loss": 0.043346650898456573, "step": 4725 }, { "epoch": 1.2268951194184838, "grad_norm": 2.426758289337158, "learning_rate": 2.045900545883768e-06, "loss": 0.021238144487142563, "step": 4726 }, { "epoch": 1.2271547248182761, "grad_norm": 3.0025546550750732, "learning_rate": 2.045002111745459e-06, "loss": 0.02314605563879013, "step": 4727 }, { "epoch": 1.2274143302180685, "grad_norm": 3.6281497478485107, "learning_rate": 2.0441038022346354e-06, "loss": 0.020144693553447723, "step": 4728 }, { "epoch": 1.2276739356178608, "grad_norm": 5.437496662139893, "learning_rate": 2.0432056175100957e-06, "loss": 0.05353795737028122, "step": 4729 }, { "epoch": 1.227933541017653, "grad_norm": 3.768650531768799, "learning_rate": 2.0423075577306194e-06, "loss": 0.05242795869708061, "step": 4730 }, { "epoch": 1.2281931464174454, "grad_norm": 2.4681897163391113, "learning_rate": 2.041409623054963e-06, "loss": 0.03988077491521835, "step": 4731 }, { "epoch": 1.2284527518172377, "grad_norm": 3.3609650135040283, "learning_rate": 2.040511813641862e-06, "loss": 0.08503907918930054, "step": 4732 }, { "epoch": 1.2287123572170302, "grad_norm": 3.4321444034576416, "learning_rate": 2.0396141296500256e-06, "loss": 0.032895322889089584, "step": 4733 }, { "epoch": 1.2289719626168225, "grad_norm": 3.1299667358398438, "learning_rate": 2.038716571238145e-06, "loss": 0.018448323011398315, "step": 4734 }, { "epoch": 1.2292315680166148, "grad_norm": 2.0119802951812744, "learning_rate": 2.037819138564889e-06, "loss": 0.11419419944286346, "step": 4735 }, { "epoch": 1.2294911734164071, "grad_norm": 2.1170525550842285, "learning_rate": 2.036921831788904e-06, "loss": 0.07908175885677338, "step": 4736 }, { "epoch": 1.2297507788161994, "grad_norm": 3.0291011333465576, "learning_rate": 2.0360246510688102e-06, "loss": 0.038931168615818024, "step": 4737 }, { "epoch": 1.2300103842159917, "grad_norm": 3.954749822616577, "learning_rate": 2.0351275965632104e-06, "loss": 0.061125800013542175, "step": 4738 }, { "epoch": 1.230269989615784, "grad_norm": 5.065027236938477, "learning_rate": 2.0342306684306834e-06, "loss": 0.02753121592104435, "step": 4739 }, { "epoch": 1.2305295950155763, "grad_norm": 3.300446033477783, "learning_rate": 2.033333866829785e-06, "loss": 0.035745952278375626, "step": 4740 }, { "epoch": 1.2307892004153687, "grad_norm": 3.6525774002075195, "learning_rate": 2.03243719191905e-06, "loss": 0.017364131286740303, "step": 4741 }, { "epoch": 1.231048805815161, "grad_norm": 3.299987554550171, "learning_rate": 2.031540643856989e-06, "loss": 0.040359318256378174, "step": 4742 }, { "epoch": 1.2313084112149533, "grad_norm": 2.929872989654541, "learning_rate": 2.0306442228020917e-06, "loss": 0.025899749249219894, "step": 4743 }, { "epoch": 1.2315680166147456, "grad_norm": 3.4732484817504883, "learning_rate": 2.0297479289128246e-06, "loss": 0.10189589858055115, "step": 4744 }, { "epoch": 1.2318276220145379, "grad_norm": 2.591549873352051, "learning_rate": 2.0288517623476324e-06, "loss": 0.024824492633342743, "step": 4745 }, { "epoch": 1.2320872274143302, "grad_norm": 2.869572639465332, "learning_rate": 2.027955723264936e-06, "loss": 0.08576315641403198, "step": 4746 }, { "epoch": 1.2323468328141225, "grad_norm": 4.2416205406188965, "learning_rate": 2.027059811823136e-06, "loss": 0.03499286249279976, "step": 4747 }, { "epoch": 1.2326064382139148, "grad_norm": 2.3789210319519043, "learning_rate": 2.026164028180608e-06, "loss": 0.09734369069337845, "step": 4748 }, { "epoch": 1.232866043613707, "grad_norm": 3.972353458404541, "learning_rate": 2.0252683724957052e-06, "loss": 0.028032150119543076, "step": 4749 }, { "epoch": 1.2331256490134994, "grad_norm": 2.3773293495178223, "learning_rate": 2.02437284492676e-06, "loss": 0.07013898342847824, "step": 4750 }, { "epoch": 1.2333852544132917, "grad_norm": 3.0005970001220703, "learning_rate": 2.0234774456320823e-06, "loss": 0.0414753332734108, "step": 4751 }, { "epoch": 1.233644859813084, "grad_norm": 2.2735512256622314, "learning_rate": 2.0225821747699566e-06, "loss": 0.05413667485117912, "step": 4752 }, { "epoch": 1.2339044652128763, "grad_norm": 4.0365986824035645, "learning_rate": 2.0216870324986466e-06, "loss": 0.04875566065311432, "step": 4753 }, { "epoch": 1.2341640706126689, "grad_norm": 2.600391387939453, "learning_rate": 2.020792018976393e-06, "loss": 0.017318442463874817, "step": 4754 }, { "epoch": 1.2344236760124612, "grad_norm": 3.207627058029175, "learning_rate": 2.0198971343614136e-06, "loss": 0.06359320878982544, "step": 4755 }, { "epoch": 1.2346832814122535, "grad_norm": 2.3455276489257812, "learning_rate": 2.0190023788119044e-06, "loss": 0.06493906676769257, "step": 4756 }, { "epoch": 1.2349428868120458, "grad_norm": 2.8562171459198, "learning_rate": 2.018107752486036e-06, "loss": 0.029071900993585587, "step": 4757 }, { "epoch": 1.235202492211838, "grad_norm": 2.784301996231079, "learning_rate": 2.01721325554196e-06, "loss": 0.022306939586997032, "step": 4758 }, { "epoch": 1.2354620976116304, "grad_norm": 4.501221179962158, "learning_rate": 2.0163188881378017e-06, "loss": 0.025528328493237495, "step": 4759 }, { "epoch": 1.2357217030114227, "grad_norm": 2.852508544921875, "learning_rate": 2.0154246504316637e-06, "loss": 0.03093409352004528, "step": 4760 }, { "epoch": 1.235981308411215, "grad_norm": 3.022855520248413, "learning_rate": 2.0145305425816287e-06, "loss": 0.02803685888648033, "step": 4761 }, { "epoch": 1.2362409138110073, "grad_norm": 3.6803324222564697, "learning_rate": 2.013636564745754e-06, "loss": 0.02927975356578827, "step": 4762 }, { "epoch": 1.2365005192107996, "grad_norm": 2.0641703605651855, "learning_rate": 2.0127427170820746e-06, "loss": 0.07211735099554062, "step": 4763 }, { "epoch": 1.236760124610592, "grad_norm": 2.223555564880371, "learning_rate": 2.0118489997486013e-06, "loss": 0.08266732096672058, "step": 4764 }, { "epoch": 1.2370197300103842, "grad_norm": 2.074030637741089, "learning_rate": 2.0109554129033225e-06, "loss": 0.08581554889678955, "step": 4765 }, { "epoch": 1.2372793354101765, "grad_norm": 2.1008358001708984, "learning_rate": 2.0100619567042068e-06, "loss": 0.006212475709617138, "step": 4766 }, { "epoch": 1.2375389408099688, "grad_norm": 2.4354543685913086, "learning_rate": 2.0091686313091933e-06, "loss": 0.016666024923324585, "step": 4767 }, { "epoch": 1.2377985462097612, "grad_norm": 3.1765518188476562, "learning_rate": 2.0082754368762036e-06, "loss": 0.02943357080221176, "step": 4768 }, { "epoch": 1.2380581516095535, "grad_norm": 2.9276347160339355, "learning_rate": 2.0073823735631324e-06, "loss": 0.012112297117710114, "step": 4769 }, { "epoch": 1.2383177570093458, "grad_norm": 3.0992579460144043, "learning_rate": 2.006489441527855e-06, "loss": 0.07356691360473633, "step": 4770 }, { "epoch": 1.238577362409138, "grad_norm": 2.8451879024505615, "learning_rate": 2.00559664092822e-06, "loss": 0.061102598905563354, "step": 4771 }, { "epoch": 1.2388369678089304, "grad_norm": 3.932462453842163, "learning_rate": 2.004703971922054e-06, "loss": 0.0407792329788208, "step": 4772 }, { "epoch": 1.2390965732087227, "grad_norm": 3.7146642208099365, "learning_rate": 2.0038114346671612e-06, "loss": 0.028389208018779755, "step": 4773 }, { "epoch": 1.239356178608515, "grad_norm": 2.918079137802124, "learning_rate": 2.0029190293213213e-06, "loss": 0.057297661900520325, "step": 4774 }, { "epoch": 1.2396157840083073, "grad_norm": 3.7600769996643066, "learning_rate": 2.002026756042291e-06, "loss": 0.036989204585552216, "step": 4775 }, { "epoch": 1.2398753894080996, "grad_norm": 2.4234273433685303, "learning_rate": 2.0011346149878032e-06, "loss": 0.05645844340324402, "step": 4776 }, { "epoch": 1.240134994807892, "grad_norm": 3.1373817920684814, "learning_rate": 2.0002426063155694e-06, "loss": 0.07036234438419342, "step": 4777 }, { "epoch": 1.2403946002076842, "grad_norm": 3.182177782058716, "learning_rate": 1.999350730183276e-06, "loss": 0.012346155010163784, "step": 4778 }, { "epoch": 1.2406542056074765, "grad_norm": 1.6951992511749268, "learning_rate": 1.998458986748585e-06, "loss": 0.10352116823196411, "step": 4779 }, { "epoch": 1.2409138110072688, "grad_norm": 2.583176612854004, "learning_rate": 1.997567376169137e-06, "loss": 0.029450807720422745, "step": 4780 }, { "epoch": 1.2411734164070614, "grad_norm": 2.0022847652435303, "learning_rate": 1.99667589860255e-06, "loss": 0.02350815385580063, "step": 4781 }, { "epoch": 1.2414330218068537, "grad_norm": 2.4385902881622314, "learning_rate": 1.995784554206414e-06, "loss": 0.021245650947093964, "step": 4782 }, { "epoch": 1.241692627206646, "grad_norm": 2.536285877227783, "learning_rate": 1.9948933431383e-06, "loss": 0.020600946620106697, "step": 4783 }, { "epoch": 1.2419522326064383, "grad_norm": 4.645227909088135, "learning_rate": 1.9940022655557527e-06, "loss": 0.04313796013593674, "step": 4784 }, { "epoch": 1.2422118380062306, "grad_norm": 2.737971544265747, "learning_rate": 1.9931113216162956e-06, "loss": 0.0313313826918602, "step": 4785 }, { "epoch": 1.242471443406023, "grad_norm": 3.120234489440918, "learning_rate": 1.992220511477426e-06, "loss": 0.04322269558906555, "step": 4786 }, { "epoch": 1.2427310488058152, "grad_norm": 2.8961567878723145, "learning_rate": 1.9913298352966193e-06, "loss": 0.05243758112192154, "step": 4787 }, { "epoch": 1.2429906542056075, "grad_norm": 2.495619297027588, "learning_rate": 1.990439293231326e-06, "loss": 0.05121603608131409, "step": 4788 }, { "epoch": 1.2432502596053998, "grad_norm": 2.5942108631134033, "learning_rate": 1.9895488854389744e-06, "loss": 0.025593269616365433, "step": 4789 }, { "epoch": 1.2435098650051921, "grad_norm": 2.4539272785186768, "learning_rate": 1.9886586120769685e-06, "loss": 0.017960723489522934, "step": 4790 }, { "epoch": 1.2437694704049844, "grad_norm": 3.2786707878112793, "learning_rate": 1.9877684733026865e-06, "loss": 0.09948860108852386, "step": 4791 }, { "epoch": 1.2440290758047767, "grad_norm": 3.7935562133789062, "learning_rate": 1.986878469273486e-06, "loss": 0.014525601640343666, "step": 4792 }, { "epoch": 1.244288681204569, "grad_norm": 1.8783352375030518, "learning_rate": 1.9859886001466988e-06, "loss": 0.0971619188785553, "step": 4793 }, { "epoch": 1.2445482866043613, "grad_norm": 3.520427703857422, "learning_rate": 1.9850988660796344e-06, "loss": 0.012887230142951012, "step": 4794 }, { "epoch": 1.2448078920041536, "grad_norm": 2.2368226051330566, "learning_rate": 1.984209267229576e-06, "loss": 0.09322115778923035, "step": 4795 }, { "epoch": 1.245067497403946, "grad_norm": 3.237971782684326, "learning_rate": 1.9833198037537844e-06, "loss": 0.07770118862390518, "step": 4796 }, { "epoch": 1.2453271028037383, "grad_norm": 3.3259410858154297, "learning_rate": 1.982430475809498e-06, "loss": 0.05182849243283272, "step": 4797 }, { "epoch": 1.2455867082035306, "grad_norm": 3.20133113861084, "learning_rate": 1.9815412835539277e-06, "loss": 0.06517849862575531, "step": 4798 }, { "epoch": 1.2458463136033229, "grad_norm": 3.7500858306884766, "learning_rate": 1.9806522271442624e-06, "loss": 0.028355322778224945, "step": 4799 }, { "epoch": 1.2461059190031152, "grad_norm": 2.8707590103149414, "learning_rate": 1.979763306737668e-06, "loss": 0.021655071526765823, "step": 4800 }, { "epoch": 1.2463655244029075, "grad_norm": 2.6320784091949463, "learning_rate": 1.978874522491286e-06, "loss": 0.06597031652927399, "step": 4801 }, { "epoch": 1.2466251298027, "grad_norm": 3.5099499225616455, "learning_rate": 1.977985874562231e-06, "loss": 0.014890103600919247, "step": 4802 }, { "epoch": 1.2468847352024923, "grad_norm": 4.026238441467285, "learning_rate": 1.977097363107597e-06, "loss": 0.026522459462285042, "step": 4803 }, { "epoch": 1.2471443406022846, "grad_norm": 3.0801515579223633, "learning_rate": 1.9762089882844517e-06, "loss": 0.04317469894886017, "step": 4804 }, { "epoch": 1.247403946002077, "grad_norm": 2.424767017364502, "learning_rate": 1.97532075024984e-06, "loss": 0.08570301532745361, "step": 4805 }, { "epoch": 1.2476635514018692, "grad_norm": 3.6981875896453857, "learning_rate": 1.974432649160782e-06, "loss": 0.02740117534995079, "step": 4806 }, { "epoch": 1.2479231568016615, "grad_norm": 2.2301559448242188, "learning_rate": 1.973544685174273e-06, "loss": 0.014867136254906654, "step": 4807 }, { "epoch": 1.2481827622014539, "grad_norm": 2.7059361934661865, "learning_rate": 1.972656858447285e-06, "loss": 0.045348625630140305, "step": 4808 }, { "epoch": 1.2484423676012462, "grad_norm": 2.801162004470825, "learning_rate": 1.971769169136767e-06, "loss": 0.008968951180577278, "step": 4809 }, { "epoch": 1.2487019730010385, "grad_norm": 2.5517823696136475, "learning_rate": 1.9708816173996392e-06, "loss": 0.08959998190402985, "step": 4810 }, { "epoch": 1.2489615784008308, "grad_norm": 2.6939690113067627, "learning_rate": 1.969994203392803e-06, "loss": 0.06774307787418365, "step": 4811 }, { "epoch": 1.249221183800623, "grad_norm": 3.1999926567077637, "learning_rate": 1.969106927273132e-06, "loss": 0.03992019221186638, "step": 4812 }, { "epoch": 1.2494807892004154, "grad_norm": 2.628568649291992, "learning_rate": 1.9682197891974753e-06, "loss": 0.03953719511628151, "step": 4813 }, { "epoch": 1.2497403946002077, "grad_norm": 4.842316627502441, "learning_rate": 1.9673327893226603e-06, "loss": 0.04079653322696686, "step": 4814 }, { "epoch": 1.25, "grad_norm": 2.7009189128875732, "learning_rate": 1.9664459278054866e-06, "loss": 0.03085993230342865, "step": 4815 }, { "epoch": 1.2502596053997923, "grad_norm": 2.9188053607940674, "learning_rate": 1.9655592048027324e-06, "loss": 0.027308503165841103, "step": 4816 }, { "epoch": 1.2505192107995846, "grad_norm": 3.4007484912872314, "learning_rate": 1.964672620471149e-06, "loss": 0.02895933948457241, "step": 4817 }, { "epoch": 1.250778816199377, "grad_norm": 2.9753360748291016, "learning_rate": 1.963786174967465e-06, "loss": 0.01940327323973179, "step": 4818 }, { "epoch": 1.2510384215991692, "grad_norm": 3.8466570377349854, "learning_rate": 1.9628998684483835e-06, "loss": 0.04102165251970291, "step": 4819 }, { "epoch": 1.2512980269989615, "grad_norm": 3.923173427581787, "learning_rate": 1.962013701070583e-06, "loss": 0.01818908378481865, "step": 4820 }, { "epoch": 1.2515576323987538, "grad_norm": 2.8014509677886963, "learning_rate": 1.9611276729907187e-06, "loss": 0.021233059465885162, "step": 4821 }, { "epoch": 1.2518172377985461, "grad_norm": 3.0808563232421875, "learning_rate": 1.960241784365417e-06, "loss": 0.06463675200939178, "step": 4822 }, { "epoch": 1.2520768431983385, "grad_norm": 3.5405094623565674, "learning_rate": 1.959356035351286e-06, "loss": 0.055328547954559326, "step": 4823 }, { "epoch": 1.2523364485981308, "grad_norm": 2.6297857761383057, "learning_rate": 1.9584704261049047e-06, "loss": 0.01639009267091751, "step": 4824 }, { "epoch": 1.252596053997923, "grad_norm": 3.089174509048462, "learning_rate": 1.957584956782828e-06, "loss": 0.04291946440935135, "step": 4825 }, { "epoch": 1.2528556593977154, "grad_norm": 2.8858284950256348, "learning_rate": 1.956699627541587e-06, "loss": 0.07072989642620087, "step": 4826 }, { "epoch": 1.2531152647975077, "grad_norm": 1.3144514560699463, "learning_rate": 1.9558144385376877e-06, "loss": 0.0037572053261101246, "step": 4827 }, { "epoch": 1.2533748701973, "grad_norm": 3.9207799434661865, "learning_rate": 1.9549293899276115e-06, "loss": 0.018051406368613243, "step": 4828 }, { "epoch": 1.2536344755970923, "grad_norm": 4.013047695159912, "learning_rate": 1.9540444818678146e-06, "loss": 0.06048068404197693, "step": 4829 }, { "epoch": 1.2538940809968846, "grad_norm": 1.9738696813583374, "learning_rate": 1.953159714514728e-06, "loss": 0.09292323887348175, "step": 4830 }, { "epoch": 1.2541536863966771, "grad_norm": 3.4579386711120605, "learning_rate": 1.952275088024758e-06, "loss": 0.0649033710360527, "step": 4831 }, { "epoch": 1.2544132917964694, "grad_norm": 2.755863666534424, "learning_rate": 1.9513906025542883e-06, "loss": 0.025362389162182808, "step": 4832 }, { "epoch": 1.2546728971962617, "grad_norm": 1.9586127996444702, "learning_rate": 1.9505062582596735e-06, "loss": 0.07365977019071579, "step": 4833 }, { "epoch": 1.254932502596054, "grad_norm": 3.490429639816284, "learning_rate": 1.9496220552972467e-06, "loss": 0.04192964732646942, "step": 4834 }, { "epoch": 1.2551921079958464, "grad_norm": 2.6014437675476074, "learning_rate": 1.9487379938233136e-06, "loss": 0.030377451330423355, "step": 4835 }, { "epoch": 1.2554517133956387, "grad_norm": 1.9745618104934692, "learning_rate": 1.9478540739941577e-06, "loss": 0.06935235857963562, "step": 4836 }, { "epoch": 1.255711318795431, "grad_norm": 3.527015447616577, "learning_rate": 1.946970295966034e-06, "loss": 0.024481311440467834, "step": 4837 }, { "epoch": 1.2559709241952233, "grad_norm": 4.757015705108643, "learning_rate": 1.9460866598951748e-06, "loss": 0.03453140705823898, "step": 4838 }, { "epoch": 1.2562305295950156, "grad_norm": 2.4810409545898438, "learning_rate": 1.9452031659377874e-06, "loss": 0.024162717163562775, "step": 4839 }, { "epoch": 1.2564901349948079, "grad_norm": 3.016263484954834, "learning_rate": 1.9443198142500537e-06, "loss": 0.038314126431941986, "step": 4840 }, { "epoch": 1.2567497403946002, "grad_norm": 2.9839184284210205, "learning_rate": 1.943436604988128e-06, "loss": 0.0686831921339035, "step": 4841 }, { "epoch": 1.2570093457943925, "grad_norm": 4.689566135406494, "learning_rate": 1.9425535383081434e-06, "loss": 0.02629014290869236, "step": 4842 }, { "epoch": 1.2572689511941848, "grad_norm": 2.3018527030944824, "learning_rate": 1.9416706143662044e-06, "loss": 0.06703983247280121, "step": 4843 }, { "epoch": 1.257528556593977, "grad_norm": 2.8288257122039795, "learning_rate": 1.9407878333183933e-06, "loss": 0.05002717673778534, "step": 4844 }, { "epoch": 1.2577881619937694, "grad_norm": 1.9786916971206665, "learning_rate": 1.9399051953207643e-06, "loss": 0.09112239629030228, "step": 4845 }, { "epoch": 1.2580477673935617, "grad_norm": 2.3922245502471924, "learning_rate": 1.939022700529348e-06, "loss": 0.013013365678489208, "step": 4846 }, { "epoch": 1.258307372793354, "grad_norm": 2.7999448776245117, "learning_rate": 1.9381403491001492e-06, "loss": 0.025809381157159805, "step": 4847 }, { "epoch": 1.2585669781931463, "grad_norm": 3.2854833602905273, "learning_rate": 1.9372581411891485e-06, "loss": 0.06026008725166321, "step": 4848 }, { "epoch": 1.2588265835929389, "grad_norm": 3.18001651763916, "learning_rate": 1.9363760769522982e-06, "loss": 0.055313464254140854, "step": 4849 }, { "epoch": 1.2590861889927312, "grad_norm": 3.1327004432678223, "learning_rate": 1.935494156545528e-06, "loss": 0.06705192476511002, "step": 4850 }, { "epoch": 1.2593457943925235, "grad_norm": 4.062257766723633, "learning_rate": 1.9346123801247422e-06, "loss": 0.01759173348546028, "step": 4851 }, { "epoch": 1.2596053997923158, "grad_norm": 4.438665866851807, "learning_rate": 1.9337307478458167e-06, "loss": 0.030839676037430763, "step": 4852 }, { "epoch": 1.259865005192108, "grad_norm": 2.6154370307922363, "learning_rate": 1.9328492598646047e-06, "loss": 0.016072629019618034, "step": 4853 }, { "epoch": 1.2601246105919004, "grad_norm": 3.305051803588867, "learning_rate": 1.9319679163369336e-06, "loss": 0.040088947862386703, "step": 4854 }, { "epoch": 1.2603842159916927, "grad_norm": 3.1904890537261963, "learning_rate": 1.9310867174186054e-06, "loss": 0.04505518078804016, "step": 4855 }, { "epoch": 1.260643821391485, "grad_norm": 4.241700649261475, "learning_rate": 1.9302056632653937e-06, "loss": 0.02242773398756981, "step": 4856 }, { "epoch": 1.2609034267912773, "grad_norm": 2.7741634845733643, "learning_rate": 1.92932475403305e-06, "loss": 0.011250087060034275, "step": 4857 }, { "epoch": 1.2611630321910696, "grad_norm": 2.7049291133880615, "learning_rate": 1.9284439898772994e-06, "loss": 0.05948024243116379, "step": 4858 }, { "epoch": 1.261422637590862, "grad_norm": 3.461333990097046, "learning_rate": 1.9275633709538403e-06, "loss": 0.05174858123064041, "step": 4859 }, { "epoch": 1.2616822429906542, "grad_norm": 3.9350011348724365, "learning_rate": 1.9266828974183456e-06, "loss": 0.032121673226356506, "step": 4860 }, { "epoch": 1.2619418483904465, "grad_norm": 3.602595806121826, "learning_rate": 1.925802569426463e-06, "loss": 0.039857152849435806, "step": 4861 }, { "epoch": 1.2622014537902388, "grad_norm": 3.261673927307129, "learning_rate": 1.9249223871338142e-06, "loss": 0.015012422576546669, "step": 4862 }, { "epoch": 1.2624610591900312, "grad_norm": 2.3810784816741943, "learning_rate": 1.924042350695997e-06, "loss": 0.024140315130352974, "step": 4863 }, { "epoch": 1.2627206645898235, "grad_norm": 2.3807790279388428, "learning_rate": 1.9231624602685787e-06, "loss": 0.09196625649929047, "step": 4864 }, { "epoch": 1.2629802699896158, "grad_norm": 3.302273750305176, "learning_rate": 1.922282716007106e-06, "loss": 0.040967606008052826, "step": 4865 }, { "epoch": 1.263239875389408, "grad_norm": 3.7953696250915527, "learning_rate": 1.9214031180670973e-06, "loss": 0.02503744140267372, "step": 4866 }, { "epoch": 1.2634994807892004, "grad_norm": 2.493664264678955, "learning_rate": 1.9205236666040443e-06, "loss": 0.06171448156237602, "step": 4867 }, { "epoch": 1.2637590861889927, "grad_norm": 4.324277400970459, "learning_rate": 1.919644361773415e-06, "loss": 0.02093760296702385, "step": 4868 }, { "epoch": 1.264018691588785, "grad_norm": 2.6784512996673584, "learning_rate": 1.9187652037306496e-06, "loss": 0.08573959022760391, "step": 4869 }, { "epoch": 1.2642782969885773, "grad_norm": 2.7160751819610596, "learning_rate": 1.917886192631163e-06, "loss": 0.021927399560809135, "step": 4870 }, { "epoch": 1.2645379023883696, "grad_norm": 2.467470645904541, "learning_rate": 1.9170073286303458e-06, "loss": 0.06762684881687164, "step": 4871 }, { "epoch": 1.264797507788162, "grad_norm": 3.3786940574645996, "learning_rate": 1.9161286118835594e-06, "loss": 0.02286742627620697, "step": 4872 }, { "epoch": 1.2650571131879542, "grad_norm": 3.0374414920806885, "learning_rate": 1.9152500425461414e-06, "loss": 0.0474967360496521, "step": 4873 }, { "epoch": 1.2653167185877465, "grad_norm": 4.53406286239624, "learning_rate": 1.914371620773402e-06, "loss": 0.02460668608546257, "step": 4874 }, { "epoch": 1.2655763239875388, "grad_norm": 2.841503858566284, "learning_rate": 1.9134933467206274e-06, "loss": 0.07393456250429153, "step": 4875 }, { "epoch": 1.2658359293873311, "grad_norm": 3.020745277404785, "learning_rate": 1.9126152205430754e-06, "loss": 0.054160043597221375, "step": 4876 }, { "epoch": 1.2660955347871234, "grad_norm": 2.146085023880005, "learning_rate": 1.9117372423959786e-06, "loss": 0.09768528491258621, "step": 4877 }, { "epoch": 1.2663551401869158, "grad_norm": 1.824026107788086, "learning_rate": 1.9108594124345444e-06, "loss": 0.08483437448740005, "step": 4878 }, { "epoch": 1.2666147455867083, "grad_norm": 3.6825311183929443, "learning_rate": 1.9099817308139516e-06, "loss": 0.021078763529658318, "step": 4879 }, { "epoch": 1.2668743509865006, "grad_norm": 1.8146876096725464, "learning_rate": 1.9091041976893547e-06, "loss": 0.09221287071704865, "step": 4880 }, { "epoch": 1.267133956386293, "grad_norm": 3.212759017944336, "learning_rate": 1.908226813215882e-06, "loss": 0.029505867511034012, "step": 4881 }, { "epoch": 1.2673935617860852, "grad_norm": 3.554051160812378, "learning_rate": 1.9073495775486356e-06, "loss": 0.060346849262714386, "step": 4882 }, { "epoch": 1.2676531671858775, "grad_norm": 4.517953395843506, "learning_rate": 1.9064724908426885e-06, "loss": 0.04736330732703209, "step": 4883 }, { "epoch": 1.2679127725856698, "grad_norm": 2.807673931121826, "learning_rate": 1.9055955532530912e-06, "loss": 0.07261912524700165, "step": 4884 }, { "epoch": 1.2681723779854621, "grad_norm": 4.0500946044921875, "learning_rate": 1.9047187649348652e-06, "loss": 0.03211558610200882, "step": 4885 }, { "epoch": 1.2684319833852544, "grad_norm": 2.376288414001465, "learning_rate": 1.903842126043009e-06, "loss": 0.02795371040701866, "step": 4886 }, { "epoch": 1.2686915887850467, "grad_norm": 2.127413511276245, "learning_rate": 1.902965636732489e-06, "loss": 0.010454737581312656, "step": 4887 }, { "epoch": 1.268951194184839, "grad_norm": 2.7720537185668945, "learning_rate": 1.9020892971582495e-06, "loss": 0.023128926753997803, "step": 4888 }, { "epoch": 1.2692107995846313, "grad_norm": 2.4660816192626953, "learning_rate": 1.9012131074752083e-06, "loss": 0.0184493325650692, "step": 4889 }, { "epoch": 1.2694704049844237, "grad_norm": 5.11779260635376, "learning_rate": 1.900337067838256e-06, "loss": 0.031013816595077515, "step": 4890 }, { "epoch": 1.269730010384216, "grad_norm": 3.222311496734619, "learning_rate": 1.8994611784022542e-06, "loss": 0.06249456852674484, "step": 4891 }, { "epoch": 1.2699896157840083, "grad_norm": 4.040714263916016, "learning_rate": 1.8985854393220414e-06, "loss": 0.035334572196006775, "step": 4892 }, { "epoch": 1.2702492211838006, "grad_norm": 3.7165048122406006, "learning_rate": 1.8977098507524283e-06, "loss": 0.0363282784819603, "step": 4893 }, { "epoch": 1.2705088265835929, "grad_norm": 3.7222135066986084, "learning_rate": 1.8968344128481992e-06, "loss": 0.04609465226531029, "step": 4894 }, { "epoch": 1.2707684319833852, "grad_norm": 3.8983147144317627, "learning_rate": 1.8959591257641107e-06, "loss": 0.0410173125565052, "step": 4895 }, { "epoch": 1.2710280373831775, "grad_norm": 2.653571605682373, "learning_rate": 1.8950839896548934e-06, "loss": 0.04865947365760803, "step": 4896 }, { "epoch": 1.27128764278297, "grad_norm": 1.9806807041168213, "learning_rate": 1.894209004675252e-06, "loss": 0.07924829423427582, "step": 4897 }, { "epoch": 1.2715472481827623, "grad_norm": 3.9461731910705566, "learning_rate": 1.8933341709798642e-06, "loss": 0.0436076857149601, "step": 4898 }, { "epoch": 1.2718068535825546, "grad_norm": 1.6420975923538208, "learning_rate": 1.8924594887233787e-06, "loss": 0.08729632198810577, "step": 4899 }, { "epoch": 1.272066458982347, "grad_norm": 1.8669373989105225, "learning_rate": 1.8915849580604213e-06, "loss": 0.08583647757768631, "step": 4900 }, { "epoch": 1.2723260643821392, "grad_norm": 2.5746166706085205, "learning_rate": 1.8907105791455877e-06, "loss": 0.0773712620139122, "step": 4901 }, { "epoch": 1.2725856697819315, "grad_norm": 2.877324104309082, "learning_rate": 1.8898363521334494e-06, "loss": 0.014730541966855526, "step": 4902 }, { "epoch": 1.2728452751817239, "grad_norm": 2.7110109329223633, "learning_rate": 1.888962277178548e-06, "loss": 0.030010949820280075, "step": 4903 }, { "epoch": 1.2731048805815162, "grad_norm": 2.739577531814575, "learning_rate": 1.8880883544354017e-06, "loss": 0.04775041341781616, "step": 4904 }, { "epoch": 1.2733644859813085, "grad_norm": 2.785888671875, "learning_rate": 1.8872145840584987e-06, "loss": 0.04143499583005905, "step": 4905 }, { "epoch": 1.2736240913811008, "grad_norm": 1.8443740606307983, "learning_rate": 1.8863409662023024e-06, "loss": 0.08878193795681, "step": 4906 }, { "epoch": 1.273883696780893, "grad_norm": 2.218317985534668, "learning_rate": 1.8854675010212476e-06, "loss": 0.0886259526014328, "step": 4907 }, { "epoch": 1.2741433021806854, "grad_norm": 2.633964776992798, "learning_rate": 1.8845941886697432e-06, "loss": 0.04967038333415985, "step": 4908 }, { "epoch": 1.2744029075804777, "grad_norm": 2.991745948791504, "learning_rate": 1.8837210293021724e-06, "loss": 0.06152038648724556, "step": 4909 }, { "epoch": 1.27466251298027, "grad_norm": 2.445068359375, "learning_rate": 1.8828480230728874e-06, "loss": 0.07394133508205414, "step": 4910 }, { "epoch": 1.2749221183800623, "grad_norm": 2.650892972946167, "learning_rate": 1.8819751701362165e-06, "loss": 0.03505498170852661, "step": 4911 }, { "epoch": 1.2751817237798546, "grad_norm": 3.926236391067505, "learning_rate": 1.8811024706464614e-06, "loss": 0.018644005060195923, "step": 4912 }, { "epoch": 1.275441329179647, "grad_norm": 2.7687857151031494, "learning_rate": 1.8802299247578943e-06, "loss": 0.08113716542720795, "step": 4913 }, { "epoch": 1.2757009345794392, "grad_norm": 5.244493007659912, "learning_rate": 1.8793575326247603e-06, "loss": 0.03194098547101021, "step": 4914 }, { "epoch": 1.2759605399792315, "grad_norm": 3.2629637718200684, "learning_rate": 1.8784852944012799e-06, "loss": 0.041439399123191833, "step": 4915 }, { "epoch": 1.2762201453790238, "grad_norm": 2.134878396987915, "learning_rate": 1.8776132102416444e-06, "loss": 0.06970615684986115, "step": 4916 }, { "epoch": 1.2764797507788161, "grad_norm": 2.874849557876587, "learning_rate": 1.8767412803000195e-06, "loss": 0.026814298704266548, "step": 4917 }, { "epoch": 1.2767393561786085, "grad_norm": 2.0872678756713867, "learning_rate": 1.875869504730541e-06, "loss": 0.07467156648635864, "step": 4918 }, { "epoch": 1.2769989615784008, "grad_norm": 3.541328191757202, "learning_rate": 1.8749978836873174e-06, "loss": 0.0353488065302372, "step": 4919 }, { "epoch": 1.277258566978193, "grad_norm": 6.142854690551758, "learning_rate": 1.874126417324434e-06, "loss": 0.023092148825526237, "step": 4920 }, { "epoch": 1.2775181723779854, "grad_norm": 2.0215506553649902, "learning_rate": 1.8732551057959459e-06, "loss": 0.0763479471206665, "step": 4921 }, { "epoch": 1.2777777777777777, "grad_norm": 3.613023042678833, "learning_rate": 1.8723839492558793e-06, "loss": 0.04565581679344177, "step": 4922 }, { "epoch": 1.27803738317757, "grad_norm": 3.198154926300049, "learning_rate": 1.8715129478582367e-06, "loss": 0.018985800445079803, "step": 4923 }, { "epoch": 1.2782969885773623, "grad_norm": 3.501007318496704, "learning_rate": 1.8706421017569888e-06, "loss": 0.045419007539749146, "step": 4924 }, { "epoch": 1.2785565939771546, "grad_norm": 3.119927167892456, "learning_rate": 1.8697714111060843e-06, "loss": 0.06174454465508461, "step": 4925 }, { "epoch": 1.278816199376947, "grad_norm": 2.7351717948913574, "learning_rate": 1.868900876059439e-06, "loss": 0.06040704995393753, "step": 4926 }, { "epoch": 1.2790758047767394, "grad_norm": 2.1687872409820557, "learning_rate": 1.868030496770944e-06, "loss": 0.08554696291685104, "step": 4927 }, { "epoch": 1.2793354101765317, "grad_norm": 3.2200605869293213, "learning_rate": 1.8671602733944633e-06, "loss": 0.0659550204873085, "step": 4928 }, { "epoch": 1.279595015576324, "grad_norm": 3.1015613079071045, "learning_rate": 1.8662902060838318e-06, "loss": 0.04540734738111496, "step": 4929 }, { "epoch": 1.2798546209761164, "grad_norm": 3.3322951793670654, "learning_rate": 1.8654202949928574e-06, "loss": 0.06529036909341812, "step": 4930 }, { "epoch": 1.2801142263759087, "grad_norm": 2.8638856410980225, "learning_rate": 1.8645505402753205e-06, "loss": 0.01664717122912407, "step": 4931 }, { "epoch": 1.280373831775701, "grad_norm": 3.1558563709259033, "learning_rate": 1.8636809420849745e-06, "loss": 0.07460133731365204, "step": 4932 }, { "epoch": 1.2806334371754933, "grad_norm": 3.0683186054229736, "learning_rate": 1.8628115005755432e-06, "loss": 0.030610959976911545, "step": 4933 }, { "epoch": 1.2808930425752856, "grad_norm": 3.3795478343963623, "learning_rate": 1.861942215900725e-06, "loss": 0.05558117479085922, "step": 4934 }, { "epoch": 1.2811526479750779, "grad_norm": 2.843385934829712, "learning_rate": 1.8610730882141893e-06, "loss": 0.03014654852449894, "step": 4935 }, { "epoch": 1.2814122533748702, "grad_norm": 2.713233709335327, "learning_rate": 1.8602041176695779e-06, "loss": 0.08135664463043213, "step": 4936 }, { "epoch": 1.2816718587746625, "grad_norm": 2.1558544635772705, "learning_rate": 1.8593353044205047e-06, "loss": 0.0914512574672699, "step": 4937 }, { "epoch": 1.2819314641744548, "grad_norm": 2.94927978515625, "learning_rate": 1.8584666486205567e-06, "loss": 0.07529295980930328, "step": 4938 }, { "epoch": 1.282191069574247, "grad_norm": 3.2361509799957275, "learning_rate": 1.857598150423291e-06, "loss": 0.016817297786474228, "step": 4939 }, { "epoch": 1.2824506749740394, "grad_norm": 5.768586158752441, "learning_rate": 1.8567298099822395e-06, "loss": 0.021715689450502396, "step": 4940 }, { "epoch": 1.2827102803738317, "grad_norm": 2.2124574184417725, "learning_rate": 1.8558616274509046e-06, "loss": 0.08155335485935211, "step": 4941 }, { "epoch": 1.282969885773624, "grad_norm": 3.454627752304077, "learning_rate": 1.8549936029827603e-06, "loss": 0.05204412341117859, "step": 4942 }, { "epoch": 1.2832294911734163, "grad_norm": 2.915825366973877, "learning_rate": 1.8541257367312543e-06, "loss": 0.08354248106479645, "step": 4943 }, { "epoch": 1.2834890965732086, "grad_norm": 2.517355442047119, "learning_rate": 1.8532580288498065e-06, "loss": 0.06976987421512604, "step": 4944 }, { "epoch": 1.2837487019730012, "grad_norm": 3.219465494155884, "learning_rate": 1.8523904794918046e-06, "loss": 0.07276017963886261, "step": 4945 }, { "epoch": 1.2840083073727935, "grad_norm": 3.340177297592163, "learning_rate": 1.851523088810614e-06, "loss": 0.019750386476516724, "step": 4946 }, { "epoch": 1.2842679127725858, "grad_norm": 3.5334312915802, "learning_rate": 1.850655856959569e-06, "loss": 0.02863927185535431, "step": 4947 }, { "epoch": 1.284527518172378, "grad_norm": 3.815887212753296, "learning_rate": 1.849788784091977e-06, "loss": 0.022969309240579605, "step": 4948 }, { "epoch": 1.2847871235721704, "grad_norm": 3.2700350284576416, "learning_rate": 1.848921870361115e-06, "loss": 0.061170756816864014, "step": 4949 }, { "epoch": 1.2850467289719627, "grad_norm": 3.5764338970184326, "learning_rate": 1.8480551159202342e-06, "loss": 0.0649617537856102, "step": 4950 }, { "epoch": 1.285306334371755, "grad_norm": 4.44542121887207, "learning_rate": 1.847188520922557e-06, "loss": 0.031983382999897, "step": 4951 }, { "epoch": 1.2855659397715473, "grad_norm": 3.013580799102783, "learning_rate": 1.8463220855212789e-06, "loss": 0.06461603194475174, "step": 4952 }, { "epoch": 1.2858255451713396, "grad_norm": 3.406494379043579, "learning_rate": 1.845455809869564e-06, "loss": 0.022456783801317215, "step": 4953 }, { "epoch": 1.286085150571132, "grad_norm": 2.779860734939575, "learning_rate": 1.8445896941205503e-06, "loss": 0.04369044303894043, "step": 4954 }, { "epoch": 1.2863447559709242, "grad_norm": 2.721144199371338, "learning_rate": 1.8437237384273471e-06, "loss": 0.0205202866345644, "step": 4955 }, { "epoch": 1.2866043613707165, "grad_norm": 3.5468857288360596, "learning_rate": 1.842857942943037e-06, "loss": 0.041295167058706284, "step": 4956 }, { "epoch": 1.2868639667705088, "grad_norm": 3.477872610092163, "learning_rate": 1.8419923078206714e-06, "loss": 0.038263607770204544, "step": 4957 }, { "epoch": 1.2871235721703012, "grad_norm": 3.9652931690216064, "learning_rate": 1.8411268332132747e-06, "loss": 0.04828566312789917, "step": 4958 }, { "epoch": 1.2873831775700935, "grad_norm": 2.957057476043701, "learning_rate": 1.8402615192738443e-06, "loss": 0.07621113210916519, "step": 4959 }, { "epoch": 1.2876427829698858, "grad_norm": 1.3905458450317383, "learning_rate": 1.8393963661553466e-06, "loss": 0.006637639366090298, "step": 4960 }, { "epoch": 1.287902388369678, "grad_norm": 2.564145565032959, "learning_rate": 1.8385313740107208e-06, "loss": 0.01502874493598938, "step": 4961 }, { "epoch": 1.2881619937694704, "grad_norm": 2.2068593502044678, "learning_rate": 1.8376665429928786e-06, "loss": 0.08424872159957886, "step": 4962 }, { "epoch": 1.2884215991692627, "grad_norm": 6.243385314941406, "learning_rate": 1.8368018732547026e-06, "loss": 0.057793017476797104, "step": 4963 }, { "epoch": 1.288681204569055, "grad_norm": 4.0126190185546875, "learning_rate": 1.8359373649490452e-06, "loss": 0.03350003808736801, "step": 4964 }, { "epoch": 1.2889408099688473, "grad_norm": 3.3738420009613037, "learning_rate": 1.8350730182287323e-06, "loss": 0.03788742050528526, "step": 4965 }, { "epoch": 1.2892004153686396, "grad_norm": 2.817436695098877, "learning_rate": 1.8342088332465612e-06, "loss": 0.03169320523738861, "step": 4966 }, { "epoch": 1.289460020768432, "grad_norm": 4.2616682052612305, "learning_rate": 1.8333448101552993e-06, "loss": 0.028541119769215584, "step": 4967 }, { "epoch": 1.2897196261682242, "grad_norm": 2.5049617290496826, "learning_rate": 1.8324809491076861e-06, "loss": 0.00828145444393158, "step": 4968 }, { "epoch": 1.2899792315680165, "grad_norm": 2.950672149658203, "learning_rate": 1.8316172502564334e-06, "loss": 0.043642401695251465, "step": 4969 }, { "epoch": 1.2902388369678088, "grad_norm": 2.0775136947631836, "learning_rate": 1.8307537137542219e-06, "loss": 0.07188089191913605, "step": 4970 }, { "epoch": 1.2904984423676011, "grad_norm": 2.660585641860962, "learning_rate": 1.829890339753706e-06, "loss": 0.01313313189893961, "step": 4971 }, { "epoch": 1.2907580477673934, "grad_norm": 3.2832181453704834, "learning_rate": 1.8290271284075106e-06, "loss": 0.036368533968925476, "step": 4972 }, { "epoch": 1.2910176531671858, "grad_norm": 2.515780448913574, "learning_rate": 1.8281640798682316e-06, "loss": 0.020703120157122612, "step": 4973 }, { "epoch": 1.291277258566978, "grad_norm": 2.423645496368408, "learning_rate": 1.8273011942884356e-06, "loss": 0.06784627586603165, "step": 4974 }, { "epoch": 1.2915368639667706, "grad_norm": 2.853278875350952, "learning_rate": 1.8264384718206624e-06, "loss": 0.04680345207452774, "step": 4975 }, { "epoch": 1.291796469366563, "grad_norm": 2.847930908203125, "learning_rate": 1.8255759126174188e-06, "loss": 0.02131059393286705, "step": 4976 }, { "epoch": 1.2920560747663552, "grad_norm": 3.0123512744903564, "learning_rate": 1.8247135168311885e-06, "loss": 0.014355268329381943, "step": 4977 }, { "epoch": 1.2923156801661475, "grad_norm": 3.123236656188965, "learning_rate": 1.823851284614422e-06, "loss": 0.030687399208545685, "step": 4978 }, { "epoch": 1.2925752855659398, "grad_norm": 3.211003541946411, "learning_rate": 1.8229892161195435e-06, "loss": 0.036740563809871674, "step": 4979 }, { "epoch": 1.2928348909657321, "grad_norm": 1.8778043985366821, "learning_rate": 1.8221273114989447e-06, "loss": 0.0820692926645279, "step": 4980 }, { "epoch": 1.2930944963655244, "grad_norm": 2.265667200088501, "learning_rate": 1.8212655709049908e-06, "loss": 0.03259197622537613, "step": 4981 }, { "epoch": 1.2933541017653167, "grad_norm": 2.5901618003845215, "learning_rate": 1.8204039944900198e-06, "loss": 0.06740175932645798, "step": 4982 }, { "epoch": 1.293613707165109, "grad_norm": 2.4867374897003174, "learning_rate": 1.8195425824063384e-06, "loss": 0.06984198838472366, "step": 4983 }, { "epoch": 1.2938733125649013, "grad_norm": 2.3088724613189697, "learning_rate": 1.8186813348062232e-06, "loss": 0.07225284725427628, "step": 4984 }, { "epoch": 1.2941329179646937, "grad_norm": 3.1302778720855713, "learning_rate": 1.817820251841923e-06, "loss": 0.030308062210679054, "step": 4985 }, { "epoch": 1.294392523364486, "grad_norm": 2.3547184467315674, "learning_rate": 1.8169593336656594e-06, "loss": 0.09683901071548462, "step": 4986 }, { "epoch": 1.2946521287642783, "grad_norm": 4.229942321777344, "learning_rate": 1.816098580429621e-06, "loss": 0.051422908902168274, "step": 4987 }, { "epoch": 1.2949117341640706, "grad_norm": 3.6271448135375977, "learning_rate": 1.8152379922859697e-06, "loss": 0.0431194007396698, "step": 4988 }, { "epoch": 1.2951713395638629, "grad_norm": 5.792318820953369, "learning_rate": 1.8143775693868387e-06, "loss": 0.01892990991473198, "step": 4989 }, { "epoch": 1.2954309449636552, "grad_norm": 4.184528827667236, "learning_rate": 1.8135173118843313e-06, "loss": 0.03861721605062485, "step": 4990 }, { "epoch": 1.2956905503634475, "grad_norm": 2.8124616146087646, "learning_rate": 1.8126572199305192e-06, "loss": 0.03322313353419304, "step": 4991 }, { "epoch": 1.2959501557632398, "grad_norm": 3.2039034366607666, "learning_rate": 1.811797293677448e-06, "loss": 0.06683096289634705, "step": 4992 }, { "epoch": 1.2962097611630323, "grad_norm": 3.1897175312042236, "learning_rate": 1.8109375332771346e-06, "loss": 0.018981698900461197, "step": 4993 }, { "epoch": 1.2964693665628246, "grad_norm": 3.0742814540863037, "learning_rate": 1.8100779388815631e-06, "loss": 0.06625348329544067, "step": 4994 }, { "epoch": 1.296728971962617, "grad_norm": 2.7624335289001465, "learning_rate": 1.8092185106426913e-06, "loss": 0.06832879781723022, "step": 4995 }, { "epoch": 1.2969885773624092, "grad_norm": 2.5406243801116943, "learning_rate": 1.808359248712444e-06, "loss": 0.057906374335289, "step": 4996 }, { "epoch": 1.2972481827622016, "grad_norm": 2.8447959423065186, "learning_rate": 1.8075001532427216e-06, "loss": 0.06702421605587006, "step": 4997 }, { "epoch": 1.2975077881619939, "grad_norm": 2.1401615142822266, "learning_rate": 1.806641224385393e-06, "loss": 0.08238525688648224, "step": 4998 }, { "epoch": 1.2977673935617862, "grad_norm": 3.4194114208221436, "learning_rate": 1.8057824622922948e-06, "loss": 0.026298774406313896, "step": 4999 }, { "epoch": 1.2980269989615785, "grad_norm": 3.6424386501312256, "learning_rate": 1.8049238671152381e-06, "loss": 0.02281968481838703, "step": 5000 }, { "epoch": 1.2982866043613708, "grad_norm": 4.436131954193115, "learning_rate": 1.804065439006001e-06, "loss": 0.036425378173589706, "step": 5001 }, { "epoch": 1.298546209761163, "grad_norm": 3.701683759689331, "learning_rate": 1.8032071781163363e-06, "loss": 0.028067901730537415, "step": 5002 }, { "epoch": 1.2988058151609554, "grad_norm": 3.078381299972534, "learning_rate": 1.802349084597964e-06, "loss": 0.023032303899526596, "step": 5003 }, { "epoch": 1.2990654205607477, "grad_norm": 3.315673351287842, "learning_rate": 1.8014911586025747e-06, "loss": 0.0579710379242897, "step": 5004 }, { "epoch": 1.29932502596054, "grad_norm": 3.3777692317962646, "learning_rate": 1.800633400281831e-06, "loss": 0.057802505791187286, "step": 5005 }, { "epoch": 1.2995846313603323, "grad_norm": 3.0296218395233154, "learning_rate": 1.7997758097873647e-06, "loss": 0.020546913146972656, "step": 5006 }, { "epoch": 1.2998442367601246, "grad_norm": 4.949371337890625, "learning_rate": 1.7989183872707767e-06, "loss": 0.04375538229942322, "step": 5007 }, { "epoch": 1.300103842159917, "grad_norm": 2.896975517272949, "learning_rate": 1.7980611328836422e-06, "loss": 0.021163906902074814, "step": 5008 }, { "epoch": 1.3003634475597092, "grad_norm": 3.8172850608825684, "learning_rate": 1.7972040467775021e-06, "loss": 0.0459519699215889, "step": 5009 }, { "epoch": 1.3006230529595015, "grad_norm": 3.594130754470825, "learning_rate": 1.7963471291038708e-06, "loss": 0.05494973808526993, "step": 5010 }, { "epoch": 1.3008826583592938, "grad_norm": 2.434234380722046, "learning_rate": 1.7954903800142321e-06, "loss": 0.03113619238138199, "step": 5011 }, { "epoch": 1.3011422637590861, "grad_norm": 3.9876744747161865, "learning_rate": 1.7946337996600371e-06, "loss": 0.03330384939908981, "step": 5012 }, { "epoch": 1.3014018691588785, "grad_norm": 4.13045072555542, "learning_rate": 1.793777388192713e-06, "loss": 0.05107324197888374, "step": 5013 }, { "epoch": 1.3016614745586708, "grad_norm": 2.7816877365112305, "learning_rate": 1.7929211457636514e-06, "loss": 0.057532697916030884, "step": 5014 }, { "epoch": 1.301921079958463, "grad_norm": 2.113116502761841, "learning_rate": 1.7920650725242167e-06, "loss": 0.08321012556552887, "step": 5015 }, { "epoch": 1.3021806853582554, "grad_norm": 2.4915108680725098, "learning_rate": 1.7912091686257438e-06, "loss": 0.07657743990421295, "step": 5016 }, { "epoch": 1.3024402907580477, "grad_norm": 3.149010181427002, "learning_rate": 1.7903534342195366e-06, "loss": 0.04703246057033539, "step": 5017 }, { "epoch": 1.30269989615784, "grad_norm": 4.038452625274658, "learning_rate": 1.7894978694568688e-06, "loss": 0.011974048800766468, "step": 5018 }, { "epoch": 1.3029595015576323, "grad_norm": 3.126469135284424, "learning_rate": 1.7886424744889852e-06, "loss": 0.025219572708010674, "step": 5019 }, { "epoch": 1.3032191069574246, "grad_norm": 2.5284509658813477, "learning_rate": 1.7877872494670996e-06, "loss": 0.019744573161005974, "step": 5020 }, { "epoch": 1.303478712357217, "grad_norm": 2.363715410232544, "learning_rate": 1.7869321945423985e-06, "loss": 0.010493336245417595, "step": 5021 }, { "epoch": 1.3037383177570092, "grad_norm": 3.531646728515625, "learning_rate": 1.7860773098660316e-06, "loss": 0.032365381717681885, "step": 5022 }, { "epoch": 1.3039979231568017, "grad_norm": 3.087256908416748, "learning_rate": 1.7852225955891256e-06, "loss": 0.059393513947725296, "step": 5023 }, { "epoch": 1.304257528556594, "grad_norm": 3.5219571590423584, "learning_rate": 1.7843680518627753e-06, "loss": 0.021186979487538338, "step": 5024 }, { "epoch": 1.3045171339563864, "grad_norm": 2.8376944065093994, "learning_rate": 1.7835136788380436e-06, "loss": 0.07005595415830612, "step": 5025 }, { "epoch": 1.3047767393561787, "grad_norm": 3.4416871070861816, "learning_rate": 1.7826594766659638e-06, "loss": 0.04769100248813629, "step": 5026 }, { "epoch": 1.305036344755971, "grad_norm": 3.7169294357299805, "learning_rate": 1.7818054454975387e-06, "loss": 0.01372448354959488, "step": 5027 }, { "epoch": 1.3052959501557633, "grad_norm": 3.5852487087249756, "learning_rate": 1.7809515854837418e-06, "loss": 0.035595186054706573, "step": 5028 }, { "epoch": 1.3055555555555556, "grad_norm": 3.0256845951080322, "learning_rate": 1.7800978967755179e-06, "loss": 0.04176686331629753, "step": 5029 }, { "epoch": 1.3058151609553479, "grad_norm": 2.3206772804260254, "learning_rate": 1.7792443795237764e-06, "loss": 0.07903486490249634, "step": 5030 }, { "epoch": 1.3060747663551402, "grad_norm": 3.7805051803588867, "learning_rate": 1.7783910338794019e-06, "loss": 0.06780779361724854, "step": 5031 }, { "epoch": 1.3063343717549325, "grad_norm": 2.648557662963867, "learning_rate": 1.7775378599932452e-06, "loss": 0.05287407711148262, "step": 5032 }, { "epoch": 1.3065939771547248, "grad_norm": 4.021817684173584, "learning_rate": 1.776684858016129e-06, "loss": 0.014298487454652786, "step": 5033 }, { "epoch": 1.3068535825545171, "grad_norm": 2.9848804473876953, "learning_rate": 1.7758320280988444e-06, "loss": 0.0526699498295784, "step": 5034 }, { "epoch": 1.3071131879543094, "grad_norm": 2.6905829906463623, "learning_rate": 1.774979370392151e-06, "loss": 0.00990389846265316, "step": 5035 }, { "epoch": 1.3073727933541017, "grad_norm": 1.9660892486572266, "learning_rate": 1.7741268850467796e-06, "loss": 0.0887647420167923, "step": 5036 }, { "epoch": 1.307632398753894, "grad_norm": 2.0137388706207275, "learning_rate": 1.7732745722134315e-06, "loss": 0.019965581595897675, "step": 5037 }, { "epoch": 1.3078920041536863, "grad_norm": 2.600766181945801, "learning_rate": 1.7724224320427741e-06, "loss": 0.06432478129863739, "step": 5038 }, { "epoch": 1.3081516095534786, "grad_norm": 4.025595664978027, "learning_rate": 1.7715704646854476e-06, "loss": 0.040486253798007965, "step": 5039 }, { "epoch": 1.308411214953271, "grad_norm": 4.5985307693481445, "learning_rate": 1.7707186702920587e-06, "loss": 0.02144092321395874, "step": 5040 }, { "epoch": 1.3086708203530635, "grad_norm": 3.2209770679473877, "learning_rate": 1.7698670490131875e-06, "loss": 0.03681732714176178, "step": 5041 }, { "epoch": 1.3089304257528558, "grad_norm": 2.9235148429870605, "learning_rate": 1.7690156009993803e-06, "loss": 0.02297649160027504, "step": 5042 }, { "epoch": 1.309190031152648, "grad_norm": 3.6052439212799072, "learning_rate": 1.7681643264011523e-06, "loss": 0.02030140347778797, "step": 5043 }, { "epoch": 1.3094496365524404, "grad_norm": 3.3578929901123047, "learning_rate": 1.7673132253689913e-06, "loss": 0.015400845557451248, "step": 5044 }, { "epoch": 1.3097092419522327, "grad_norm": 3.4059789180755615, "learning_rate": 1.7664622980533508e-06, "loss": 0.0650663673877716, "step": 5045 }, { "epoch": 1.309968847352025, "grad_norm": 3.3737220764160156, "learning_rate": 1.7656115446046558e-06, "loss": 0.016249487176537514, "step": 5046 }, { "epoch": 1.3102284527518173, "grad_norm": 3.1892664432525635, "learning_rate": 1.764760965173301e-06, "loss": 0.04650064557790756, "step": 5047 }, { "epoch": 1.3104880581516096, "grad_norm": 3.5296175479888916, "learning_rate": 1.7639105599096484e-06, "loss": 0.04947546869516373, "step": 5048 }, { "epoch": 1.310747663551402, "grad_norm": 3.9688446521759033, "learning_rate": 1.7630603289640298e-06, "loss": 0.02734941989183426, "step": 5049 }, { "epoch": 1.3110072689511942, "grad_norm": 3.7536706924438477, "learning_rate": 1.7622102724867473e-06, "loss": 0.025021320208907127, "step": 5050 }, { "epoch": 1.3112668743509865, "grad_norm": 3.0415568351745605, "learning_rate": 1.7613603906280708e-06, "loss": 0.04064884036779404, "step": 5051 }, { "epoch": 1.3115264797507789, "grad_norm": 5.286158561706543, "learning_rate": 1.7605106835382424e-06, "loss": 0.01787458173930645, "step": 5052 }, { "epoch": 1.3117860851505712, "grad_norm": 5.220255374908447, "learning_rate": 1.7596611513674667e-06, "loss": 0.047674812376499176, "step": 5053 }, { "epoch": 1.3120456905503635, "grad_norm": 2.9339516162872314, "learning_rate": 1.7588117942659239e-06, "loss": 0.06630728393793106, "step": 5054 }, { "epoch": 1.3123052959501558, "grad_norm": 2.8300929069519043, "learning_rate": 1.7579626123837612e-06, "loss": 0.0655001699924469, "step": 5055 }, { "epoch": 1.312564901349948, "grad_norm": 2.063500165939331, "learning_rate": 1.7571136058710942e-06, "loss": 0.07977773249149323, "step": 5056 }, { "epoch": 1.3128245067497404, "grad_norm": 1.5404354333877563, "learning_rate": 1.7562647748780076e-06, "loss": 0.097772516310215, "step": 5057 }, { "epoch": 1.3130841121495327, "grad_norm": 2.8463618755340576, "learning_rate": 1.7554161195545545e-06, "loss": 0.049584779888391495, "step": 5058 }, { "epoch": 1.313343717549325, "grad_norm": 3.7109317779541016, "learning_rate": 1.754567640050759e-06, "loss": 0.022833239287137985, "step": 5059 }, { "epoch": 1.3136033229491173, "grad_norm": 1.756618857383728, "learning_rate": 1.7537193365166133e-06, "loss": 0.0907064899802208, "step": 5060 }, { "epoch": 1.3138629283489096, "grad_norm": 2.416555881500244, "learning_rate": 1.7528712091020756e-06, "loss": 0.01779431477189064, "step": 5061 }, { "epoch": 1.314122533748702, "grad_norm": 2.843947649002075, "learning_rate": 1.7520232579570768e-06, "loss": 0.044010892510414124, "step": 5062 }, { "epoch": 1.3143821391484942, "grad_norm": 2.44077205657959, "learning_rate": 1.7511754832315165e-06, "loss": 0.0546024851500988, "step": 5063 }, { "epoch": 1.3146417445482865, "grad_norm": 3.6872336864471436, "learning_rate": 1.7503278850752604e-06, "loss": 0.03654859587550163, "step": 5064 }, { "epoch": 1.3149013499480788, "grad_norm": 1.9863330125808716, "learning_rate": 1.7494804636381446e-06, "loss": 0.07795984297990799, "step": 5065 }, { "epoch": 1.3151609553478711, "grad_norm": 2.8782522678375244, "learning_rate": 1.748633219069974e-06, "loss": 0.02427593432366848, "step": 5066 }, { "epoch": 1.3154205607476634, "grad_norm": 2.8847970962524414, "learning_rate": 1.7477861515205213e-06, "loss": 0.05565240979194641, "step": 5067 }, { "epoch": 1.3156801661474558, "grad_norm": 3.417431354522705, "learning_rate": 1.7469392611395309e-06, "loss": 0.026256822049617767, "step": 5068 }, { "epoch": 1.315939771547248, "grad_norm": 2.2919349670410156, "learning_rate": 1.7460925480767107e-06, "loss": 0.015704048797488213, "step": 5069 }, { "epoch": 1.3161993769470404, "grad_norm": 3.7077722549438477, "learning_rate": 1.7452460124817424e-06, "loss": 0.04681278020143509, "step": 5070 }, { "epoch": 1.316458982346833, "grad_norm": 3.1926283836364746, "learning_rate": 1.7443996545042734e-06, "loss": 0.06049932539463043, "step": 5071 }, { "epoch": 1.3167185877466252, "grad_norm": 3.0968515872955322, "learning_rate": 1.7435534742939193e-06, "loss": 0.027058003470301628, "step": 5072 }, { "epoch": 1.3169781931464175, "grad_norm": 2.7949764728546143, "learning_rate": 1.7427074720002673e-06, "loss": 0.06241659075021744, "step": 5073 }, { "epoch": 1.3172377985462098, "grad_norm": 3.3248848915100098, "learning_rate": 1.7418616477728694e-06, "loss": 0.04876895248889923, "step": 5074 }, { "epoch": 1.3174974039460021, "grad_norm": 3.1694483757019043, "learning_rate": 1.74101600176125e-06, "loss": 0.04862051457166672, "step": 5075 }, { "epoch": 1.3177570093457944, "grad_norm": 3.183504581451416, "learning_rate": 1.740170534114898e-06, "loss": 0.03133748844265938, "step": 5076 }, { "epoch": 1.3180166147455867, "grad_norm": 3.3678863048553467, "learning_rate": 1.7393252449832734e-06, "loss": 0.029667843133211136, "step": 5077 }, { "epoch": 1.318276220145379, "grad_norm": 2.836073398590088, "learning_rate": 1.7384801345158048e-06, "loss": 0.044806741178035736, "step": 5078 }, { "epoch": 1.3185358255451713, "grad_norm": 2.7168614864349365, "learning_rate": 1.7376352028618877e-06, "loss": 0.013524316251277924, "step": 5079 }, { "epoch": 1.3187954309449637, "grad_norm": 1.7934929132461548, "learning_rate": 1.7367904501708856e-06, "loss": 0.07856844365596771, "step": 5080 }, { "epoch": 1.319055036344756, "grad_norm": 3.8451573848724365, "learning_rate": 1.7359458765921334e-06, "loss": 0.030296064913272858, "step": 5081 }, { "epoch": 1.3193146417445483, "grad_norm": 2.3989579677581787, "learning_rate": 1.7351014822749308e-06, "loss": 0.05408741906285286, "step": 5082 }, { "epoch": 1.3195742471443406, "grad_norm": 2.347623586654663, "learning_rate": 1.734257267368549e-06, "loss": 0.006875915452837944, "step": 5083 }, { "epoch": 1.3198338525441329, "grad_norm": 3.8896701335906982, "learning_rate": 1.7334132320222244e-06, "loss": 0.03597857803106308, "step": 5084 }, { "epoch": 1.3200934579439252, "grad_norm": 3.4824795722961426, "learning_rate": 1.732569376385163e-06, "loss": 0.04418565332889557, "step": 5085 }, { "epoch": 1.3203530633437175, "grad_norm": 3.010624408721924, "learning_rate": 1.7317257006065407e-06, "loss": 0.05244050174951553, "step": 5086 }, { "epoch": 1.3206126687435098, "grad_norm": 2.5538759231567383, "learning_rate": 1.7308822048354996e-06, "loss": 0.016208652406930923, "step": 5087 }, { "epoch": 1.320872274143302, "grad_norm": 2.208956718444824, "learning_rate": 1.7300388892211483e-06, "loss": 0.09831664711236954, "step": 5088 }, { "epoch": 1.3211318795430946, "grad_norm": 2.137234687805176, "learning_rate": 1.7291957539125686e-06, "loss": 0.07859758287668228, "step": 5089 }, { "epoch": 1.321391484942887, "grad_norm": 3.1716418266296387, "learning_rate": 1.7283527990588055e-06, "loss": 0.04358731210231781, "step": 5090 }, { "epoch": 1.3216510903426792, "grad_norm": 2.416213035583496, "learning_rate": 1.7275100248088766e-06, "loss": 0.011227241717278957, "step": 5091 }, { "epoch": 1.3219106957424716, "grad_norm": 3.852548122406006, "learning_rate": 1.7266674313117615e-06, "loss": 0.018349183723330498, "step": 5092 }, { "epoch": 1.3221703011422639, "grad_norm": 2.1017234325408936, "learning_rate": 1.7258250187164139e-06, "loss": 0.03553561866283417, "step": 5093 }, { "epoch": 1.3224299065420562, "grad_norm": 2.2780654430389404, "learning_rate": 1.7249827871717526e-06, "loss": 0.02490193024277687, "step": 5094 }, { "epoch": 1.3226895119418485, "grad_norm": 4.17454195022583, "learning_rate": 1.7241407368266649e-06, "loss": 0.035954855382442474, "step": 5095 }, { "epoch": 1.3229491173416408, "grad_norm": 1.7780739068984985, "learning_rate": 1.7232988678300062e-06, "loss": 0.08563972264528275, "step": 5096 }, { "epoch": 1.323208722741433, "grad_norm": 2.6943702697753906, "learning_rate": 1.7224571803305984e-06, "loss": 0.049721501767635345, "step": 5097 }, { "epoch": 1.3234683281412254, "grad_norm": 1.8175301551818848, "learning_rate": 1.721615674477234e-06, "loss": 0.0751522034406662, "step": 5098 }, { "epoch": 1.3237279335410177, "grad_norm": 1.687609314918518, "learning_rate": 1.720774350418672e-06, "loss": 0.07412709295749664, "step": 5099 }, { "epoch": 1.32398753894081, "grad_norm": 2.7952075004577637, "learning_rate": 1.7199332083036374e-06, "loss": 0.056221894919872284, "step": 5100 }, { "epoch": 1.3242471443406023, "grad_norm": 1.0408380031585693, "learning_rate": 1.719092248280827e-06, "loss": 0.002444813959300518, "step": 5101 }, { "epoch": 1.3245067497403946, "grad_norm": 2.7281765937805176, "learning_rate": 1.7182514704989029e-06, "loss": 0.06575191766023636, "step": 5102 }, { "epoch": 1.324766355140187, "grad_norm": 3.227531909942627, "learning_rate": 1.7174108751064939e-06, "loss": 0.02553911879658699, "step": 5103 }, { "epoch": 1.3250259605399792, "grad_norm": 1.9098668098449707, "learning_rate": 1.7165704622521994e-06, "loss": 0.077634297311306, "step": 5104 }, { "epoch": 1.3252855659397715, "grad_norm": 2.4896483421325684, "learning_rate": 1.7157302320845848e-06, "loss": 0.07488807290792465, "step": 5105 }, { "epoch": 1.3255451713395638, "grad_norm": 2.870185375213623, "learning_rate": 1.7148901847521837e-06, "loss": 0.034321170300245285, "step": 5106 }, { "epoch": 1.3258047767393562, "grad_norm": 2.2661280632019043, "learning_rate": 1.7140503204034973e-06, "loss": 0.07721766829490662, "step": 5107 }, { "epoch": 1.3260643821391485, "grad_norm": 2.6436564922332764, "learning_rate": 1.713210639186993e-06, "loss": 0.0647699385881424, "step": 5108 }, { "epoch": 1.3263239875389408, "grad_norm": 3.2890920639038086, "learning_rate": 1.7123711412511096e-06, "loss": 0.029419396072626114, "step": 5109 }, { "epoch": 1.326583592938733, "grad_norm": 3.2512428760528564, "learning_rate": 1.7115318267442496e-06, "loss": 0.06715287268161774, "step": 5110 }, { "epoch": 1.3268431983385254, "grad_norm": 2.8392271995544434, "learning_rate": 1.7106926958147841e-06, "loss": 0.03500286862254143, "step": 5111 }, { "epoch": 1.3271028037383177, "grad_norm": 1.815352439880371, "learning_rate": 1.7098537486110539e-06, "loss": 0.08030281960964203, "step": 5112 }, { "epoch": 1.32736240913811, "grad_norm": 3.4147229194641113, "learning_rate": 1.709014985281364e-06, "loss": 0.033808961510658264, "step": 5113 }, { "epoch": 1.3276220145379023, "grad_norm": 2.468648910522461, "learning_rate": 1.70817640597399e-06, "loss": 0.08214739710092545, "step": 5114 }, { "epoch": 1.3278816199376946, "grad_norm": 2.786073684692383, "learning_rate": 1.7073380108371728e-06, "loss": 0.062048155814409256, "step": 5115 }, { "epoch": 1.328141225337487, "grad_norm": 3.368312120437622, "learning_rate": 1.7064998000191207e-06, "loss": 0.027994023635983467, "step": 5116 }, { "epoch": 1.3284008307372792, "grad_norm": 2.157836675643921, "learning_rate": 1.705661773668012e-06, "loss": 0.06318862736225128, "step": 5117 }, { "epoch": 1.3286604361370715, "grad_norm": 3.3187711238861084, "learning_rate": 1.7048239319319894e-06, "loss": 0.03242437541484833, "step": 5118 }, { "epoch": 1.328920041536864, "grad_norm": 3.8105766773223877, "learning_rate": 1.7039862749591635e-06, "loss": 0.033079344779253006, "step": 5119 }, { "epoch": 1.3291796469366564, "grad_norm": 3.2692344188690186, "learning_rate": 1.703148802897614e-06, "loss": 0.07336148619651794, "step": 5120 }, { "epoch": 1.3294392523364487, "grad_norm": 3.0280580520629883, "learning_rate": 1.7023115158953862e-06, "loss": 0.05622216686606407, "step": 5121 }, { "epoch": 1.329698857736241, "grad_norm": 2.894537925720215, "learning_rate": 1.7014744141004948e-06, "loss": 0.013667307794094086, "step": 5122 }, { "epoch": 1.3299584631360333, "grad_norm": 2.927985191345215, "learning_rate": 1.7006374976609174e-06, "loss": 0.008826844394207, "step": 5123 }, { "epoch": 1.3302180685358256, "grad_norm": 1.7201396226882935, "learning_rate": 1.6998007667246037e-06, "loss": 0.08557968586683273, "step": 5124 }, { "epoch": 1.3304776739356179, "grad_norm": 2.2463927268981934, "learning_rate": 1.698964221439469e-06, "loss": 0.061770085245370865, "step": 5125 }, { "epoch": 1.3307372793354102, "grad_norm": 3.5589921474456787, "learning_rate": 1.698127861953393e-06, "loss": 0.06907764077186584, "step": 5126 }, { "epoch": 1.3309968847352025, "grad_norm": 2.8414812088012695, "learning_rate": 1.697291688414227e-06, "loss": 0.047978438436985016, "step": 5127 }, { "epoch": 1.3312564901349948, "grad_norm": 2.783205509185791, "learning_rate": 1.6964557009697858e-06, "loss": 0.021250877529382706, "step": 5128 }, { "epoch": 1.3315160955347871, "grad_norm": 3.683664560317993, "learning_rate": 1.6956198997678546e-06, "loss": 0.056524112820625305, "step": 5129 }, { "epoch": 1.3317757009345794, "grad_norm": 3.3483505249023438, "learning_rate": 1.6947842849561835e-06, "loss": 0.019067851826548576, "step": 5130 }, { "epoch": 1.3320353063343717, "grad_norm": 3.7076430320739746, "learning_rate": 1.6939488566824885e-06, "loss": 0.04905059561133385, "step": 5131 }, { "epoch": 1.332294911734164, "grad_norm": 5.360375881195068, "learning_rate": 1.6931136150944561e-06, "loss": 0.017835646867752075, "step": 5132 }, { "epoch": 1.3325545171339563, "grad_norm": 3.636131525039673, "learning_rate": 1.692278560339737e-06, "loss": 0.013578365556895733, "step": 5133 }, { "epoch": 1.3328141225337486, "grad_norm": 3.652322292327881, "learning_rate": 1.6914436925659488e-06, "loss": 0.014197695069015026, "step": 5134 }, { "epoch": 1.333073727933541, "grad_norm": 2.28391695022583, "learning_rate": 1.6906090119206792e-06, "loss": 0.08055491745471954, "step": 5135 }, { "epoch": 1.3333333333333333, "grad_norm": 3.476743221282959, "learning_rate": 1.689774518551478e-06, "loss": 0.06061220541596413, "step": 5136 }, { "epoch": 1.3335929387331256, "grad_norm": 2.8565096855163574, "learning_rate": 1.6889402126058674e-06, "loss": 0.06235573813319206, "step": 5137 }, { "epoch": 1.333852544132918, "grad_norm": 2.2718968391418457, "learning_rate": 1.6881060942313322e-06, "loss": 0.015224728733301163, "step": 5138 }, { "epoch": 1.3341121495327104, "grad_norm": 3.5222079753875732, "learning_rate": 1.687272163575324e-06, "loss": 0.057731591165065765, "step": 5139 }, { "epoch": 1.3343717549325027, "grad_norm": 4.09586763381958, "learning_rate": 1.6864384207852655e-06, "loss": 0.05141026899218559, "step": 5140 }, { "epoch": 1.334631360332295, "grad_norm": 3.619034767150879, "learning_rate": 1.6856048660085418e-06, "loss": 0.053848374634981155, "step": 5141 }, { "epoch": 1.3348909657320873, "grad_norm": 2.1725332736968994, "learning_rate": 1.6847714993925051e-06, "loss": 0.020651856437325478, "step": 5142 }, { "epoch": 1.3351505711318796, "grad_norm": 2.8801701068878174, "learning_rate": 1.6839383210844779e-06, "loss": 0.04916530102491379, "step": 5143 }, { "epoch": 1.335410176531672, "grad_norm": 2.4924352169036865, "learning_rate": 1.683105331231745e-06, "loss": 0.057635169476270676, "step": 5144 }, { "epoch": 1.3356697819314642, "grad_norm": 2.6670165061950684, "learning_rate": 1.6822725299815618e-06, "loss": 0.05365944653749466, "step": 5145 }, { "epoch": 1.3359293873312565, "grad_norm": 3.930433511734009, "learning_rate": 1.6814399174811475e-06, "loss": 0.012457028962671757, "step": 5146 }, { "epoch": 1.3361889927310489, "grad_norm": 1.9045614004135132, "learning_rate": 1.6806074938776884e-06, "loss": 0.08150210976600647, "step": 5147 }, { "epoch": 1.3364485981308412, "grad_norm": 2.8113300800323486, "learning_rate": 1.6797752593183393e-06, "loss": 0.016734734177589417, "step": 5148 }, { "epoch": 1.3367082035306335, "grad_norm": 0.481524795293808, "learning_rate": 1.6789432139502192e-06, "loss": 0.0020107561722397804, "step": 5149 }, { "epoch": 1.3369678089304258, "grad_norm": 3.964550018310547, "learning_rate": 1.6781113579204145e-06, "loss": 0.07902558147907257, "step": 5150 }, { "epoch": 1.337227414330218, "grad_norm": 4.401334762573242, "learning_rate": 1.6772796913759792e-06, "loss": 0.039408281445503235, "step": 5151 }, { "epoch": 1.3374870197300104, "grad_norm": 3.9660873413085938, "learning_rate": 1.6764482144639326e-06, "loss": 0.018693383783102036, "step": 5152 }, { "epoch": 1.3377466251298027, "grad_norm": 3.242215156555176, "learning_rate": 1.6756169273312606e-06, "loss": 0.019244752824306488, "step": 5153 }, { "epoch": 1.338006230529595, "grad_norm": 3.7431230545043945, "learning_rate": 1.6747858301249148e-06, "loss": 0.025557206943631172, "step": 5154 }, { "epoch": 1.3382658359293873, "grad_norm": 2.4031591415405273, "learning_rate": 1.6739549229918154e-06, "loss": 0.07299382984638214, "step": 5155 }, { "epoch": 1.3385254413291796, "grad_norm": 3.098736047744751, "learning_rate": 1.6731242060788489e-06, "loss": 0.020304707810282707, "step": 5156 }, { "epoch": 1.338785046728972, "grad_norm": 3.3457021713256836, "learning_rate": 1.6722936795328637e-06, "loss": 0.024673480540513992, "step": 5157 }, { "epoch": 1.3390446521287642, "grad_norm": 4.332711696624756, "learning_rate": 1.671463343500681e-06, "loss": 0.030013086274266243, "step": 5158 }, { "epoch": 1.3393042575285565, "grad_norm": 2.076504707336426, "learning_rate": 1.670633198129083e-06, "loss": 0.09470918774604797, "step": 5159 }, { "epoch": 1.3395638629283488, "grad_norm": 4.782994747161865, "learning_rate": 1.669803243564822e-06, "loss": 0.018526609987020493, "step": 5160 }, { "epoch": 1.3398234683281411, "grad_norm": 3.046804189682007, "learning_rate": 1.6689734799546145e-06, "loss": 0.02193407341837883, "step": 5161 }, { "epoch": 1.3400830737279334, "grad_norm": 3.8853626251220703, "learning_rate": 1.6681439074451424e-06, "loss": 0.03325625881552696, "step": 5162 }, { "epoch": 1.3403426791277258, "grad_norm": 2.9683566093444824, "learning_rate": 1.6673145261830577e-06, "loss": 0.0476653128862381, "step": 5163 }, { "epoch": 1.340602284527518, "grad_norm": 3.6392617225646973, "learning_rate": 1.6664853363149739e-06, "loss": 0.02134581282734871, "step": 5164 }, { "epoch": 1.3408618899273104, "grad_norm": 1.5036990642547607, "learning_rate": 1.6656563379874724e-06, "loss": 0.0038465477991849184, "step": 5165 }, { "epoch": 1.3411214953271027, "grad_norm": 2.8764662742614746, "learning_rate": 1.6648275313471039e-06, "loss": 0.01741502806544304, "step": 5166 }, { "epoch": 1.341381100726895, "grad_norm": 2.4997994899749756, "learning_rate": 1.6639989165403794e-06, "loss": 0.0689658522605896, "step": 5167 }, { "epoch": 1.3416407061266875, "grad_norm": 2.9174768924713135, "learning_rate": 1.6631704937137811e-06, "loss": 0.04456786811351776, "step": 5168 }, { "epoch": 1.3419003115264798, "grad_norm": 3.4269886016845703, "learning_rate": 1.6623422630137542e-06, "loss": 0.01795257069170475, "step": 5169 }, { "epoch": 1.3421599169262721, "grad_norm": 1.7118364572525024, "learning_rate": 1.6615142245867107e-06, "loss": 0.08670338988304138, "step": 5170 }, { "epoch": 1.3424195223260644, "grad_norm": 2.3635783195495605, "learning_rate": 1.66068637857903e-06, "loss": 0.06357835978269577, "step": 5171 }, { "epoch": 1.3426791277258567, "grad_norm": 2.5482773780822754, "learning_rate": 1.6598587251370557e-06, "loss": 0.05459996312856674, "step": 5172 }, { "epoch": 1.342938733125649, "grad_norm": 2.38144588470459, "learning_rate": 1.659031264407097e-06, "loss": 0.013304618187248707, "step": 5173 }, { "epoch": 1.3431983385254413, "grad_norm": 2.8081464767456055, "learning_rate": 1.6582039965354313e-06, "loss": 0.019010845571756363, "step": 5174 }, { "epoch": 1.3434579439252337, "grad_norm": 3.702463388442993, "learning_rate": 1.6573769216683e-06, "loss": 0.030609840527176857, "step": 5175 }, { "epoch": 1.343717549325026, "grad_norm": 3.45302677154541, "learning_rate": 1.6565500399519116e-06, "loss": 0.0324622243642807, "step": 5176 }, { "epoch": 1.3439771547248183, "grad_norm": 3.0871875286102295, "learning_rate": 1.6557233515324398e-06, "loss": 0.03553486615419388, "step": 5177 }, { "epoch": 1.3442367601246106, "grad_norm": 2.951120376586914, "learning_rate": 1.6548968565560226e-06, "loss": 0.07739856839179993, "step": 5178 }, { "epoch": 1.3444963655244029, "grad_norm": 3.5612998008728027, "learning_rate": 1.6540705551687686e-06, "loss": 0.06777061522006989, "step": 5179 }, { "epoch": 1.3447559709241952, "grad_norm": 3.788158893585205, "learning_rate": 1.653244447516745e-06, "loss": 0.042850375175476074, "step": 5180 }, { "epoch": 1.3450155763239875, "grad_norm": 5.636967182159424, "learning_rate": 1.6524185337459909e-06, "loss": 0.011874613352119923, "step": 5181 }, { "epoch": 1.3452751817237798, "grad_norm": 2.3996331691741943, "learning_rate": 1.6515928140025095e-06, "loss": 0.07274257391691208, "step": 5182 }, { "epoch": 1.345534787123572, "grad_norm": 3.1155662536621094, "learning_rate": 1.6507672884322685e-06, "loss": 0.05281934142112732, "step": 5183 }, { "epoch": 1.3457943925233644, "grad_norm": 2.670152187347412, "learning_rate": 1.649941957181202e-06, "loss": 0.05506581813097, "step": 5184 }, { "epoch": 1.3460539979231567, "grad_norm": 3.5316972732543945, "learning_rate": 1.6491168203952085e-06, "loss": 0.03756154328584671, "step": 5185 }, { "epoch": 1.3463136033229492, "grad_norm": 2.1405928134918213, "learning_rate": 1.6482918782201537e-06, "loss": 0.06509090960025787, "step": 5186 }, { "epoch": 1.3465732087227416, "grad_norm": 2.090911626815796, "learning_rate": 1.6474671308018714e-06, "loss": 0.07563529908657074, "step": 5187 }, { "epoch": 1.3468328141225339, "grad_norm": 3.1241455078125, "learning_rate": 1.646642578286154e-06, "loss": 0.018586181104183197, "step": 5188 }, { "epoch": 1.3470924195223262, "grad_norm": 2.4485325813293457, "learning_rate": 1.645818220818766e-06, "loss": 0.007837989367544651, "step": 5189 }, { "epoch": 1.3473520249221185, "grad_norm": 3.2708370685577393, "learning_rate": 1.644994058545433e-06, "loss": 0.058842070400714874, "step": 5190 }, { "epoch": 1.3476116303219108, "grad_norm": 3.14465594291687, "learning_rate": 1.6441700916118505e-06, "loss": 0.029039612039923668, "step": 5191 }, { "epoch": 1.347871235721703, "grad_norm": 1.3383995294570923, "learning_rate": 1.6433463201636755e-06, "loss": 0.009414936415851116, "step": 5192 }, { "epoch": 1.3481308411214954, "grad_norm": 2.862588405609131, "learning_rate": 1.642522744346531e-06, "loss": 0.03415023908019066, "step": 5193 }, { "epoch": 1.3483904465212877, "grad_norm": 3.8156745433807373, "learning_rate": 1.6416993643060081e-06, "loss": 0.026019811630249023, "step": 5194 }, { "epoch": 1.34865005192108, "grad_norm": 3.114069700241089, "learning_rate": 1.6408761801876612e-06, "loss": 0.05034932494163513, "step": 5195 }, { "epoch": 1.3489096573208723, "grad_norm": 4.802491664886475, "learning_rate": 1.6400531921370095e-06, "loss": 0.02671331912279129, "step": 5196 }, { "epoch": 1.3491692627206646, "grad_norm": 4.64399528503418, "learning_rate": 1.6392304002995397e-06, "loss": 0.021126531064510345, "step": 5197 }, { "epoch": 1.349428868120457, "grad_norm": 2.545741558074951, "learning_rate": 1.6384078048207008e-06, "loss": 0.07552672922611237, "step": 5198 }, { "epoch": 1.3496884735202492, "grad_norm": 3.8809452056884766, "learning_rate": 1.6375854058459107e-06, "loss": 0.046934619545936584, "step": 5199 }, { "epoch": 1.3499480789200415, "grad_norm": 3.6391735076904297, "learning_rate": 1.63676320352055e-06, "loss": 0.036903075873851776, "step": 5200 }, { "epoch": 1.3502076843198338, "grad_norm": 3.193204164505005, "learning_rate": 1.6359411979899648e-06, "loss": 0.06588970124721527, "step": 5201 }, { "epoch": 1.3504672897196262, "grad_norm": 3.12282133102417, "learning_rate": 1.6351193893994675e-06, "loss": 0.03740557283163071, "step": 5202 }, { "epoch": 1.3507268951194185, "grad_norm": 3.5879509449005127, "learning_rate": 1.6342977778943353e-06, "loss": 0.047739364206790924, "step": 5203 }, { "epoch": 1.3509865005192108, "grad_norm": 3.0727155208587646, "learning_rate": 1.6334763636198083e-06, "loss": 0.028964979574084282, "step": 5204 }, { "epoch": 1.351246105919003, "grad_norm": 3.2566487789154053, "learning_rate": 1.6326551467210965e-06, "loss": 0.05294695496559143, "step": 5205 }, { "epoch": 1.3515057113187954, "grad_norm": 4.651160717010498, "learning_rate": 1.6318341273433704e-06, "loss": 0.028163570910692215, "step": 5206 }, { "epoch": 1.3517653167185877, "grad_norm": 3.16257643699646, "learning_rate": 1.6310133056317678e-06, "loss": 0.04592739790678024, "step": 5207 }, { "epoch": 1.35202492211838, "grad_norm": 2.8817858695983887, "learning_rate": 1.6301926817313923e-06, "loss": 0.011719447560608387, "step": 5208 }, { "epoch": 1.3522845275181723, "grad_norm": 3.187788963317871, "learning_rate": 1.629372255787309e-06, "loss": 0.04186534136533737, "step": 5209 }, { "epoch": 1.3525441329179646, "grad_norm": 3.077523708343506, "learning_rate": 1.6285520279445539e-06, "loss": 0.048281699419021606, "step": 5210 }, { "epoch": 1.352803738317757, "grad_norm": 3.2674078941345215, "learning_rate": 1.627731998348121e-06, "loss": 0.038816556334495544, "step": 5211 }, { "epoch": 1.3530633437175492, "grad_norm": 3.5304863452911377, "learning_rate": 1.626912167142974e-06, "loss": 0.039077043533325195, "step": 5212 }, { "epoch": 1.3533229491173415, "grad_norm": 2.698913812637329, "learning_rate": 1.6260925344740414e-06, "loss": 0.008769161067903042, "step": 5213 }, { "epoch": 1.3535825545171338, "grad_norm": 2.6460821628570557, "learning_rate": 1.6252731004862149e-06, "loss": 0.0418495237827301, "step": 5214 }, { "epoch": 1.3538421599169261, "grad_norm": 2.9883365631103516, "learning_rate": 1.6244538653243508e-06, "loss": 0.02088012360036373, "step": 5215 }, { "epoch": 1.3541017653167187, "grad_norm": 3.0717251300811768, "learning_rate": 1.6236348291332716e-06, "loss": 0.020965158939361572, "step": 5216 }, { "epoch": 1.354361370716511, "grad_norm": 2.1587624549865723, "learning_rate": 1.6228159920577638e-06, "loss": 0.07639804482460022, "step": 5217 }, { "epoch": 1.3546209761163033, "grad_norm": 2.685249090194702, "learning_rate": 1.6219973542425816e-06, "loss": 0.05250214785337448, "step": 5218 }, { "epoch": 1.3548805815160956, "grad_norm": 2.449671745300293, "learning_rate": 1.621178915832437e-06, "loss": 0.02419300004839897, "step": 5219 }, { "epoch": 1.355140186915888, "grad_norm": 2.0023481845855713, "learning_rate": 1.6203606769720148e-06, "loss": 0.0767461508512497, "step": 5220 }, { "epoch": 1.3553997923156802, "grad_norm": 3.425903081893921, "learning_rate": 1.6195426378059587e-06, "loss": 0.074874147772789, "step": 5221 }, { "epoch": 1.3556593977154725, "grad_norm": 2.8871984481811523, "learning_rate": 1.6187247984788811e-06, "loss": 0.06172852963209152, "step": 5222 }, { "epoch": 1.3559190031152648, "grad_norm": 2.971620559692383, "learning_rate": 1.6179071591353563e-06, "loss": 0.030244305729866028, "step": 5223 }, { "epoch": 1.3561786085150571, "grad_norm": 2.425156593322754, "learning_rate": 1.6170897199199231e-06, "loss": 0.003943970892578363, "step": 5224 }, { "epoch": 1.3564382139148494, "grad_norm": 2.8297383785247803, "learning_rate": 1.6162724809770885e-06, "loss": 0.05450982227921486, "step": 5225 }, { "epoch": 1.3566978193146417, "grad_norm": 3.6276681423187256, "learning_rate": 1.6154554424513203e-06, "loss": 0.03222626447677612, "step": 5226 }, { "epoch": 1.356957424714434, "grad_norm": 3.766725778579712, "learning_rate": 1.6146386044870514e-06, "loss": 0.020584968850016594, "step": 5227 }, { "epoch": 1.3572170301142263, "grad_norm": 3.379150867462158, "learning_rate": 1.6138219672286814e-06, "loss": 0.035971302539110184, "step": 5228 }, { "epoch": 1.3574766355140186, "grad_norm": 4.115869522094727, "learning_rate": 1.6130055308205723e-06, "loss": 0.015453767962753773, "step": 5229 }, { "epoch": 1.357736240913811, "grad_norm": 2.759155035018921, "learning_rate": 1.612189295407052e-06, "loss": 0.05781536176800728, "step": 5230 }, { "epoch": 1.3579958463136033, "grad_norm": 2.597038507461548, "learning_rate": 1.6113732611324118e-06, "loss": 0.07567235082387924, "step": 5231 }, { "epoch": 1.3582554517133956, "grad_norm": 2.2931833267211914, "learning_rate": 1.6105574281409073e-06, "loss": 0.004884254187345505, "step": 5232 }, { "epoch": 1.3585150571131879, "grad_norm": 2.9870567321777344, "learning_rate": 1.6097417965767608e-06, "loss": 0.053922660648822784, "step": 5233 }, { "epoch": 1.3587746625129804, "grad_norm": 2.002335786819458, "learning_rate": 1.608926366584156e-06, "loss": 0.07643938809633255, "step": 5234 }, { "epoch": 1.3590342679127727, "grad_norm": 4.278123378753662, "learning_rate": 1.6081111383072414e-06, "loss": 0.018030069768428802, "step": 5235 }, { "epoch": 1.359293873312565, "grad_norm": 2.9110629558563232, "learning_rate": 1.6072961118901328e-06, "loss": 0.06715907156467438, "step": 5236 }, { "epoch": 1.3595534787123573, "grad_norm": 2.238400459289551, "learning_rate": 1.6064812874769073e-06, "loss": 0.082084521651268, "step": 5237 }, { "epoch": 1.3598130841121496, "grad_norm": 2.301973581314087, "learning_rate": 1.605666665211606e-06, "loss": 0.01916005089879036, "step": 5238 }, { "epoch": 1.360072689511942, "grad_norm": 3.2457175254821777, "learning_rate": 1.6048522452382375e-06, "loss": 0.04655686765909195, "step": 5239 }, { "epoch": 1.3603322949117342, "grad_norm": 2.807217597961426, "learning_rate": 1.604038027700771e-06, "loss": 0.048777177929878235, "step": 5240 }, { "epoch": 1.3605919003115265, "grad_norm": 2.9690661430358887, "learning_rate": 1.6032240127431433e-06, "loss": 0.05919215828180313, "step": 5241 }, { "epoch": 1.3608515057113189, "grad_norm": 3.9457929134368896, "learning_rate": 1.6024102005092513e-06, "loss": 0.063929483294487, "step": 5242 }, { "epoch": 1.3611111111111112, "grad_norm": 5.243505477905273, "learning_rate": 1.6015965911429593e-06, "loss": 0.013697944581508636, "step": 5243 }, { "epoch": 1.3613707165109035, "grad_norm": 2.252431631088257, "learning_rate": 1.600783184788096e-06, "loss": 0.09039824455976486, "step": 5244 }, { "epoch": 1.3616303219106958, "grad_norm": 2.470184564590454, "learning_rate": 1.5999699815884518e-06, "loss": 0.07214237749576569, "step": 5245 }, { "epoch": 1.361889927310488, "grad_norm": 3.1288962364196777, "learning_rate": 1.5991569816877827e-06, "loss": 0.04096943512558937, "step": 5246 }, { "epoch": 1.3621495327102804, "grad_norm": 2.649033784866333, "learning_rate": 1.5983441852298075e-06, "loss": 0.05524279177188873, "step": 5247 }, { "epoch": 1.3624091381100727, "grad_norm": 3.822190284729004, "learning_rate": 1.5975315923582113e-06, "loss": 0.03358392417430878, "step": 5248 }, { "epoch": 1.362668743509865, "grad_norm": 2.997925043106079, "learning_rate": 1.5967192032166429e-06, "loss": 0.062379270792007446, "step": 5249 }, { "epoch": 1.3629283489096573, "grad_norm": 2.164217948913574, "learning_rate": 1.5959070179487112e-06, "loss": 0.006587847135961056, "step": 5250 }, { "epoch": 1.3631879543094496, "grad_norm": 2.913154363632202, "learning_rate": 1.5950950366979943e-06, "loss": 0.04064025729894638, "step": 5251 }, { "epoch": 1.363447559709242, "grad_norm": 2.373485565185547, "learning_rate": 1.5942832596080304e-06, "loss": 0.01044674962759018, "step": 5252 }, { "epoch": 1.3637071651090342, "grad_norm": 3.175673723220825, "learning_rate": 1.5934716868223244e-06, "loss": 0.036824487149715424, "step": 5253 }, { "epoch": 1.3639667705088265, "grad_norm": 3.8347198963165283, "learning_rate": 1.592660318484343e-06, "loss": 0.027544980868697166, "step": 5254 }, { "epoch": 1.3642263759086188, "grad_norm": 3.220114231109619, "learning_rate": 1.5918491547375175e-06, "loss": 0.019630342721939087, "step": 5255 }, { "epoch": 1.3644859813084111, "grad_norm": 5.311201095581055, "learning_rate": 1.591038195725244e-06, "loss": 0.017421042546629906, "step": 5256 }, { "epoch": 1.3647455867082035, "grad_norm": 1.8040295839309692, "learning_rate": 1.5902274415908802e-06, "loss": 0.08558328449726105, "step": 5257 }, { "epoch": 1.3650051921079958, "grad_norm": 2.592050790786743, "learning_rate": 1.5894168924777493e-06, "loss": 0.07521691918373108, "step": 5258 }, { "epoch": 1.365264797507788, "grad_norm": 2.2813801765441895, "learning_rate": 1.5886065485291385e-06, "loss": 0.07736937701702118, "step": 5259 }, { "epoch": 1.3655244029075804, "grad_norm": 3.3899900913238525, "learning_rate": 1.587796409888297e-06, "loss": 0.05108652263879776, "step": 5260 }, { "epoch": 1.3657840083073727, "grad_norm": 2.9677605628967285, "learning_rate": 1.5869864766984397e-06, "loss": 0.06579794734716415, "step": 5261 }, { "epoch": 1.366043613707165, "grad_norm": 3.5074963569641113, "learning_rate": 1.586176749102744e-06, "loss": 0.043067459017038345, "step": 5262 }, { "epoch": 1.3663032191069573, "grad_norm": 3.923954963684082, "learning_rate": 1.5853672272443499e-06, "loss": 0.06290445476770401, "step": 5263 }, { "epoch": 1.3665628245067498, "grad_norm": 3.532248020172119, "learning_rate": 1.5845579112663651e-06, "loss": 0.07048611342906952, "step": 5264 }, { "epoch": 1.3668224299065421, "grad_norm": 1.4082735776901245, "learning_rate": 1.5837488013118558e-06, "loss": 0.0030057623516768217, "step": 5265 }, { "epoch": 1.3670820353063344, "grad_norm": 3.126589775085449, "learning_rate": 1.5829398975238542e-06, "loss": 0.05616685375571251, "step": 5266 }, { "epoch": 1.3673416407061267, "grad_norm": 3.169633626937866, "learning_rate": 1.5821312000453575e-06, "loss": 0.028945285826921463, "step": 5267 }, { "epoch": 1.367601246105919, "grad_norm": 3.898789405822754, "learning_rate": 1.581322709019324e-06, "loss": 0.018937280401587486, "step": 5268 }, { "epoch": 1.3678608515057114, "grad_norm": 2.1538033485412598, "learning_rate": 1.580514424588675e-06, "loss": 0.011686346493661404, "step": 5269 }, { "epoch": 1.3681204569055037, "grad_norm": 2.1155896186828613, "learning_rate": 1.5797063468962998e-06, "loss": 0.004839427303522825, "step": 5270 }, { "epoch": 1.368380062305296, "grad_norm": 3.497385025024414, "learning_rate": 1.5788984760850445e-06, "loss": 0.011541169136762619, "step": 5271 }, { "epoch": 1.3686396677050883, "grad_norm": 2.7993509769439697, "learning_rate": 1.5780908122977265e-06, "loss": 0.08329811692237854, "step": 5272 }, { "epoch": 1.3688992731048806, "grad_norm": 3.6451849937438965, "learning_rate": 1.5772833556771177e-06, "loss": 0.01021417137235403, "step": 5273 }, { "epoch": 1.3691588785046729, "grad_norm": 1.7150734663009644, "learning_rate": 1.57647610636596e-06, "loss": 0.07404275238513947, "step": 5274 }, { "epoch": 1.3694184839044652, "grad_norm": 2.1882030963897705, "learning_rate": 1.5756690645069573e-06, "loss": 0.011853272095322609, "step": 5275 }, { "epoch": 1.3696780893042575, "grad_norm": 3.8172543048858643, "learning_rate": 1.5748622302427755e-06, "loss": 0.044156186282634735, "step": 5276 }, { "epoch": 1.3699376947040498, "grad_norm": 5.527970790863037, "learning_rate": 1.5740556037160443e-06, "loss": 0.05073793977499008, "step": 5277 }, { "epoch": 1.370197300103842, "grad_norm": 2.6645312309265137, "learning_rate": 1.5732491850693557e-06, "loss": 0.021123453974723816, "step": 5278 }, { "epoch": 1.3704569055036344, "grad_norm": 3.507014751434326, "learning_rate": 1.5724429744452674e-06, "loss": 0.05216170847415924, "step": 5279 }, { "epoch": 1.3707165109034267, "grad_norm": 3.7674896717071533, "learning_rate": 1.5716369719863003e-06, "loss": 0.043219346553087234, "step": 5280 }, { "epoch": 1.370976116303219, "grad_norm": 3.2912545204162598, "learning_rate": 1.5708311778349341e-06, "loss": 0.036050811409950256, "step": 5281 }, { "epoch": 1.3712357217030116, "grad_norm": 3.0875113010406494, "learning_rate": 1.5700255921336168e-06, "loss": 0.034689005464315414, "step": 5282 }, { "epoch": 1.3714953271028039, "grad_norm": 3.855508327484131, "learning_rate": 1.5692202150247566e-06, "loss": 0.03222096711397171, "step": 5283 }, { "epoch": 1.3717549325025962, "grad_norm": 2.5716490745544434, "learning_rate": 1.5684150466507263e-06, "loss": 0.0858309268951416, "step": 5284 }, { "epoch": 1.3720145379023885, "grad_norm": 3.301665782928467, "learning_rate": 1.567610087153862e-06, "loss": 0.06230367720127106, "step": 5285 }, { "epoch": 1.3722741433021808, "grad_norm": 3.2007346153259277, "learning_rate": 1.5668053366764596e-06, "loss": 0.04893533140420914, "step": 5286 }, { "epoch": 1.372533748701973, "grad_norm": 2.568632125854492, "learning_rate": 1.566000795360783e-06, "loss": 0.015488417819142342, "step": 5287 }, { "epoch": 1.3727933541017654, "grad_norm": 2.251884937286377, "learning_rate": 1.565196463349056e-06, "loss": 0.06640296429395676, "step": 5288 }, { "epoch": 1.3730529595015577, "grad_norm": 6.198606014251709, "learning_rate": 1.564392340783465e-06, "loss": 0.031067315489053726, "step": 5289 }, { "epoch": 1.37331256490135, "grad_norm": 3.1076879501342773, "learning_rate": 1.563588427806162e-06, "loss": 0.022579936310648918, "step": 5290 }, { "epoch": 1.3735721703011423, "grad_norm": 2.9486191272735596, "learning_rate": 1.5627847245592598e-06, "loss": 0.08200503885746002, "step": 5291 }, { "epoch": 1.3738317757009346, "grad_norm": 3.341075897216797, "learning_rate": 1.5619812311848336e-06, "loss": 0.07742609828710556, "step": 5292 }, { "epoch": 1.374091381100727, "grad_norm": 2.8642096519470215, "learning_rate": 1.561177947824925e-06, "loss": 0.08599798381328583, "step": 5293 }, { "epoch": 1.3743509865005192, "grad_norm": 3.4062488079071045, "learning_rate": 1.5603748746215338e-06, "loss": 0.016818998381495476, "step": 5294 }, { "epoch": 1.3746105919003115, "grad_norm": 2.4108331203460693, "learning_rate": 1.5595720117166268e-06, "loss": 0.07341644167900085, "step": 5295 }, { "epoch": 1.3748701973001038, "grad_norm": 3.4783544540405273, "learning_rate": 1.5587693592521307e-06, "loss": 0.029146524146199226, "step": 5296 }, { "epoch": 1.3751298026998962, "grad_norm": 2.8518857955932617, "learning_rate": 1.557966917369936e-06, "loss": 0.01944649964570999, "step": 5297 }, { "epoch": 1.3753894080996885, "grad_norm": 5.378554344177246, "learning_rate": 1.5571646862118968e-06, "loss": 0.015223385766148567, "step": 5298 }, { "epoch": 1.3756490134994808, "grad_norm": 5.111140727996826, "learning_rate": 1.5563626659198289e-06, "loss": 0.01222232636064291, "step": 5299 }, { "epoch": 1.375908618899273, "grad_norm": 3.5906982421875, "learning_rate": 1.55556085663551e-06, "loss": 0.06196519359946251, "step": 5300 }, { "epoch": 1.3761682242990654, "grad_norm": 3.8233091831207275, "learning_rate": 1.5547592585006839e-06, "loss": 0.047698985785245895, "step": 5301 }, { "epoch": 1.3764278296988577, "grad_norm": 3.291454792022705, "learning_rate": 1.5539578716570524e-06, "loss": 0.059240248054265976, "step": 5302 }, { "epoch": 1.37668743509865, "grad_norm": 2.4375827312469482, "learning_rate": 1.5531566962462853e-06, "loss": 0.06737002730369568, "step": 5303 }, { "epoch": 1.3769470404984423, "grad_norm": 7.288602828979492, "learning_rate": 1.552355732410008e-06, "loss": 0.01895730569958687, "step": 5304 }, { "epoch": 1.3772066458982346, "grad_norm": 4.2491655349731445, "learning_rate": 1.551554980289815e-06, "loss": 0.03349617123603821, "step": 5305 }, { "epoch": 1.377466251298027, "grad_norm": 2.796678066253662, "learning_rate": 1.550754440027262e-06, "loss": 0.07148011028766632, "step": 5306 }, { "epoch": 1.3777258566978192, "grad_norm": 4.022212028503418, "learning_rate": 1.5499541117638645e-06, "loss": 0.017667576670646667, "step": 5307 }, { "epoch": 1.3779854620976115, "grad_norm": 2.830585479736328, "learning_rate": 1.5491539956411027e-06, "loss": 0.07961385697126389, "step": 5308 }, { "epoch": 1.3782450674974038, "grad_norm": 3.058788299560547, "learning_rate": 1.548354091800418e-06, "loss": 0.012850997969508171, "step": 5309 }, { "epoch": 1.3785046728971961, "grad_norm": 3.3656246662139893, "learning_rate": 1.5475544003832161e-06, "loss": 0.06520101428031921, "step": 5310 }, { "epoch": 1.3787642782969884, "grad_norm": 2.739067792892456, "learning_rate": 1.546754921530865e-06, "loss": 0.04073311388492584, "step": 5311 }, { "epoch": 1.379023883696781, "grad_norm": 3.379702568054199, "learning_rate": 1.5459556553846915e-06, "loss": 0.021461378782987595, "step": 5312 }, { "epoch": 1.3792834890965733, "grad_norm": 3.6524085998535156, "learning_rate": 1.5451566020859903e-06, "loss": 0.05558651685714722, "step": 5313 }, { "epoch": 1.3795430944963656, "grad_norm": 3.8375234603881836, "learning_rate": 1.5443577617760137e-06, "loss": 0.026035238057374954, "step": 5314 }, { "epoch": 1.379802699896158, "grad_norm": 2.502340078353882, "learning_rate": 1.54355913459598e-06, "loss": 0.06450574845075607, "step": 5315 }, { "epoch": 1.3800623052959502, "grad_norm": 3.8365609645843506, "learning_rate": 1.5427607206870677e-06, "loss": 0.016096655279397964, "step": 5316 }, { "epoch": 1.3803219106957425, "grad_norm": 3.079092025756836, "learning_rate": 1.5419625201904164e-06, "loss": 0.025210008025169373, "step": 5317 }, { "epoch": 1.3805815160955348, "grad_norm": 3.2972047328948975, "learning_rate": 1.5411645332471326e-06, "loss": 0.05549423396587372, "step": 5318 }, { "epoch": 1.3808411214953271, "grad_norm": 3.9083986282348633, "learning_rate": 1.5403667599982808e-06, "loss": 0.044058896601200104, "step": 5319 }, { "epoch": 1.3811007268951194, "grad_norm": 6.0994462966918945, "learning_rate": 1.5395692005848876e-06, "loss": 0.02614310197532177, "step": 5320 }, { "epoch": 1.3813603322949117, "grad_norm": 2.965008497238159, "learning_rate": 1.5387718551479452e-06, "loss": 0.04650701582431793, "step": 5321 }, { "epoch": 1.381619937694704, "grad_norm": 3.4110934734344482, "learning_rate": 1.5379747238284059e-06, "loss": 0.015066438354551792, "step": 5322 }, { "epoch": 1.3818795430944963, "grad_norm": 3.2526040077209473, "learning_rate": 1.5371778067671826e-06, "loss": 0.017978670075535774, "step": 5323 }, { "epoch": 1.3821391484942886, "grad_norm": 3.52300763130188, "learning_rate": 1.5363811041051541e-06, "loss": 0.01381206326186657, "step": 5324 }, { "epoch": 1.382398753894081, "grad_norm": 1.9561254978179932, "learning_rate": 1.5355846159831572e-06, "loss": 0.07839018106460571, "step": 5325 }, { "epoch": 1.3826583592938733, "grad_norm": 4.231991291046143, "learning_rate": 1.5347883425419945e-06, "loss": 0.006867261603474617, "step": 5326 }, { "epoch": 1.3829179646936656, "grad_norm": 3.7543904781341553, "learning_rate": 1.5339922839224285e-06, "loss": 0.018792444840073586, "step": 5327 }, { "epoch": 1.3831775700934579, "grad_norm": 2.766085624694824, "learning_rate": 1.5331964402651827e-06, "loss": 0.049861565232276917, "step": 5328 }, { "epoch": 1.3834371754932502, "grad_norm": 3.730243682861328, "learning_rate": 1.5324008117109463e-06, "loss": 0.021035678684711456, "step": 5329 }, { "epoch": 1.3836967808930427, "grad_norm": 3.7893190383911133, "learning_rate": 1.531605398400367e-06, "loss": 0.061114292591810226, "step": 5330 }, { "epoch": 1.383956386292835, "grad_norm": 2.7605507373809814, "learning_rate": 1.5308102004740549e-06, "loss": 0.01277434267103672, "step": 5331 }, { "epoch": 1.3842159916926273, "grad_norm": 2.2703969478607178, "learning_rate": 1.530015218072584e-06, "loss": 0.06793738156557083, "step": 5332 }, { "epoch": 1.3844755970924196, "grad_norm": 2.490906238555908, "learning_rate": 1.5292204513364884e-06, "loss": 0.020315740257501602, "step": 5333 }, { "epoch": 1.384735202492212, "grad_norm": 2.5962212085723877, "learning_rate": 1.528425900406266e-06, "loss": 0.06601846218109131, "step": 5334 }, { "epoch": 1.3849948078920042, "grad_norm": 2.484159469604492, "learning_rate": 1.5276315654223728e-06, "loss": 0.02650158479809761, "step": 5335 }, { "epoch": 1.3852544132917965, "grad_norm": 2.3912580013275146, "learning_rate": 1.5268374465252301e-06, "loss": 0.07968975603580475, "step": 5336 }, { "epoch": 1.3855140186915889, "grad_norm": 2.4598917961120605, "learning_rate": 1.5260435438552208e-06, "loss": 0.0663350522518158, "step": 5337 }, { "epoch": 1.3857736240913812, "grad_norm": 2.7110166549682617, "learning_rate": 1.5252498575526887e-06, "loss": 0.033838629722595215, "step": 5338 }, { "epoch": 1.3860332294911735, "grad_norm": 2.356527805328369, "learning_rate": 1.524456387757938e-06, "loss": 0.061258867383003235, "step": 5339 }, { "epoch": 1.3862928348909658, "grad_norm": 1.2106214761734009, "learning_rate": 1.5236631346112364e-06, "loss": 0.0037987176328897476, "step": 5340 }, { "epoch": 1.386552440290758, "grad_norm": 2.4109060764312744, "learning_rate": 1.5228700982528127e-06, "loss": 0.009992795065045357, "step": 5341 }, { "epoch": 1.3868120456905504, "grad_norm": 3.7226414680480957, "learning_rate": 1.5220772788228599e-06, "loss": 0.036537788808345795, "step": 5342 }, { "epoch": 1.3870716510903427, "grad_norm": 2.823201894760132, "learning_rate": 1.5212846764615271e-06, "loss": 0.03990427404642105, "step": 5343 }, { "epoch": 1.387331256490135, "grad_norm": 3.8271353244781494, "learning_rate": 1.5204922913089305e-06, "loss": 0.04820641130208969, "step": 5344 }, { "epoch": 1.3875908618899273, "grad_norm": 2.645369052886963, "learning_rate": 1.5197001235051446e-06, "loss": 0.009552521631121635, "step": 5345 }, { "epoch": 1.3878504672897196, "grad_norm": 3.0466532707214355, "learning_rate": 1.5189081731902062e-06, "loss": 0.06076488643884659, "step": 5346 }, { "epoch": 1.388110072689512, "grad_norm": 3.6761815547943115, "learning_rate": 1.5181164405041149e-06, "loss": 0.06811410188674927, "step": 5347 }, { "epoch": 1.3883696780893042, "grad_norm": 2.4792299270629883, "learning_rate": 1.5173249255868299e-06, "loss": 0.05523666739463806, "step": 5348 }, { "epoch": 1.3886292834890965, "grad_norm": 2.8236660957336426, "learning_rate": 1.5165336285782739e-06, "loss": 0.04480328410863876, "step": 5349 }, { "epoch": 1.3888888888888888, "grad_norm": 6.399196147918701, "learning_rate": 1.5157425496183303e-06, "loss": 0.02831718884408474, "step": 5350 }, { "epoch": 1.3891484942886811, "grad_norm": 2.7924110889434814, "learning_rate": 1.5149516888468423e-06, "loss": 0.028160957619547844, "step": 5351 }, { "epoch": 1.3894080996884735, "grad_norm": 3.0858583450317383, "learning_rate": 1.5141610464036177e-06, "loss": 0.01591339521110058, "step": 5352 }, { "epoch": 1.3896677050882658, "grad_norm": 2.8183388710021973, "learning_rate": 1.513370622428423e-06, "loss": 0.03051932156085968, "step": 5353 }, { "epoch": 1.389927310488058, "grad_norm": 4.371247291564941, "learning_rate": 1.5125804170609865e-06, "loss": 0.023190824314951897, "step": 5354 }, { "epoch": 1.3901869158878504, "grad_norm": 2.770169258117676, "learning_rate": 1.5117904304409997e-06, "loss": 0.038972146809101105, "step": 5355 }, { "epoch": 1.3904465212876427, "grad_norm": 2.826413869857788, "learning_rate": 1.511000662708113e-06, "loss": 0.01738240383565426, "step": 5356 }, { "epoch": 1.390706126687435, "grad_norm": 2.70200252532959, "learning_rate": 1.5102111140019404e-06, "loss": 0.031137267127633095, "step": 5357 }, { "epoch": 1.3909657320872273, "grad_norm": 3.4043209552764893, "learning_rate": 1.5094217844620558e-06, "loss": 0.027919556945562363, "step": 5358 }, { "epoch": 1.3912253374870196, "grad_norm": 2.9070701599121094, "learning_rate": 1.508632674227993e-06, "loss": 0.07009387016296387, "step": 5359 }, { "epoch": 1.3914849428868121, "grad_norm": 2.788498640060425, "learning_rate": 1.5078437834392508e-06, "loss": 0.042143814265728, "step": 5360 }, { "epoch": 1.3917445482866044, "grad_norm": 3.0333237648010254, "learning_rate": 1.507055112235286e-06, "loss": 0.016536619514226913, "step": 5361 }, { "epoch": 1.3920041536863967, "grad_norm": 2.088031768798828, "learning_rate": 1.5062666607555175e-06, "loss": 0.06260021030902863, "step": 5362 }, { "epoch": 1.392263759086189, "grad_norm": 2.396599292755127, "learning_rate": 1.505478429139326e-06, "loss": 0.07279188185930252, "step": 5363 }, { "epoch": 1.3925233644859814, "grad_norm": 3.555347442626953, "learning_rate": 1.5046904175260513e-06, "loss": 0.02424030378460884, "step": 5364 }, { "epoch": 1.3927829698857737, "grad_norm": 2.8026857376098633, "learning_rate": 1.5039026260549988e-06, "loss": 0.016021694988012314, "step": 5365 }, { "epoch": 1.393042575285566, "grad_norm": 1.8659201860427856, "learning_rate": 1.5031150548654288e-06, "loss": 0.0743987187743187, "step": 5366 }, { "epoch": 1.3933021806853583, "grad_norm": 2.3234314918518066, "learning_rate": 1.502327704096567e-06, "loss": 0.08785532414913177, "step": 5367 }, { "epoch": 1.3935617860851506, "grad_norm": 1.5566115379333496, "learning_rate": 1.5015405738876002e-06, "loss": 0.0036096139810979366, "step": 5368 }, { "epoch": 1.3938213914849429, "grad_norm": 3.0752174854278564, "learning_rate": 1.5007536643776737e-06, "loss": 0.055428396910429, "step": 5369 }, { "epoch": 1.3940809968847352, "grad_norm": 1.175519585609436, "learning_rate": 1.4999669757058956e-06, "loss": 0.001991058699786663, "step": 5370 }, { "epoch": 1.3943406022845275, "grad_norm": 1.9381595849990845, "learning_rate": 1.4991805080113337e-06, "loss": 0.06925161182880402, "step": 5371 }, { "epoch": 1.3946002076843198, "grad_norm": 4.2400898933410645, "learning_rate": 1.4983942614330182e-06, "loss": 0.033837851136922836, "step": 5372 }, { "epoch": 1.394859813084112, "grad_norm": 2.5684683322906494, "learning_rate": 1.4976082361099397e-06, "loss": 0.0648224726319313, "step": 5373 }, { "epoch": 1.3951194184839044, "grad_norm": 4.397650718688965, "learning_rate": 1.496822432181048e-06, "loss": 0.015931377187371254, "step": 5374 }, { "epoch": 1.3953790238836967, "grad_norm": 2.6268041133880615, "learning_rate": 1.4960368497852568e-06, "loss": 0.06892918795347214, "step": 5375 }, { "epoch": 1.395638629283489, "grad_norm": 2.838736057281494, "learning_rate": 1.49525148906144e-06, "loss": 0.052999380975961685, "step": 5376 }, { "epoch": 1.3958982346832813, "grad_norm": 2.4667561054229736, "learning_rate": 1.4944663501484285e-06, "loss": 0.07116887718439102, "step": 5377 }, { "epoch": 1.3961578400830739, "grad_norm": 3.041529655456543, "learning_rate": 1.4936814331850188e-06, "loss": 0.03261156380176544, "step": 5378 }, { "epoch": 1.3964174454828662, "grad_norm": 3.024746894836426, "learning_rate": 1.4928967383099655e-06, "loss": 0.05030198022723198, "step": 5379 }, { "epoch": 1.3966770508826585, "grad_norm": 3.035888195037842, "learning_rate": 1.4921122656619855e-06, "loss": 0.046883873641490936, "step": 5380 }, { "epoch": 1.3969366562824508, "grad_norm": 2.684823513031006, "learning_rate": 1.4913280153797553e-06, "loss": 0.06142863258719444, "step": 5381 }, { "epoch": 1.397196261682243, "grad_norm": 2.297592878341675, "learning_rate": 1.4905439876019113e-06, "loss": 0.0764896422624588, "step": 5382 }, { "epoch": 1.3974558670820354, "grad_norm": 3.0092310905456543, "learning_rate": 1.4897601824670534e-06, "loss": 0.017804624512791634, "step": 5383 }, { "epoch": 1.3977154724818277, "grad_norm": 4.492912769317627, "learning_rate": 1.4889766001137395e-06, "loss": 0.04317062720656395, "step": 5384 }, { "epoch": 1.39797507788162, "grad_norm": 2.4191102981567383, "learning_rate": 1.4881932406804883e-06, "loss": 0.06543823331594467, "step": 5385 }, { "epoch": 1.3982346832814123, "grad_norm": 2.1685400009155273, "learning_rate": 1.4874101043057814e-06, "loss": 0.06696341931819916, "step": 5386 }, { "epoch": 1.3984942886812046, "grad_norm": 4.1764116287231445, "learning_rate": 1.4866271911280578e-06, "loss": 0.019687138497829437, "step": 5387 }, { "epoch": 1.398753894080997, "grad_norm": 4.643912315368652, "learning_rate": 1.4858445012857203e-06, "loss": 0.024425501003861427, "step": 5388 }, { "epoch": 1.3990134994807892, "grad_norm": 2.9240477085113525, "learning_rate": 1.4850620349171297e-06, "loss": 0.06780295073986053, "step": 5389 }, { "epoch": 1.3992731048805815, "grad_norm": 2.6664249897003174, "learning_rate": 1.4842797921606073e-06, "loss": 0.0547240786254406, "step": 5390 }, { "epoch": 1.3995327102803738, "grad_norm": 4.233505725860596, "learning_rate": 1.483497773154437e-06, "loss": 0.0163573045283556, "step": 5391 }, { "epoch": 1.3997923156801662, "grad_norm": 2.8739705085754395, "learning_rate": 1.4827159780368616e-06, "loss": 0.017734376713633537, "step": 5392 }, { "epoch": 1.4000519210799585, "grad_norm": 4.255805969238281, "learning_rate": 1.4819344069460833e-06, "loss": 0.02969679981470108, "step": 5393 }, { "epoch": 1.4003115264797508, "grad_norm": 3.334554433822632, "learning_rate": 1.4811530600202683e-06, "loss": 0.018078725785017014, "step": 5394 }, { "epoch": 1.400571131879543, "grad_norm": 3.419524669647217, "learning_rate": 1.4803719373975382e-06, "loss": 0.01561475545167923, "step": 5395 }, { "epoch": 1.4008307372793354, "grad_norm": 2.6481473445892334, "learning_rate": 1.4795910392159798e-06, "loss": 0.00932624563574791, "step": 5396 }, { "epoch": 1.4010903426791277, "grad_norm": 2.4854319095611572, "learning_rate": 1.4788103656136372e-06, "loss": 0.02251804620027542, "step": 5397 }, { "epoch": 1.40134994807892, "grad_norm": 2.304548978805542, "learning_rate": 1.478029916728515e-06, "loss": 0.057884879410266876, "step": 5398 }, { "epoch": 1.4016095534787123, "grad_norm": 2.7175326347351074, "learning_rate": 1.4772496926985808e-06, "loss": 0.011337324045598507, "step": 5399 }, { "epoch": 1.4018691588785046, "grad_norm": 3.2252233028411865, "learning_rate": 1.476469693661757e-06, "loss": 0.011745406314730644, "step": 5400 }, { "epoch": 1.402128764278297, "grad_norm": 2.955659866333008, "learning_rate": 1.4756899197559317e-06, "loss": 0.01689283363521099, "step": 5401 }, { "epoch": 1.4023883696780892, "grad_norm": 3.243467330932617, "learning_rate": 1.4749103711189517e-06, "loss": 0.05031602457165718, "step": 5402 }, { "epoch": 1.4026479750778815, "grad_norm": 3.4548287391662598, "learning_rate": 1.4741310478886217e-06, "loss": 0.04626919701695442, "step": 5403 }, { "epoch": 1.4029075804776738, "grad_norm": 2.6179614067077637, "learning_rate": 1.4733519502027093e-06, "loss": 0.06504489481449127, "step": 5404 }, { "epoch": 1.4031671858774661, "grad_norm": 3.252415180206299, "learning_rate": 1.47257307819894e-06, "loss": 0.017186474055051804, "step": 5405 }, { "epoch": 1.4034267912772584, "grad_norm": 3.5888261795043945, "learning_rate": 1.4717944320150013e-06, "loss": 0.04069022089242935, "step": 5406 }, { "epoch": 1.4036863966770508, "grad_norm": 3.1800012588500977, "learning_rate": 1.4710160117885414e-06, "loss": 0.06633182615041733, "step": 5407 }, { "epoch": 1.4039460020768433, "grad_norm": 5.312698841094971, "learning_rate": 1.4702378176571641e-06, "loss": 0.03305753320455551, "step": 5408 }, { "epoch": 1.4042056074766356, "grad_norm": 3.254829168319702, "learning_rate": 1.4694598497584392e-06, "loss": 0.030720874667167664, "step": 5409 }, { "epoch": 1.404465212876428, "grad_norm": 2.6611015796661377, "learning_rate": 1.4686821082298914e-06, "loss": 0.059677645564079285, "step": 5410 }, { "epoch": 1.4047248182762202, "grad_norm": 3.615443706512451, "learning_rate": 1.4679045932090092e-06, "loss": 0.041871823370456696, "step": 5411 }, { "epoch": 1.4049844236760125, "grad_norm": 3.861833095550537, "learning_rate": 1.4671273048332389e-06, "loss": 0.010331682860851288, "step": 5412 }, { "epoch": 1.4052440290758048, "grad_norm": 2.0114941596984863, "learning_rate": 1.4663502432399861e-06, "loss": 0.07927116751670837, "step": 5413 }, { "epoch": 1.4055036344755971, "grad_norm": 3.278226137161255, "learning_rate": 1.4655734085666198e-06, "loss": 0.07173125445842743, "step": 5414 }, { "epoch": 1.4057632398753894, "grad_norm": 3.8627352714538574, "learning_rate": 1.4647968009504654e-06, "loss": 0.014247672632336617, "step": 5415 }, { "epoch": 1.4060228452751817, "grad_norm": 3.157599449157715, "learning_rate": 1.4640204205288084e-06, "loss": 0.04601980745792389, "step": 5416 }, { "epoch": 1.406282450674974, "grad_norm": 3.197028398513794, "learning_rate": 1.4632442674388969e-06, "loss": 0.045035477727651596, "step": 5417 }, { "epoch": 1.4065420560747663, "grad_norm": 2.8461735248565674, "learning_rate": 1.4624683418179353e-06, "loss": 0.017166350036859512, "step": 5418 }, { "epoch": 1.4068016614745587, "grad_norm": 3.2830593585968018, "learning_rate": 1.461692643803091e-06, "loss": 0.01809411682188511, "step": 5419 }, { "epoch": 1.407061266874351, "grad_norm": 2.957994222640991, "learning_rate": 1.4609171735314892e-06, "loss": 0.03226044028997421, "step": 5420 }, { "epoch": 1.4073208722741433, "grad_norm": 3.150418281555176, "learning_rate": 1.4601419311402142e-06, "loss": 0.03666083887219429, "step": 5421 }, { "epoch": 1.4075804776739356, "grad_norm": 2.594421863555908, "learning_rate": 1.4593669167663126e-06, "loss": 0.024974210187792778, "step": 5422 }, { "epoch": 1.4078400830737279, "grad_norm": 3.295433521270752, "learning_rate": 1.4585921305467887e-06, "loss": 0.03327850252389908, "step": 5423 }, { "epoch": 1.4080996884735202, "grad_norm": 3.3008856773376465, "learning_rate": 1.4578175726186056e-06, "loss": 0.024380572140216827, "step": 5424 }, { "epoch": 1.4083592938733125, "grad_norm": 2.6871771812438965, "learning_rate": 1.45704324311869e-06, "loss": 0.07416678965091705, "step": 5425 }, { "epoch": 1.408618899273105, "grad_norm": 3.556887626647949, "learning_rate": 1.4562691421839237e-06, "loss": 0.025169702246785164, "step": 5426 }, { "epoch": 1.4088785046728973, "grad_norm": 3.842233419418335, "learning_rate": 1.4554952699511498e-06, "loss": 0.0410592257976532, "step": 5427 }, { "epoch": 1.4091381100726896, "grad_norm": 2.7361831665039062, "learning_rate": 1.454721626557173e-06, "loss": 0.05694560334086418, "step": 5428 }, { "epoch": 1.409397715472482, "grad_norm": 3.704206943511963, "learning_rate": 1.4539482121387539e-06, "loss": 0.050014980137348175, "step": 5429 }, { "epoch": 1.4096573208722742, "grad_norm": 3.3917338848114014, "learning_rate": 1.4531750268326166e-06, "loss": 0.04731392860412598, "step": 5430 }, { "epoch": 1.4099169262720666, "grad_norm": 3.1400504112243652, "learning_rate": 1.45240207077544e-06, "loss": 0.027105528861284256, "step": 5431 }, { "epoch": 1.4101765316718589, "grad_norm": 2.0759544372558594, "learning_rate": 1.4516293441038659e-06, "loss": 0.009286142885684967, "step": 5432 }, { "epoch": 1.4104361370716512, "grad_norm": 5.158533096313477, "learning_rate": 1.450856846954496e-06, "loss": 0.01726001873612404, "step": 5433 }, { "epoch": 1.4106957424714435, "grad_norm": 3.7693052291870117, "learning_rate": 1.4500845794638886e-06, "loss": 0.06294573098421097, "step": 5434 }, { "epoch": 1.4109553478712358, "grad_norm": 2.819340705871582, "learning_rate": 1.449312541768564e-06, "loss": 0.011995197273790836, "step": 5435 }, { "epoch": 1.411214953271028, "grad_norm": 3.1651268005371094, "learning_rate": 1.4485407340049991e-06, "loss": 0.0070152198895812035, "step": 5436 }, { "epoch": 1.4114745586708204, "grad_norm": 2.6317296028137207, "learning_rate": 1.447769156309633e-06, "loss": 0.06499123573303223, "step": 5437 }, { "epoch": 1.4117341640706127, "grad_norm": 4.594099998474121, "learning_rate": 1.4469978088188646e-06, "loss": 0.03520290181040764, "step": 5438 }, { "epoch": 1.411993769470405, "grad_norm": 3.981776237487793, "learning_rate": 1.4462266916690474e-06, "loss": 0.020177310332655907, "step": 5439 }, { "epoch": 1.4122533748701973, "grad_norm": 3.740358829498291, "learning_rate": 1.4454558049964993e-06, "loss": 0.03636191412806511, "step": 5440 }, { "epoch": 1.4125129802699896, "grad_norm": 2.2025234699249268, "learning_rate": 1.4446851489374942e-06, "loss": 0.026368308812379837, "step": 5441 }, { "epoch": 1.412772585669782, "grad_norm": 2.9407734870910645, "learning_rate": 1.4439147236282674e-06, "loss": 0.04332476481795311, "step": 5442 }, { "epoch": 1.4130321910695742, "grad_norm": 2.3994438648223877, "learning_rate": 1.4431445292050123e-06, "loss": 0.06586075574159622, "step": 5443 }, { "epoch": 1.4132917964693665, "grad_norm": 3.1101620197296143, "learning_rate": 1.4423745658038808e-06, "loss": 0.06273720413446426, "step": 5444 }, { "epoch": 1.4135514018691588, "grad_norm": 3.0635581016540527, "learning_rate": 1.4416048335609856e-06, "loss": 0.0174468532204628, "step": 5445 }, { "epoch": 1.4138110072689511, "grad_norm": 2.4374494552612305, "learning_rate": 1.4408353326123981e-06, "loss": 0.01628846675157547, "step": 5446 }, { "epoch": 1.4140706126687435, "grad_norm": 2.519624710083008, "learning_rate": 1.4400660630941469e-06, "loss": 0.02777247317135334, "step": 5447 }, { "epoch": 1.4143302180685358, "grad_norm": 2.965423583984375, "learning_rate": 1.439297025142223e-06, "loss": 0.029064133763313293, "step": 5448 }, { "epoch": 1.414589823468328, "grad_norm": 5.108669757843018, "learning_rate": 1.4385282188925732e-06, "loss": 0.036634430289268494, "step": 5449 }, { "epoch": 1.4148494288681204, "grad_norm": 3.001251697540283, "learning_rate": 1.4377596444811064e-06, "loss": 0.04028099775314331, "step": 5450 }, { "epoch": 1.4151090342679127, "grad_norm": 2.80958890914917, "learning_rate": 1.4369913020436882e-06, "loss": 0.03205485641956329, "step": 5451 }, { "epoch": 1.415368639667705, "grad_norm": 3.436013698577881, "learning_rate": 1.436223191716143e-06, "loss": 0.029795892536640167, "step": 5452 }, { "epoch": 1.4156282450674973, "grad_norm": 3.0325794219970703, "learning_rate": 1.4354553136342569e-06, "loss": 0.026009850203990936, "step": 5453 }, { "epoch": 1.4158878504672896, "grad_norm": 3.0547120571136475, "learning_rate": 1.4346876679337721e-06, "loss": 0.05472986027598381, "step": 5454 }, { "epoch": 1.416147455867082, "grad_norm": 2.523702621459961, "learning_rate": 1.43392025475039e-06, "loss": 0.07505179941654205, "step": 5455 }, { "epoch": 1.4164070612668744, "grad_norm": 2.6803805828094482, "learning_rate": 1.4331530742197738e-06, "loss": 0.04775252938270569, "step": 5456 }, { "epoch": 1.4166666666666667, "grad_norm": 2.693448781967163, "learning_rate": 1.432386126477542e-06, "loss": 0.04443725571036339, "step": 5457 }, { "epoch": 1.416926272066459, "grad_norm": 3.1592681407928467, "learning_rate": 1.431619411659273e-06, "loss": 0.01677391305565834, "step": 5458 }, { "epoch": 1.4171858774662514, "grad_norm": 2.386061906814575, "learning_rate": 1.4308529299005059e-06, "loss": 0.011559110134840012, "step": 5459 }, { "epoch": 1.4174454828660437, "grad_norm": 2.4105234146118164, "learning_rate": 1.4300866813367356e-06, "loss": 0.020313918590545654, "step": 5460 }, { "epoch": 1.417705088265836, "grad_norm": 3.037870407104492, "learning_rate": 1.4293206661034198e-06, "loss": 0.012145254760980606, "step": 5461 }, { "epoch": 1.4179646936656283, "grad_norm": 2.244454860687256, "learning_rate": 1.4285548843359686e-06, "loss": 0.013151806779205799, "step": 5462 }, { "epoch": 1.4182242990654206, "grad_norm": 2.332810878753662, "learning_rate": 1.427789336169757e-06, "loss": 0.013237416744232178, "step": 5463 }, { "epoch": 1.4184839044652129, "grad_norm": 2.8841426372528076, "learning_rate": 1.4270240217401171e-06, "loss": 0.014539400115609169, "step": 5464 }, { "epoch": 1.4187435098650052, "grad_norm": 6.134956359863281, "learning_rate": 1.426258941182338e-06, "loss": 0.02753485180437565, "step": 5465 }, { "epoch": 1.4190031152647975, "grad_norm": 3.7253265380859375, "learning_rate": 1.4254940946316684e-06, "loss": 0.03317816182971001, "step": 5466 }, { "epoch": 1.4192627206645898, "grad_norm": 3.3597593307495117, "learning_rate": 1.4247294822233146e-06, "loss": 0.0453786626458168, "step": 5467 }, { "epoch": 1.4195223260643821, "grad_norm": 3.4129579067230225, "learning_rate": 1.423965104092444e-06, "loss": 0.012479370459914207, "step": 5468 }, { "epoch": 1.4197819314641744, "grad_norm": 3.4140326976776123, "learning_rate": 1.4232009603741822e-06, "loss": 0.05572313442826271, "step": 5469 }, { "epoch": 1.4200415368639667, "grad_norm": 3.3881402015686035, "learning_rate": 1.4224370512036095e-06, "loss": 0.011574753560125828, "step": 5470 }, { "epoch": 1.420301142263759, "grad_norm": 2.521561622619629, "learning_rate": 1.4216733767157698e-06, "loss": 0.021386098116636276, "step": 5471 }, { "epoch": 1.4205607476635513, "grad_norm": 4.293235778808594, "learning_rate": 1.420909937045662e-06, "loss": 0.042342156171798706, "step": 5472 }, { "epoch": 1.4208203530633436, "grad_norm": 2.665548801422119, "learning_rate": 1.4201467323282453e-06, "loss": 0.07439989596605301, "step": 5473 }, { "epoch": 1.421079958463136, "grad_norm": 3.2649991512298584, "learning_rate": 1.419383762698437e-06, "loss": 0.019886301830410957, "step": 5474 }, { "epoch": 1.4213395638629285, "grad_norm": 3.2311627864837646, "learning_rate": 1.4186210282911118e-06, "loss": 0.0331772156059742, "step": 5475 }, { "epoch": 1.4215991692627208, "grad_norm": 3.2105157375335693, "learning_rate": 1.4178585292411046e-06, "loss": 0.046258896589279175, "step": 5476 }, { "epoch": 1.421858774662513, "grad_norm": 4.279628753662109, "learning_rate": 1.4170962656832077e-06, "loss": 0.0312599316239357, "step": 5477 }, { "epoch": 1.4221183800623054, "grad_norm": 3.513059377670288, "learning_rate": 1.416334237752171e-06, "loss": 0.05226254463195801, "step": 5478 }, { "epoch": 1.4223779854620977, "grad_norm": 2.9844138622283936, "learning_rate": 1.4155724455827046e-06, "loss": 0.0633285865187645, "step": 5479 }, { "epoch": 1.42263759086189, "grad_norm": 4.166800498962402, "learning_rate": 1.4148108893094756e-06, "loss": 0.07176434248685837, "step": 5480 }, { "epoch": 1.4228971962616823, "grad_norm": 2.2736854553222656, "learning_rate": 1.414049569067109e-06, "loss": 0.011716838926076889, "step": 5481 }, { "epoch": 1.4231568016614746, "grad_norm": 2.3380861282348633, "learning_rate": 1.41328848499019e-06, "loss": 0.07813187688589096, "step": 5482 }, { "epoch": 1.423416407061267, "grad_norm": 2.6913442611694336, "learning_rate": 1.4125276372132594e-06, "loss": 0.05817342922091484, "step": 5483 }, { "epoch": 1.4236760124610592, "grad_norm": 5.836007595062256, "learning_rate": 1.4117670258708194e-06, "loss": 0.030968010425567627, "step": 5484 }, { "epoch": 1.4239356178608515, "grad_norm": 2.2673983573913574, "learning_rate": 1.4110066510973275e-06, "loss": 0.06020163744688034, "step": 5485 }, { "epoch": 1.4241952232606438, "grad_norm": 3.6528728008270264, "learning_rate": 1.4102465130272003e-06, "loss": 0.03981510549783707, "step": 5486 }, { "epoch": 1.4244548286604362, "grad_norm": 3.9217255115509033, "learning_rate": 1.4094866117948136e-06, "loss": 0.03427422419190407, "step": 5487 }, { "epoch": 1.4247144340602285, "grad_norm": 2.9798238277435303, "learning_rate": 1.4087269475345009e-06, "loss": 0.037225402891635895, "step": 5488 }, { "epoch": 1.4249740394600208, "grad_norm": 1.0364850759506226, "learning_rate": 1.4079675203805516e-06, "loss": 0.003542162012308836, "step": 5489 }, { "epoch": 1.425233644859813, "grad_norm": 3.6139605045318604, "learning_rate": 1.4072083304672174e-06, "loss": 0.02401045709848404, "step": 5490 }, { "epoch": 1.4254932502596054, "grad_norm": 3.13960337638855, "learning_rate": 1.4064493779287033e-06, "loss": 0.029804876074194908, "step": 5491 }, { "epoch": 1.4257528556593977, "grad_norm": 4.131729602813721, "learning_rate": 1.4056906628991775e-06, "loss": 0.02384304255247116, "step": 5492 }, { "epoch": 1.42601246105919, "grad_norm": 4.105720520019531, "learning_rate": 1.4049321855127608e-06, "loss": 0.02956751361489296, "step": 5493 }, { "epoch": 1.4262720664589823, "grad_norm": 2.2689294815063477, "learning_rate": 1.4041739459035352e-06, "loss": 0.012687236070632935, "step": 5494 }, { "epoch": 1.4265316718587746, "grad_norm": 3.186906099319458, "learning_rate": 1.4034159442055417e-06, "loss": 0.01405656524002552, "step": 5495 }, { "epoch": 1.426791277258567, "grad_norm": 3.375120162963867, "learning_rate": 1.4026581805527767e-06, "loss": 0.04702045023441315, "step": 5496 }, { "epoch": 1.4270508826583592, "grad_norm": 2.3289005756378174, "learning_rate": 1.4019006550791953e-06, "loss": 0.07778863608837128, "step": 5497 }, { "epoch": 1.4273104880581515, "grad_norm": 3.1508545875549316, "learning_rate": 1.4011433679187098e-06, "loss": 0.025516778230667114, "step": 5498 }, { "epoch": 1.4275700934579438, "grad_norm": 0.8217019438743591, "learning_rate": 1.400386319205192e-06, "loss": 0.003182134358212352, "step": 5499 }, { "epoch": 1.4278296988577361, "grad_norm": 2.075331687927246, "learning_rate": 1.3996295090724723e-06, "loss": 0.08753097802400589, "step": 5500 }, { "epoch": 1.4280893042575284, "grad_norm": 2.792144536972046, "learning_rate": 1.3988729376543347e-06, "loss": 0.05300259590148926, "step": 5501 }, { "epoch": 1.4283489096573208, "grad_norm": 2.164043664932251, "learning_rate": 1.3981166050845258e-06, "loss": 0.08252309262752533, "step": 5502 }, { "epoch": 1.428608515057113, "grad_norm": 2.969619035720825, "learning_rate": 1.3973605114967465e-06, "loss": 0.03224165737628937, "step": 5503 }, { "epoch": 1.4288681204569054, "grad_norm": 2.8188605308532715, "learning_rate": 1.396604657024658e-06, "loss": 0.05794621258974075, "step": 5504 }, { "epoch": 1.429127725856698, "grad_norm": 2.205106258392334, "learning_rate": 1.3958490418018778e-06, "loss": 0.06738189607858658, "step": 5505 }, { "epoch": 1.4293873312564902, "grad_norm": 2.2711803913116455, "learning_rate": 1.3950936659619795e-06, "loss": 0.026335114613175392, "step": 5506 }, { "epoch": 1.4296469366562825, "grad_norm": 2.612271785736084, "learning_rate": 1.3943385296384996e-06, "loss": 0.012207991443574429, "step": 5507 }, { "epoch": 1.4299065420560748, "grad_norm": 2.7423899173736572, "learning_rate": 1.3935836329649266e-06, "loss": 0.015231526456773281, "step": 5508 }, { "epoch": 1.4301661474558671, "grad_norm": 3.176413059234619, "learning_rate": 1.3928289760747088e-06, "loss": 0.027182886376976967, "step": 5509 }, { "epoch": 1.4304257528556594, "grad_norm": 2.9889442920684814, "learning_rate": 1.3920745591012536e-06, "loss": 0.08863961696624756, "step": 5510 }, { "epoch": 1.4306853582554517, "grad_norm": 2.6354241371154785, "learning_rate": 1.3913203821779239e-06, "loss": 0.08409613370895386, "step": 5511 }, { "epoch": 1.430944963655244, "grad_norm": 2.3130199909210205, "learning_rate": 1.3905664454380401e-06, "loss": 0.055857934057712555, "step": 5512 }, { "epoch": 1.4312045690550363, "grad_norm": 3.1323843002319336, "learning_rate": 1.3898127490148832e-06, "loss": 0.008961346931755543, "step": 5513 }, { "epoch": 1.4314641744548287, "grad_norm": 2.798983097076416, "learning_rate": 1.389059293041687e-06, "loss": 0.01339508406817913, "step": 5514 }, { "epoch": 1.431723779854621, "grad_norm": 3.1577959060668945, "learning_rate": 1.3883060776516473e-06, "loss": 0.0410565622150898, "step": 5515 }, { "epoch": 1.4319833852544133, "grad_norm": 2.963961601257324, "learning_rate": 1.3875531029779137e-06, "loss": 0.092691570520401, "step": 5516 }, { "epoch": 1.4322429906542056, "grad_norm": 3.628683090209961, "learning_rate": 1.3868003691535958e-06, "loss": 0.0310360137373209, "step": 5517 }, { "epoch": 1.4325025960539979, "grad_norm": 3.3183302879333496, "learning_rate": 1.3860478763117596e-06, "loss": 0.07330864667892456, "step": 5518 }, { "epoch": 1.4327622014537902, "grad_norm": 3.7012391090393066, "learning_rate": 1.3852956245854288e-06, "loss": 0.03602685034275055, "step": 5519 }, { "epoch": 1.4330218068535825, "grad_norm": 2.9225261211395264, "learning_rate": 1.3845436141075832e-06, "loss": 0.012915808707475662, "step": 5520 }, { "epoch": 1.4332814122533748, "grad_norm": 5.389255523681641, "learning_rate": 1.3837918450111626e-06, "loss": 0.015823476016521454, "step": 5521 }, { "epoch": 1.433541017653167, "grad_norm": 5.003693580627441, "learning_rate": 1.3830403174290611e-06, "loss": 0.03777237981557846, "step": 5522 }, { "epoch": 1.4338006230529596, "grad_norm": 3.30387020111084, "learning_rate": 1.382289031494134e-06, "loss": 0.02024359628558159, "step": 5523 }, { "epoch": 1.434060228452752, "grad_norm": 3.3035871982574463, "learning_rate": 1.3815379873391876e-06, "loss": 0.04503867030143738, "step": 5524 }, { "epoch": 1.4343198338525442, "grad_norm": 2.7282629013061523, "learning_rate": 1.3807871850969918e-06, "loss": 0.020083051174879074, "step": 5525 }, { "epoch": 1.4345794392523366, "grad_norm": 2.439960479736328, "learning_rate": 1.3800366249002715e-06, "loss": 0.04635339230298996, "step": 5526 }, { "epoch": 1.4348390446521289, "grad_norm": 3.3281922340393066, "learning_rate": 1.3792863068817079e-06, "loss": 0.03044823184609413, "step": 5527 }, { "epoch": 1.4350986500519212, "grad_norm": 2.9127583503723145, "learning_rate": 1.3785362311739397e-06, "loss": 0.014249825850129128, "step": 5528 }, { "epoch": 1.4353582554517135, "grad_norm": 4.007819652557373, "learning_rate": 1.377786397909563e-06, "loss": 0.03217654675245285, "step": 5529 }, { "epoch": 1.4356178608515058, "grad_norm": 3.778369665145874, "learning_rate": 1.377036807221132e-06, "loss": 0.04378335177898407, "step": 5530 }, { "epoch": 1.435877466251298, "grad_norm": 3.052081823348999, "learning_rate": 1.376287459241158e-06, "loss": 0.055280838161706924, "step": 5531 }, { "epoch": 1.4361370716510904, "grad_norm": 2.8735876083374023, "learning_rate": 1.3755383541021056e-06, "loss": 0.046708233654499054, "step": 5532 }, { "epoch": 1.4363966770508827, "grad_norm": 2.968817710876465, "learning_rate": 1.3747894919364024e-06, "loss": 0.019683796912431717, "step": 5533 }, { "epoch": 1.436656282450675, "grad_norm": 2.4931042194366455, "learning_rate": 1.3740408728764277e-06, "loss": 0.052478596568107605, "step": 5534 }, { "epoch": 1.4369158878504673, "grad_norm": 3.600841760635376, "learning_rate": 1.3732924970545225e-06, "loss": 0.04376871883869171, "step": 5535 }, { "epoch": 1.4371754932502596, "grad_norm": 3.4740474224090576, "learning_rate": 1.3725443646029811e-06, "loss": 0.04426364600658417, "step": 5536 }, { "epoch": 1.437435098650052, "grad_norm": 3.164607048034668, "learning_rate": 1.3717964756540557e-06, "loss": 0.05209365487098694, "step": 5537 }, { "epoch": 1.4376947040498442, "grad_norm": 4.971129417419434, "learning_rate": 1.3710488303399577e-06, "loss": 0.022272320464253426, "step": 5538 }, { "epoch": 1.4379543094496365, "grad_norm": 2.1299312114715576, "learning_rate": 1.3703014287928528e-06, "loss": 0.06880383938550949, "step": 5539 }, { "epoch": 1.4382139148494288, "grad_norm": 3.554436206817627, "learning_rate": 1.3695542711448636e-06, "loss": 0.014990506693720818, "step": 5540 }, { "epoch": 1.4384735202492211, "grad_norm": 2.34053897857666, "learning_rate": 1.368807357528072e-06, "loss": 0.007963062264025211, "step": 5541 }, { "epoch": 1.4387331256490135, "grad_norm": 2.854259729385376, "learning_rate": 1.368060688074514e-06, "loss": 0.02167411893606186, "step": 5542 }, { "epoch": 1.4389927310488058, "grad_norm": 2.87783145904541, "learning_rate": 1.367314262916184e-06, "loss": 0.04989102482795715, "step": 5543 }, { "epoch": 1.439252336448598, "grad_norm": 3.2020363807678223, "learning_rate": 1.3665680821850335e-06, "loss": 0.03734380006790161, "step": 5544 }, { "epoch": 1.4395119418483904, "grad_norm": 4.317654609680176, "learning_rate": 1.365822146012969e-06, "loss": 0.04165171831846237, "step": 5545 }, { "epoch": 1.4397715472481827, "grad_norm": 2.962331533432007, "learning_rate": 1.3650764545318562e-06, "loss": 0.06088909134268761, "step": 5546 }, { "epoch": 1.440031152647975, "grad_norm": 3.1530017852783203, "learning_rate": 1.3643310078735156e-06, "loss": 0.03165967017412186, "step": 5547 }, { "epoch": 1.4402907580477673, "grad_norm": 2.461361885070801, "learning_rate": 1.3635858061697246e-06, "loss": 0.017158852890133858, "step": 5548 }, { "epoch": 1.4405503634475596, "grad_norm": 3.637171506881714, "learning_rate": 1.3628408495522195e-06, "loss": 0.02889712154865265, "step": 5549 }, { "epoch": 1.440809968847352, "grad_norm": 3.027811288833618, "learning_rate": 1.36209613815269e-06, "loss": 0.011019546538591385, "step": 5550 }, { "epoch": 1.4410695742471442, "grad_norm": 3.537632703781128, "learning_rate": 1.3613516721027837e-06, "loss": 0.04861186817288399, "step": 5551 }, { "epoch": 1.4413291796469365, "grad_norm": 3.0206422805786133, "learning_rate": 1.360607451534107e-06, "loss": 0.07635419070720673, "step": 5552 }, { "epoch": 1.441588785046729, "grad_norm": 2.841587543487549, "learning_rate": 1.3598634765782192e-06, "loss": 0.028681248426437378, "step": 5553 }, { "epoch": 1.4418483904465214, "grad_norm": 2.7127575874328613, "learning_rate": 1.3591197473666401e-06, "loss": 0.016409995034337044, "step": 5554 }, { "epoch": 1.4421079958463137, "grad_norm": 3.369187116622925, "learning_rate": 1.3583762640308412e-06, "loss": 0.020741179585456848, "step": 5555 }, { "epoch": 1.442367601246106, "grad_norm": 3.4866278171539307, "learning_rate": 1.3576330267022552e-06, "loss": 0.03188781440258026, "step": 5556 }, { "epoch": 1.4426272066458983, "grad_norm": 2.161935806274414, "learning_rate": 1.3568900355122697e-06, "loss": 0.008304700255393982, "step": 5557 }, { "epoch": 1.4428868120456906, "grad_norm": 2.8992762565612793, "learning_rate": 1.356147290592228e-06, "loss": 0.02426084131002426, "step": 5558 }, { "epoch": 1.4431464174454829, "grad_norm": 5.0361809730529785, "learning_rate": 1.3554047920734301e-06, "loss": 0.02075262740254402, "step": 5559 }, { "epoch": 1.4434060228452752, "grad_norm": 3.2857584953308105, "learning_rate": 1.3546625400871321e-06, "loss": 0.022935183718800545, "step": 5560 }, { "epoch": 1.4436656282450675, "grad_norm": 3.370650291442871, "learning_rate": 1.3539205347645485e-06, "loss": 0.05493592470884323, "step": 5561 }, { "epoch": 1.4439252336448598, "grad_norm": 3.840369939804077, "learning_rate": 1.3531787762368498e-06, "loss": 0.03760518133640289, "step": 5562 }, { "epoch": 1.4441848390446521, "grad_norm": 3.1000888347625732, "learning_rate": 1.3524372646351585e-06, "loss": 0.047616079449653625, "step": 5563 }, { "epoch": 1.4444444444444444, "grad_norm": 4.750823497772217, "learning_rate": 1.3516960000905601e-06, "loss": 0.026174131780862808, "step": 5564 }, { "epoch": 1.4447040498442367, "grad_norm": 4.979226112365723, "learning_rate": 1.3509549827340917e-06, "loss": 0.01647092029452324, "step": 5565 }, { "epoch": 1.444963655244029, "grad_norm": 2.4472875595092773, "learning_rate": 1.350214212696748e-06, "loss": 0.019792642444372177, "step": 5566 }, { "epoch": 1.4452232606438213, "grad_norm": 3.736905336380005, "learning_rate": 1.3494736901094814e-06, "loss": 0.012467524968087673, "step": 5567 }, { "epoch": 1.4454828660436136, "grad_norm": 2.7668278217315674, "learning_rate": 1.3487334151031975e-06, "loss": 0.019203253090381622, "step": 5568 }, { "epoch": 1.445742471443406, "grad_norm": 3.8021509647369385, "learning_rate": 1.347993387808762e-06, "loss": 0.028513886034488678, "step": 5569 }, { "epoch": 1.4460020768431983, "grad_norm": 2.9855289459228516, "learning_rate": 1.3472536083569937e-06, "loss": 0.018711576238274574, "step": 5570 }, { "epoch": 1.4462616822429908, "grad_norm": 3.3182497024536133, "learning_rate": 1.3465140768786683e-06, "loss": 0.02054009586572647, "step": 5571 }, { "epoch": 1.446521287642783, "grad_norm": 2.5853240489959717, "learning_rate": 1.3457747935045193e-06, "loss": 0.02462918497622013, "step": 5572 }, { "epoch": 1.4467808930425754, "grad_norm": 2.2784693241119385, "learning_rate": 1.345035758365234e-06, "loss": 0.07980221509933472, "step": 5573 }, { "epoch": 1.4470404984423677, "grad_norm": 2.2863073348999023, "learning_rate": 1.3442969715914568e-06, "loss": 0.061994679272174835, "step": 5574 }, { "epoch": 1.44730010384216, "grad_norm": 2.464240550994873, "learning_rate": 1.343558433313789e-06, "loss": 0.011608809232711792, "step": 5575 }, { "epoch": 1.4475597092419523, "grad_norm": 2.7734477519989014, "learning_rate": 1.3428201436627863e-06, "loss": 0.03958674892783165, "step": 5576 }, { "epoch": 1.4478193146417446, "grad_norm": 2.458625078201294, "learning_rate": 1.342082102768963e-06, "loss": 0.06490109860897064, "step": 5577 }, { "epoch": 1.448078920041537, "grad_norm": 2.473761796951294, "learning_rate": 1.3413443107627872e-06, "loss": 0.019312433898448944, "step": 5578 }, { "epoch": 1.4483385254413292, "grad_norm": 2.9066684246063232, "learning_rate": 1.340606767774682e-06, "loss": 0.029979683458805084, "step": 5579 }, { "epoch": 1.4485981308411215, "grad_norm": 2.8905112743377686, "learning_rate": 1.3398694739350303e-06, "loss": 0.036435581743717194, "step": 5580 }, { "epoch": 1.4488577362409139, "grad_norm": 1.8556071519851685, "learning_rate": 1.3391324293741681e-06, "loss": 0.07826616615056992, "step": 5581 }, { "epoch": 1.4491173416407062, "grad_norm": 2.57700252532959, "learning_rate": 1.3383956342223872e-06, "loss": 0.022306213155388832, "step": 5582 }, { "epoch": 1.4493769470404985, "grad_norm": 3.2030510902404785, "learning_rate": 1.3376590886099372e-06, "loss": 0.0516701377928257, "step": 5583 }, { "epoch": 1.4496365524402908, "grad_norm": 3.573655605316162, "learning_rate": 1.3369227926670215e-06, "loss": 0.019809987396001816, "step": 5584 }, { "epoch": 1.449896157840083, "grad_norm": 0.7426949739456177, "learning_rate": 1.336186746523802e-06, "loss": 0.008048705756664276, "step": 5585 }, { "epoch": 1.4501557632398754, "grad_norm": 2.9131593704223633, "learning_rate": 1.3354509503103922e-06, "loss": 0.04994025081396103, "step": 5586 }, { "epoch": 1.4504153686396677, "grad_norm": 3.2721383571624756, "learning_rate": 1.3347154041568655e-06, "loss": 0.04670603573322296, "step": 5587 }, { "epoch": 1.45067497403946, "grad_norm": 2.9555675983428955, "learning_rate": 1.3339801081932507e-06, "loss": 0.050543412566185, "step": 5588 }, { "epoch": 1.4509345794392523, "grad_norm": 3.0420289039611816, "learning_rate": 1.3332450625495298e-06, "loss": 0.02661786414682865, "step": 5589 }, { "epoch": 1.4511941848390446, "grad_norm": 3.7138521671295166, "learning_rate": 1.332510267355643e-06, "loss": 0.028503920882940292, "step": 5590 }, { "epoch": 1.451453790238837, "grad_norm": 2.545107841491699, "learning_rate": 1.3317757227414835e-06, "loss": 0.05630854144692421, "step": 5591 }, { "epoch": 1.4517133956386292, "grad_norm": 2.3698112964630127, "learning_rate": 1.331041428836904e-06, "loss": 0.005838282406330109, "step": 5592 }, { "epoch": 1.4519730010384215, "grad_norm": 1.677958369255066, "learning_rate": 1.3303073857717103e-06, "loss": 0.017667047679424286, "step": 5593 }, { "epoch": 1.4522326064382138, "grad_norm": 3.385061264038086, "learning_rate": 1.3295735936756631e-06, "loss": 0.03302912414073944, "step": 5594 }, { "epoch": 1.4524922118380061, "grad_norm": 3.7192273139953613, "learning_rate": 1.328840052678482e-06, "loss": 0.032926302403211594, "step": 5595 }, { "epoch": 1.4527518172377984, "grad_norm": 2.7025749683380127, "learning_rate": 1.3281067629098394e-06, "loss": 0.01701299473643303, "step": 5596 }, { "epoch": 1.4530114226375908, "grad_norm": 3.4877004623413086, "learning_rate": 1.3273737244993632e-06, "loss": 0.0768735259771347, "step": 5597 }, { "epoch": 1.453271028037383, "grad_norm": 2.9578311443328857, "learning_rate": 1.3266409375766392e-06, "loss": 0.011308277025818825, "step": 5598 }, { "epoch": 1.4535306334371754, "grad_norm": 2.295374631881714, "learning_rate": 1.3259084022712065e-06, "loss": 0.01113523542881012, "step": 5599 }, { "epoch": 1.4537902388369677, "grad_norm": 4.953676223754883, "learning_rate": 1.3251761187125614e-06, "loss": 0.029059330001473427, "step": 5600 }, { "epoch": 1.4540498442367602, "grad_norm": 2.7824833393096924, "learning_rate": 1.3244440870301545e-06, "loss": 0.04272804781794548, "step": 5601 }, { "epoch": 1.4543094496365525, "grad_norm": 3.216758966445923, "learning_rate": 1.3237123073533915e-06, "loss": 0.01624995470046997, "step": 5602 }, { "epoch": 1.4545690550363448, "grad_norm": 3.000317335128784, "learning_rate": 1.3229807798116351e-06, "loss": 0.05216282978653908, "step": 5603 }, { "epoch": 1.4548286604361371, "grad_norm": 3.3883755207061768, "learning_rate": 1.3222495045342032e-06, "loss": 0.024574942886829376, "step": 5604 }, { "epoch": 1.4550882658359294, "grad_norm": 3.3809139728546143, "learning_rate": 1.3215184816503663e-06, "loss": 0.04720041900873184, "step": 5605 }, { "epoch": 1.4553478712357217, "grad_norm": 4.584168910980225, "learning_rate": 1.3207877112893553e-06, "loss": 0.010697178542613983, "step": 5606 }, { "epoch": 1.455607476635514, "grad_norm": 2.820350170135498, "learning_rate": 1.3200571935803513e-06, "loss": 0.018204033374786377, "step": 5607 }, { "epoch": 1.4558670820353063, "grad_norm": 3.3553428649902344, "learning_rate": 1.3193269286524945e-06, "loss": 0.016481071710586548, "step": 5608 }, { "epoch": 1.4561266874350987, "grad_norm": 3.8650259971618652, "learning_rate": 1.318596916634879e-06, "loss": 0.06374470144510269, "step": 5609 }, { "epoch": 1.456386292834891, "grad_norm": 2.5074381828308105, "learning_rate": 1.3178671576565533e-06, "loss": 0.012873200699687004, "step": 5610 }, { "epoch": 1.4566458982346833, "grad_norm": 2.6643571853637695, "learning_rate": 1.3171376518465228e-06, "loss": 0.01404520682990551, "step": 5611 }, { "epoch": 1.4569055036344756, "grad_norm": 2.7177109718322754, "learning_rate": 1.3164083993337478e-06, "loss": 0.011899342760443687, "step": 5612 }, { "epoch": 1.4571651090342679, "grad_norm": 2.1439218521118164, "learning_rate": 1.3156794002471417e-06, "loss": 0.08614011853933334, "step": 5613 }, { "epoch": 1.4574247144340602, "grad_norm": 2.4440338611602783, "learning_rate": 1.314950654715577e-06, "loss": 0.0746769905090332, "step": 5614 }, { "epoch": 1.4576843198338525, "grad_norm": 3.4665169715881348, "learning_rate": 1.3142221628678778e-06, "loss": 0.023839861154556274, "step": 5615 }, { "epoch": 1.4579439252336448, "grad_norm": 3.3228862285614014, "learning_rate": 1.3134939248328265e-06, "loss": 0.062145888805389404, "step": 5616 }, { "epoch": 1.458203530633437, "grad_norm": 3.8663437366485596, "learning_rate": 1.3127659407391564e-06, "loss": 0.026967331767082214, "step": 5617 }, { "epoch": 1.4584631360332294, "grad_norm": 2.401305675506592, "learning_rate": 1.3120382107155595e-06, "loss": 0.054030075669288635, "step": 5618 }, { "epoch": 1.458722741433022, "grad_norm": 3.0547075271606445, "learning_rate": 1.3113107348906833e-06, "loss": 0.027119850739836693, "step": 5619 }, { "epoch": 1.4589823468328142, "grad_norm": 4.106033802032471, "learning_rate": 1.3105835133931264e-06, "loss": 0.02789643220603466, "step": 5620 }, { "epoch": 1.4592419522326066, "grad_norm": 3.052814245223999, "learning_rate": 1.3098565463514467e-06, "loss": 0.03096553310751915, "step": 5621 }, { "epoch": 1.4595015576323989, "grad_norm": 3.988229274749756, "learning_rate": 1.3091298338941538e-06, "loss": 0.031890690326690674, "step": 5622 }, { "epoch": 1.4597611630321912, "grad_norm": 0.24614407122135162, "learning_rate": 1.3084033761497156e-06, "loss": 0.0007861350895836949, "step": 5623 }, { "epoch": 1.4600207684319835, "grad_norm": 2.4067540168762207, "learning_rate": 1.3076771732465521e-06, "loss": 0.023022936657071114, "step": 5624 }, { "epoch": 1.4602803738317758, "grad_norm": 2.3500468730926514, "learning_rate": 1.306951225313039e-06, "loss": 0.07633309066295624, "step": 5625 }, { "epoch": 1.460539979231568, "grad_norm": 1.9812920093536377, "learning_rate": 1.3062255324775086e-06, "loss": 0.01154734194278717, "step": 5626 }, { "epoch": 1.4607995846313604, "grad_norm": 2.7590324878692627, "learning_rate": 1.3055000948682458e-06, "loss": 0.012963579967617989, "step": 5627 }, { "epoch": 1.4610591900311527, "grad_norm": 3.8659772872924805, "learning_rate": 1.3047749126134912e-06, "loss": 0.027579069137573242, "step": 5628 }, { "epoch": 1.461318795430945, "grad_norm": 3.350071907043457, "learning_rate": 1.304049985841441e-06, "loss": 0.01227511279284954, "step": 5629 }, { "epoch": 1.4615784008307373, "grad_norm": 3.421457290649414, "learning_rate": 1.303325314680245e-06, "loss": 0.01754750870168209, "step": 5630 }, { "epoch": 1.4618380062305296, "grad_norm": 3.2848451137542725, "learning_rate": 1.3026008992580096e-06, "loss": 0.032670486718416214, "step": 5631 }, { "epoch": 1.462097611630322, "grad_norm": 3.85939621925354, "learning_rate": 1.3018767397027942e-06, "loss": 0.012718187645077705, "step": 5632 }, { "epoch": 1.4623572170301142, "grad_norm": 3.7218611240386963, "learning_rate": 1.301152836142613e-06, "loss": 0.042102985084056854, "step": 5633 }, { "epoch": 1.4626168224299065, "grad_norm": 3.638282060623169, "learning_rate": 1.3004291887054366e-06, "loss": 0.03727177530527115, "step": 5634 }, { "epoch": 1.4628764278296988, "grad_norm": 2.2052760124206543, "learning_rate": 1.299705797519189e-06, "loss": 0.07606840133666992, "step": 5635 }, { "epoch": 1.4631360332294912, "grad_norm": 2.59858775138855, "learning_rate": 1.2989826627117482e-06, "loss": 0.04189707338809967, "step": 5636 }, { "epoch": 1.4633956386292835, "grad_norm": 2.8608715534210205, "learning_rate": 1.2982597844109498e-06, "loss": 0.019231576472520828, "step": 5637 }, { "epoch": 1.4636552440290758, "grad_norm": 2.1744332313537598, "learning_rate": 1.29753716274458e-06, "loss": 0.005921089090406895, "step": 5638 }, { "epoch": 1.463914849428868, "grad_norm": 3.0960867404937744, "learning_rate": 1.2968147978403835e-06, "loss": 0.023674270138144493, "step": 5639 }, { "epoch": 1.4641744548286604, "grad_norm": 2.263108491897583, "learning_rate": 1.2960926898260575e-06, "loss": 0.07566536962985992, "step": 5640 }, { "epoch": 1.4644340602284527, "grad_norm": 3.684022903442383, "learning_rate": 1.2953708388292533e-06, "loss": 0.012857552617788315, "step": 5641 }, { "epoch": 1.464693665628245, "grad_norm": 3.1961381435394287, "learning_rate": 1.294649244977579e-06, "loss": 0.05012768507003784, "step": 5642 }, { "epoch": 1.4649532710280373, "grad_norm": 3.281507730484009, "learning_rate": 1.2939279083985945e-06, "loss": 0.019813358783721924, "step": 5643 }, { "epoch": 1.4652128764278296, "grad_norm": 3.0907249450683594, "learning_rate": 1.2932068292198157e-06, "loss": 0.06853516399860382, "step": 5644 }, { "epoch": 1.465472481827622, "grad_norm": 3.380527973175049, "learning_rate": 1.292486007568714e-06, "loss": 0.05610665678977966, "step": 5645 }, { "epoch": 1.4657320872274142, "grad_norm": 3.281630516052246, "learning_rate": 1.2917654435727137e-06, "loss": 0.03906415402889252, "step": 5646 }, { "epoch": 1.4659916926272065, "grad_norm": 3.0326900482177734, "learning_rate": 1.2910451373591936e-06, "loss": 0.05597802996635437, "step": 5647 }, { "epoch": 1.4662512980269988, "grad_norm": 3.4904139041900635, "learning_rate": 1.290325089055487e-06, "loss": 0.02801968716084957, "step": 5648 }, { "epoch": 1.4665109034267914, "grad_norm": 3.41601824760437, "learning_rate": 1.2896052987888825e-06, "loss": 0.038017187267541885, "step": 5649 }, { "epoch": 1.4667705088265837, "grad_norm": 3.3638031482696533, "learning_rate": 1.2888857666866243e-06, "loss": 0.028638597577810287, "step": 5650 }, { "epoch": 1.467030114226376, "grad_norm": 3.6364574432373047, "learning_rate": 1.2881664928759051e-06, "loss": 0.05463099479675293, "step": 5651 }, { "epoch": 1.4672897196261683, "grad_norm": 2.8172149658203125, "learning_rate": 1.2874474774838802e-06, "loss": 0.07238151133060455, "step": 5652 }, { "epoch": 1.4675493250259606, "grad_norm": 2.514209270477295, "learning_rate": 1.2867287206376522e-06, "loss": 0.03802824020385742, "step": 5653 }, { "epoch": 1.467808930425753, "grad_norm": 2.098435401916504, "learning_rate": 1.2860102224642823e-06, "loss": 0.06848631799221039, "step": 5654 }, { "epoch": 1.4680685358255452, "grad_norm": 2.8081114292144775, "learning_rate": 1.2852919830907846e-06, "loss": 0.04737629368901253, "step": 5655 }, { "epoch": 1.4683281412253375, "grad_norm": 4.801427841186523, "learning_rate": 1.284574002644126e-06, "loss": 0.022893676534295082, "step": 5656 }, { "epoch": 1.4685877466251298, "grad_norm": 3.025996208190918, "learning_rate": 1.2838562812512306e-06, "loss": 0.010979786515235901, "step": 5657 }, { "epoch": 1.4688473520249221, "grad_norm": 3.8474225997924805, "learning_rate": 1.2831388190389747e-06, "loss": 0.04119924455881119, "step": 5658 }, { "epoch": 1.4691069574247144, "grad_norm": 2.2931325435638428, "learning_rate": 1.2824216161341879e-06, "loss": 0.07503673434257507, "step": 5659 }, { "epoch": 1.4693665628245067, "grad_norm": 2.9900338649749756, "learning_rate": 1.2817046726636573e-06, "loss": 0.02830197475850582, "step": 5660 }, { "epoch": 1.469626168224299, "grad_norm": 2.496701717376709, "learning_rate": 1.28098798875412e-06, "loss": 0.06482379138469696, "step": 5661 }, { "epoch": 1.4698857736240913, "grad_norm": 2.891265630722046, "learning_rate": 1.280271564532271e-06, "loss": 0.024697259068489075, "step": 5662 }, { "epoch": 1.4701453790238836, "grad_norm": 2.866478204727173, "learning_rate": 1.2795554001247574e-06, "loss": 0.02496374025940895, "step": 5663 }, { "epoch": 1.470404984423676, "grad_norm": 2.24411940574646, "learning_rate": 1.2788394956581795e-06, "loss": 0.08278223872184753, "step": 5664 }, { "epoch": 1.4706645898234683, "grad_norm": 2.014235734939575, "learning_rate": 1.2781238512590944e-06, "loss": 0.06910324096679688, "step": 5665 }, { "epoch": 1.4709241952232606, "grad_norm": 2.984532594680786, "learning_rate": 1.277408467054011e-06, "loss": 0.02423151209950447, "step": 5666 }, { "epoch": 1.471183800623053, "grad_norm": 2.440396785736084, "learning_rate": 1.2766933431693919e-06, "loss": 0.01295602135360241, "step": 5667 }, { "epoch": 1.4714434060228454, "grad_norm": 2.128826379776001, "learning_rate": 1.275978479731656e-06, "loss": 0.0066741397604346275, "step": 5668 }, { "epoch": 1.4717030114226377, "grad_norm": 3.0541112422943115, "learning_rate": 1.2752638768671738e-06, "loss": 0.0319041907787323, "step": 5669 }, { "epoch": 1.47196261682243, "grad_norm": 4.383377552032471, "learning_rate": 1.2745495347022722e-06, "loss": 0.04290516674518585, "step": 5670 }, { "epoch": 1.4722222222222223, "grad_norm": 3.830972194671631, "learning_rate": 1.273835453363229e-06, "loss": 0.016028128564357758, "step": 5671 }, { "epoch": 1.4724818276220146, "grad_norm": 2.6787002086639404, "learning_rate": 1.2731216329762775e-06, "loss": 0.04012325033545494, "step": 5672 }, { "epoch": 1.472741433021807, "grad_norm": 2.3416707515716553, "learning_rate": 1.2724080736676064e-06, "loss": 0.024345332756638527, "step": 5673 }, { "epoch": 1.4730010384215992, "grad_norm": 6.463390350341797, "learning_rate": 1.2716947755633547e-06, "loss": 0.021787656471133232, "step": 5674 }, { "epoch": 1.4732606438213915, "grad_norm": 3.6693179607391357, "learning_rate": 1.2709817387896173e-06, "loss": 0.005358321592211723, "step": 5675 }, { "epoch": 1.4735202492211839, "grad_norm": 3.161750555038452, "learning_rate": 1.2702689634724446e-06, "loss": 0.04121840000152588, "step": 5676 }, { "epoch": 1.4737798546209762, "grad_norm": 2.306196451187134, "learning_rate": 1.2695564497378377e-06, "loss": 0.015870334580540657, "step": 5677 }, { "epoch": 1.4740394600207685, "grad_norm": 3.2229597568511963, "learning_rate": 1.2688441977117526e-06, "loss": 0.04525560885667801, "step": 5678 }, { "epoch": 1.4742990654205608, "grad_norm": 2.7227020263671875, "learning_rate": 1.2681322075200988e-06, "loss": 0.08387388288974762, "step": 5679 }, { "epoch": 1.474558670820353, "grad_norm": 2.7915029525756836, "learning_rate": 1.2674204792887406e-06, "loss": 0.06706103682518005, "step": 5680 }, { "epoch": 1.4748182762201454, "grad_norm": 3.5741848945617676, "learning_rate": 1.2667090131434961e-06, "loss": 0.010649814270436764, "step": 5681 }, { "epoch": 1.4750778816199377, "grad_norm": 1.6632115840911865, "learning_rate": 1.2659978092101343e-06, "loss": 0.01026606373488903, "step": 5682 }, { "epoch": 1.47533748701973, "grad_norm": 2.516544818878174, "learning_rate": 1.2652868676143806e-06, "loss": 0.00837092287838459, "step": 5683 }, { "epoch": 1.4755970924195223, "grad_norm": 3.103147506713867, "learning_rate": 1.2645761884819136e-06, "loss": 0.016163744032382965, "step": 5684 }, { "epoch": 1.4758566978193146, "grad_norm": 3.0583913326263428, "learning_rate": 1.263865771938365e-06, "loss": 0.07911628484725952, "step": 5685 }, { "epoch": 1.476116303219107, "grad_norm": 3.3984575271606445, "learning_rate": 1.2631556181093202e-06, "loss": 0.04730922728776932, "step": 5686 }, { "epoch": 1.4763759086188992, "grad_norm": 3.274928569793701, "learning_rate": 1.2624457271203172e-06, "loss": 0.018755819648504257, "step": 5687 }, { "epoch": 1.4766355140186915, "grad_norm": 1.476504921913147, "learning_rate": 1.2617360990968486e-06, "loss": 0.0017407590057700872, "step": 5688 }, { "epoch": 1.4768951194184838, "grad_norm": 3.3494691848754883, "learning_rate": 1.2610267341643628e-06, "loss": 0.035720326006412506, "step": 5689 }, { "epoch": 1.4771547248182761, "grad_norm": 2.5131783485412598, "learning_rate": 1.260317632448256e-06, "loss": 0.06826398521661758, "step": 5690 }, { "epoch": 1.4774143302180685, "grad_norm": 2.7654037475585938, "learning_rate": 1.2596087940738832e-06, "loss": 0.06507483124732971, "step": 5691 }, { "epoch": 1.4776739356178608, "grad_norm": 2.2517285346984863, "learning_rate": 1.2589002191665495e-06, "loss": 0.06138203293085098, "step": 5692 }, { "epoch": 1.477933541017653, "grad_norm": 1.8241020441055298, "learning_rate": 1.2581919078515161e-06, "loss": 0.009923228994011879, "step": 5693 }, { "epoch": 1.4781931464174454, "grad_norm": 2.909043550491333, "learning_rate": 1.2574838602539955e-06, "loss": 0.05766008049249649, "step": 5694 }, { "epoch": 1.4784527518172377, "grad_norm": 3.204176664352417, "learning_rate": 1.2567760764991536e-06, "loss": 0.06912857294082642, "step": 5695 }, { "epoch": 1.47871235721703, "grad_norm": 3.6781985759735107, "learning_rate": 1.2560685567121116e-06, "loss": 0.011400489136576653, "step": 5696 }, { "epoch": 1.4789719626168225, "grad_norm": 2.7607738971710205, "learning_rate": 1.2553613010179427e-06, "loss": 0.012663488276302814, "step": 5697 }, { "epoch": 1.4792315680166148, "grad_norm": 3.7975046634674072, "learning_rate": 1.2546543095416718e-06, "loss": 0.020113125443458557, "step": 5698 }, { "epoch": 1.4794911734164071, "grad_norm": 2.2523436546325684, "learning_rate": 1.2539475824082813e-06, "loss": 0.012195728719234467, "step": 5699 }, { "epoch": 1.4797507788161994, "grad_norm": 2.641871213912964, "learning_rate": 1.2532411197427034e-06, "loss": 0.00961492769420147, "step": 5700 }, { "epoch": 1.4800103842159917, "grad_norm": 2.5609965324401855, "learning_rate": 1.2525349216698229e-06, "loss": 0.05807751417160034, "step": 5701 }, { "epoch": 1.480269989615784, "grad_norm": 2.28363037109375, "learning_rate": 1.2518289883144819e-06, "loss": 0.07087504863739014, "step": 5702 }, { "epoch": 1.4805295950155763, "grad_norm": 2.8420016765594482, "learning_rate": 1.2511233198014716e-06, "loss": 0.010793551802635193, "step": 5703 }, { "epoch": 1.4807892004153687, "grad_norm": 3.774203062057495, "learning_rate": 1.2504179162555392e-06, "loss": 0.03969400376081467, "step": 5704 }, { "epoch": 1.481048805815161, "grad_norm": 3.5678606033325195, "learning_rate": 1.2497127778013833e-06, "loss": 0.052854202687740326, "step": 5705 }, { "epoch": 1.4813084112149533, "grad_norm": 1.4151113033294678, "learning_rate": 1.2490079045636554e-06, "loss": 0.0027961120940744877, "step": 5706 }, { "epoch": 1.4815680166147456, "grad_norm": 2.6847434043884277, "learning_rate": 1.248303296666963e-06, "loss": 0.012571994215250015, "step": 5707 }, { "epoch": 1.4818276220145379, "grad_norm": 3.230311870574951, "learning_rate": 1.247598954235863e-06, "loss": 0.029261425137519836, "step": 5708 }, { "epoch": 1.4820872274143302, "grad_norm": 2.9771499633789062, "learning_rate": 1.2468948773948669e-06, "loss": 0.014318069443106651, "step": 5709 }, { "epoch": 1.4823468328141225, "grad_norm": 3.7717604637145996, "learning_rate": 1.2461910662684404e-06, "loss": 0.013312248513102531, "step": 5710 }, { "epoch": 1.4826064382139148, "grad_norm": 2.962834358215332, "learning_rate": 1.2454875209810005e-06, "loss": 0.0724666491150856, "step": 5711 }, { "epoch": 1.482866043613707, "grad_norm": 4.182078838348389, "learning_rate": 1.2447842416569192e-06, "loss": 0.03972841054201126, "step": 5712 }, { "epoch": 1.4831256490134994, "grad_norm": 3.5180063247680664, "learning_rate": 1.244081228420518e-06, "loss": 0.012692023999989033, "step": 5713 }, { "epoch": 1.4833852544132917, "grad_norm": 3.644559860229492, "learning_rate": 1.2433784813960743e-06, "loss": 0.015200439840555191, "step": 5714 }, { "epoch": 1.4836448598130842, "grad_norm": 2.280226945877075, "learning_rate": 1.2426760007078188e-06, "loss": 0.0172518752515316, "step": 5715 }, { "epoch": 1.4839044652128766, "grad_norm": 3.3247640132904053, "learning_rate": 1.2419737864799334e-06, "loss": 0.05213788524270058, "step": 5716 }, { "epoch": 1.4841640706126689, "grad_norm": 3.2446913719177246, "learning_rate": 1.2412718388365532e-06, "loss": 0.03270568326115608, "step": 5717 }, { "epoch": 1.4844236760124612, "grad_norm": 3.512308120727539, "learning_rate": 1.2405701579017658e-06, "loss": 0.015170164406299591, "step": 5718 }, { "epoch": 1.4846832814122535, "grad_norm": 1.998274564743042, "learning_rate": 1.2398687437996133e-06, "loss": 0.005640925373882055, "step": 5719 }, { "epoch": 1.4849428868120458, "grad_norm": 3.1448214054107666, "learning_rate": 1.239167596654091e-06, "loss": 0.04237372428178787, "step": 5720 }, { "epoch": 1.485202492211838, "grad_norm": 2.778479814529419, "learning_rate": 1.2384667165891424e-06, "loss": 0.05872299522161484, "step": 5721 }, { "epoch": 1.4854620976116304, "grad_norm": 3.3329060077667236, "learning_rate": 1.2377661037286698e-06, "loss": 0.01274645235389471, "step": 5722 }, { "epoch": 1.4857217030114227, "grad_norm": 4.598614692687988, "learning_rate": 1.2370657581965236e-06, "loss": 0.03927575424313545, "step": 5723 }, { "epoch": 1.485981308411215, "grad_norm": 2.2093496322631836, "learning_rate": 1.2363656801165104e-06, "loss": 0.007231594063341618, "step": 5724 }, { "epoch": 1.4862409138110073, "grad_norm": 3.5963141918182373, "learning_rate": 1.2356658696123873e-06, "loss": 0.023587463423609734, "step": 5725 }, { "epoch": 1.4865005192107996, "grad_norm": 5.043284893035889, "learning_rate": 1.234966326807864e-06, "loss": 0.026009757071733475, "step": 5726 }, { "epoch": 1.486760124610592, "grad_norm": 2.5796759128570557, "learning_rate": 1.2342670518266057e-06, "loss": 0.009758107364177704, "step": 5727 }, { "epoch": 1.4870197300103842, "grad_norm": 3.1885530948638916, "learning_rate": 1.2335680447922266e-06, "loss": 0.004457726143300533, "step": 5728 }, { "epoch": 1.4872793354101765, "grad_norm": 3.600877523422241, "learning_rate": 1.2328693058282945e-06, "loss": 0.0679779201745987, "step": 5729 }, { "epoch": 1.4875389408099688, "grad_norm": 3.2262558937072754, "learning_rate": 1.2321708350583324e-06, "loss": 0.025720397010445595, "step": 5730 }, { "epoch": 1.4877985462097612, "grad_norm": 3.731330156326294, "learning_rate": 1.231472632605813e-06, "loss": 0.03341691195964813, "step": 5731 }, { "epoch": 1.4880581516095535, "grad_norm": 4.5873870849609375, "learning_rate": 1.230774698594162e-06, "loss": 0.03658095747232437, "step": 5732 }, { "epoch": 1.4883177570093458, "grad_norm": 3.6251566410064697, "learning_rate": 1.230077033146759e-06, "loss": 0.039687179028987885, "step": 5733 }, { "epoch": 1.488577362409138, "grad_norm": 2.6979920864105225, "learning_rate": 1.2293796363869345e-06, "loss": 0.06760502606630325, "step": 5734 }, { "epoch": 1.4888369678089304, "grad_norm": 3.1961753368377686, "learning_rate": 1.2286825084379733e-06, "loss": 0.059539955109357834, "step": 5735 }, { "epoch": 1.4890965732087227, "grad_norm": 3.063847541809082, "learning_rate": 1.227985649423111e-06, "loss": 0.048677489161491394, "step": 5736 }, { "epoch": 1.489356178608515, "grad_norm": 3.369086265563965, "learning_rate": 1.2272890594655356e-06, "loss": 0.02229205146431923, "step": 5737 }, { "epoch": 1.4896157840083073, "grad_norm": 4.871644973754883, "learning_rate": 1.2265927386883898e-06, "loss": 0.0178153607994318, "step": 5738 }, { "epoch": 1.4898753894080996, "grad_norm": 3.1351351737976074, "learning_rate": 1.2258966872147664e-06, "loss": 0.033313024789094925, "step": 5739 }, { "epoch": 1.490134994807892, "grad_norm": 2.7139477729797363, "learning_rate": 1.2252009051677106e-06, "loss": 0.07205891609191895, "step": 5740 }, { "epoch": 1.4903946002076842, "grad_norm": 4.410611152648926, "learning_rate": 1.2245053926702224e-06, "loss": 0.025264767929911613, "step": 5741 }, { "epoch": 1.4906542056074765, "grad_norm": 2.437192678451538, "learning_rate": 1.2238101498452507e-06, "loss": 0.06328118592500687, "step": 5742 }, { "epoch": 1.4909138110072688, "grad_norm": 2.795896530151367, "learning_rate": 1.2231151768157008e-06, "loss": 0.06718496978282928, "step": 5743 }, { "epoch": 1.4911734164070611, "grad_norm": 2.645540714263916, "learning_rate": 1.2224204737044252e-06, "loss": 0.004510780796408653, "step": 5744 }, { "epoch": 1.4914330218068537, "grad_norm": 3.425400733947754, "learning_rate": 1.221726040634233e-06, "loss": 0.015102043747901917, "step": 5745 }, { "epoch": 1.491692627206646, "grad_norm": 2.4488108158111572, "learning_rate": 1.2210318777278844e-06, "loss": 0.008938306011259556, "step": 5746 }, { "epoch": 1.4919522326064383, "grad_norm": 2.994657039642334, "learning_rate": 1.2203379851080914e-06, "loss": 0.0620756670832634, "step": 5747 }, { "epoch": 1.4922118380062306, "grad_norm": 2.245035409927368, "learning_rate": 1.219644362897518e-06, "loss": 0.014796005561947823, "step": 5748 }, { "epoch": 1.492471443406023, "grad_norm": 3.326707363128662, "learning_rate": 1.21895101121878e-06, "loss": 0.04403781145811081, "step": 5749 }, { "epoch": 1.4927310488058152, "grad_norm": 2.142963409423828, "learning_rate": 1.218257930194447e-06, "loss": 0.011305319145321846, "step": 5750 }, { "epoch": 1.4929906542056075, "grad_norm": 2.7296180725097656, "learning_rate": 1.2175651199470408e-06, "loss": 0.021091053262352943, "step": 5751 }, { "epoch": 1.4932502596053998, "grad_norm": 3.8368256092071533, "learning_rate": 1.2168725805990319e-06, "loss": 0.033172838389873505, "step": 5752 }, { "epoch": 1.4935098650051921, "grad_norm": 2.459038496017456, "learning_rate": 1.2161803122728476e-06, "loss": 0.012122288346290588, "step": 5753 }, { "epoch": 1.4937694704049844, "grad_norm": 3.283615827560425, "learning_rate": 1.2154883150908635e-06, "loss": 0.04497780278325081, "step": 5754 }, { "epoch": 1.4940290758047767, "grad_norm": 3.6792104244232178, "learning_rate": 1.2147965891754104e-06, "loss": 0.008308577351272106, "step": 5755 }, { "epoch": 1.494288681204569, "grad_norm": 2.8905444145202637, "learning_rate": 1.2141051346487692e-06, "loss": 0.019875582307577133, "step": 5756 }, { "epoch": 1.4945482866043613, "grad_norm": 3.339442014694214, "learning_rate": 1.2134139516331715e-06, "loss": 0.0585363432765007, "step": 5757 }, { "epoch": 1.4948078920041536, "grad_norm": 3.2059974670410156, "learning_rate": 1.212723040250805e-06, "loss": 0.0490235798060894, "step": 5758 }, { "epoch": 1.495067497403946, "grad_norm": 9.816157341003418, "learning_rate": 1.212032400623806e-06, "loss": 0.0461495965719223, "step": 5759 }, { "epoch": 1.4953271028037383, "grad_norm": 3.5069234371185303, "learning_rate": 1.2113420328742629e-06, "loss": 0.035058487206697464, "step": 5760 }, { "epoch": 1.4955867082035306, "grad_norm": 4.6298017501831055, "learning_rate": 1.2106519371242188e-06, "loss": 0.010046019218862057, "step": 5761 }, { "epoch": 1.4958463136033229, "grad_norm": 3.2118661403656006, "learning_rate": 1.2099621134956656e-06, "loss": 0.030118361115455627, "step": 5762 }, { "epoch": 1.4961059190031154, "grad_norm": 5.14497184753418, "learning_rate": 1.209272562110548e-06, "loss": 0.010778933763504028, "step": 5763 }, { "epoch": 1.4963655244029077, "grad_norm": 2.6932692527770996, "learning_rate": 1.2085832830907642e-06, "loss": 0.008445004932582378, "step": 5764 }, { "epoch": 1.4966251298027, "grad_norm": 3.198401927947998, "learning_rate": 1.2078942765581611e-06, "loss": 0.0206961240619421, "step": 5765 }, { "epoch": 1.4968847352024923, "grad_norm": 2.471867561340332, "learning_rate": 1.2072055426345414e-06, "loss": 0.007395833730697632, "step": 5766 }, { "epoch": 1.4971443406022846, "grad_norm": 2.5242767333984375, "learning_rate": 1.2065170814416564e-06, "loss": 0.02375493198633194, "step": 5767 }, { "epoch": 1.497403946002077, "grad_norm": 4.367153167724609, "learning_rate": 1.2058288931012095e-06, "loss": 0.048437487334012985, "step": 5768 }, { "epoch": 1.4976635514018692, "grad_norm": 2.4381535053253174, "learning_rate": 1.205140977734858e-06, "loss": 0.089644655585289, "step": 5769 }, { "epoch": 1.4979231568016615, "grad_norm": 2.610635995864868, "learning_rate": 1.2044533354642095e-06, "loss": 0.017441436648368835, "step": 5770 }, { "epoch": 1.4981827622014539, "grad_norm": 3.061462163925171, "learning_rate": 1.2037659664108222e-06, "loss": 0.048718031495809555, "step": 5771 }, { "epoch": 1.4984423676012462, "grad_norm": 2.028364658355713, "learning_rate": 1.2030788706962085e-06, "loss": 0.003169652074575424, "step": 5772 }, { "epoch": 1.4987019730010385, "grad_norm": 2.4215543270111084, "learning_rate": 1.20239204844183e-06, "loss": 0.0677376389503479, "step": 5773 }, { "epoch": 1.4989615784008308, "grad_norm": 2.9585702419281006, "learning_rate": 1.2017054997691032e-06, "loss": 0.016534756869077682, "step": 5774 }, { "epoch": 1.499221183800623, "grad_norm": 3.492462396621704, "learning_rate": 1.2010192247993915e-06, "loss": 0.03774246945977211, "step": 5775 }, { "epoch": 1.4994807892004154, "grad_norm": 2.326472043991089, "learning_rate": 1.200333223654014e-06, "loss": 0.06607542932033539, "step": 5776 }, { "epoch": 1.4997403946002077, "grad_norm": 1.8344504833221436, "learning_rate": 1.19964749645424e-06, "loss": 0.006107655353844166, "step": 5777 }, { "epoch": 1.5, "grad_norm": 3.217616319656372, "learning_rate": 1.1989620433212908e-06, "loss": 0.02899688109755516, "step": 5778 }, { "epoch": 1.5002596053997923, "grad_norm": 3.2804946899414062, "learning_rate": 1.1982768643763382e-06, "loss": 0.0492829829454422, "step": 5779 }, { "epoch": 1.5005192107995846, "grad_norm": 2.9389607906341553, "learning_rate": 1.1975919597405054e-06, "loss": 0.04581580311059952, "step": 5780 }, { "epoch": 1.500778816199377, "grad_norm": 3.6959173679351807, "learning_rate": 1.1969073295348686e-06, "loss": 0.055841706693172455, "step": 5781 }, { "epoch": 1.5010384215991692, "grad_norm": 3.139291286468506, "learning_rate": 1.1962229738804565e-06, "loss": 0.022199004888534546, "step": 5782 }, { "epoch": 1.5012980269989615, "grad_norm": 2.0920908451080322, "learning_rate": 1.1955388928982444e-06, "loss": 0.01946653053164482, "step": 5783 }, { "epoch": 1.5015576323987538, "grad_norm": 3.5300726890563965, "learning_rate": 1.1948550867091643e-06, "loss": 0.02793831378221512, "step": 5784 }, { "epoch": 1.5018172377985461, "grad_norm": 3.5405325889587402, "learning_rate": 1.1941715554340965e-06, "loss": 0.025646720081567764, "step": 5785 }, { "epoch": 1.5020768431983385, "grad_norm": 3.134455680847168, "learning_rate": 1.1934882991938737e-06, "loss": 0.039480842649936676, "step": 5786 }, { "epoch": 1.5023364485981308, "grad_norm": 2.9411840438842773, "learning_rate": 1.1928053181092804e-06, "loss": 0.08001668006181717, "step": 5787 }, { "epoch": 1.502596053997923, "grad_norm": 3.100590229034424, "learning_rate": 1.1921226123010515e-06, "loss": 0.019322363659739494, "step": 5788 }, { "epoch": 1.5028556593977154, "grad_norm": 3.806201696395874, "learning_rate": 1.1914401818898745e-06, "loss": 0.03650045767426491, "step": 5789 }, { "epoch": 1.5031152647975077, "grad_norm": 1.9871293306350708, "learning_rate": 1.1907580269963874e-06, "loss": 0.013895587995648384, "step": 5790 }, { "epoch": 1.5033748701973, "grad_norm": 2.8713889122009277, "learning_rate": 1.1900761477411782e-06, "loss": 0.03571634739637375, "step": 5791 }, { "epoch": 1.5036344755970923, "grad_norm": 4.212135314941406, "learning_rate": 1.1893945442447892e-06, "loss": 0.040482297539711, "step": 5792 }, { "epoch": 1.5038940809968846, "grad_norm": 3.0338354110717773, "learning_rate": 1.188713216627712e-06, "loss": 0.046884749084711075, "step": 5793 }, { "epoch": 1.504153686396677, "grad_norm": 3.392336130142212, "learning_rate": 1.1880321650103884e-06, "loss": 0.062321096658706665, "step": 5794 }, { "epoch": 1.5044132917964692, "grad_norm": 3.103687047958374, "learning_rate": 1.1873513895132145e-06, "loss": 0.05476059392094612, "step": 5795 }, { "epoch": 1.5046728971962615, "grad_norm": 3.1429293155670166, "learning_rate": 1.1866708902565346e-06, "loss": 0.03256668895483017, "step": 5796 }, { "epoch": 1.504932502596054, "grad_norm": 3.474369764328003, "learning_rate": 1.1859906673606462e-06, "loss": 0.03513437509536743, "step": 5797 }, { "epoch": 1.5051921079958464, "grad_norm": 2.710721492767334, "learning_rate": 1.1853107209457966e-06, "loss": 0.06058942526578903, "step": 5798 }, { "epoch": 1.5054517133956387, "grad_norm": 3.9227240085601807, "learning_rate": 1.1846310511321845e-06, "loss": 0.06156359612941742, "step": 5799 }, { "epoch": 1.505711318795431, "grad_norm": 3.5134031772613525, "learning_rate": 1.1839516580399607e-06, "loss": 0.014398854225873947, "step": 5800 }, { "epoch": 1.5059709241952233, "grad_norm": 3.6887106895446777, "learning_rate": 1.1832725417892263e-06, "loss": 0.010451210662722588, "step": 5801 }, { "epoch": 1.5062305295950156, "grad_norm": 2.9599101543426514, "learning_rate": 1.1825937025000325e-06, "loss": 0.040439292788505554, "step": 5802 }, { "epoch": 1.5064901349948079, "grad_norm": 2.9875175952911377, "learning_rate": 1.181915140292384e-06, "loss": 0.022783106192946434, "step": 5803 }, { "epoch": 1.5067497403946002, "grad_norm": 3.012317419052124, "learning_rate": 1.1812368552862336e-06, "loss": 0.033187806606292725, "step": 5804 }, { "epoch": 1.5070093457943925, "grad_norm": 4.293739318847656, "learning_rate": 1.1805588476014886e-06, "loss": 0.02210313454270363, "step": 5805 }, { "epoch": 1.5072689511941848, "grad_norm": 2.9257426261901855, "learning_rate": 1.1798811173580024e-06, "loss": 0.04956849291920662, "step": 5806 }, { "epoch": 1.507528556593977, "grad_norm": 2.685997724533081, "learning_rate": 1.1792036646755838e-06, "loss": 0.012573360465466976, "step": 5807 }, { "epoch": 1.5077881619937694, "grad_norm": 3.375978708267212, "learning_rate": 1.1785264896739914e-06, "loss": 0.044043637812137604, "step": 5808 }, { "epoch": 1.508047767393562, "grad_norm": 2.176368236541748, "learning_rate": 1.177849592472934e-06, "loss": 0.007941343821585178, "step": 5809 }, { "epoch": 1.5083073727933543, "grad_norm": 2.173344612121582, "learning_rate": 1.1771729731920707e-06, "loss": 0.06519637256860733, "step": 5810 }, { "epoch": 1.5085669781931466, "grad_norm": 2.2376415729522705, "learning_rate": 1.1764966319510124e-06, "loss": 0.0752631351351738, "step": 5811 }, { "epoch": 1.5088265835929389, "grad_norm": 2.5031228065490723, "learning_rate": 1.1758205688693219e-06, "loss": 0.06682145595550537, "step": 5812 }, { "epoch": 1.5090861889927312, "grad_norm": 3.080662727355957, "learning_rate": 1.1751447840665105e-06, "loss": 0.021470502018928528, "step": 5813 }, { "epoch": 1.5093457943925235, "grad_norm": 3.588252067565918, "learning_rate": 1.1744692776620414e-06, "loss": 0.03641190379858017, "step": 5814 }, { "epoch": 1.5096053997923158, "grad_norm": 2.566959857940674, "learning_rate": 1.17379404977533e-06, "loss": 0.017545092850923538, "step": 5815 }, { "epoch": 1.509865005192108, "grad_norm": 3.119546413421631, "learning_rate": 1.1731191005257399e-06, "loss": 0.020749010145664215, "step": 5816 }, { "epoch": 1.5101246105919004, "grad_norm": 2.522639274597168, "learning_rate": 1.1724444300325864e-06, "loss": 0.005773248616605997, "step": 5817 }, { "epoch": 1.5103842159916927, "grad_norm": 4.008302211761475, "learning_rate": 1.1717700384151375e-06, "loss": 0.039653923362493515, "step": 5818 }, { "epoch": 1.510643821391485, "grad_norm": 2.385284662246704, "learning_rate": 1.1710959257926084e-06, "loss": 0.07541614770889282, "step": 5819 }, { "epoch": 1.5109034267912773, "grad_norm": 3.846834182739258, "learning_rate": 1.1704220922841678e-06, "loss": 0.021793460473418236, "step": 5820 }, { "epoch": 1.5111630321910696, "grad_norm": 3.473569393157959, "learning_rate": 1.169748538008934e-06, "loss": 0.03654153272509575, "step": 5821 }, { "epoch": 1.511422637590862, "grad_norm": 3.232428550720215, "learning_rate": 1.1690752630859748e-06, "loss": 0.02671245113015175, "step": 5822 }, { "epoch": 1.5116822429906542, "grad_norm": 3.1110758781433105, "learning_rate": 1.1684022676343118e-06, "loss": 0.0314708836376667, "step": 5823 }, { "epoch": 1.5119418483904465, "grad_norm": 3.2169291973114014, "learning_rate": 1.1677295517729138e-06, "loss": 0.040247052907943726, "step": 5824 }, { "epoch": 1.5122014537902388, "grad_norm": 3.0089504718780518, "learning_rate": 1.1670571156207013e-06, "loss": 0.01765754632651806, "step": 5825 }, { "epoch": 1.5124610591900312, "grad_norm": 3.7242255210876465, "learning_rate": 1.1663849592965466e-06, "loss": 0.03250645101070404, "step": 5826 }, { "epoch": 1.5127206645898235, "grad_norm": 3.2046713829040527, "learning_rate": 1.1657130829192708e-06, "loss": 0.05355687066912651, "step": 5827 }, { "epoch": 1.5129802699896158, "grad_norm": 3.4674129486083984, "learning_rate": 1.1650414866076472e-06, "loss": 0.04508782550692558, "step": 5828 }, { "epoch": 1.513239875389408, "grad_norm": 3.385378837585449, "learning_rate": 1.1643701704803977e-06, "loss": 0.04524723440408707, "step": 5829 }, { "epoch": 1.5134994807892004, "grad_norm": 3.4502575397491455, "learning_rate": 1.1636991346561956e-06, "loss": 0.02697659656405449, "step": 5830 }, { "epoch": 1.5137590861889927, "grad_norm": 3.668416976928711, "learning_rate": 1.1630283792536657e-06, "loss": 0.026760129258036613, "step": 5831 }, { "epoch": 1.514018691588785, "grad_norm": 2.983189582824707, "learning_rate": 1.1623579043913814e-06, "loss": 0.017656531184911728, "step": 5832 }, { "epoch": 1.5142782969885773, "grad_norm": 2.785245418548584, "learning_rate": 1.1616877101878665e-06, "loss": 0.020145071670413017, "step": 5833 }, { "epoch": 1.5145379023883696, "grad_norm": 2.4029974937438965, "learning_rate": 1.161017796761598e-06, "loss": 0.05835771560668945, "step": 5834 }, { "epoch": 1.514797507788162, "grad_norm": 3.482235908508301, "learning_rate": 1.1603481642309992e-06, "loss": 0.015597711317241192, "step": 5835 }, { "epoch": 1.5150571131879542, "grad_norm": 3.5803275108337402, "learning_rate": 1.1596788127144485e-06, "loss": 0.04875823110342026, "step": 5836 }, { "epoch": 1.5153167185877465, "grad_norm": 3.752309560775757, "learning_rate": 1.1590097423302683e-06, "loss": 0.016361579298973083, "step": 5837 }, { "epoch": 1.5155763239875388, "grad_norm": 2.45013165473938, "learning_rate": 1.1583409531967372e-06, "loss": 0.06577788293361664, "step": 5838 }, { "epoch": 1.5158359293873311, "grad_norm": 2.091688871383667, "learning_rate": 1.1576724454320828e-06, "loss": 0.007232510484755039, "step": 5839 }, { "epoch": 1.5160955347871234, "grad_norm": 2.2388408184051514, "learning_rate": 1.1570042191544787e-06, "loss": 0.07920731604099274, "step": 5840 }, { "epoch": 1.5163551401869158, "grad_norm": 3.9745054244995117, "learning_rate": 1.156336274482055e-06, "loss": 0.017847789451479912, "step": 5841 }, { "epoch": 1.516614745586708, "grad_norm": 2.7765421867370605, "learning_rate": 1.1556686115328872e-06, "loss": 0.010131465271115303, "step": 5842 }, { "epoch": 1.5168743509865004, "grad_norm": 3.6395320892333984, "learning_rate": 1.1550012304250043e-06, "loss": 0.00840020552277565, "step": 5843 }, { "epoch": 1.5171339563862927, "grad_norm": 1.9818265438079834, "learning_rate": 1.154334131276383e-06, "loss": 0.008887174539268017, "step": 5844 }, { "epoch": 1.5173935617860852, "grad_norm": 2.990837812423706, "learning_rate": 1.1536673142049507e-06, "loss": 0.022325439378619194, "step": 5845 }, { "epoch": 1.5176531671858775, "grad_norm": 2.7926647663116455, "learning_rate": 1.153000779328587e-06, "loss": 0.03561854735016823, "step": 5846 }, { "epoch": 1.5179127725856698, "grad_norm": 3.2615458965301514, "learning_rate": 1.1523345267651188e-06, "loss": 0.018494918942451477, "step": 5847 }, { "epoch": 1.5181723779854621, "grad_norm": 3.467766523361206, "learning_rate": 1.1516685566323243e-06, "loss": 0.030251268297433853, "step": 5848 }, { "epoch": 1.5184319833852544, "grad_norm": 3.6417057514190674, "learning_rate": 1.1510028690479323e-06, "loss": 0.030334942042827606, "step": 5849 }, { "epoch": 1.5186915887850467, "grad_norm": 1.7529380321502686, "learning_rate": 1.1503374641296204e-06, "loss": 0.01313760969787836, "step": 5850 }, { "epoch": 1.518951194184839, "grad_norm": 2.7005269527435303, "learning_rate": 1.1496723419950178e-06, "loss": 0.06664001941680908, "step": 5851 }, { "epoch": 1.5192107995846313, "grad_norm": 2.976409912109375, "learning_rate": 1.149007502761703e-06, "loss": 0.038270384073257446, "step": 5852 }, { "epoch": 1.5194704049844237, "grad_norm": 3.0785458087921143, "learning_rate": 1.1483429465472026e-06, "loss": 0.05225815996527672, "step": 5853 }, { "epoch": 1.519730010384216, "grad_norm": 3.1315720081329346, "learning_rate": 1.1476786734689973e-06, "loss": 0.018861787393689156, "step": 5854 }, { "epoch": 1.5199896157840083, "grad_norm": 2.943499803543091, "learning_rate": 1.1470146836445141e-06, "loss": 0.054814841598272324, "step": 5855 }, { "epoch": 1.5202492211838006, "grad_norm": 2.4165236949920654, "learning_rate": 1.1463509771911305e-06, "loss": 0.09083057194948196, "step": 5856 }, { "epoch": 1.520508826583593, "grad_norm": 3.7699406147003174, "learning_rate": 1.1456875542261765e-06, "loss": 0.0413057878613472, "step": 5857 }, { "epoch": 1.5207684319833854, "grad_norm": 3.1741414070129395, "learning_rate": 1.145024414866928e-06, "loss": 0.05042139068245888, "step": 5858 }, { "epoch": 1.5210280373831777, "grad_norm": 3.3652327060699463, "learning_rate": 1.1443615592306147e-06, "loss": 0.024815257638692856, "step": 5859 }, { "epoch": 1.52128764278297, "grad_norm": 3.1493630409240723, "learning_rate": 1.1436989874344135e-06, "loss": 0.013266335241496563, "step": 5860 }, { "epoch": 1.5215472481827623, "grad_norm": 3.397082567214966, "learning_rate": 1.143036699595451e-06, "loss": 0.016529981046915054, "step": 5861 }, { "epoch": 1.5218068535825546, "grad_norm": 2.418262004852295, "learning_rate": 1.1423746958308066e-06, "loss": 0.05052381008863449, "step": 5862 }, { "epoch": 1.522066458982347, "grad_norm": 2.8507914543151855, "learning_rate": 1.1417129762575057e-06, "loss": 0.028578976169228554, "step": 5863 }, { "epoch": 1.5223260643821392, "grad_norm": 2.65006685256958, "learning_rate": 1.1410515409925252e-06, "loss": 0.05484924837946892, "step": 5864 }, { "epoch": 1.5225856697819315, "grad_norm": 2.8795254230499268, "learning_rate": 1.1403903901527929e-06, "loss": 0.04126124456524849, "step": 5865 }, { "epoch": 1.5228452751817239, "grad_norm": 2.324016571044922, "learning_rate": 1.1397295238551847e-06, "loss": 0.06377570331096649, "step": 5866 }, { "epoch": 1.5231048805815162, "grad_norm": 1.7729110717773438, "learning_rate": 1.1390689422165262e-06, "loss": 0.007861233316361904, "step": 5867 }, { "epoch": 1.5233644859813085, "grad_norm": 2.541954755783081, "learning_rate": 1.1384086453535925e-06, "loss": 0.05786486715078354, "step": 5868 }, { "epoch": 1.5236240913811008, "grad_norm": 2.800994873046875, "learning_rate": 1.1377486333831099e-06, "loss": 0.01383329089730978, "step": 5869 }, { "epoch": 1.523883696780893, "grad_norm": 3.2409427165985107, "learning_rate": 1.1370889064217545e-06, "loss": 0.017983589321374893, "step": 5870 }, { "epoch": 1.5241433021806854, "grad_norm": 4.118898868560791, "learning_rate": 1.1364294645861483e-06, "loss": 0.03223484009504318, "step": 5871 }, { "epoch": 1.5244029075804777, "grad_norm": 2.7943742275238037, "learning_rate": 1.1357703079928676e-06, "loss": 0.017498616129159927, "step": 5872 }, { "epoch": 1.52466251298027, "grad_norm": 3.043269634246826, "learning_rate": 1.1351114367584347e-06, "loss": 0.05202166736125946, "step": 5873 }, { "epoch": 1.5249221183800623, "grad_norm": 3.711871385574341, "learning_rate": 1.1344528509993246e-06, "loss": 0.023267215117812157, "step": 5874 }, { "epoch": 1.5251817237798546, "grad_norm": 4.149894714355469, "learning_rate": 1.1337945508319593e-06, "loss": 0.04756813868880272, "step": 5875 }, { "epoch": 1.525441329179647, "grad_norm": 3.5504403114318848, "learning_rate": 1.1331365363727106e-06, "loss": 0.018220432102680206, "step": 5876 }, { "epoch": 1.5257009345794392, "grad_norm": 2.36854887008667, "learning_rate": 1.1324788077379015e-06, "loss": 0.006851114332675934, "step": 5877 }, { "epoch": 1.5259605399792315, "grad_norm": 3.8289268016815186, "learning_rate": 1.1318213650438028e-06, "loss": 0.033151187002658844, "step": 5878 }, { "epoch": 1.5262201453790238, "grad_norm": 3.4150636196136475, "learning_rate": 1.131164208406635e-06, "loss": 0.02266603708267212, "step": 5879 }, { "epoch": 1.5264797507788161, "grad_norm": 4.166406154632568, "learning_rate": 1.130507337942569e-06, "loss": 0.022111352533102036, "step": 5880 }, { "epoch": 1.5267393561786085, "grad_norm": 3.7297606468200684, "learning_rate": 1.129850753767724e-06, "loss": 0.03352087736129761, "step": 5881 }, { "epoch": 1.5269989615784008, "grad_norm": 3.475661039352417, "learning_rate": 1.1291944559981697e-06, "loss": 0.018941251561045647, "step": 5882 }, { "epoch": 1.527258566978193, "grad_norm": 3.6323955059051514, "learning_rate": 1.128538444749924e-06, "loss": 0.017774496227502823, "step": 5883 }, { "epoch": 1.5275181723779854, "grad_norm": 3.694056510925293, "learning_rate": 1.1278827201389539e-06, "loss": 0.027571551501750946, "step": 5884 }, { "epoch": 1.5277777777777777, "grad_norm": 2.98040509223938, "learning_rate": 1.1272272822811784e-06, "loss": 0.01240449771285057, "step": 5885 }, { "epoch": 1.52803738317757, "grad_norm": 3.785050630569458, "learning_rate": 1.1265721312924624e-06, "loss": 0.014017906971275806, "step": 5886 }, { "epoch": 1.5282969885773623, "grad_norm": 2.5819551944732666, "learning_rate": 1.1259172672886216e-06, "loss": 0.04609810560941696, "step": 5887 }, { "epoch": 1.5285565939771546, "grad_norm": 2.7499215602874756, "learning_rate": 1.1252626903854218e-06, "loss": 0.01446227915585041, "step": 5888 }, { "epoch": 1.528816199376947, "grad_norm": 3.5793213844299316, "learning_rate": 1.1246084006985764e-06, "loss": 0.05163326859474182, "step": 5889 }, { "epoch": 1.5290758047767392, "grad_norm": 2.8467578887939453, "learning_rate": 1.1239543983437498e-06, "loss": 0.04298338294029236, "step": 5890 }, { "epoch": 1.5293354101765315, "grad_norm": 2.6782357692718506, "learning_rate": 1.1233006834365542e-06, "loss": 0.051191676408052444, "step": 5891 }, { "epoch": 1.5295950155763238, "grad_norm": 2.4012231826782227, "learning_rate": 1.1226472560925502e-06, "loss": 0.08577841520309448, "step": 5892 }, { "epoch": 1.5298546209761164, "grad_norm": 2.697341203689575, "learning_rate": 1.1219941164272515e-06, "loss": 0.011484208516776562, "step": 5893 }, { "epoch": 1.5301142263759087, "grad_norm": 2.7541935443878174, "learning_rate": 1.1213412645561154e-06, "loss": 0.01362624578177929, "step": 5894 }, { "epoch": 1.530373831775701, "grad_norm": 3.341464042663574, "learning_rate": 1.1206887005945521e-06, "loss": 0.021385839208960533, "step": 5895 }, { "epoch": 1.5306334371754933, "grad_norm": 3.060429096221924, "learning_rate": 1.1200364246579213e-06, "loss": 0.027516372501850128, "step": 5896 }, { "epoch": 1.5308930425752856, "grad_norm": 3.2814888954162598, "learning_rate": 1.119384436861529e-06, "loss": 0.024846451357007027, "step": 5897 }, { "epoch": 1.5311526479750779, "grad_norm": 3.924819231033325, "learning_rate": 1.1187327373206323e-06, "loss": 0.008275222033262253, "step": 5898 }, { "epoch": 1.5314122533748702, "grad_norm": 3.9987375736236572, "learning_rate": 1.118081326150436e-06, "loss": 0.02461778186261654, "step": 5899 }, { "epoch": 1.5316718587746625, "grad_norm": 2.544034004211426, "learning_rate": 1.117430203466095e-06, "loss": 0.013100758194923401, "step": 5900 }, { "epoch": 1.5319314641744548, "grad_norm": 3.324747323989868, "learning_rate": 1.1167793693827145e-06, "loss": 0.02965763956308365, "step": 5901 }, { "epoch": 1.532191069574247, "grad_norm": 3.6658949851989746, "learning_rate": 1.1161288240153446e-06, "loss": 0.014334321022033691, "step": 5902 }, { "epoch": 1.5324506749740394, "grad_norm": 3.8848371505737305, "learning_rate": 1.1154785674789883e-06, "loss": 0.036729846149683, "step": 5903 }, { "epoch": 1.5327102803738317, "grad_norm": 2.1565749645233154, "learning_rate": 1.1148285998885956e-06, "loss": 0.004049520939588547, "step": 5904 }, { "epoch": 1.5329698857736243, "grad_norm": 3.079608917236328, "learning_rate": 1.114178921359066e-06, "loss": 0.02131674811244011, "step": 5905 }, { "epoch": 1.5332294911734166, "grad_norm": 3.112781524658203, "learning_rate": 1.113529532005248e-06, "loss": 0.04322507977485657, "step": 5906 }, { "epoch": 1.5334890965732089, "grad_norm": 3.081530809402466, "learning_rate": 1.112880431941938e-06, "loss": 0.04248815029859543, "step": 5907 }, { "epoch": 1.5337487019730012, "grad_norm": 2.5660006999969482, "learning_rate": 1.112231621283883e-06, "loss": 0.03857989236712456, "step": 5908 }, { "epoch": 1.5340083073727935, "grad_norm": 2.772744655609131, "learning_rate": 1.1115831001457775e-06, "loss": 0.04990009963512421, "step": 5909 }, { "epoch": 1.5342679127725858, "grad_norm": 4.071662902832031, "learning_rate": 1.1109348686422643e-06, "loss": 0.03293571621179581, "step": 5910 }, { "epoch": 1.534527518172378, "grad_norm": 3.828979730606079, "learning_rate": 1.1102869268879379e-06, "loss": 0.023773085325956345, "step": 5911 }, { "epoch": 1.5347871235721704, "grad_norm": 2.5166821479797363, "learning_rate": 1.1096392749973373e-06, "loss": 0.06305243819952011, "step": 5912 }, { "epoch": 1.5350467289719627, "grad_norm": 2.873974323272705, "learning_rate": 1.1089919130849543e-06, "loss": 0.014373221434652805, "step": 5913 }, { "epoch": 1.535306334371755, "grad_norm": 2.1271824836730957, "learning_rate": 1.108344841265227e-06, "loss": 0.010575995780527592, "step": 5914 }, { "epoch": 1.5355659397715473, "grad_norm": 6.789743900299072, "learning_rate": 1.1076980596525422e-06, "loss": 0.027998175472021103, "step": 5915 }, { "epoch": 1.5358255451713396, "grad_norm": 1.6128038167953491, "learning_rate": 1.1070515683612373e-06, "loss": 0.004380626603960991, "step": 5916 }, { "epoch": 1.536085150571132, "grad_norm": 6.887283802032471, "learning_rate": 1.1064053675055967e-06, "loss": 0.017194611951708794, "step": 5917 }, { "epoch": 1.5363447559709242, "grad_norm": 2.3681833744049072, "learning_rate": 1.1057594571998528e-06, "loss": 0.007251486647874117, "step": 5918 }, { "epoch": 1.5366043613707165, "grad_norm": 2.6961476802825928, "learning_rate": 1.10511383755819e-06, "loss": 0.01633729785680771, "step": 5919 }, { "epoch": 1.5368639667705088, "grad_norm": 2.528278350830078, "learning_rate": 1.1044685086947376e-06, "loss": 0.061371687799692154, "step": 5920 }, { "epoch": 1.5371235721703012, "grad_norm": 3.4626877307891846, "learning_rate": 1.1038234707235742e-06, "loss": 0.02167516015470028, "step": 5921 }, { "epoch": 1.5373831775700935, "grad_norm": 3.63873028755188, "learning_rate": 1.1031787237587296e-06, "loss": 0.03962811455130577, "step": 5922 }, { "epoch": 1.5376427829698858, "grad_norm": 3.2897493839263916, "learning_rate": 1.102534267914179e-06, "loss": 0.015233078971505165, "step": 5923 }, { "epoch": 1.537902388369678, "grad_norm": 2.4494216442108154, "learning_rate": 1.101890103303849e-06, "loss": 0.012740007601678371, "step": 5924 }, { "epoch": 1.5381619937694704, "grad_norm": 2.9475300312042236, "learning_rate": 1.1012462300416104e-06, "loss": 0.006567790172994137, "step": 5925 }, { "epoch": 1.5384215991692627, "grad_norm": 3.4284517765045166, "learning_rate": 1.100602648241287e-06, "loss": 0.011388196609914303, "step": 5926 }, { "epoch": 1.538681204569055, "grad_norm": 2.960329294204712, "learning_rate": 1.0999593580166501e-06, "loss": 0.053233616054058075, "step": 5927 }, { "epoch": 1.5389408099688473, "grad_norm": 2.4702131748199463, "learning_rate": 1.099316359481418e-06, "loss": 0.013176534324884415, "step": 5928 }, { "epoch": 1.5392004153686396, "grad_norm": 2.9803762435913086, "learning_rate": 1.0986736527492573e-06, "loss": 0.07270682603120804, "step": 5929 }, { "epoch": 1.539460020768432, "grad_norm": 4.075784206390381, "learning_rate": 1.0980312379337844e-06, "loss": 0.05035470798611641, "step": 5930 }, { "epoch": 1.5397196261682242, "grad_norm": 3.5710294246673584, "learning_rate": 1.097389115148563e-06, "loss": 0.03901968151330948, "step": 5931 }, { "epoch": 1.5399792315680165, "grad_norm": 1.5934594869613647, "learning_rate": 1.0967472845071078e-06, "loss": 0.003513597883284092, "step": 5932 }, { "epoch": 1.5402388369678088, "grad_norm": 3.7451438903808594, "learning_rate": 1.0961057461228768e-06, "loss": 0.03226010128855705, "step": 5933 }, { "epoch": 1.5404984423676011, "grad_norm": 5.4228973388671875, "learning_rate": 1.0954645001092818e-06, "loss": 0.018758244812488556, "step": 5934 }, { "epoch": 1.5407580477673934, "grad_norm": 2.9474356174468994, "learning_rate": 1.0948235465796784e-06, "loss": 0.028780827298760414, "step": 5935 }, { "epoch": 1.5410176531671858, "grad_norm": 5.293044567108154, "learning_rate": 1.0941828856473738e-06, "loss": 0.02048862725496292, "step": 5936 }, { "epoch": 1.541277258566978, "grad_norm": 3.836970329284668, "learning_rate": 1.0935425174256224e-06, "loss": 0.011890966445207596, "step": 5937 }, { "epoch": 1.5415368639667704, "grad_norm": 3.354802370071411, "learning_rate": 1.092902442027625e-06, "loss": 0.01455449964851141, "step": 5938 }, { "epoch": 1.5417964693665627, "grad_norm": 2.8922696113586426, "learning_rate": 1.0922626595665343e-06, "loss": 0.011650580912828445, "step": 5939 }, { "epoch": 1.542056074766355, "grad_norm": 3.412388801574707, "learning_rate": 1.091623170155448e-06, "loss": 0.05465855076909065, "step": 5940 }, { "epoch": 1.5423156801661475, "grad_norm": 3.2316322326660156, "learning_rate": 1.0909839739074126e-06, "loss": 0.012182437814772129, "step": 5941 }, { "epoch": 1.5425752855659398, "grad_norm": 2.9497039318084717, "learning_rate": 1.0903450709354253e-06, "loss": 0.013891180977225304, "step": 5942 }, { "epoch": 1.5428348909657321, "grad_norm": 2.834291458129883, "learning_rate": 1.0897064613524273e-06, "loss": 0.016894495114684105, "step": 5943 }, { "epoch": 1.5430944963655244, "grad_norm": 3.266740560531616, "learning_rate": 1.0890681452713123e-06, "loss": 0.0071730283088982105, "step": 5944 }, { "epoch": 1.5433541017653167, "grad_norm": 2.484445810317993, "learning_rate": 1.0884301228049187e-06, "loss": 0.02791060321033001, "step": 5945 }, { "epoch": 1.543613707165109, "grad_norm": 2.9735147953033447, "learning_rate": 1.0877923940660338e-06, "loss": 0.025556685402989388, "step": 5946 }, { "epoch": 1.5438733125649013, "grad_norm": 4.148834705352783, "learning_rate": 1.087154959167395e-06, "loss": 0.028689367696642876, "step": 5947 }, { "epoch": 1.5441329179646937, "grad_norm": 3.1122868061065674, "learning_rate": 1.0865178182216852e-06, "loss": 0.012097139842808247, "step": 5948 }, { "epoch": 1.544392523364486, "grad_norm": 3.2898733615875244, "learning_rate": 1.0858809713415355e-06, "loss": 0.017389945685863495, "step": 5949 }, { "epoch": 1.5446521287642783, "grad_norm": 2.2835025787353516, "learning_rate": 1.0852444186395279e-06, "loss": 0.012848848477005959, "step": 5950 }, { "epoch": 1.5449117341640706, "grad_norm": 3.279489278793335, "learning_rate": 1.0846081602281889e-06, "loss": 0.03857088088989258, "step": 5951 }, { "epoch": 1.5451713395638629, "grad_norm": 4.122847557067871, "learning_rate": 1.0839721962199942e-06, "loss": 0.014642994850873947, "step": 5952 }, { "epoch": 1.5454309449636554, "grad_norm": 2.7695107460021973, "learning_rate": 1.083336526727369e-06, "loss": 0.059321314096450806, "step": 5953 }, { "epoch": 1.5456905503634477, "grad_norm": 3.1730663776397705, "learning_rate": 1.0827011518626836e-06, "loss": 0.02664191462099552, "step": 5954 }, { "epoch": 1.54595015576324, "grad_norm": 2.87562894821167, "learning_rate": 1.0820660717382598e-06, "loss": 0.011237302795052528, "step": 5955 }, { "epoch": 1.5462097611630323, "grad_norm": 3.4448580741882324, "learning_rate": 1.0814312864663625e-06, "loss": 0.029949471354484558, "step": 5956 }, { "epoch": 1.5464693665628246, "grad_norm": 3.323120594024658, "learning_rate": 1.0807967961592086e-06, "loss": 0.007435671053826809, "step": 5957 }, { "epoch": 1.546728971962617, "grad_norm": 2.698356866836548, "learning_rate": 1.0801626009289621e-06, "loss": 0.0886525884270668, "step": 5958 }, { "epoch": 1.5469885773624092, "grad_norm": 1.5418096780776978, "learning_rate": 1.0795287008877334e-06, "loss": 0.012370597571134567, "step": 5959 }, { "epoch": 1.5472481827622016, "grad_norm": 3.9526565074920654, "learning_rate": 1.0788950961475819e-06, "loss": 0.05577130615711212, "step": 5960 }, { "epoch": 1.5475077881619939, "grad_norm": 3.0225391387939453, "learning_rate": 1.0782617868205138e-06, "loss": 0.003439157037064433, "step": 5961 }, { "epoch": 1.5477673935617862, "grad_norm": 2.314069986343384, "learning_rate": 1.0776287730184837e-06, "loss": 0.08195759356021881, "step": 5962 }, { "epoch": 1.5480269989615785, "grad_norm": 4.385042667388916, "learning_rate": 1.0769960548533955e-06, "loss": 0.012291313149034977, "step": 5963 }, { "epoch": 1.5482866043613708, "grad_norm": 3.5320403575897217, "learning_rate": 1.076363632437097e-06, "loss": 0.05150085315108299, "step": 5964 }, { "epoch": 1.548546209761163, "grad_norm": 3.6797025203704834, "learning_rate": 1.0757315058813877e-06, "loss": 0.021111007779836655, "step": 5965 }, { "epoch": 1.5488058151609554, "grad_norm": 3.6908838748931885, "learning_rate": 1.0750996752980117e-06, "loss": 0.03878699243068695, "step": 5966 }, { "epoch": 1.5490654205607477, "grad_norm": 2.2306227684020996, "learning_rate": 1.0744681407986636e-06, "loss": 0.09063901007175446, "step": 5967 }, { "epoch": 1.54932502596054, "grad_norm": 3.1501543521881104, "learning_rate": 1.073836902494984e-06, "loss": 0.049494218081235886, "step": 5968 }, { "epoch": 1.5495846313603323, "grad_norm": 2.421182870864868, "learning_rate": 1.0732059604985602e-06, "loss": 0.0620611198246479, "step": 5969 }, { "epoch": 1.5498442367601246, "grad_norm": 4.643164157867432, "learning_rate": 1.0725753149209295e-06, "loss": 0.027811748906970024, "step": 5970 }, { "epoch": 1.550103842159917, "grad_norm": 3.4281084537506104, "learning_rate": 1.0719449658735754e-06, "loss": 0.007331306114792824, "step": 5971 }, { "epoch": 1.5503634475597092, "grad_norm": 6.131853103637695, "learning_rate": 1.0713149134679281e-06, "loss": 0.031091101467609406, "step": 5972 }, { "epoch": 1.5506230529595015, "grad_norm": 1.4248390197753906, "learning_rate": 1.0706851578153681e-06, "loss": 0.004615104291588068, "step": 5973 }, { "epoch": 1.5508826583592938, "grad_norm": 2.7840898036956787, "learning_rate": 1.0700556990272203e-06, "loss": 0.059496741741895676, "step": 5974 }, { "epoch": 1.5511422637590861, "grad_norm": 2.2190046310424805, "learning_rate": 1.06942653721476e-06, "loss": 0.006611331831663847, "step": 5975 }, { "epoch": 1.5514018691588785, "grad_norm": 2.6382901668548584, "learning_rate": 1.0687976724892078e-06, "loss": 0.053744345903396606, "step": 5976 }, { "epoch": 1.5516614745586708, "grad_norm": 0.9341540932655334, "learning_rate": 1.0681691049617322e-06, "loss": 0.0035230880603194237, "step": 5977 }, { "epoch": 1.551921079958463, "grad_norm": 2.988525629043579, "learning_rate": 1.0675408347434504e-06, "loss": 0.03851496800780296, "step": 5978 }, { "epoch": 1.5521806853582554, "grad_norm": 2.0889272689819336, "learning_rate": 1.066912861945426e-06, "loss": 0.006159774027764797, "step": 5979 }, { "epoch": 1.5524402907580477, "grad_norm": 4.803558826446533, "learning_rate": 1.0662851866786694e-06, "loss": 0.03481912612915039, "step": 5980 }, { "epoch": 1.55269989615784, "grad_norm": 3.5278146266937256, "learning_rate": 1.0656578090541402e-06, "loss": 0.0432179719209671, "step": 5981 }, { "epoch": 1.5529595015576323, "grad_norm": 2.281902551651001, "learning_rate": 1.0650307291827443e-06, "loss": 0.06305908411741257, "step": 5982 }, { "epoch": 1.5532191069574246, "grad_norm": 3.3287227153778076, "learning_rate": 1.0644039471753343e-06, "loss": 0.04124075919389725, "step": 5983 }, { "epoch": 1.553478712357217, "grad_norm": 2.6717848777770996, "learning_rate": 1.0637774631427117e-06, "loss": 0.018580220639705658, "step": 5984 }, { "epoch": 1.5537383177570092, "grad_norm": 1.4864970445632935, "learning_rate": 1.0631512771956238e-06, "loss": 0.004978463985025883, "step": 5985 }, { "epoch": 1.5539979231568015, "grad_norm": 3.304771661758423, "learning_rate": 1.0625253894447676e-06, "loss": 0.04651980474591255, "step": 5986 }, { "epoch": 1.5542575285565938, "grad_norm": 2.3373453617095947, "learning_rate": 1.0618998000007832e-06, "loss": 0.01584557257592678, "step": 5987 }, { "epoch": 1.5545171339563861, "grad_norm": 4.21213436126709, "learning_rate": 1.061274508974262e-06, "loss": 0.014536900445818901, "step": 5988 }, { "epoch": 1.5547767393561787, "grad_norm": 3.483581781387329, "learning_rate": 1.060649516475741e-06, "loss": 0.045409195125103, "step": 5989 }, { "epoch": 1.555036344755971, "grad_norm": 1.8257497549057007, "learning_rate": 1.0600248226157051e-06, "loss": 0.010033663362264633, "step": 5990 }, { "epoch": 1.5552959501557633, "grad_norm": 1.8130930662155151, "learning_rate": 1.0594004275045849e-06, "loss": 0.011984898708760738, "step": 5991 }, { "epoch": 1.5555555555555556, "grad_norm": 2.874058723449707, "learning_rate": 1.058776331252759e-06, "loss": 0.014200393110513687, "step": 5992 }, { "epoch": 1.5558151609553479, "grad_norm": 3.7888126373291016, "learning_rate": 1.058152533970554e-06, "loss": 0.018937911838293076, "step": 5993 }, { "epoch": 1.5560747663551402, "grad_norm": 3.3128151893615723, "learning_rate": 1.0575290357682439e-06, "loss": 0.03094058483839035, "step": 5994 }, { "epoch": 1.5563343717549325, "grad_norm": 1.832593321800232, "learning_rate": 1.0569058367560468e-06, "loss": 0.009496623650193214, "step": 5995 }, { "epoch": 1.5565939771547248, "grad_norm": 2.894681692123413, "learning_rate": 1.0562829370441304e-06, "loss": 0.020697742700576782, "step": 5996 }, { "epoch": 1.5568535825545171, "grad_norm": 2.215606451034546, "learning_rate": 1.055660336742611e-06, "loss": 0.010667365975677967, "step": 5997 }, { "epoch": 1.5571131879543094, "grad_norm": 2.792120933532715, "learning_rate": 1.0550380359615487e-06, "loss": 0.06003246456384659, "step": 5998 }, { "epoch": 1.5573727933541017, "grad_norm": 1.6430641412734985, "learning_rate": 1.054416034810952e-06, "loss": 0.0026031858287751675, "step": 5999 }, { "epoch": 1.557632398753894, "grad_norm": 3.52690052986145, "learning_rate": 1.0537943334007762e-06, "loss": 0.019734874367713928, "step": 6000 }, { "epoch": 1.5578920041536866, "grad_norm": 2.4740042686462402, "learning_rate": 1.0531729318409242e-06, "loss": 0.0030430187471210957, "step": 6001 }, { "epoch": 1.5581516095534789, "grad_norm": 3.9435083866119385, "learning_rate": 1.052551830241247e-06, "loss": 0.05228842422366142, "step": 6002 }, { "epoch": 1.5584112149532712, "grad_norm": 3.417067289352417, "learning_rate": 1.0519310287115383e-06, "loss": 0.009538101963698864, "step": 6003 }, { "epoch": 1.5586708203530635, "grad_norm": 4.633639335632324, "learning_rate": 1.051310527361544e-06, "loss": 0.019425228238105774, "step": 6004 }, { "epoch": 1.5589304257528558, "grad_norm": 3.345468044281006, "learning_rate": 1.0506903263009536e-06, "loss": 0.018661610782146454, "step": 6005 }, { "epoch": 1.559190031152648, "grad_norm": 3.558122158050537, "learning_rate": 1.0500704256394037e-06, "loss": 0.046585142612457275, "step": 6006 }, { "epoch": 1.5594496365524404, "grad_norm": 3.8676323890686035, "learning_rate": 1.04945082548648e-06, "loss": 0.010239707306027412, "step": 6007 }, { "epoch": 1.5597092419522327, "grad_norm": 3.1111669540405273, "learning_rate": 1.0488315259517124e-06, "loss": 0.06460879743099213, "step": 6008 }, { "epoch": 1.559968847352025, "grad_norm": 1.8188085556030273, "learning_rate": 1.0482125271445798e-06, "loss": 0.013983279466629028, "step": 6009 }, { "epoch": 1.5602284527518173, "grad_norm": 3.712860345840454, "learning_rate": 1.0475938291745069e-06, "loss": 0.011204374954104424, "step": 6010 }, { "epoch": 1.5604880581516096, "grad_norm": 3.9242541790008545, "learning_rate": 1.0469754321508641e-06, "loss": 0.03702811524271965, "step": 6011 }, { "epoch": 1.560747663551402, "grad_norm": 2.849442481994629, "learning_rate": 1.046357336182972e-06, "loss": 0.08289183676242828, "step": 6012 }, { "epoch": 1.5610072689511942, "grad_norm": 3.253671169281006, "learning_rate": 1.045739541380094e-06, "loss": 0.0232237596064806, "step": 6013 }, { "epoch": 1.5612668743509865, "grad_norm": 2.5983080863952637, "learning_rate": 1.0451220478514424e-06, "loss": 0.022166460752487183, "step": 6014 }, { "epoch": 1.5615264797507789, "grad_norm": 3.6826164722442627, "learning_rate": 1.044504855706177e-06, "loss": 0.026493791490793228, "step": 6015 }, { "epoch": 1.5617860851505712, "grad_norm": 4.277162075042725, "learning_rate": 1.0438879650534016e-06, "loss": 0.013263793662190437, "step": 6016 }, { "epoch": 1.5620456905503635, "grad_norm": 2.8319883346557617, "learning_rate": 1.0432713760021696e-06, "loss": 0.014978813007473946, "step": 6017 }, { "epoch": 1.5623052959501558, "grad_norm": 4.130578994750977, "learning_rate": 1.0426550886614797e-06, "loss": 0.02888006903231144, "step": 6018 }, { "epoch": 1.562564901349948, "grad_norm": 2.8533785343170166, "learning_rate": 1.0420391031402763e-06, "loss": 0.01760982908308506, "step": 6019 }, { "epoch": 1.5628245067497404, "grad_norm": 3.3643033504486084, "learning_rate": 1.041423419547453e-06, "loss": 0.04809294641017914, "step": 6020 }, { "epoch": 1.5630841121495327, "grad_norm": 4.509416103363037, "learning_rate": 1.0408080379918481e-06, "loss": 0.028857922181487083, "step": 6021 }, { "epoch": 1.563343717549325, "grad_norm": 4.0345458984375, "learning_rate": 1.0401929585822459e-06, "loss": 0.03575378283858299, "step": 6022 }, { "epoch": 1.5636033229491173, "grad_norm": 4.292766571044922, "learning_rate": 1.0395781814273801e-06, "loss": 0.028540533035993576, "step": 6023 }, { "epoch": 1.5638629283489096, "grad_norm": 2.7896523475646973, "learning_rate": 1.0389637066359278e-06, "loss": 0.02790146693587303, "step": 6024 }, { "epoch": 1.564122533748702, "grad_norm": 3.1561739444732666, "learning_rate": 1.0383495343165154e-06, "loss": 0.011744383722543716, "step": 6025 }, { "epoch": 1.5643821391484942, "grad_norm": 4.133475303649902, "learning_rate": 1.037735664577713e-06, "loss": 0.02292957901954651, "step": 6026 }, { "epoch": 1.5646417445482865, "grad_norm": 3.210843563079834, "learning_rate": 1.0371220975280393e-06, "loss": 0.02109566703438759, "step": 6027 }, { "epoch": 1.5649013499480788, "grad_norm": 2.8667571544647217, "learning_rate": 1.0365088332759596e-06, "loss": 0.006093067117035389, "step": 6028 }, { "epoch": 1.5651609553478711, "grad_norm": 2.346506357192993, "learning_rate": 1.0358958719298845e-06, "loss": 0.0069222827441990376, "step": 6029 }, { "epoch": 1.5654205607476634, "grad_norm": 1.0839704275131226, "learning_rate": 1.0352832135981711e-06, "loss": 0.009260784834623337, "step": 6030 }, { "epoch": 1.5656801661474558, "grad_norm": 3.41593337059021, "learning_rate": 1.0346708583891237e-06, "loss": 0.03976923227310181, "step": 6031 }, { "epoch": 1.565939771547248, "grad_norm": 3.2327637672424316, "learning_rate": 1.034058806410993e-06, "loss": 0.04925837740302086, "step": 6032 }, { "epoch": 1.5661993769470404, "grad_norm": 4.645875453948975, "learning_rate": 1.0334470577719755e-06, "loss": 0.026971517130732536, "step": 6033 }, { "epoch": 1.5664589823468327, "grad_norm": 2.479020833969116, "learning_rate": 1.0328356125802136e-06, "loss": 0.07036156952381134, "step": 6034 }, { "epoch": 1.566718587746625, "grad_norm": 3.2822253704071045, "learning_rate": 1.0322244709437983e-06, "loss": 0.033101871609687805, "step": 6035 }, { "epoch": 1.5669781931464173, "grad_norm": 3.6861014366149902, "learning_rate": 1.0316136329707649e-06, "loss": 0.0189609844237566, "step": 6036 }, { "epoch": 1.5672377985462098, "grad_norm": 4.209863185882568, "learning_rate": 1.0310030987690945e-06, "loss": 0.028111033141613007, "step": 6037 }, { "epoch": 1.5674974039460021, "grad_norm": 3.7667367458343506, "learning_rate": 1.030392868446717e-06, "loss": 0.033383600413799286, "step": 6038 }, { "epoch": 1.5677570093457944, "grad_norm": 3.3268280029296875, "learning_rate": 1.029782942111506e-06, "loss": 0.025702359154820442, "step": 6039 }, { "epoch": 1.5680166147455867, "grad_norm": 3.791375160217285, "learning_rate": 1.0291733198712838e-06, "loss": 0.030155519023537636, "step": 6040 }, { "epoch": 1.568276220145379, "grad_norm": 3.0979158878326416, "learning_rate": 1.0285640018338166e-06, "loss": 0.04091858118772507, "step": 6041 }, { "epoch": 1.5685358255451713, "grad_norm": 5.632453918457031, "learning_rate": 1.027954988106818e-06, "loss": 0.03398680314421654, "step": 6042 }, { "epoch": 1.5687954309449637, "grad_norm": 3.4315574169158936, "learning_rate": 1.0273462787979485e-06, "loss": 0.02172502502799034, "step": 6043 }, { "epoch": 1.569055036344756, "grad_norm": 3.0921785831451416, "learning_rate": 1.0267378740148131e-06, "loss": 0.06610605120658875, "step": 6044 }, { "epoch": 1.5693146417445483, "grad_norm": 3.416264772415161, "learning_rate": 1.0261297738649637e-06, "loss": 0.013549447059631348, "step": 6045 }, { "epoch": 1.5695742471443406, "grad_norm": 3.8332841396331787, "learning_rate": 1.0255219784558992e-06, "loss": 0.03640022128820419, "step": 6046 }, { "epoch": 1.5698338525441329, "grad_norm": 3.9438624382019043, "learning_rate": 1.0249144878950634e-06, "loss": 0.015433716587722301, "step": 6047 }, { "epoch": 1.5700934579439252, "grad_norm": 2.656564950942993, "learning_rate": 1.0243073022898472e-06, "loss": 0.018354032188653946, "step": 6048 }, { "epoch": 1.5703530633437177, "grad_norm": 3.780195474624634, "learning_rate": 1.023700421747587e-06, "loss": 0.00845298357307911, "step": 6049 }, { "epoch": 1.57061266874351, "grad_norm": 2.9780266284942627, "learning_rate": 1.0230938463755648e-06, "loss": 0.022716393694281578, "step": 6050 }, { "epoch": 1.5708722741433023, "grad_norm": 2.3737645149230957, "learning_rate": 1.0224875762810102e-06, "loss": 0.08142545074224472, "step": 6051 }, { "epoch": 1.5711318795430946, "grad_norm": 4.308854103088379, "learning_rate": 1.021881611571097e-06, "loss": 0.023497479036450386, "step": 6052 }, { "epoch": 1.571391484942887, "grad_norm": 3.2573306560516357, "learning_rate": 1.0212759523529456e-06, "loss": 0.034080054610967636, "step": 6053 }, { "epoch": 1.5716510903426792, "grad_norm": 3.000516176223755, "learning_rate": 1.020670598733624e-06, "loss": 0.06511278450489044, "step": 6054 }, { "epoch": 1.5719106957424716, "grad_norm": 2.302377700805664, "learning_rate": 1.0200655508201438e-06, "loss": 0.008604870177805424, "step": 6055 }, { "epoch": 1.5721703011422639, "grad_norm": 3.1646640300750732, "learning_rate": 1.019460808719465e-06, "loss": 0.05486108362674713, "step": 6056 }, { "epoch": 1.5724299065420562, "grad_norm": 2.9201819896698, "learning_rate": 1.0188563725384898e-06, "loss": 0.06626080721616745, "step": 6057 }, { "epoch": 1.5726895119418485, "grad_norm": 4.133954048156738, "learning_rate": 1.01825224238407e-06, "loss": 0.0393059141933918, "step": 6058 }, { "epoch": 1.5729491173416408, "grad_norm": 2.958925485610962, "learning_rate": 1.0176484183630035e-06, "loss": 0.01302148774266243, "step": 6059 }, { "epoch": 1.573208722741433, "grad_norm": 2.377572536468506, "learning_rate": 1.0170449005820296e-06, "loss": 0.07289588451385498, "step": 6060 }, { "epoch": 1.5734683281412254, "grad_norm": 4.090234279632568, "learning_rate": 1.0164416891478382e-06, "loss": 0.005740927532315254, "step": 6061 }, { "epoch": 1.5737279335410177, "grad_norm": 2.598135232925415, "learning_rate": 1.0158387841670626e-06, "loss": 0.015602339059114456, "step": 6062 }, { "epoch": 1.57398753894081, "grad_norm": 3.2221906185150146, "learning_rate": 1.0152361857462836e-06, "loss": 0.010346505790948868, "step": 6063 }, { "epoch": 1.5742471443406023, "grad_norm": 3.022017240524292, "learning_rate": 1.0146338939920262e-06, "loss": 0.03122929111123085, "step": 6064 }, { "epoch": 1.5745067497403946, "grad_norm": 3.6041324138641357, "learning_rate": 1.014031909010761e-06, "loss": 0.013058090582489967, "step": 6065 }, { "epoch": 1.574766355140187, "grad_norm": 3.660510301589966, "learning_rate": 1.0134302309089065e-06, "loss": 0.043531328439712524, "step": 6066 }, { "epoch": 1.5750259605399792, "grad_norm": 3.7434067726135254, "learning_rate": 1.0128288597928252e-06, "loss": 0.03781682997941971, "step": 6067 }, { "epoch": 1.5752855659397715, "grad_norm": 3.3473448753356934, "learning_rate": 1.0122277957688244e-06, "loss": 0.020049866288900375, "step": 6068 }, { "epoch": 1.5755451713395638, "grad_norm": 2.44987154006958, "learning_rate": 1.0116270389431606e-06, "loss": 0.02084505930542946, "step": 6069 }, { "epoch": 1.5758047767393562, "grad_norm": 1.263347864151001, "learning_rate": 1.0110265894220322e-06, "loss": 0.007714473642408848, "step": 6070 }, { "epoch": 1.5760643821391485, "grad_norm": 3.3147778511047363, "learning_rate": 1.010426447311586e-06, "loss": 0.03307636082172394, "step": 6071 }, { "epoch": 1.5763239875389408, "grad_norm": 3.9402942657470703, "learning_rate": 1.0098266127179132e-06, "loss": 0.040869150310754776, "step": 6072 }, { "epoch": 1.576583592938733, "grad_norm": 3.2215824127197266, "learning_rate": 1.00922708574705e-06, "loss": 0.01015331782400608, "step": 6073 }, { "epoch": 1.5768431983385254, "grad_norm": 3.691666603088379, "learning_rate": 1.0086278665049801e-06, "loss": 0.02541925013065338, "step": 6074 }, { "epoch": 1.5771028037383177, "grad_norm": 2.8997886180877686, "learning_rate": 1.0080289550976314e-06, "loss": 0.007056279107928276, "step": 6075 }, { "epoch": 1.57736240913811, "grad_norm": 3.5013906955718994, "learning_rate": 1.0074303516308767e-06, "loss": 0.035153523087501526, "step": 6076 }, { "epoch": 1.5776220145379023, "grad_norm": 4.321535110473633, "learning_rate": 1.0068320562105369e-06, "loss": 0.047707315534353256, "step": 6077 }, { "epoch": 1.5778816199376946, "grad_norm": 2.9327197074890137, "learning_rate": 1.0062340689423762e-06, "loss": 0.011274624615907669, "step": 6078 }, { "epoch": 1.578141225337487, "grad_norm": 2.9898276329040527, "learning_rate": 1.0056363899321053e-06, "loss": 0.033935919404029846, "step": 6079 }, { "epoch": 1.5784008307372792, "grad_norm": 4.574464321136475, "learning_rate": 1.0050390192853801e-06, "loss": 0.013998925685882568, "step": 6080 }, { "epoch": 1.5786604361370715, "grad_norm": 2.9837350845336914, "learning_rate": 1.0044419571078016e-06, "loss": 0.011782238259911537, "step": 6081 }, { "epoch": 1.5789200415368638, "grad_norm": 3.4072659015655518, "learning_rate": 1.0038452035049174e-06, "loss": 0.045390017330646515, "step": 6082 }, { "epoch": 1.5791796469366561, "grad_norm": 4.0714569091796875, "learning_rate": 1.0032487585822196e-06, "loss": 0.03994377702474594, "step": 6083 }, { "epoch": 1.5794392523364484, "grad_norm": 2.4907286167144775, "learning_rate": 1.0026526224451455e-06, "loss": 0.013387092389166355, "step": 6084 }, { "epoch": 1.579698857736241, "grad_norm": 3.349996328353882, "learning_rate": 1.0020567951990794e-06, "loss": 0.03855457156896591, "step": 6085 }, { "epoch": 1.5799584631360333, "grad_norm": 3.062270164489746, "learning_rate": 1.0014612769493494e-06, "loss": 0.03272054344415665, "step": 6086 }, { "epoch": 1.5802180685358256, "grad_norm": 3.0150556564331055, "learning_rate": 1.000866067801229e-06, "loss": 0.01623612642288208, "step": 6087 }, { "epoch": 1.5804776739356179, "grad_norm": 2.9613232612609863, "learning_rate": 1.0002711678599378e-06, "loss": 0.01608687825500965, "step": 6088 }, { "epoch": 1.5807372793354102, "grad_norm": 3.812791585922241, "learning_rate": 9.996765772306405e-07, "loss": 0.035468485206365585, "step": 6089 }, { "epoch": 1.5809968847352025, "grad_norm": 2.539552688598633, "learning_rate": 9.990822960184488e-07, "loss": 0.013834653422236443, "step": 6090 }, { "epoch": 1.5812564901349948, "grad_norm": 3.6461195945739746, "learning_rate": 9.984883243284146e-07, "loss": 0.021709654480218887, "step": 6091 }, { "epoch": 1.5815160955347871, "grad_norm": 3.7073352336883545, "learning_rate": 9.978946622655414e-07, "loss": 0.010830138809978962, "step": 6092 }, { "epoch": 1.5817757009345794, "grad_norm": 4.249434947967529, "learning_rate": 9.973013099347734e-07, "loss": 0.019131213426589966, "step": 6093 }, { "epoch": 1.5820353063343717, "grad_norm": 3.9698269367218018, "learning_rate": 9.967082674410028e-07, "loss": 0.0389595627784729, "step": 6094 }, { "epoch": 1.582294911734164, "grad_norm": 3.7025632858276367, "learning_rate": 9.961155348890654e-07, "loss": 0.03819163888692856, "step": 6095 }, { "epoch": 1.5825545171339563, "grad_norm": 3.995223045349121, "learning_rate": 9.955231123837422e-07, "loss": 0.03226301074028015, "step": 6096 }, { "epoch": 1.5828141225337489, "grad_norm": 2.5740246772766113, "learning_rate": 9.94931000029761e-07, "loss": 0.053830817341804504, "step": 6097 }, { "epoch": 1.5830737279335412, "grad_norm": 3.0563583374023438, "learning_rate": 9.943391979317938e-07, "loss": 0.04002480208873749, "step": 6098 }, { "epoch": 1.5833333333333335, "grad_norm": 1.1268669366836548, "learning_rate": 9.93747706194456e-07, "loss": 0.0015269549330696464, "step": 6099 }, { "epoch": 1.5835929387331258, "grad_norm": 2.7978906631469727, "learning_rate": 9.931565249223116e-07, "loss": 0.020408209413290024, "step": 6100 }, { "epoch": 1.583852544132918, "grad_norm": 3.4240775108337402, "learning_rate": 9.925656542198666e-07, "loss": 0.029333394020795822, "step": 6101 }, { "epoch": 1.5841121495327104, "grad_norm": 3.6265363693237305, "learning_rate": 9.919750941915748e-07, "loss": 0.04470162093639374, "step": 6102 }, { "epoch": 1.5843717549325027, "grad_norm": 3.7147648334503174, "learning_rate": 9.91384844941833e-07, "loss": 0.03764510154724121, "step": 6103 }, { "epoch": 1.584631360332295, "grad_norm": 2.3279266357421875, "learning_rate": 9.907949065749827e-07, "loss": 0.014181885868310928, "step": 6104 }, { "epoch": 1.5848909657320873, "grad_norm": 3.6347620487213135, "learning_rate": 9.902052791953131e-07, "loss": 0.024950627237558365, "step": 6105 }, { "epoch": 1.5851505711318796, "grad_norm": 3.579183340072632, "learning_rate": 9.896159629070564e-07, "loss": 0.031244751065969467, "step": 6106 }, { "epoch": 1.585410176531672, "grad_norm": 3.343208074569702, "learning_rate": 9.890269578143894e-07, "loss": 0.062449708580970764, "step": 6107 }, { "epoch": 1.5856697819314642, "grad_norm": 3.2939093112945557, "learning_rate": 9.884382640214357e-07, "loss": 0.046701375395059586, "step": 6108 }, { "epoch": 1.5859293873312565, "grad_norm": 4.924783706665039, "learning_rate": 9.878498816322621e-07, "loss": 0.02791442722082138, "step": 6109 }, { "epoch": 1.5861889927310489, "grad_norm": 3.455521583557129, "learning_rate": 9.87261810750882e-07, "loss": 0.03437615931034088, "step": 6110 }, { "epoch": 1.5864485981308412, "grad_norm": 3.446047306060791, "learning_rate": 9.866740514812524e-07, "loss": 0.0506872832775116, "step": 6111 }, { "epoch": 1.5867082035306335, "grad_norm": 3.981293201446533, "learning_rate": 9.86086603927275e-07, "loss": 0.01739545166492462, "step": 6112 }, { "epoch": 1.5869678089304258, "grad_norm": 4.2158427238464355, "learning_rate": 9.854994681927992e-07, "loss": 0.019366804510354996, "step": 6113 }, { "epoch": 1.587227414330218, "grad_norm": 5.805853843688965, "learning_rate": 9.849126443816146e-07, "loss": 0.015548162162303925, "step": 6114 }, { "epoch": 1.5874870197300104, "grad_norm": 1.4792026281356812, "learning_rate": 9.843261325974591e-07, "loss": 0.004913422744721174, "step": 6115 }, { "epoch": 1.5877466251298027, "grad_norm": 3.1464860439300537, "learning_rate": 9.837399329440155e-07, "loss": 0.0786900594830513, "step": 6116 }, { "epoch": 1.588006230529595, "grad_norm": 2.3875927925109863, "learning_rate": 9.831540455249097e-07, "loss": 0.0701623260974884, "step": 6117 }, { "epoch": 1.5882658359293873, "grad_norm": 3.117837429046631, "learning_rate": 9.825684704437132e-07, "loss": 0.007099773734807968, "step": 6118 }, { "epoch": 1.5885254413291796, "grad_norm": 3.3271284103393555, "learning_rate": 9.819832078039419e-07, "loss": 0.07491271197795868, "step": 6119 }, { "epoch": 1.588785046728972, "grad_norm": 3.4275963306427, "learning_rate": 9.813982577090572e-07, "loss": 0.03457970917224884, "step": 6120 }, { "epoch": 1.5890446521287642, "grad_norm": 4.564988136291504, "learning_rate": 9.808136202624657e-07, "loss": 0.04701969027519226, "step": 6121 }, { "epoch": 1.5893042575285565, "grad_norm": 3.0806291103363037, "learning_rate": 9.802292955675163e-07, "loss": 0.015475122258067131, "step": 6122 }, { "epoch": 1.5895638629283488, "grad_norm": 3.2262697219848633, "learning_rate": 9.796452837275055e-07, "loss": 0.028721746057271957, "step": 6123 }, { "epoch": 1.5898234683281411, "grad_norm": 2.8514227867126465, "learning_rate": 9.790615848456725e-07, "loss": 0.05407094210386276, "step": 6124 }, { "epoch": 1.5900830737279334, "grad_norm": 3.6031415462493896, "learning_rate": 9.784781990252027e-07, "loss": 0.031188348308205605, "step": 6125 }, { "epoch": 1.5903426791277258, "grad_norm": 2.884504795074463, "learning_rate": 9.778951263692247e-07, "loss": 0.06207672879099846, "step": 6126 }, { "epoch": 1.590602284527518, "grad_norm": 3.2394790649414062, "learning_rate": 9.773123669808118e-07, "loss": 0.0200716070830822, "step": 6127 }, { "epoch": 1.5908618899273104, "grad_norm": 3.0639865398406982, "learning_rate": 9.767299209629841e-07, "loss": 0.05452297255396843, "step": 6128 }, { "epoch": 1.5911214953271027, "grad_norm": 1.5762020349502563, "learning_rate": 9.761477884187038e-07, "loss": 0.007317018695175648, "step": 6129 }, { "epoch": 1.591381100726895, "grad_norm": 3.0825765132904053, "learning_rate": 9.755659694508781e-07, "loss": 0.0287192203104496, "step": 6130 }, { "epoch": 1.5916407061266873, "grad_norm": 2.6351253986358643, "learning_rate": 9.749844641623605e-07, "loss": 0.02161761000752449, "step": 6131 }, { "epoch": 1.5919003115264796, "grad_norm": 2.710698127746582, "learning_rate": 9.744032726559465e-07, "loss": 0.05612438917160034, "step": 6132 }, { "epoch": 1.592159916926272, "grad_norm": 2.5747158527374268, "learning_rate": 9.738223950343789e-07, "loss": 0.011745794676244259, "step": 6133 }, { "epoch": 1.5924195223260644, "grad_norm": 3.1793482303619385, "learning_rate": 9.732418314003422e-07, "loss": 0.03990611806511879, "step": 6134 }, { "epoch": 1.5926791277258567, "grad_norm": 3.0279762744903564, "learning_rate": 9.726615818564675e-07, "loss": 0.02609900012612343, "step": 6135 }, { "epoch": 1.592938733125649, "grad_norm": 1.16177499294281, "learning_rate": 9.720816465053296e-07, "loss": 0.003556267125532031, "step": 6136 }, { "epoch": 1.5931983385254413, "grad_norm": 3.9180169105529785, "learning_rate": 9.715020254494475e-07, "loss": 0.02098000980913639, "step": 6137 }, { "epoch": 1.5934579439252337, "grad_norm": 3.8451690673828125, "learning_rate": 9.709227187912852e-07, "loss": 0.06397750973701477, "step": 6138 }, { "epoch": 1.593717549325026, "grad_norm": 3.4602668285369873, "learning_rate": 9.703437266332506e-07, "loss": 0.03288882225751877, "step": 6139 }, { "epoch": 1.5939771547248183, "grad_norm": 2.738712787628174, "learning_rate": 9.697650490776968e-07, "loss": 0.06297329068183899, "step": 6140 }, { "epoch": 1.5942367601246106, "grad_norm": 2.3162968158721924, "learning_rate": 9.691866862269197e-07, "loss": 0.05135940760374069, "step": 6141 }, { "epoch": 1.5944963655244029, "grad_norm": 1.6714270114898682, "learning_rate": 9.686086381831617e-07, "loss": 0.004981907084584236, "step": 6142 }, { "epoch": 1.5947559709241952, "grad_norm": 3.401933193206787, "learning_rate": 9.680309050486075e-07, "loss": 0.02387426607310772, "step": 6143 }, { "epoch": 1.5950155763239875, "grad_norm": 2.9921939373016357, "learning_rate": 9.674534869253887e-07, "loss": 0.0067929113283753395, "step": 6144 }, { "epoch": 1.5952751817237798, "grad_norm": 3.64900803565979, "learning_rate": 9.668763839155776e-07, "loss": 0.011880754493176937, "step": 6145 }, { "epoch": 1.5955347871235723, "grad_norm": 3.560075521469116, "learning_rate": 9.662995961211935e-07, "loss": 0.032756853848695755, "step": 6146 }, { "epoch": 1.5957943925233646, "grad_norm": 2.5461394786834717, "learning_rate": 9.657231236442002e-07, "loss": 0.006315036676824093, "step": 6147 }, { "epoch": 1.596053997923157, "grad_norm": 3.6047675609588623, "learning_rate": 9.65146966586504e-07, "loss": 0.04572659730911255, "step": 6148 }, { "epoch": 1.5963136033229492, "grad_norm": 4.071210861206055, "learning_rate": 9.645711250499565e-07, "loss": 0.038148991763591766, "step": 6149 }, { "epoch": 1.5965732087227416, "grad_norm": 3.5436878204345703, "learning_rate": 9.639955991363526e-07, "loss": 0.0676218718290329, "step": 6150 }, { "epoch": 1.5968328141225339, "grad_norm": 3.0903518199920654, "learning_rate": 9.634203889474324e-07, "loss": 0.04577340558171272, "step": 6151 }, { "epoch": 1.5970924195223262, "grad_norm": 2.8885700702667236, "learning_rate": 9.628454945848815e-07, "loss": 0.07375422865152359, "step": 6152 }, { "epoch": 1.5973520249221185, "grad_norm": 1.7499548196792603, "learning_rate": 9.622709161503257e-07, "loss": 0.007607205305248499, "step": 6153 }, { "epoch": 1.5976116303219108, "grad_norm": 2.5724008083343506, "learning_rate": 9.616966537453389e-07, "loss": 0.016221294179558754, "step": 6154 }, { "epoch": 1.597871235721703, "grad_norm": 3.0420775413513184, "learning_rate": 9.611227074714364e-07, "loss": 0.013847021386027336, "step": 6155 }, { "epoch": 1.5981308411214954, "grad_norm": 2.937791109085083, "learning_rate": 9.6054907743008e-07, "loss": 0.05006840080022812, "step": 6156 }, { "epoch": 1.5983904465212877, "grad_norm": 1.0871108770370483, "learning_rate": 9.599757637226739e-07, "loss": 0.004674352705478668, "step": 6157 }, { "epoch": 1.59865005192108, "grad_norm": 3.5588598251342773, "learning_rate": 9.594027664505656e-07, "loss": 0.049501270055770874, "step": 6158 }, { "epoch": 1.5989096573208723, "grad_norm": 1.4541774988174438, "learning_rate": 9.588300857150498e-07, "loss": 0.007693331688642502, "step": 6159 }, { "epoch": 1.5991692627206646, "grad_norm": 3.5984816551208496, "learning_rate": 9.582577216173626e-07, "loss": 0.03305778652429581, "step": 6160 }, { "epoch": 1.599428868120457, "grad_norm": 2.35241436958313, "learning_rate": 9.576856742586837e-07, "loss": 0.009744955226778984, "step": 6161 }, { "epoch": 1.5996884735202492, "grad_norm": 2.3255722522735596, "learning_rate": 9.5711394374014e-07, "loss": 0.005882419180124998, "step": 6162 }, { "epoch": 1.5999480789200415, "grad_norm": 3.658604383468628, "learning_rate": 9.565425301627987e-07, "loss": 0.035117007791996, "step": 6163 }, { "epoch": 1.6002076843198338, "grad_norm": 2.6436831951141357, "learning_rate": 9.55971433627674e-07, "loss": 0.07211184501647949, "step": 6164 }, { "epoch": 1.6004672897196262, "grad_norm": 3.713113307952881, "learning_rate": 9.554006542357219e-07, "loss": 0.028422273695468903, "step": 6165 }, { "epoch": 1.6007268951194185, "grad_norm": 4.339831352233887, "learning_rate": 9.548301920878428e-07, "loss": 0.01314133033156395, "step": 6166 }, { "epoch": 1.6009865005192108, "grad_norm": 3.122154712677002, "learning_rate": 9.542600472848825e-07, "loss": 0.02693517506122589, "step": 6167 }, { "epoch": 1.601246105919003, "grad_norm": 2.9400384426116943, "learning_rate": 9.536902199276284e-07, "loss": 0.05952860787510872, "step": 6168 }, { "epoch": 1.6015057113187954, "grad_norm": 3.2920823097229004, "learning_rate": 9.531207101168132e-07, "loss": 0.025773674249649048, "step": 6169 }, { "epoch": 1.6017653167185877, "grad_norm": 2.82757568359375, "learning_rate": 9.525515179531139e-07, "loss": 0.050613999366760254, "step": 6170 }, { "epoch": 1.60202492211838, "grad_norm": 3.5715179443359375, "learning_rate": 9.519826435371502e-07, "loss": 0.023259514942765236, "step": 6171 }, { "epoch": 1.6022845275181723, "grad_norm": 3.6966264247894287, "learning_rate": 9.514140869694852e-07, "loss": 0.02864641509950161, "step": 6172 }, { "epoch": 1.6025441329179646, "grad_norm": 2.4778759479522705, "learning_rate": 9.508458483506281e-07, "loss": 0.008200882002711296, "step": 6173 }, { "epoch": 1.602803738317757, "grad_norm": 3.3586792945861816, "learning_rate": 9.502779277810292e-07, "loss": 0.024035118520259857, "step": 6174 }, { "epoch": 1.6030633437175492, "grad_norm": 3.281242847442627, "learning_rate": 9.497103253610857e-07, "loss": 0.006528923287987709, "step": 6175 }, { "epoch": 1.6033229491173415, "grad_norm": 3.747467279434204, "learning_rate": 9.491430411911343e-07, "loss": 0.020266054198145866, "step": 6176 }, { "epoch": 1.6035825545171338, "grad_norm": 3.9480600357055664, "learning_rate": 9.485760753714592e-07, "loss": 0.019540930166840553, "step": 6177 }, { "epoch": 1.6038421599169261, "grad_norm": 2.6927573680877686, "learning_rate": 9.480094280022873e-07, "loss": 0.057803843170404434, "step": 6178 }, { "epoch": 1.6041017653167184, "grad_norm": 4.066867828369141, "learning_rate": 9.474430991837886e-07, "loss": 0.031145373359322548, "step": 6179 }, { "epoch": 1.6043613707165107, "grad_norm": 3.2455294132232666, "learning_rate": 9.468770890160767e-07, "loss": 0.021648483350872993, "step": 6180 }, { "epoch": 1.604620976116303, "grad_norm": 3.5177910327911377, "learning_rate": 9.463113975992086e-07, "loss": 0.02782377228140831, "step": 6181 }, { "epoch": 1.6048805815160956, "grad_norm": 2.674816370010376, "learning_rate": 9.457460250331868e-07, "loss": 0.008483339101076126, "step": 6182 }, { "epoch": 1.605140186915888, "grad_norm": 3.6727142333984375, "learning_rate": 9.451809714179569e-07, "loss": 0.033333756029605865, "step": 6183 }, { "epoch": 1.6053997923156802, "grad_norm": 4.225372314453125, "learning_rate": 9.446162368534051e-07, "loss": 0.027679577469825745, "step": 6184 }, { "epoch": 1.6056593977154725, "grad_norm": 2.9743592739105225, "learning_rate": 9.440518214393657e-07, "loss": 0.05768146365880966, "step": 6185 }, { "epoch": 1.6059190031152648, "grad_norm": 4.782276630401611, "learning_rate": 9.434877252756125e-07, "loss": 0.026413410902023315, "step": 6186 }, { "epoch": 1.6061786085150571, "grad_norm": 5.275618076324463, "learning_rate": 9.429239484618669e-07, "loss": 0.016492653638124466, "step": 6187 }, { "epoch": 1.6064382139148494, "grad_norm": 4.1701812744140625, "learning_rate": 9.423604910977904e-07, "loss": 0.028635919094085693, "step": 6188 }, { "epoch": 1.6066978193146417, "grad_norm": 4.027394771575928, "learning_rate": 9.41797353282989e-07, "loss": 0.037835124880075455, "step": 6189 }, { "epoch": 1.606957424714434, "grad_norm": 2.332808256149292, "learning_rate": 9.412345351170138e-07, "loss": 0.008317019790410995, "step": 6190 }, { "epoch": 1.6072170301142263, "grad_norm": 4.483226299285889, "learning_rate": 9.406720366993575e-07, "loss": 0.004679577890783548, "step": 6191 }, { "epoch": 1.6074766355140186, "grad_norm": 3.364941120147705, "learning_rate": 9.401098581294563e-07, "loss": 0.046278778463602066, "step": 6192 }, { "epoch": 1.607736240913811, "grad_norm": 2.817265033721924, "learning_rate": 9.395479995066919e-07, "loss": 0.03702060878276825, "step": 6193 }, { "epoch": 1.6079958463136035, "grad_norm": 0.6426070928573608, "learning_rate": 9.38986460930387e-07, "loss": 0.0014876355417072773, "step": 6194 }, { "epoch": 1.6082554517133958, "grad_norm": 3.1023190021514893, "learning_rate": 9.384252424998094e-07, "loss": 0.056667231023311615, "step": 6195 }, { "epoch": 1.608515057113188, "grad_norm": 2.959223747253418, "learning_rate": 9.378643443141694e-07, "loss": 0.0559745728969574, "step": 6196 }, { "epoch": 1.6087746625129804, "grad_norm": 3.5830535888671875, "learning_rate": 9.373037664726204e-07, "loss": 0.04824138432741165, "step": 6197 }, { "epoch": 1.6090342679127727, "grad_norm": 1.718123197555542, "learning_rate": 9.367435090742608e-07, "loss": 0.0027534065302461386, "step": 6198 }, { "epoch": 1.609293873312565, "grad_norm": 2.051856517791748, "learning_rate": 9.361835722181306e-07, "loss": 0.008244694210588932, "step": 6199 }, { "epoch": 1.6095534787123573, "grad_norm": 3.843604803085327, "learning_rate": 9.356239560032135e-07, "loss": 0.036279045045375824, "step": 6200 }, { "epoch": 1.6098130841121496, "grad_norm": 2.9673941135406494, "learning_rate": 9.350646605284378e-07, "loss": 0.006567058153450489, "step": 6201 }, { "epoch": 1.610072689511942, "grad_norm": 2.53251314163208, "learning_rate": 9.345056858926735e-07, "loss": 0.07495008409023285, "step": 6202 }, { "epoch": 1.6103322949117342, "grad_norm": 0.6858431696891785, "learning_rate": 9.339470321947341e-07, "loss": 0.0016857997979968786, "step": 6203 }, { "epoch": 1.6105919003115265, "grad_norm": 3.0014455318450928, "learning_rate": 9.333886995333782e-07, "loss": 0.019237861037254333, "step": 6204 }, { "epoch": 1.6108515057113189, "grad_norm": 3.2394614219665527, "learning_rate": 9.328306880073046e-07, "loss": 0.0067319683730602264, "step": 6205 }, { "epoch": 1.6111111111111112, "grad_norm": 3.019988775253296, "learning_rate": 9.322729977151587e-07, "loss": 0.027800820767879486, "step": 6206 }, { "epoch": 1.6113707165109035, "grad_norm": 2.5572900772094727, "learning_rate": 9.317156287555253e-07, "loss": 0.04497101902961731, "step": 6207 }, { "epoch": 1.6116303219106958, "grad_norm": 3.2865145206451416, "learning_rate": 9.311585812269356e-07, "loss": 0.01649293303489685, "step": 6208 }, { "epoch": 1.611889927310488, "grad_norm": 2.357295274734497, "learning_rate": 9.306018552278631e-07, "loss": 0.007945861667394638, "step": 6209 }, { "epoch": 1.6121495327102804, "grad_norm": 3.1459293365478516, "learning_rate": 9.300454508567242e-07, "loss": 0.05626485124230385, "step": 6210 }, { "epoch": 1.6124091381100727, "grad_norm": 2.742854595184326, "learning_rate": 9.294893682118783e-07, "loss": 0.006928187794983387, "step": 6211 }, { "epoch": 1.612668743509865, "grad_norm": 5.312188625335693, "learning_rate": 9.28933607391627e-07, "loss": 0.0063664596527814865, "step": 6212 }, { "epoch": 1.6129283489096573, "grad_norm": 3.955261468887329, "learning_rate": 9.28378168494217e-07, "loss": 0.019492726773023605, "step": 6213 }, { "epoch": 1.6131879543094496, "grad_norm": 3.6986637115478516, "learning_rate": 9.278230516178385e-07, "loss": 0.012490193359553814, "step": 6214 }, { "epoch": 1.613447559709242, "grad_norm": 3.1661267280578613, "learning_rate": 9.272682568606209e-07, "loss": 0.04723729193210602, "step": 6215 }, { "epoch": 1.6137071651090342, "grad_norm": 1.6668403148651123, "learning_rate": 9.267137843206411e-07, "loss": 0.004816896747797728, "step": 6216 }, { "epoch": 1.6139667705088265, "grad_norm": 2.9189352989196777, "learning_rate": 9.261596340959161e-07, "loss": 0.06609559059143066, "step": 6217 }, { "epoch": 1.6142263759086188, "grad_norm": 2.04362416267395, "learning_rate": 9.256058062844078e-07, "loss": 0.009664937853813171, "step": 6218 }, { "epoch": 1.6144859813084111, "grad_norm": 3.184504508972168, "learning_rate": 9.250523009840201e-07, "loss": 0.028778500854969025, "step": 6219 }, { "epoch": 1.6147455867082035, "grad_norm": 2.2763915061950684, "learning_rate": 9.244991182925987e-07, "loss": 0.00479259854182601, "step": 6220 }, { "epoch": 1.6150051921079958, "grad_norm": 2.868591070175171, "learning_rate": 9.239462583079359e-07, "loss": 0.035609081387519836, "step": 6221 }, { "epoch": 1.615264797507788, "grad_norm": 3.811065435409546, "learning_rate": 9.233937211277633e-07, "loss": 0.02285638637840748, "step": 6222 }, { "epoch": 1.6155244029075804, "grad_norm": 2.834181547164917, "learning_rate": 9.228415068497565e-07, "loss": 0.07227650284767151, "step": 6223 }, { "epoch": 1.6157840083073727, "grad_norm": 2.917646646499634, "learning_rate": 9.222896155715352e-07, "loss": 0.054927997291088104, "step": 6224 }, { "epoch": 1.616043613707165, "grad_norm": 2.6676738262176514, "learning_rate": 9.217380473906612e-07, "loss": 0.007666202262043953, "step": 6225 }, { "epoch": 1.6163032191069573, "grad_norm": 0.9234708547592163, "learning_rate": 9.211868024046381e-07, "loss": 0.003753119148313999, "step": 6226 }, { "epoch": 1.6165628245067496, "grad_norm": 3.758085012435913, "learning_rate": 9.206358807109147e-07, "loss": 0.03931509703397751, "step": 6227 }, { "epoch": 1.616822429906542, "grad_norm": 1.7480509281158447, "learning_rate": 9.200852824068801e-07, "loss": 0.009887486696243286, "step": 6228 }, { "epoch": 1.6170820353063342, "grad_norm": 2.7803406715393066, "learning_rate": 9.195350075898684e-07, "loss": 0.06155514717102051, "step": 6229 }, { "epoch": 1.6173416407061267, "grad_norm": 2.361821174621582, "learning_rate": 9.189850563571556e-07, "loss": 0.0157487615942955, "step": 6230 }, { "epoch": 1.617601246105919, "grad_norm": 3.3324708938598633, "learning_rate": 9.184354288059596e-07, "loss": 0.032848723232746124, "step": 6231 }, { "epoch": 1.6178608515057114, "grad_norm": 3.9237208366394043, "learning_rate": 9.178861250334429e-07, "loss": 0.012711971066892147, "step": 6232 }, { "epoch": 1.6181204569055037, "grad_norm": 2.6714978218078613, "learning_rate": 9.173371451367097e-07, "loss": 0.05099751800298691, "step": 6233 }, { "epoch": 1.618380062305296, "grad_norm": 3.183952569961548, "learning_rate": 9.167884892128059e-07, "loss": 0.035750195384025574, "step": 6234 }, { "epoch": 1.6186396677050883, "grad_norm": 3.7940173149108887, "learning_rate": 9.162401573587229e-07, "loss": 0.042609263211488724, "step": 6235 }, { "epoch": 1.6188992731048806, "grad_norm": 2.1368584632873535, "learning_rate": 9.156921496713922e-07, "loss": 0.0043907975777983665, "step": 6236 }, { "epoch": 1.6191588785046729, "grad_norm": 3.6709444522857666, "learning_rate": 9.151444662476902e-07, "loss": 0.034335821866989136, "step": 6237 }, { "epoch": 1.6194184839044652, "grad_norm": 2.274949550628662, "learning_rate": 9.145971071844332e-07, "loss": 0.007516063749790192, "step": 6238 }, { "epoch": 1.6196780893042575, "grad_norm": 2.368410110473633, "learning_rate": 9.140500725783818e-07, "loss": 0.020379656925797462, "step": 6239 }, { "epoch": 1.6199376947040498, "grad_norm": 2.46932315826416, "learning_rate": 9.135033625262412e-07, "loss": 0.011290271766483784, "step": 6240 }, { "epoch": 1.620197300103842, "grad_norm": 1.7034687995910645, "learning_rate": 9.129569771246555e-07, "loss": 0.01437437254935503, "step": 6241 }, { "epoch": 1.6204569055036346, "grad_norm": 2.6270298957824707, "learning_rate": 9.124109164702135e-07, "loss": 0.06896883249282837, "step": 6242 }, { "epoch": 1.620716510903427, "grad_norm": 2.7440145015716553, "learning_rate": 9.118651806594459e-07, "loss": 0.05698486045002937, "step": 6243 }, { "epoch": 1.6209761163032192, "grad_norm": 2.8920812606811523, "learning_rate": 9.113197697888265e-07, "loss": 0.06457921862602234, "step": 6244 }, { "epoch": 1.6212357217030116, "grad_norm": 3.537692070007324, "learning_rate": 9.107746839547729e-07, "loss": 0.051381926983594894, "step": 6245 }, { "epoch": 1.6214953271028039, "grad_norm": 2.9562947750091553, "learning_rate": 9.102299232536413e-07, "loss": 0.03937417268753052, "step": 6246 }, { "epoch": 1.6217549325025962, "grad_norm": 3.131598711013794, "learning_rate": 9.096854877817346e-07, "loss": 0.04753103107213974, "step": 6247 }, { "epoch": 1.6220145379023885, "grad_norm": 3.1766841411590576, "learning_rate": 9.091413776352954e-07, "loss": 0.02167120762169361, "step": 6248 }, { "epoch": 1.6222741433021808, "grad_norm": 2.9658966064453125, "learning_rate": 9.085975929105113e-07, "loss": 0.010276466608047485, "step": 6249 }, { "epoch": 1.622533748701973, "grad_norm": 2.6925981044769287, "learning_rate": 9.080541337035102e-07, "loss": 0.025022421032190323, "step": 6250 }, { "epoch": 1.6227933541017654, "grad_norm": 3.6560120582580566, "learning_rate": 9.075110001103627e-07, "loss": 0.028425980359315872, "step": 6251 }, { "epoch": 1.6230529595015577, "grad_norm": 2.316894292831421, "learning_rate": 9.069681922270837e-07, "loss": 0.0035849777050316334, "step": 6252 }, { "epoch": 1.62331256490135, "grad_norm": 3.1015267372131348, "learning_rate": 9.064257101496285e-07, "loss": 0.02450426295399666, "step": 6253 }, { "epoch": 1.6235721703011423, "grad_norm": 3.042353868484497, "learning_rate": 9.058835539738947e-07, "loss": 0.021091055124998093, "step": 6254 }, { "epoch": 1.6238317757009346, "grad_norm": 3.54522705078125, "learning_rate": 9.053417237957243e-07, "loss": 0.015529695898294449, "step": 6255 }, { "epoch": 1.624091381100727, "grad_norm": 1.447175145149231, "learning_rate": 9.048002197109004e-07, "loss": 0.0029062489047646523, "step": 6256 }, { "epoch": 1.6243509865005192, "grad_norm": 3.6230435371398926, "learning_rate": 9.042590418151474e-07, "loss": 0.024567365646362305, "step": 6257 }, { "epoch": 1.6246105919003115, "grad_norm": 1.8398650884628296, "learning_rate": 9.037181902041348e-07, "loss": 0.004939794074743986, "step": 6258 }, { "epoch": 1.6248701973001038, "grad_norm": 2.1571991443634033, "learning_rate": 9.031776649734708e-07, "loss": 0.006141575053334236, "step": 6259 }, { "epoch": 1.6251298026998962, "grad_norm": 2.658195734024048, "learning_rate": 9.026374662187102e-07, "loss": 0.011104485020041466, "step": 6260 }, { "epoch": 1.6253894080996885, "grad_norm": 3.386737823486328, "learning_rate": 9.020975940353463e-07, "loss": 0.015640297904610634, "step": 6261 }, { "epoch": 1.6256490134994808, "grad_norm": 4.396691799163818, "learning_rate": 9.015580485188158e-07, "loss": 0.012754354625940323, "step": 6262 }, { "epoch": 1.625908618899273, "grad_norm": 5.21802282333374, "learning_rate": 9.010188297644994e-07, "loss": 0.02588333748281002, "step": 6263 }, { "epoch": 1.6261682242990654, "grad_norm": 4.7111735343933105, "learning_rate": 9.004799378677176e-07, "loss": 0.014559745788574219, "step": 6264 }, { "epoch": 1.6264278296988577, "grad_norm": 2.9426047801971436, "learning_rate": 8.999413729237339e-07, "loss": 0.019939327612519264, "step": 6265 }, { "epoch": 1.62668743509865, "grad_norm": 3.725756883621216, "learning_rate": 8.994031350277552e-07, "loss": 0.048981040716171265, "step": 6266 }, { "epoch": 1.6269470404984423, "grad_norm": 3.8127541542053223, "learning_rate": 8.988652242749288e-07, "loss": 0.047140564769506454, "step": 6267 }, { "epoch": 1.6272066458982346, "grad_norm": 3.287944793701172, "learning_rate": 8.983276407603464e-07, "loss": 0.005471562035381794, "step": 6268 }, { "epoch": 1.627466251298027, "grad_norm": 3.062156915664673, "learning_rate": 8.977903845790384e-07, "loss": 0.043533921241760254, "step": 6269 }, { "epoch": 1.6277258566978192, "grad_norm": 4.000792980194092, "learning_rate": 8.972534558259803e-07, "loss": 0.027836576104164124, "step": 6270 }, { "epoch": 1.6279854620976115, "grad_norm": 2.5164620876312256, "learning_rate": 8.967168545960897e-07, "loss": 0.01357973087579012, "step": 6271 }, { "epoch": 1.6282450674974038, "grad_norm": 2.8197858333587646, "learning_rate": 8.961805809842244e-07, "loss": 0.009395929053425789, "step": 6272 }, { "epoch": 1.6285046728971961, "grad_norm": 2.609239101409912, "learning_rate": 8.956446350851855e-07, "loss": 0.008456122130155563, "step": 6273 }, { "epoch": 1.6287642782969884, "grad_norm": 3.4229719638824463, "learning_rate": 8.951090169937158e-07, "loss": 0.017575344070792198, "step": 6274 }, { "epoch": 1.6290238836967808, "grad_norm": 3.3477327823638916, "learning_rate": 8.945737268045004e-07, "loss": 0.015393871814012527, "step": 6275 }, { "epoch": 1.629283489096573, "grad_norm": 3.4507551193237305, "learning_rate": 8.940387646121673e-07, "loss": 0.011313240975141525, "step": 6276 }, { "epoch": 1.6295430944963654, "grad_norm": 3.4926602840423584, "learning_rate": 8.935041305112838e-07, "loss": 0.039253491908311844, "step": 6277 }, { "epoch": 1.629802699896158, "grad_norm": 3.5831797122955322, "learning_rate": 8.929698245963625e-07, "loss": 0.024649638682603836, "step": 6278 }, { "epoch": 1.6300623052959502, "grad_norm": 2.4427759647369385, "learning_rate": 8.92435846961856e-07, "loss": 0.008866630494594574, "step": 6279 }, { "epoch": 1.6303219106957425, "grad_norm": 4.71986722946167, "learning_rate": 8.919021977021584e-07, "loss": 0.01571401208639145, "step": 6280 }, { "epoch": 1.6305815160955348, "grad_norm": 2.8875253200531006, "learning_rate": 8.91368876911608e-07, "loss": 0.04437744617462158, "step": 6281 }, { "epoch": 1.6308411214953271, "grad_norm": 3.580568313598633, "learning_rate": 8.908358846844828e-07, "loss": 0.01114700362086296, "step": 6282 }, { "epoch": 1.6311007268951194, "grad_norm": 2.9746036529541016, "learning_rate": 8.903032211150042e-07, "loss": 0.02786211296916008, "step": 6283 }, { "epoch": 1.6313603322949117, "grad_norm": 2.4249236583709717, "learning_rate": 8.89770886297335e-07, "loss": 0.006378725171089172, "step": 6284 }, { "epoch": 1.631619937694704, "grad_norm": 0.410601407289505, "learning_rate": 8.892388803255788e-07, "loss": 0.0010796219576150179, "step": 6285 }, { "epoch": 1.6318795430944963, "grad_norm": 3.7859251499176025, "learning_rate": 8.887072032937833e-07, "loss": 0.028226371854543686, "step": 6286 }, { "epoch": 1.6321391484942886, "grad_norm": 2.997303009033203, "learning_rate": 8.881758552959366e-07, "loss": 0.03884521871805191, "step": 6287 }, { "epoch": 1.632398753894081, "grad_norm": 2.158824920654297, "learning_rate": 8.876448364259677e-07, "loss": 0.009088147431612015, "step": 6288 }, { "epoch": 1.6326583592938733, "grad_norm": 5.201671600341797, "learning_rate": 8.871141467777501e-07, "loss": 0.020644908770918846, "step": 6289 }, { "epoch": 1.6329179646936658, "grad_norm": 3.432248592376709, "learning_rate": 8.865837864450962e-07, "loss": 0.04624351114034653, "step": 6290 }, { "epoch": 1.633177570093458, "grad_norm": 3.9033632278442383, "learning_rate": 8.86053755521763e-07, "loss": 0.045653682202100754, "step": 6291 }, { "epoch": 1.6334371754932504, "grad_norm": 3.7154080867767334, "learning_rate": 8.855240541014471e-07, "loss": 0.02854989469051361, "step": 6292 }, { "epoch": 1.6336967808930427, "grad_norm": 3.0004239082336426, "learning_rate": 8.849946822777871e-07, "loss": 0.011606963351368904, "step": 6293 }, { "epoch": 1.633956386292835, "grad_norm": 3.976388931274414, "learning_rate": 8.844656401443646e-07, "loss": 0.01699461229145527, "step": 6294 }, { "epoch": 1.6342159916926273, "grad_norm": 3.897986888885498, "learning_rate": 8.839369277947021e-07, "loss": 0.02220633625984192, "step": 6295 }, { "epoch": 1.6344755970924196, "grad_norm": 3.1660332679748535, "learning_rate": 8.83408545322263e-07, "loss": 0.0277128666639328, "step": 6296 }, { "epoch": 1.634735202492212, "grad_norm": 2.6016623973846436, "learning_rate": 8.828804928204543e-07, "loss": 0.016847437247633934, "step": 6297 }, { "epoch": 1.6349948078920042, "grad_norm": 3.4318692684173584, "learning_rate": 8.823527703826227e-07, "loss": 0.022851809859275818, "step": 6298 }, { "epoch": 1.6352544132917965, "grad_norm": 2.847742795944214, "learning_rate": 8.818253781020586e-07, "loss": 0.009180044755339622, "step": 6299 }, { "epoch": 1.6355140186915889, "grad_norm": 2.729849100112915, "learning_rate": 8.812983160719912e-07, "loss": 0.06187994033098221, "step": 6300 }, { "epoch": 1.6357736240913812, "grad_norm": 3.139369487762451, "learning_rate": 8.807715843855941e-07, "loss": 0.023941440507769585, "step": 6301 }, { "epoch": 1.6360332294911735, "grad_norm": 3.618575096130371, "learning_rate": 8.802451831359815e-07, "loss": 0.03519328683614731, "step": 6302 }, { "epoch": 1.6362928348909658, "grad_norm": 2.8635077476501465, "learning_rate": 8.797191124162091e-07, "loss": 0.05726255476474762, "step": 6303 }, { "epoch": 1.636552440290758, "grad_norm": 3.4401113986968994, "learning_rate": 8.791933723192731e-07, "loss": 0.027883116155862808, "step": 6304 }, { "epoch": 1.6368120456905504, "grad_norm": 4.342580795288086, "learning_rate": 8.786679629381136e-07, "loss": 0.006072216667234898, "step": 6305 }, { "epoch": 1.6370716510903427, "grad_norm": 3.292604923248291, "learning_rate": 8.781428843656106e-07, "loss": 0.047544389963150024, "step": 6306 }, { "epoch": 1.637331256490135, "grad_norm": 0.8587899804115295, "learning_rate": 8.776181366945858e-07, "loss": 0.0014299445319920778, "step": 6307 }, { "epoch": 1.6375908618899273, "grad_norm": 0.8368101716041565, "learning_rate": 8.77093720017802e-07, "loss": 0.0036278925836086273, "step": 6308 }, { "epoch": 1.6378504672897196, "grad_norm": 1.141581416130066, "learning_rate": 8.765696344279645e-07, "loss": 0.0017416037153452635, "step": 6309 }, { "epoch": 1.638110072689512, "grad_norm": 3.023789405822754, "learning_rate": 8.76045880017721e-07, "loss": 0.04120317101478577, "step": 6310 }, { "epoch": 1.6383696780893042, "grad_norm": 2.749032497406006, "learning_rate": 8.755224568796569e-07, "loss": 0.01461855135858059, "step": 6311 }, { "epoch": 1.6386292834890965, "grad_norm": 3.439227819442749, "learning_rate": 8.749993651063033e-07, "loss": 0.025294912979006767, "step": 6312 }, { "epoch": 1.6388888888888888, "grad_norm": 2.363504648208618, "learning_rate": 8.744766047901291e-07, "loss": 0.013760576955974102, "step": 6313 }, { "epoch": 1.6391484942886811, "grad_norm": 3.1583926677703857, "learning_rate": 8.739541760235484e-07, "loss": 0.007505042944103479, "step": 6314 }, { "epoch": 1.6394080996884735, "grad_norm": 3.9251515865325928, "learning_rate": 8.734320788989133e-07, "loss": 0.03681086003780365, "step": 6315 }, { "epoch": 1.6396677050882658, "grad_norm": 2.313992500305176, "learning_rate": 8.729103135085184e-07, "loss": 0.01555941067636013, "step": 6316 }, { "epoch": 1.639927310488058, "grad_norm": 1.3436585664749146, "learning_rate": 8.723888799446012e-07, "loss": 0.0063570411875844, "step": 6317 }, { "epoch": 1.6401869158878504, "grad_norm": 4.1735968589782715, "learning_rate": 8.718677782993383e-07, "loss": 0.031824931502342224, "step": 6318 }, { "epoch": 1.6404465212876427, "grad_norm": 3.3681299686431885, "learning_rate": 8.713470086648482e-07, "loss": 0.06648774445056915, "step": 6319 }, { "epoch": 1.640706126687435, "grad_norm": 2.9430782794952393, "learning_rate": 8.70826571133192e-07, "loss": 0.010990357026457787, "step": 6320 }, { "epoch": 1.6409657320872273, "grad_norm": 3.6612746715545654, "learning_rate": 8.703064657963703e-07, "loss": 0.026419198140501976, "step": 6321 }, { "epoch": 1.6412253374870196, "grad_norm": 2.2443177700042725, "learning_rate": 8.697866927463264e-07, "loss": 0.009917947463691235, "step": 6322 }, { "epoch": 1.641484942886812, "grad_norm": 2.0635547637939453, "learning_rate": 8.692672520749443e-07, "loss": 0.007559997029602528, "step": 6323 }, { "epoch": 1.6417445482866042, "grad_norm": 4.2647809982299805, "learning_rate": 8.687481438740481e-07, "loss": 0.014413380064070225, "step": 6324 }, { "epoch": 1.6420041536863965, "grad_norm": 1.7857095003128052, "learning_rate": 8.682293682354059e-07, "loss": 0.005099168047308922, "step": 6325 }, { "epoch": 1.642263759086189, "grad_norm": 2.517604351043701, "learning_rate": 8.677109252507246e-07, "loss": 0.009045030921697617, "step": 6326 }, { "epoch": 1.6425233644859814, "grad_norm": 3.13413405418396, "learning_rate": 8.671928150116523e-07, "loss": 0.05615713447332382, "step": 6327 }, { "epoch": 1.6427829698857737, "grad_norm": 3.638561964035034, "learning_rate": 8.666750376097805e-07, "loss": 0.04416210949420929, "step": 6328 }, { "epoch": 1.643042575285566, "grad_norm": 3.847442626953125, "learning_rate": 8.661575931366387e-07, "loss": 0.043404076248407364, "step": 6329 }, { "epoch": 1.6433021806853583, "grad_norm": 3.149111747741699, "learning_rate": 8.656404816837011e-07, "loss": 0.017331626266241074, "step": 6330 }, { "epoch": 1.6435617860851506, "grad_norm": 3.9450531005859375, "learning_rate": 8.651237033423801e-07, "loss": 0.018025418743491173, "step": 6331 }, { "epoch": 1.6438213914849429, "grad_norm": 3.3783066272735596, "learning_rate": 8.646072582040296e-07, "loss": 0.054481055587530136, "step": 6332 }, { "epoch": 1.6440809968847352, "grad_norm": 3.2512922286987305, "learning_rate": 8.640911463599475e-07, "loss": 0.01584460213780403, "step": 6333 }, { "epoch": 1.6443406022845275, "grad_norm": 3.3019931316375732, "learning_rate": 8.635753679013681e-07, "loss": 0.035334907472133636, "step": 6334 }, { "epoch": 1.6446002076843198, "grad_norm": 3.7710251808166504, "learning_rate": 8.6305992291947e-07, "loss": 0.01027017179876566, "step": 6335 }, { "epoch": 1.644859813084112, "grad_norm": 1.9515002965927124, "learning_rate": 8.62544811505373e-07, "loss": 0.004551254212856293, "step": 6336 }, { "epoch": 1.6451194184839044, "grad_norm": 3.298999786376953, "learning_rate": 8.620300337501361e-07, "loss": 0.047576695680618286, "step": 6337 }, { "epoch": 1.645379023883697, "grad_norm": 1.5014456510543823, "learning_rate": 8.615155897447606e-07, "loss": 0.003755336394533515, "step": 6338 }, { "epoch": 1.6456386292834893, "grad_norm": 2.8875892162323, "learning_rate": 8.610014795801879e-07, "loss": 0.0050178589299321175, "step": 6339 }, { "epoch": 1.6458982346832816, "grad_norm": 1.6749504804611206, "learning_rate": 8.604877033473011e-07, "loss": 0.01574006862938404, "step": 6340 }, { "epoch": 1.6461578400830739, "grad_norm": 3.6251235008239746, "learning_rate": 8.599742611369255e-07, "loss": 0.035362862050533295, "step": 6341 }, { "epoch": 1.6464174454828662, "grad_norm": 2.3080949783325195, "learning_rate": 8.594611530398235e-07, "loss": 0.004984197206795216, "step": 6342 }, { "epoch": 1.6466770508826585, "grad_norm": 3.597709894180298, "learning_rate": 8.589483791467024e-07, "loss": 0.03142646327614784, "step": 6343 }, { "epoch": 1.6469366562824508, "grad_norm": 2.7425763607025146, "learning_rate": 8.584359395482081e-07, "loss": 0.06174864619970322, "step": 6344 }, { "epoch": 1.647196261682243, "grad_norm": 2.5177857875823975, "learning_rate": 8.579238343349295e-07, "loss": 0.011102167889475822, "step": 6345 }, { "epoch": 1.6474558670820354, "grad_norm": 2.7420902252197266, "learning_rate": 8.574120635973937e-07, "loss": 0.06021719425916672, "step": 6346 }, { "epoch": 1.6477154724818277, "grad_norm": 2.4370594024658203, "learning_rate": 8.569006274260702e-07, "loss": 0.02767001837491989, "step": 6347 }, { "epoch": 1.64797507788162, "grad_norm": 3.267451286315918, "learning_rate": 8.563895259113703e-07, "loss": 0.04511465132236481, "step": 6348 }, { "epoch": 1.6482346832814123, "grad_norm": 2.332310914993286, "learning_rate": 8.558787591436437e-07, "loss": 0.008899644017219543, "step": 6349 }, { "epoch": 1.6484942886812046, "grad_norm": 0.5035515427589417, "learning_rate": 8.553683272131827e-07, "loss": 0.0014798083575442433, "step": 6350 }, { "epoch": 1.648753894080997, "grad_norm": 3.5371015071868896, "learning_rate": 8.548582302102207e-07, "loss": 0.05055924504995346, "step": 6351 }, { "epoch": 1.6490134994807892, "grad_norm": 3.6977121829986572, "learning_rate": 8.543484682249301e-07, "loss": 0.035880815237760544, "step": 6352 }, { "epoch": 1.6492731048805815, "grad_norm": 3.6645450592041016, "learning_rate": 8.53839041347426e-07, "loss": 0.04274669289588928, "step": 6353 }, { "epoch": 1.6495327102803738, "grad_norm": 3.295379877090454, "learning_rate": 8.53329949667763e-07, "loss": 0.017424654215574265, "step": 6354 }, { "epoch": 1.6497923156801662, "grad_norm": 3.8035316467285156, "learning_rate": 8.528211932759367e-07, "loss": 0.016894588246941566, "step": 6355 }, { "epoch": 1.6500519210799585, "grad_norm": 3.705716371536255, "learning_rate": 8.52312772261884e-07, "loss": 0.016507446765899658, "step": 6356 }, { "epoch": 1.6503115264797508, "grad_norm": 2.842618227005005, "learning_rate": 8.518046867154823e-07, "loss": 0.013775360770523548, "step": 6357 }, { "epoch": 1.650571131879543, "grad_norm": 2.678469657897949, "learning_rate": 8.512969367265483e-07, "loss": 0.06854385137557983, "step": 6358 }, { "epoch": 1.6508307372793354, "grad_norm": 3.5981593132019043, "learning_rate": 8.507895223848417e-07, "loss": 0.013454851694405079, "step": 6359 }, { "epoch": 1.6510903426791277, "grad_norm": 3.293412446975708, "learning_rate": 8.502824437800615e-07, "loss": 0.02759130671620369, "step": 6360 }, { "epoch": 1.65134994807892, "grad_norm": 4.087477207183838, "learning_rate": 8.49775701001847e-07, "loss": 0.03103368729352951, "step": 6361 }, { "epoch": 1.6516095534787123, "grad_norm": 3.6948843002319336, "learning_rate": 8.492692941397798e-07, "loss": 0.0499681755900383, "step": 6362 }, { "epoch": 1.6518691588785046, "grad_norm": 4.0989556312561035, "learning_rate": 8.487632232833798e-07, "loss": 0.036490827798843384, "step": 6363 }, { "epoch": 1.652128764278297, "grad_norm": 2.9589767456054688, "learning_rate": 8.482574885221106e-07, "loss": 0.033988337963819504, "step": 6364 }, { "epoch": 1.6523883696780892, "grad_norm": 4.026688575744629, "learning_rate": 8.477520899453719e-07, "loss": 0.017553914338350296, "step": 6365 }, { "epoch": 1.6526479750778815, "grad_norm": 0.514191746711731, "learning_rate": 8.472470276425082e-07, "loss": 0.0012089229421690106, "step": 6366 }, { "epoch": 1.6529075804776738, "grad_norm": 3.059170722961426, "learning_rate": 8.467423017028035e-07, "loss": 0.02716069296002388, "step": 6367 }, { "epoch": 1.6531671858774661, "grad_norm": 3.173590898513794, "learning_rate": 8.462379122154807e-07, "loss": 0.0517645888030529, "step": 6368 }, { "epoch": 1.6534267912772584, "grad_norm": 2.4393718242645264, "learning_rate": 8.457338592697046e-07, "loss": 0.004292589146643877, "step": 6369 }, { "epoch": 1.6536863966770508, "grad_norm": 1.0938456058502197, "learning_rate": 8.452301429545798e-07, "loss": 0.011415124870836735, "step": 6370 }, { "epoch": 1.653946002076843, "grad_norm": 3.4617016315460205, "learning_rate": 8.447267633591525e-07, "loss": 0.02117186039686203, "step": 6371 }, { "epoch": 1.6542056074766354, "grad_norm": 3.533029794692993, "learning_rate": 8.442237205724094e-07, "loss": 0.015994228422641754, "step": 6372 }, { "epoch": 1.6544652128764277, "grad_norm": 2.7180287837982178, "learning_rate": 8.437210146832749e-07, "loss": 0.009317140094935894, "step": 6373 }, { "epoch": 1.6547248182762202, "grad_norm": 3.0537006855010986, "learning_rate": 8.432186457806175e-07, "loss": 0.027346858754754066, "step": 6374 }, { "epoch": 1.6549844236760125, "grad_norm": 3.411881923675537, "learning_rate": 8.427166139532436e-07, "loss": 0.027287829667329788, "step": 6375 }, { "epoch": 1.6552440290758048, "grad_norm": 3.9046976566314697, "learning_rate": 8.422149192899018e-07, "loss": 0.02133345603942871, "step": 6376 }, { "epoch": 1.6555036344755971, "grad_norm": 3.788942575454712, "learning_rate": 8.417135618792799e-07, "loss": 0.013495313003659248, "step": 6377 }, { "epoch": 1.6557632398753894, "grad_norm": 3.814992904663086, "learning_rate": 8.412125418100055e-07, "loss": 0.01739915832877159, "step": 6378 }, { "epoch": 1.6560228452751817, "grad_norm": 2.410881519317627, "learning_rate": 8.407118591706493e-07, "loss": 0.008453717455267906, "step": 6379 }, { "epoch": 1.656282450674974, "grad_norm": 3.646122455596924, "learning_rate": 8.402115140497189e-07, "loss": 0.03436795994639397, "step": 6380 }, { "epoch": 1.6565420560747663, "grad_norm": 2.579481840133667, "learning_rate": 8.397115065356641e-07, "loss": 0.0705704465508461, "step": 6381 }, { "epoch": 1.6568016614745587, "grad_norm": 3.7703604698181152, "learning_rate": 8.392118367168757e-07, "loss": 0.038419004529714584, "step": 6382 }, { "epoch": 1.657061266874351, "grad_norm": 3.7893784046173096, "learning_rate": 8.38712504681683e-07, "loss": 0.021586477756500244, "step": 6383 }, { "epoch": 1.6573208722741433, "grad_norm": 3.16825795173645, "learning_rate": 8.38213510518357e-07, "loss": 0.0460653081536293, "step": 6384 }, { "epoch": 1.6575804776739356, "grad_norm": 3.075956344604492, "learning_rate": 8.377148543151084e-07, "loss": 0.025198090821504593, "step": 6385 }, { "epoch": 1.657840083073728, "grad_norm": 2.949957847595215, "learning_rate": 8.372165361600876e-07, "loss": 0.01668640226125717, "step": 6386 }, { "epoch": 1.6580996884735204, "grad_norm": 4.920401573181152, "learning_rate": 8.367185561413866e-07, "loss": 0.03022783435881138, "step": 6387 }, { "epoch": 1.6583592938733127, "grad_norm": 1.5516546964645386, "learning_rate": 8.362209143470368e-07, "loss": 0.008534779772162437, "step": 6388 }, { "epoch": 1.658618899273105, "grad_norm": 2.944765329360962, "learning_rate": 8.357236108650091e-07, "loss": 0.007721484173089266, "step": 6389 }, { "epoch": 1.6588785046728973, "grad_norm": 3.059643030166626, "learning_rate": 8.352266457832158e-07, "loss": 0.01809374988079071, "step": 6390 }, { "epoch": 1.6591381100726896, "grad_norm": 3.071413278579712, "learning_rate": 8.347300191895097e-07, "loss": 0.02136079967021942, "step": 6391 }, { "epoch": 1.659397715472482, "grad_norm": 3.40004563331604, "learning_rate": 8.342337311716817e-07, "loss": 0.03131866455078125, "step": 6392 }, { "epoch": 1.6596573208722742, "grad_norm": 2.941183090209961, "learning_rate": 8.337377818174651e-07, "loss": 0.059756070375442505, "step": 6393 }, { "epoch": 1.6599169262720666, "grad_norm": 1.3262931108474731, "learning_rate": 8.332421712145316e-07, "loss": 0.0029156552627682686, "step": 6394 }, { "epoch": 1.6601765316718589, "grad_norm": 3.516577959060669, "learning_rate": 8.327468994504953e-07, "loss": 0.02571270242333412, "step": 6395 }, { "epoch": 1.6604361370716512, "grad_norm": 2.8761651515960693, "learning_rate": 8.32251966612907e-07, "loss": 0.018348783254623413, "step": 6396 }, { "epoch": 1.6606957424714435, "grad_norm": 2.802600622177124, "learning_rate": 8.317573727892603e-07, "loss": 0.07162581384181976, "step": 6397 }, { "epoch": 1.6609553478712358, "grad_norm": 2.582895040512085, "learning_rate": 8.312631180669889e-07, "loss": 0.04731127619743347, "step": 6398 }, { "epoch": 1.661214953271028, "grad_norm": 4.488314628601074, "learning_rate": 8.307692025334648e-07, "loss": 0.021234318614006042, "step": 6399 }, { "epoch": 1.6614745586708204, "grad_norm": 2.998818874359131, "learning_rate": 8.302756262760009e-07, "loss": 0.08116941154003143, "step": 6400 }, { "epoch": 1.6617341640706127, "grad_norm": 4.0204291343688965, "learning_rate": 8.297823893818502e-07, "loss": 0.02326127514243126, "step": 6401 }, { "epoch": 1.661993769470405, "grad_norm": 2.8996448516845703, "learning_rate": 8.292894919382058e-07, "loss": 0.007728715892881155, "step": 6402 }, { "epoch": 1.6622533748701973, "grad_norm": 3.2764089107513428, "learning_rate": 8.287969340322017e-07, "loss": 0.057874895632267, "step": 6403 }, { "epoch": 1.6625129802699896, "grad_norm": 2.2937352657318115, "learning_rate": 8.283047157509091e-07, "loss": 0.002855458529666066, "step": 6404 }, { "epoch": 1.662772585669782, "grad_norm": 3.698678970336914, "learning_rate": 8.278128371813423e-07, "loss": 0.032035503536462784, "step": 6405 }, { "epoch": 1.6630321910695742, "grad_norm": 3.937847137451172, "learning_rate": 8.273212984104531e-07, "loss": 0.04953676462173462, "step": 6406 }, { "epoch": 1.6632917964693665, "grad_norm": 3.492189645767212, "learning_rate": 8.268300995251353e-07, "loss": 0.008991837501525879, "step": 6407 }, { "epoch": 1.6635514018691588, "grad_norm": 3.439605474472046, "learning_rate": 8.263392406122212e-07, "loss": 0.041175223886966705, "step": 6408 }, { "epoch": 1.6638110072689511, "grad_norm": 3.497586250305176, "learning_rate": 8.258487217584827e-07, "loss": 0.02986108511686325, "step": 6409 }, { "epoch": 1.6640706126687435, "grad_norm": 3.2574455738067627, "learning_rate": 8.253585430506333e-07, "loss": 0.016523754224181175, "step": 6410 }, { "epoch": 1.6643302180685358, "grad_norm": 3.7814292907714844, "learning_rate": 8.248687045753256e-07, "loss": 0.01509048230946064, "step": 6411 }, { "epoch": 1.664589823468328, "grad_norm": 3.1299338340759277, "learning_rate": 8.243792064191503e-07, "loss": 0.06249808892607689, "step": 6412 }, { "epoch": 1.6648494288681204, "grad_norm": 3.392608880996704, "learning_rate": 8.238900486686413e-07, "loss": 0.049981191754341125, "step": 6413 }, { "epoch": 1.6651090342679127, "grad_norm": 3.5153000354766846, "learning_rate": 8.234012314102695e-07, "loss": 0.05853217467665672, "step": 6414 }, { "epoch": 1.665368639667705, "grad_norm": 3.591724157333374, "learning_rate": 8.22912754730446e-07, "loss": 0.024088682606816292, "step": 6415 }, { "epoch": 1.6656282450674973, "grad_norm": 2.8451411724090576, "learning_rate": 8.224246187155238e-07, "loss": 0.012351058423519135, "step": 6416 }, { "epoch": 1.6658878504672896, "grad_norm": 2.458343267440796, "learning_rate": 8.219368234517927e-07, "loss": 0.0163066815584898, "step": 6417 }, { "epoch": 1.666147455867082, "grad_norm": 3.7635676860809326, "learning_rate": 8.21449369025485e-07, "loss": 0.033181071281433105, "step": 6418 }, { "epoch": 1.6664070612668742, "grad_norm": 2.075559377670288, "learning_rate": 8.209622555227709e-07, "loss": 0.006124665029346943, "step": 6419 }, { "epoch": 1.6666666666666665, "grad_norm": 3.1679277420043945, "learning_rate": 8.204754830297602e-07, "loss": 0.032652631402015686, "step": 6420 }, { "epoch": 1.6669262720664588, "grad_norm": 2.3350799083709717, "learning_rate": 8.199890516325046e-07, "loss": 0.005859936587512493, "step": 6421 }, { "epoch": 1.6671858774662511, "grad_norm": 4.399423122406006, "learning_rate": 8.195029614169934e-07, "loss": 0.023223966360092163, "step": 6422 }, { "epoch": 1.6674454828660437, "grad_norm": 3.287996768951416, "learning_rate": 8.190172124691553e-07, "loss": 0.05013316124677658, "step": 6423 }, { "epoch": 1.667705088265836, "grad_norm": 3.819312334060669, "learning_rate": 8.185318048748611e-07, "loss": 0.014060924760997295, "step": 6424 }, { "epoch": 1.6679646936656283, "grad_norm": 3.563363790512085, "learning_rate": 8.180467387199184e-07, "loss": 0.06278781592845917, "step": 6425 }, { "epoch": 1.6682242990654206, "grad_norm": 3.6232247352600098, "learning_rate": 8.175620140900774e-07, "loss": 0.030207790434360504, "step": 6426 }, { "epoch": 1.6684839044652129, "grad_norm": 2.948836088180542, "learning_rate": 8.170776310710246e-07, "loss": 0.00927453488111496, "step": 6427 }, { "epoch": 1.6687435098650052, "grad_norm": 4.533563613891602, "learning_rate": 8.165935897483881e-07, "loss": 0.005416356958448887, "step": 6428 }, { "epoch": 1.6690031152647975, "grad_norm": 3.914975643157959, "learning_rate": 8.161098902077366e-07, "loss": 0.029836688190698624, "step": 6429 }, { "epoch": 1.6692627206645898, "grad_norm": 4.380002975463867, "learning_rate": 8.156265325345764e-07, "loss": 0.026512596756219864, "step": 6430 }, { "epoch": 1.6695223260643821, "grad_norm": 3.3832926750183105, "learning_rate": 8.151435168143537e-07, "loss": 0.046794451773166656, "step": 6431 }, { "epoch": 1.6697819314641744, "grad_norm": 3.1920812129974365, "learning_rate": 8.146608431324548e-07, "loss": 0.009179509244859219, "step": 6432 }, { "epoch": 1.6700415368639667, "grad_norm": 2.6231963634490967, "learning_rate": 8.141785115742049e-07, "loss": 0.06926266103982925, "step": 6433 }, { "epoch": 1.670301142263759, "grad_norm": 3.9855849742889404, "learning_rate": 8.136965222248713e-07, "loss": 0.011426689103245735, "step": 6434 }, { "epoch": 1.6705607476635516, "grad_norm": 3.3079516887664795, "learning_rate": 8.132148751696556e-07, "loss": 0.04281264543533325, "step": 6435 }, { "epoch": 1.6708203530633439, "grad_norm": 2.8216753005981445, "learning_rate": 8.127335704937042e-07, "loss": 0.039746351540088654, "step": 6436 }, { "epoch": 1.6710799584631362, "grad_norm": 3.1396584510803223, "learning_rate": 8.122526082820994e-07, "loss": 0.06781072914600372, "step": 6437 }, { "epoch": 1.6713395638629285, "grad_norm": 3.5360195636749268, "learning_rate": 8.117719886198655e-07, "loss": 0.029353348538279533, "step": 6438 }, { "epoch": 1.6715991692627208, "grad_norm": 3.2879645824432373, "learning_rate": 8.112917115919644e-07, "loss": 0.011655902490019798, "step": 6439 }, { "epoch": 1.671858774662513, "grad_norm": 3.770695209503174, "learning_rate": 8.108117772832977e-07, "loss": 0.022937925532460213, "step": 6440 }, { "epoch": 1.6721183800623054, "grad_norm": 4.043923854827881, "learning_rate": 8.103321857787077e-07, "loss": 0.04155948758125305, "step": 6441 }, { "epoch": 1.6723779854620977, "grad_norm": 4.234904766082764, "learning_rate": 8.09852937162975e-07, "loss": 0.026457544416189194, "step": 6442 }, { "epoch": 1.67263759086189, "grad_norm": 2.9133529663085938, "learning_rate": 8.093740315208188e-07, "loss": 0.04073674604296684, "step": 6443 }, { "epoch": 1.6728971962616823, "grad_norm": 4.332221508026123, "learning_rate": 8.088954689368999e-07, "loss": 0.020524218678474426, "step": 6444 }, { "epoch": 1.6731568016614746, "grad_norm": 3.7939577102661133, "learning_rate": 8.084172494958168e-07, "loss": 0.020669283345341682, "step": 6445 }, { "epoch": 1.673416407061267, "grad_norm": 4.044254779815674, "learning_rate": 8.07939373282107e-07, "loss": 0.025325916707515717, "step": 6446 }, { "epoch": 1.6736760124610592, "grad_norm": 5.488681316375732, "learning_rate": 8.074618403802494e-07, "loss": 0.017375795170664787, "step": 6447 }, { "epoch": 1.6739356178608515, "grad_norm": 2.9648969173431396, "learning_rate": 8.069846508746594e-07, "loss": 0.03197059780359268, "step": 6448 }, { "epoch": 1.6741952232606438, "grad_norm": 7.055971145629883, "learning_rate": 8.065078048496946e-07, "loss": 0.010024366900324821, "step": 6449 }, { "epoch": 1.6744548286604362, "grad_norm": 3.6620588302612305, "learning_rate": 8.060313023896499e-07, "loss": 0.019788749516010284, "step": 6450 }, { "epoch": 1.6747144340602285, "grad_norm": 4.4524149894714355, "learning_rate": 8.055551435787595e-07, "loss": 0.029208127409219742, "step": 6451 }, { "epoch": 1.6749740394600208, "grad_norm": 2.562331438064575, "learning_rate": 8.050793285011983e-07, "loss": 0.007051711902022362, "step": 6452 }, { "epoch": 1.675233644859813, "grad_norm": 3.414541006088257, "learning_rate": 8.04603857241079e-07, "loss": 0.009603879414498806, "step": 6453 }, { "epoch": 1.6754932502596054, "grad_norm": 3.254833936691284, "learning_rate": 8.041287298824535e-07, "loss": 0.009171650744974613, "step": 6454 }, { "epoch": 1.6757528556593977, "grad_norm": 4.009494781494141, "learning_rate": 8.036539465093146e-07, "loss": 0.014457877725362778, "step": 6455 }, { "epoch": 1.67601246105919, "grad_norm": 3.897747755050659, "learning_rate": 8.031795072055921e-07, "loss": 0.021351411938667297, "step": 6456 }, { "epoch": 1.6762720664589823, "grad_norm": 2.4627318382263184, "learning_rate": 8.027054120551572e-07, "loss": 0.007059779483824968, "step": 6457 }, { "epoch": 1.6765316718587746, "grad_norm": 3.2024853229522705, "learning_rate": 8.022316611418176e-07, "loss": 0.02801438421010971, "step": 6458 }, { "epoch": 1.676791277258567, "grad_norm": 2.8949391841888428, "learning_rate": 8.017582545493221e-07, "loss": 0.010270798578858376, "step": 6459 }, { "epoch": 1.6770508826583592, "grad_norm": 3.7658822536468506, "learning_rate": 8.01285192361359e-07, "loss": 0.033197611570358276, "step": 6460 }, { "epoch": 1.6773104880581515, "grad_norm": 3.5904200077056885, "learning_rate": 8.008124746615542e-07, "loss": 0.006544800475239754, "step": 6461 }, { "epoch": 1.6775700934579438, "grad_norm": 2.842435598373413, "learning_rate": 8.003401015334734e-07, "loss": 0.009001845493912697, "step": 6462 }, { "epoch": 1.6778296988577361, "grad_norm": 4.684884548187256, "learning_rate": 7.99868073060621e-07, "loss": 0.015316665172576904, "step": 6463 }, { "epoch": 1.6780893042575284, "grad_norm": 3.268289804458618, "learning_rate": 7.99396389326441e-07, "loss": 0.025460628792643547, "step": 6464 }, { "epoch": 1.6783489096573208, "grad_norm": 3.137387752532959, "learning_rate": 7.989250504143174e-07, "loss": 0.05998745188117027, "step": 6465 }, { "epoch": 1.678608515057113, "grad_norm": 2.58427095413208, "learning_rate": 7.984540564075702e-07, "loss": 0.01162781286984682, "step": 6466 }, { "epoch": 1.6788681204569054, "grad_norm": 4.238248348236084, "learning_rate": 7.979834073894621e-07, "loss": 0.020081553608179092, "step": 6467 }, { "epoch": 1.6791277258566977, "grad_norm": 3.6286251544952393, "learning_rate": 7.975131034431917e-07, "loss": 0.02562359720468521, "step": 6468 }, { "epoch": 1.67938733125649, "grad_norm": 3.534125566482544, "learning_rate": 7.970431446518991e-07, "loss": 0.04252849519252777, "step": 6469 }, { "epoch": 1.6796469366562823, "grad_norm": 4.2911458015441895, "learning_rate": 7.965735310986615e-07, "loss": 0.010464709252119064, "step": 6470 }, { "epoch": 1.6799065420560748, "grad_norm": 3.4413599967956543, "learning_rate": 7.961042628664958e-07, "loss": 0.028642211109399796, "step": 6471 }, { "epoch": 1.6801661474558671, "grad_norm": 5.733522891998291, "learning_rate": 7.956353400383586e-07, "loss": 0.0370512530207634, "step": 6472 }, { "epoch": 1.6804257528556594, "grad_norm": 3.9914681911468506, "learning_rate": 7.951667626971443e-07, "loss": 0.022224027663469315, "step": 6473 }, { "epoch": 1.6806853582554517, "grad_norm": 3.4585137367248535, "learning_rate": 7.94698530925686e-07, "loss": 0.025600802153348923, "step": 6474 }, { "epoch": 1.680944963655244, "grad_norm": 2.8441519737243652, "learning_rate": 7.942306448067574e-07, "loss": 0.032792747020721436, "step": 6475 }, { "epoch": 1.6812045690550363, "grad_norm": 2.1438233852386475, "learning_rate": 7.937631044230699e-07, "loss": 0.007686486467719078, "step": 6476 }, { "epoch": 1.6814641744548287, "grad_norm": 3.360712766647339, "learning_rate": 7.932959098572731e-07, "loss": 0.02003222517669201, "step": 6477 }, { "epoch": 1.681723779854621, "grad_norm": 1.5457613468170166, "learning_rate": 7.928290611919573e-07, "loss": 0.005671720486134291, "step": 6478 }, { "epoch": 1.6819833852544133, "grad_norm": 3.068373918533325, "learning_rate": 7.923625585096496e-07, "loss": 0.03626076877117157, "step": 6479 }, { "epoch": 1.6822429906542056, "grad_norm": 1.9280670881271362, "learning_rate": 7.91896401892818e-07, "loss": 0.006845663767307997, "step": 6480 }, { "epoch": 1.6825025960539979, "grad_norm": 4.797774314880371, "learning_rate": 7.914305914238682e-07, "loss": 0.018324553966522217, "step": 6481 }, { "epoch": 1.6827622014537902, "grad_norm": 2.905057430267334, "learning_rate": 7.909651271851438e-07, "loss": 0.014799488708376884, "step": 6482 }, { "epoch": 1.6830218068535827, "grad_norm": 3.213758945465088, "learning_rate": 7.905000092589296e-07, "loss": 0.023829031735658646, "step": 6483 }, { "epoch": 1.683281412253375, "grad_norm": 5.088649272918701, "learning_rate": 7.900352377274471e-07, "loss": 0.029738133773207664, "step": 6484 }, { "epoch": 1.6835410176531673, "grad_norm": 0.5413783192634583, "learning_rate": 7.895708126728569e-07, "loss": 0.0015057278797030449, "step": 6485 }, { "epoch": 1.6838006230529596, "grad_norm": 3.152911424636841, "learning_rate": 7.891067341772596e-07, "loss": 0.019730836153030396, "step": 6486 }, { "epoch": 1.684060228452752, "grad_norm": 2.4365336894989014, "learning_rate": 7.886430023226927e-07, "loss": 0.004780878778547049, "step": 6487 }, { "epoch": 1.6843198338525442, "grad_norm": 1.8772950172424316, "learning_rate": 7.881796171911349e-07, "loss": 0.009211225435137749, "step": 6488 }, { "epoch": 1.6845794392523366, "grad_norm": 2.6714680194854736, "learning_rate": 7.877165788644998e-07, "loss": 0.008123097009956837, "step": 6489 }, { "epoch": 1.6848390446521289, "grad_norm": 4.03562593460083, "learning_rate": 7.872538874246436e-07, "loss": 0.022161278873682022, "step": 6490 }, { "epoch": 1.6850986500519212, "grad_norm": 3.2598459720611572, "learning_rate": 7.867915429533596e-07, "loss": 0.011496136896312237, "step": 6491 }, { "epoch": 1.6853582554517135, "grad_norm": 2.746079444885254, "learning_rate": 7.863295455323791e-07, "loss": 0.08405162394046783, "step": 6492 }, { "epoch": 1.6856178608515058, "grad_norm": 3.60962176322937, "learning_rate": 7.85867895243373e-07, "loss": 0.04434242844581604, "step": 6493 }, { "epoch": 1.685877466251298, "grad_norm": 2.4116363525390625, "learning_rate": 7.854065921679499e-07, "loss": 0.01500263437628746, "step": 6494 }, { "epoch": 1.6861370716510904, "grad_norm": 2.7057645320892334, "learning_rate": 7.849456363876579e-07, "loss": 0.04647335410118103, "step": 6495 }, { "epoch": 1.6863966770508827, "grad_norm": 2.55985426902771, "learning_rate": 7.844850279839847e-07, "loss": 0.06716324388980865, "step": 6496 }, { "epoch": 1.686656282450675, "grad_norm": 2.8664889335632324, "learning_rate": 7.84024767038353e-07, "loss": 0.030228886753320694, "step": 6497 }, { "epoch": 1.6869158878504673, "grad_norm": 3.704404830932617, "learning_rate": 7.835648536321285e-07, "loss": 0.025870824232697487, "step": 6498 }, { "epoch": 1.6871754932502596, "grad_norm": 3.6839029788970947, "learning_rate": 7.831052878466124e-07, "loss": 0.040916286408901215, "step": 6499 }, { "epoch": 1.687435098650052, "grad_norm": 2.9171676635742188, "learning_rate": 7.826460697630449e-07, "loss": 0.007044179365038872, "step": 6500 }, { "epoch": 1.6876947040498442, "grad_norm": 2.9895529747009277, "learning_rate": 7.821871994626062e-07, "loss": 0.012942342087626457, "step": 6501 }, { "epoch": 1.6879543094496365, "grad_norm": 3.968629837036133, "learning_rate": 7.817286770264134e-07, "loss": 0.009827328845858574, "step": 6502 }, { "epoch": 1.6882139148494288, "grad_norm": 2.651787519454956, "learning_rate": 7.812705025355234e-07, "loss": 0.05423708260059357, "step": 6503 }, { "epoch": 1.6884735202492211, "grad_norm": 1.3109040260314941, "learning_rate": 7.808126760709309e-07, "loss": 0.002715685637667775, "step": 6504 }, { "epoch": 1.6887331256490135, "grad_norm": 2.407301664352417, "learning_rate": 7.803551977135681e-07, "loss": 0.01059767697006464, "step": 6505 }, { "epoch": 1.6889927310488058, "grad_norm": 3.022339344024658, "learning_rate": 7.798980675443079e-07, "loss": 0.021799206733703613, "step": 6506 }, { "epoch": 1.689252336448598, "grad_norm": 2.995145320892334, "learning_rate": 7.794412856439602e-07, "loss": 0.009785367175936699, "step": 6507 }, { "epoch": 1.6895119418483904, "grad_norm": 1.9184205532073975, "learning_rate": 7.789848520932729e-07, "loss": 0.008317962288856506, "step": 6508 }, { "epoch": 1.6897715472481827, "grad_norm": 2.947964906692505, "learning_rate": 7.785287669729338e-07, "loss": 0.011109404265880585, "step": 6509 }, { "epoch": 1.690031152647975, "grad_norm": 3.073394298553467, "learning_rate": 7.780730303635675e-07, "loss": 0.025037512183189392, "step": 6510 }, { "epoch": 1.6902907580477673, "grad_norm": 2.330690622329712, "learning_rate": 7.776176423457391e-07, "loss": 0.0039629763923585415, "step": 6511 }, { "epoch": 1.6905503634475596, "grad_norm": 3.9745993614196777, "learning_rate": 7.771626029999496e-07, "loss": 0.025053709745407104, "step": 6512 }, { "epoch": 1.690809968847352, "grad_norm": 3.9357640743255615, "learning_rate": 7.767079124066397e-07, "loss": 0.004751599393785, "step": 6513 }, { "epoch": 1.6910695742471442, "grad_norm": 3.8232688903808594, "learning_rate": 7.762535706461887e-07, "loss": 0.030349723994731903, "step": 6514 }, { "epoch": 1.6913291796469365, "grad_norm": 3.1409032344818115, "learning_rate": 7.757995777989139e-07, "loss": 0.014550453051924706, "step": 6515 }, { "epoch": 1.6915887850467288, "grad_norm": 2.9986069202423096, "learning_rate": 7.753459339450699e-07, "loss": 0.041622523218393326, "step": 6516 }, { "epoch": 1.6918483904465211, "grad_norm": 3.7551732063293457, "learning_rate": 7.748926391648516e-07, "loss": 0.023769326508045197, "step": 6517 }, { "epoch": 1.6921079958463134, "grad_norm": 0.5107168555259705, "learning_rate": 7.744396935383904e-07, "loss": 0.0012025481555610895, "step": 6518 }, { "epoch": 1.692367601246106, "grad_norm": 0.2991635203361511, "learning_rate": 7.739870971457581e-07, "loss": 0.0006825716700404882, "step": 6519 }, { "epoch": 1.6926272066458983, "grad_norm": 0.7498235106468201, "learning_rate": 7.735348500669611e-07, "loss": 0.0012422151630744338, "step": 6520 }, { "epoch": 1.6928868120456906, "grad_norm": 1.9989551305770874, "learning_rate": 7.730829523819477e-07, "loss": 0.006097543053328991, "step": 6521 }, { "epoch": 1.6931464174454829, "grad_norm": 2.6223976612091064, "learning_rate": 7.726314041706031e-07, "loss": 0.02034960314631462, "step": 6522 }, { "epoch": 1.6934060228452752, "grad_norm": 4.16682243347168, "learning_rate": 7.721802055127509e-07, "loss": 0.003655883949249983, "step": 6523 }, { "epoch": 1.6936656282450675, "grad_norm": 1.269741415977478, "learning_rate": 7.717293564881521e-07, "loss": 0.0017583619337528944, "step": 6524 }, { "epoch": 1.6939252336448598, "grad_norm": 2.9706592559814453, "learning_rate": 7.712788571765065e-07, "loss": 0.05300583317875862, "step": 6525 }, { "epoch": 1.6941848390446521, "grad_norm": 3.6480937004089355, "learning_rate": 7.708287076574523e-07, "loss": 0.016095489263534546, "step": 6526 }, { "epoch": 1.6944444444444444, "grad_norm": 2.198697566986084, "learning_rate": 7.703789080105657e-07, "loss": 0.012711238116025925, "step": 6527 }, { "epoch": 1.6947040498442367, "grad_norm": 3.4649221897125244, "learning_rate": 7.699294583153605e-07, "loss": 0.03352709859609604, "step": 6528 }, { "epoch": 1.694963655244029, "grad_norm": 2.669904947280884, "learning_rate": 7.694803586512899e-07, "loss": 0.026481367647647858, "step": 6529 }, { "epoch": 1.6952232606438213, "grad_norm": 1.9251773357391357, "learning_rate": 7.690316090977437e-07, "loss": 0.012551076710224152, "step": 6530 }, { "epoch": 1.6954828660436139, "grad_norm": 5.138202667236328, "learning_rate": 7.685832097340508e-07, "loss": 0.018053535372018814, "step": 6531 }, { "epoch": 1.6957424714434062, "grad_norm": 4.411810398101807, "learning_rate": 7.681351606394784e-07, "loss": 0.0431458055973053, "step": 6532 }, { "epoch": 1.6960020768431985, "grad_norm": 3.9128854274749756, "learning_rate": 7.676874618932302e-07, "loss": 0.03143465146422386, "step": 6533 }, { "epoch": 1.6962616822429908, "grad_norm": 3.560201406478882, "learning_rate": 7.672401135744503e-07, "loss": 0.03657964617013931, "step": 6534 }, { "epoch": 1.696521287642783, "grad_norm": 5.838589191436768, "learning_rate": 7.667931157622192e-07, "loss": 0.02391681633889675, "step": 6535 }, { "epoch": 1.6967808930425754, "grad_norm": 4.047171592712402, "learning_rate": 7.663464685355555e-07, "loss": 0.027970360592007637, "step": 6536 }, { "epoch": 1.6970404984423677, "grad_norm": 3.641639232635498, "learning_rate": 7.65900171973417e-07, "loss": 0.02050912193953991, "step": 6537 }, { "epoch": 1.69730010384216, "grad_norm": 2.916269302368164, "learning_rate": 7.65454226154698e-07, "loss": 0.02450954169034958, "step": 6538 }, { "epoch": 1.6975597092419523, "grad_norm": 1.6371315717697144, "learning_rate": 7.650086311582314e-07, "loss": 0.004615868907421827, "step": 6539 }, { "epoch": 1.6978193146417446, "grad_norm": 3.1839089393615723, "learning_rate": 7.645633870627889e-07, "loss": 0.04106344282627106, "step": 6540 }, { "epoch": 1.698078920041537, "grad_norm": 3.232689619064331, "learning_rate": 7.641184939470788e-07, "loss": 0.018845513463020325, "step": 6541 }, { "epoch": 1.6983385254413292, "grad_norm": 3.8386754989624023, "learning_rate": 7.636739518897488e-07, "loss": 0.012836456298828125, "step": 6542 }, { "epoch": 1.6985981308411215, "grad_norm": 3.4455692768096924, "learning_rate": 7.632297609693828e-07, "loss": 0.025433383882045746, "step": 6543 }, { "epoch": 1.6988577362409139, "grad_norm": 3.8486077785491943, "learning_rate": 7.627859212645041e-07, "loss": 0.016514241695404053, "step": 6544 }, { "epoch": 1.6991173416407062, "grad_norm": 1.7339839935302734, "learning_rate": 7.623424328535738e-07, "loss": 0.012952241115272045, "step": 6545 }, { "epoch": 1.6993769470404985, "grad_norm": 2.1433522701263428, "learning_rate": 7.618992958149897e-07, "loss": 0.008439709432423115, "step": 6546 }, { "epoch": 1.6996365524402908, "grad_norm": 2.9488658905029297, "learning_rate": 7.614565102270885e-07, "loss": 0.03194650635123253, "step": 6547 }, { "epoch": 1.699896157840083, "grad_norm": 0.7540781497955322, "learning_rate": 7.610140761681449e-07, "loss": 0.0010178082156926394, "step": 6548 }, { "epoch": 1.7001557632398754, "grad_norm": 2.834531545639038, "learning_rate": 7.605719937163705e-07, "loss": 0.014652024954557419, "step": 6549 }, { "epoch": 1.7004153686396677, "grad_norm": 3.7669677734375, "learning_rate": 7.601302629499168e-07, "loss": 0.02264569327235222, "step": 6550 }, { "epoch": 1.70067497403946, "grad_norm": 1.9616937637329102, "learning_rate": 7.596888839468696e-07, "loss": 0.006670181639492512, "step": 6551 }, { "epoch": 1.7009345794392523, "grad_norm": 2.4374802112579346, "learning_rate": 7.592478567852557e-07, "loss": 0.022746309638023376, "step": 6552 }, { "epoch": 1.7011941848390446, "grad_norm": 4.606997966766357, "learning_rate": 7.588071815430393e-07, "loss": 0.021836692467331886, "step": 6553 }, { "epoch": 1.701453790238837, "grad_norm": 3.402745485305786, "learning_rate": 7.583668582981202e-07, "loss": 0.04008140414953232, "step": 6554 }, { "epoch": 1.7017133956386292, "grad_norm": 3.1461005210876465, "learning_rate": 7.579268871283385e-07, "loss": 0.05656751990318298, "step": 6555 }, { "epoch": 1.7019730010384215, "grad_norm": 0.40374135971069336, "learning_rate": 7.574872681114705e-07, "loss": 0.0013069829437881708, "step": 6556 }, { "epoch": 1.7022326064382138, "grad_norm": 2.9053890705108643, "learning_rate": 7.570480013252313e-07, "loss": 0.012918537482619286, "step": 6557 }, { "epoch": 1.7024922118380061, "grad_norm": 4.679522514343262, "learning_rate": 7.566090868472728e-07, "loss": 0.021584108471870422, "step": 6558 }, { "epoch": 1.7027518172377984, "grad_norm": 2.411167860031128, "learning_rate": 7.56170524755185e-07, "loss": 0.003031836822628975, "step": 6559 }, { "epoch": 1.7030114226375908, "grad_norm": 2.5702435970306396, "learning_rate": 7.557323151264958e-07, "loss": 0.01581927388906479, "step": 6560 }, { "epoch": 1.703271028037383, "grad_norm": 3.5515129566192627, "learning_rate": 7.552944580386709e-07, "loss": 0.008203777484595776, "step": 6561 }, { "epoch": 1.7035306334371754, "grad_norm": 0.6781403422355652, "learning_rate": 7.548569535691127e-07, "loss": 0.0008359078201465309, "step": 6562 }, { "epoch": 1.7037902388369677, "grad_norm": 0.5342404842376709, "learning_rate": 7.544198017951626e-07, "loss": 0.0027134669944643974, "step": 6563 }, { "epoch": 1.70404984423676, "grad_norm": 2.048628568649292, "learning_rate": 7.539830027940986e-07, "loss": 0.0031228363513946533, "step": 6564 }, { "epoch": 1.7043094496365523, "grad_norm": 2.7031352519989014, "learning_rate": 7.535465566431374e-07, "loss": 0.0039431690238416195, "step": 6565 }, { "epoch": 1.7045690550363446, "grad_norm": 3.4759554862976074, "learning_rate": 7.531104634194323e-07, "loss": 0.05804567039012909, "step": 6566 }, { "epoch": 1.7048286604361371, "grad_norm": 2.8812012672424316, "learning_rate": 7.526747232000737e-07, "loss": 0.06418096274137497, "step": 6567 }, { "epoch": 1.7050882658359294, "grad_norm": 3.3459413051605225, "learning_rate": 7.522393360620922e-07, "loss": 0.014966041781008244, "step": 6568 }, { "epoch": 1.7053478712357217, "grad_norm": 2.6520469188690186, "learning_rate": 7.518043020824535e-07, "loss": 0.023160260170698166, "step": 6569 }, { "epoch": 1.705607476635514, "grad_norm": 4.5580058097839355, "learning_rate": 7.513696213380608e-07, "loss": 0.02439269982278347, "step": 6570 }, { "epoch": 1.7058670820353063, "grad_norm": 3.56819748878479, "learning_rate": 7.509352939057573e-07, "loss": 0.048054151237010956, "step": 6571 }, { "epoch": 1.7061266874350987, "grad_norm": 3.881624937057495, "learning_rate": 7.505013198623207e-07, "loss": 0.030082670971751213, "step": 6572 }, { "epoch": 1.706386292834891, "grad_norm": 3.8096277713775635, "learning_rate": 7.500676992844686e-07, "loss": 0.012058096006512642, "step": 6573 }, { "epoch": 1.7066458982346833, "grad_norm": 1.0320452451705933, "learning_rate": 7.49634432248855e-07, "loss": 0.0017307419329881668, "step": 6574 }, { "epoch": 1.7069055036344756, "grad_norm": 4.707848072052002, "learning_rate": 7.492015188320711e-07, "loss": 0.041093721985816956, "step": 6575 }, { "epoch": 1.7071651090342679, "grad_norm": 3.531310796737671, "learning_rate": 7.487689591106467e-07, "loss": 0.009267332032322884, "step": 6576 }, { "epoch": 1.7074247144340602, "grad_norm": 4.165587902069092, "learning_rate": 7.483367531610483e-07, "loss": 0.029662443324923515, "step": 6577 }, { "epoch": 1.7076843198338525, "grad_norm": 2.890850782394409, "learning_rate": 7.479049010596795e-07, "loss": 0.05836740508675575, "step": 6578 }, { "epoch": 1.707943925233645, "grad_norm": 3.1127586364746094, "learning_rate": 7.474734028828828e-07, "loss": 0.07925517112016678, "step": 6579 }, { "epoch": 1.7082035306334373, "grad_norm": 3.512006998062134, "learning_rate": 7.470422587069368e-07, "loss": 0.05965358763933182, "step": 6580 }, { "epoch": 1.7084631360332296, "grad_norm": 3.6046388149261475, "learning_rate": 7.466114686080576e-07, "loss": 0.02834632620215416, "step": 6581 }, { "epoch": 1.708722741433022, "grad_norm": 3.4808871746063232, "learning_rate": 7.461810326623988e-07, "loss": 0.010232850909233093, "step": 6582 }, { "epoch": 1.7089823468328142, "grad_norm": 2.726074457168579, "learning_rate": 7.457509509460522e-07, "loss": 0.020712606608867645, "step": 6583 }, { "epoch": 1.7092419522326066, "grad_norm": 4.213994026184082, "learning_rate": 7.453212235350471e-07, "loss": 0.016089990735054016, "step": 6584 }, { "epoch": 1.7095015576323989, "grad_norm": 3.737675428390503, "learning_rate": 7.448918505053476e-07, "loss": 0.011396411806344986, "step": 6585 }, { "epoch": 1.7097611630321912, "grad_norm": 2.4568400382995605, "learning_rate": 7.444628319328586e-07, "loss": 0.007723875343799591, "step": 6586 }, { "epoch": 1.7100207684319835, "grad_norm": 2.1961967945098877, "learning_rate": 7.440341678934197e-07, "loss": 0.01190720684826374, "step": 6587 }, { "epoch": 1.7102803738317758, "grad_norm": 3.8073573112487793, "learning_rate": 7.436058584628098e-07, "loss": 0.025723638013005257, "step": 6588 }, { "epoch": 1.710539979231568, "grad_norm": 3.007739543914795, "learning_rate": 7.431779037167437e-07, "loss": 0.011731518432497978, "step": 6589 }, { "epoch": 1.7107995846313604, "grad_norm": 2.8306984901428223, "learning_rate": 7.42750303730874e-07, "loss": 0.016357483342289925, "step": 6590 }, { "epoch": 1.7110591900311527, "grad_norm": 3.42004656791687, "learning_rate": 7.423230585807907e-07, "loss": 0.03092346526682377, "step": 6591 }, { "epoch": 1.711318795430945, "grad_norm": 3.8006277084350586, "learning_rate": 7.418961683420211e-07, "loss": 0.011460820212960243, "step": 6592 }, { "epoch": 1.7115784008307373, "grad_norm": 2.381793975830078, "learning_rate": 7.414696330900291e-07, "loss": 0.00829631183296442, "step": 6593 }, { "epoch": 1.7118380062305296, "grad_norm": 2.8555831909179688, "learning_rate": 7.410434529002172e-07, "loss": 0.054282575845718384, "step": 6594 }, { "epoch": 1.712097611630322, "grad_norm": 2.308224678039551, "learning_rate": 7.406176278479232e-07, "loss": 0.010223094373941422, "step": 6595 }, { "epoch": 1.7123572170301142, "grad_norm": 4.01390266418457, "learning_rate": 7.401921580084244e-07, "loss": 0.018520154058933258, "step": 6596 }, { "epoch": 1.7126168224299065, "grad_norm": 2.7530055046081543, "learning_rate": 7.397670434569335e-07, "loss": 0.018077446147799492, "step": 6597 }, { "epoch": 1.7128764278296988, "grad_norm": 2.7471776008605957, "learning_rate": 7.393422842686008e-07, "loss": 0.008530521765351295, "step": 6598 }, { "epoch": 1.7131360332294912, "grad_norm": 2.727847099304199, "learning_rate": 7.389178805185145e-07, "loss": 0.022440755739808083, "step": 6599 }, { "epoch": 1.7133956386292835, "grad_norm": 2.766157627105713, "learning_rate": 7.384938322816995e-07, "loss": 0.06234140321612358, "step": 6600 }, { "epoch": 1.7136552440290758, "grad_norm": 2.763861656188965, "learning_rate": 7.380701396331168e-07, "loss": 0.07325172424316406, "step": 6601 }, { "epoch": 1.713914849428868, "grad_norm": 3.0085041522979736, "learning_rate": 7.376468026476669e-07, "loss": 0.012895490974187851, "step": 6602 }, { "epoch": 1.7141744548286604, "grad_norm": 2.4685823917388916, "learning_rate": 7.372238214001852e-07, "loss": 0.006718166638165712, "step": 6603 }, { "epoch": 1.7144340602284527, "grad_norm": 3.1495471000671387, "learning_rate": 7.368011959654456e-07, "loss": 0.01853763312101364, "step": 6604 }, { "epoch": 1.714693665628245, "grad_norm": 1.871596336364746, "learning_rate": 7.363789264181586e-07, "loss": 0.006797512993216515, "step": 6605 }, { "epoch": 1.7149532710280373, "grad_norm": 3.267129898071289, "learning_rate": 7.359570128329711e-07, "loss": 0.02016032487154007, "step": 6606 }, { "epoch": 1.7152128764278296, "grad_norm": 2.793922185897827, "learning_rate": 7.355354552844693e-07, "loss": 0.018136415630578995, "step": 6607 }, { "epoch": 1.715472481827622, "grad_norm": 2.9162120819091797, "learning_rate": 7.351142538471728e-07, "loss": 0.05365937575697899, "step": 6608 }, { "epoch": 1.7157320872274142, "grad_norm": 1.9421186447143555, "learning_rate": 7.346934085955416e-07, "loss": 0.007248209323734045, "step": 6609 }, { "epoch": 1.7159916926272065, "grad_norm": 2.8611247539520264, "learning_rate": 7.342729196039716e-07, "loss": 0.013061607256531715, "step": 6610 }, { "epoch": 1.7162512980269988, "grad_norm": 2.976346492767334, "learning_rate": 7.338527869467954e-07, "loss": 0.05776645243167877, "step": 6611 }, { "epoch": 1.7165109034267911, "grad_norm": 3.064929246902466, "learning_rate": 7.33433010698283e-07, "loss": 0.0229109488427639, "step": 6612 }, { "epoch": 1.7167705088265834, "grad_norm": 3.3815817832946777, "learning_rate": 7.330135909326405e-07, "loss": 0.05483725294470787, "step": 6613 }, { "epoch": 1.7170301142263757, "grad_norm": 3.056535243988037, "learning_rate": 7.325945277240124e-07, "loss": 0.01975514367222786, "step": 6614 }, { "epoch": 1.7172897196261683, "grad_norm": 3.6482741832733154, "learning_rate": 7.3217582114648e-07, "loss": 0.031594276428222656, "step": 6615 }, { "epoch": 1.7175493250259606, "grad_norm": 3.4221601486206055, "learning_rate": 7.317574712740595e-07, "loss": 0.03995085507631302, "step": 6616 }, { "epoch": 1.717808930425753, "grad_norm": 2.6219217777252197, "learning_rate": 7.313394781807066e-07, "loss": 0.07951144874095917, "step": 6617 }, { "epoch": 1.7180685358255452, "grad_norm": 5.644211292266846, "learning_rate": 7.309218419403132e-07, "loss": 0.006559263449162245, "step": 6618 }, { "epoch": 1.7183281412253375, "grad_norm": 2.65795636177063, "learning_rate": 7.305045626267074e-07, "loss": 0.010918496176600456, "step": 6619 }, { "epoch": 1.7185877466251298, "grad_norm": 2.2758376598358154, "learning_rate": 7.300876403136544e-07, "loss": 0.008300527930259705, "step": 6620 }, { "epoch": 1.7188473520249221, "grad_norm": 3.1966679096221924, "learning_rate": 7.296710750748563e-07, "loss": 0.06479578465223312, "step": 6621 }, { "epoch": 1.7191069574247144, "grad_norm": 3.538160562515259, "learning_rate": 7.292548669839531e-07, "loss": 0.012680625542998314, "step": 6622 }, { "epoch": 1.7193665628245067, "grad_norm": 1.7771635055541992, "learning_rate": 7.288390161145209e-07, "loss": 0.010055904276669025, "step": 6623 }, { "epoch": 1.719626168224299, "grad_norm": 3.6445693969726562, "learning_rate": 7.284235225400716e-07, "loss": 0.02155512385070324, "step": 6624 }, { "epoch": 1.7198857736240913, "grad_norm": 1.5645381212234497, "learning_rate": 7.280083863340559e-07, "loss": 0.005433727987110615, "step": 6625 }, { "epoch": 1.7201453790238836, "grad_norm": 4.558318614959717, "learning_rate": 7.275936075698597e-07, "loss": 0.013717739842832088, "step": 6626 }, { "epoch": 1.7204049844236762, "grad_norm": 1.7552204132080078, "learning_rate": 7.271791863208067e-07, "loss": 0.008008705452084541, "step": 6627 }, { "epoch": 1.7206645898234685, "grad_norm": 2.9374775886535645, "learning_rate": 7.267651226601577e-07, "loss": 0.06595411151647568, "step": 6628 }, { "epoch": 1.7209241952232608, "grad_norm": 0.8434232473373413, "learning_rate": 7.263514166611083e-07, "loss": 0.0021050875075161457, "step": 6629 }, { "epoch": 1.721183800623053, "grad_norm": 2.847670078277588, "learning_rate": 7.259380683967936e-07, "loss": 0.07661080360412598, "step": 6630 }, { "epoch": 1.7214434060228454, "grad_norm": 2.4209089279174805, "learning_rate": 7.255250779402835e-07, "loss": 0.009748511016368866, "step": 6631 }, { "epoch": 1.7217030114226377, "grad_norm": 2.0647878646850586, "learning_rate": 7.251124453645849e-07, "loss": 0.008985479362308979, "step": 6632 }, { "epoch": 1.72196261682243, "grad_norm": 3.0290420055389404, "learning_rate": 7.247001707426424e-07, "loss": 0.030395766720175743, "step": 6633 }, { "epoch": 1.7222222222222223, "grad_norm": 3.6615796089172363, "learning_rate": 7.242882541473367e-07, "loss": 0.02840578928589821, "step": 6634 }, { "epoch": 1.7224818276220146, "grad_norm": 5.555388927459717, "learning_rate": 7.238766956514844e-07, "loss": 0.008679637685418129, "step": 6635 }, { "epoch": 1.722741433021807, "grad_norm": 0.9850879311561584, "learning_rate": 7.23465495327841e-07, "loss": 0.006298703607171774, "step": 6636 }, { "epoch": 1.7230010384215992, "grad_norm": 3.293332099914551, "learning_rate": 7.230546532490958e-07, "loss": 0.020547252148389816, "step": 6637 }, { "epoch": 1.7232606438213915, "grad_norm": 2.7889606952667236, "learning_rate": 7.226441694878774e-07, "loss": 0.051890041679143906, "step": 6638 }, { "epoch": 1.7235202492211839, "grad_norm": 3.885807991027832, "learning_rate": 7.222340441167498e-07, "loss": 0.02357194945216179, "step": 6639 }, { "epoch": 1.7237798546209762, "grad_norm": 3.4299862384796143, "learning_rate": 7.218242772082131e-07, "loss": 0.009265373460948467, "step": 6640 }, { "epoch": 1.7240394600207685, "grad_norm": 2.1387338638305664, "learning_rate": 7.214148688347055e-07, "loss": 0.010082235559821129, "step": 6641 }, { "epoch": 1.7242990654205608, "grad_norm": 3.424471378326416, "learning_rate": 7.210058190686003e-07, "loss": 0.026074586436152458, "step": 6642 }, { "epoch": 1.724558670820353, "grad_norm": 4.884028911590576, "learning_rate": 7.205971279822084e-07, "loss": 0.026254722848534584, "step": 6643 }, { "epoch": 1.7248182762201454, "grad_norm": 3.5366313457489014, "learning_rate": 7.201887956477774e-07, "loss": 0.04297886788845062, "step": 6644 }, { "epoch": 1.7250778816199377, "grad_norm": 4.168588161468506, "learning_rate": 7.197808221374904e-07, "loss": 0.019673466682434082, "step": 6645 }, { "epoch": 1.72533748701973, "grad_norm": 3.068510055541992, "learning_rate": 7.193732075234693e-07, "loss": 0.005096053704619408, "step": 6646 }, { "epoch": 1.7255970924195223, "grad_norm": 5.177803993225098, "learning_rate": 7.189659518777687e-07, "loss": 0.008962918072938919, "step": 6647 }, { "epoch": 1.7258566978193146, "grad_norm": 0.8303733468055725, "learning_rate": 7.185590552723834e-07, "loss": 0.0017169343773275614, "step": 6648 }, { "epoch": 1.726116303219107, "grad_norm": 2.3683042526245117, "learning_rate": 7.181525177792436e-07, "loss": 0.00761116947978735, "step": 6649 }, { "epoch": 1.7263759086188992, "grad_norm": 4.266497611999512, "learning_rate": 7.177463394702157e-07, "loss": 0.022008193656802177, "step": 6650 }, { "epoch": 1.7266355140186915, "grad_norm": 0.5525133013725281, "learning_rate": 7.173405204171028e-07, "loss": 0.0013822319451719522, "step": 6651 }, { "epoch": 1.7268951194184838, "grad_norm": 0.5270809531211853, "learning_rate": 7.169350606916434e-07, "loss": 0.0037581268697977066, "step": 6652 }, { "epoch": 1.7271547248182761, "grad_norm": 2.4366836547851562, "learning_rate": 7.165299603655141e-07, "loss": 0.022255338728427887, "step": 6653 }, { "epoch": 1.7274143302180685, "grad_norm": 2.53783917427063, "learning_rate": 7.161252195103286e-07, "loss": 0.017530124634504318, "step": 6654 }, { "epoch": 1.7276739356178608, "grad_norm": 1.5955066680908203, "learning_rate": 7.157208381976338e-07, "loss": 0.005001408513635397, "step": 6655 }, { "epoch": 1.727933541017653, "grad_norm": 3.127810001373291, "learning_rate": 7.153168164989166e-07, "loss": 0.07389503717422485, "step": 6656 }, { "epoch": 1.7281931464174454, "grad_norm": 3.422802686691284, "learning_rate": 7.149131544855976e-07, "loss": 0.012796558439731598, "step": 6657 }, { "epoch": 1.7284527518172377, "grad_norm": 2.842836856842041, "learning_rate": 7.145098522290362e-07, "loss": 0.008926269598305225, "step": 6658 }, { "epoch": 1.72871235721703, "grad_norm": 3.276503801345825, "learning_rate": 7.141069098005259e-07, "loss": 0.022656472399830818, "step": 6659 }, { "epoch": 1.7289719626168223, "grad_norm": 3.0496554374694824, "learning_rate": 7.137043272712981e-07, "loss": 0.012049670331180096, "step": 6660 }, { "epoch": 1.7292315680166146, "grad_norm": 1.3432831764221191, "learning_rate": 7.133021047125206e-07, "loss": 0.004726434592157602, "step": 6661 }, { "epoch": 1.729491173416407, "grad_norm": 0.5783061981201172, "learning_rate": 7.129002421952968e-07, "loss": 0.0015100024174898863, "step": 6662 }, { "epoch": 1.7297507788161994, "grad_norm": 0.7216220498085022, "learning_rate": 7.124987397906661e-07, "loss": 0.0017855095211416483, "step": 6663 }, { "epoch": 1.7300103842159917, "grad_norm": 3.2652060985565186, "learning_rate": 7.120975975696064e-07, "loss": 0.02783513069152832, "step": 6664 }, { "epoch": 1.730269989615784, "grad_norm": 3.4161975383758545, "learning_rate": 7.116968156030297e-07, "loss": 0.038915351033210754, "step": 6665 }, { "epoch": 1.7305295950155763, "grad_norm": 3.0233914852142334, "learning_rate": 7.112963939617845e-07, "loss": 0.01522626169025898, "step": 6666 }, { "epoch": 1.7307892004153687, "grad_norm": 1.0102920532226562, "learning_rate": 7.108963327166571e-07, "loss": 0.00468176044523716, "step": 6667 }, { "epoch": 1.731048805815161, "grad_norm": 1.074552297592163, "learning_rate": 7.104966319383684e-07, "loss": 0.010803609155118465, "step": 6668 }, { "epoch": 1.7313084112149533, "grad_norm": 3.080839157104492, "learning_rate": 7.100972916975771e-07, "loss": 0.020648911595344543, "step": 6669 }, { "epoch": 1.7315680166147456, "grad_norm": 2.8872737884521484, "learning_rate": 7.096983120648771e-07, "loss": 0.008695060387253761, "step": 6670 }, { "epoch": 1.7318276220145379, "grad_norm": 2.9434213638305664, "learning_rate": 7.092996931107983e-07, "loss": 0.03028450347483158, "step": 6671 }, { "epoch": 1.7320872274143302, "grad_norm": 0.4709576368331909, "learning_rate": 7.089014349058085e-07, "loss": 0.0006397793767973781, "step": 6672 }, { "epoch": 1.7323468328141225, "grad_norm": 3.088012933731079, "learning_rate": 7.085035375203098e-07, "loss": 0.02394164726138115, "step": 6673 }, { "epoch": 1.7326064382139148, "grad_norm": 2.8667802810668945, "learning_rate": 7.081060010246415e-07, "loss": 0.07916560769081116, "step": 6674 }, { "epoch": 1.7328660436137073, "grad_norm": 2.502373695373535, "learning_rate": 7.077088254890792e-07, "loss": 0.05414792150259018, "step": 6675 }, { "epoch": 1.7331256490134996, "grad_norm": 4.085355281829834, "learning_rate": 7.07312010983834e-07, "loss": 0.03061913140118122, "step": 6676 }, { "epoch": 1.733385254413292, "grad_norm": 3.548349380493164, "learning_rate": 7.069155575790547e-07, "loss": 0.014435618184506893, "step": 6677 }, { "epoch": 1.7336448598130842, "grad_norm": 3.1246933937072754, "learning_rate": 7.065194653448234e-07, "loss": 0.005414453335106373, "step": 6678 }, { "epoch": 1.7339044652128766, "grad_norm": 1.7591595649719238, "learning_rate": 7.061237343511613e-07, "loss": 0.004536230117082596, "step": 6679 }, { "epoch": 1.7341640706126689, "grad_norm": 3.0462090969085693, "learning_rate": 7.057283646680249e-07, "loss": 0.08820571005344391, "step": 6680 }, { "epoch": 1.7344236760124612, "grad_norm": 3.4178707599639893, "learning_rate": 7.05333356365306e-07, "loss": 0.02305430918931961, "step": 6681 }, { "epoch": 1.7346832814122535, "grad_norm": 3.016430616378784, "learning_rate": 7.04938709512833e-07, "loss": 0.05594779551029205, "step": 6682 }, { "epoch": 1.7349428868120458, "grad_norm": 3.0294268131256104, "learning_rate": 7.045444241803699e-07, "loss": 0.037424877285957336, "step": 6683 }, { "epoch": 1.735202492211838, "grad_norm": 3.316411018371582, "learning_rate": 7.04150500437618e-07, "loss": 0.03678946942090988, "step": 6684 }, { "epoch": 1.7354620976116304, "grad_norm": 3.510054111480713, "learning_rate": 7.037569383542147e-07, "loss": 0.04032941907644272, "step": 6685 }, { "epoch": 1.7357217030114227, "grad_norm": 3.445323944091797, "learning_rate": 7.033637379997308e-07, "loss": 0.025426918640732765, "step": 6686 }, { "epoch": 1.735981308411215, "grad_norm": 3.170907497406006, "learning_rate": 7.029708994436771e-07, "loss": 0.016775114461779594, "step": 6687 }, { "epoch": 1.7362409138110073, "grad_norm": 4.104689121246338, "learning_rate": 7.02578422755497e-07, "loss": 0.006896584294736385, "step": 6688 }, { "epoch": 1.7365005192107996, "grad_norm": 3.62239670753479, "learning_rate": 7.021863080045726e-07, "loss": 0.009126989170908928, "step": 6689 }, { "epoch": 1.736760124610592, "grad_norm": 1.7889665365219116, "learning_rate": 7.017945552602201e-07, "loss": 0.0033496394753456116, "step": 6690 }, { "epoch": 1.7370197300103842, "grad_norm": 3.762885093688965, "learning_rate": 7.014031645916923e-07, "loss": 0.029151368886232376, "step": 6691 }, { "epoch": 1.7372793354101765, "grad_norm": 2.566988229751587, "learning_rate": 7.010121360681787e-07, "loss": 0.004349598195403814, "step": 6692 }, { "epoch": 1.7375389408099688, "grad_norm": 3.4818544387817383, "learning_rate": 7.006214697588036e-07, "loss": 0.015151036903262138, "step": 6693 }, { "epoch": 1.7377985462097612, "grad_norm": 2.0543971061706543, "learning_rate": 7.002311657326279e-07, "loss": 0.002496160566806793, "step": 6694 }, { "epoch": 1.7380581516095535, "grad_norm": 3.613309383392334, "learning_rate": 6.99841224058649e-07, "loss": 0.03721163421869278, "step": 6695 }, { "epoch": 1.7383177570093458, "grad_norm": 3.2755208015441895, "learning_rate": 6.994516448057993e-07, "loss": 0.01022060215473175, "step": 6696 }, { "epoch": 1.738577362409138, "grad_norm": 3.3293423652648926, "learning_rate": 6.99062428042947e-07, "loss": 0.04183022677898407, "step": 6697 }, { "epoch": 1.7388369678089304, "grad_norm": 3.9477779865264893, "learning_rate": 6.986735738388976e-07, "loss": 0.014002572745084763, "step": 6698 }, { "epoch": 1.7390965732087227, "grad_norm": 2.0136938095092773, "learning_rate": 6.982850822623907e-07, "loss": 0.020566482096910477, "step": 6699 }, { "epoch": 1.739356178608515, "grad_norm": 0.6071699261665344, "learning_rate": 6.978969533821036e-07, "loss": 0.00775280874222517, "step": 6700 }, { "epoch": 1.7396157840083073, "grad_norm": 4.696200370788574, "learning_rate": 6.975091872666484e-07, "loss": 0.015654731541872025, "step": 6701 }, { "epoch": 1.7398753894080996, "grad_norm": 3.2973241806030273, "learning_rate": 6.971217839845726e-07, "loss": 0.0705944299697876, "step": 6702 }, { "epoch": 1.740134994807892, "grad_norm": 4.5175395011901855, "learning_rate": 6.96734743604361e-07, "loss": 0.01084351260215044, "step": 6703 }, { "epoch": 1.7403946002076842, "grad_norm": 4.659790992736816, "learning_rate": 6.963480661944335e-07, "loss": 0.008118915371596813, "step": 6704 }, { "epoch": 1.7406542056074765, "grad_norm": 3.9743454456329346, "learning_rate": 6.959617518231448e-07, "loss": 0.01943543367087841, "step": 6705 }, { "epoch": 1.7409138110072688, "grad_norm": 4.041593074798584, "learning_rate": 6.955758005587876e-07, "loss": 0.05012732371687889, "step": 6706 }, { "epoch": 1.7411734164070611, "grad_norm": 2.442209005355835, "learning_rate": 6.951902124695885e-07, "loss": 0.005568670108914375, "step": 6707 }, { "epoch": 1.7414330218068534, "grad_norm": 3.287553071975708, "learning_rate": 6.948049876237116e-07, "loss": 0.031164437532424927, "step": 6708 }, { "epoch": 1.7416926272066457, "grad_norm": 2.1362464427948, "learning_rate": 6.944201260892544e-07, "loss": 0.01121087558567524, "step": 6709 }, { "epoch": 1.741952232606438, "grad_norm": 3.368941068649292, "learning_rate": 6.940356279342524e-07, "loss": 0.06164686754345894, "step": 6710 }, { "epoch": 1.7422118380062306, "grad_norm": 3.6117467880249023, "learning_rate": 6.936514932266763e-07, "loss": 0.0651518777012825, "step": 6711 }, { "epoch": 1.742471443406023, "grad_norm": 3.05320143699646, "learning_rate": 6.932677220344321e-07, "loss": 0.007215531077235937, "step": 6712 }, { "epoch": 1.7427310488058152, "grad_norm": 3.3429760932922363, "learning_rate": 6.928843144253619e-07, "loss": 0.019268861040472984, "step": 6713 }, { "epoch": 1.7429906542056075, "grad_norm": 3.8360161781311035, "learning_rate": 6.925012704672426e-07, "loss": 0.011772692203521729, "step": 6714 }, { "epoch": 1.7432502596053998, "grad_norm": 2.778813123703003, "learning_rate": 6.921185902277882e-07, "loss": 0.010662509128451347, "step": 6715 }, { "epoch": 1.7435098650051921, "grad_norm": 2.6044933795928955, "learning_rate": 6.917362737746487e-07, "loss": 0.010370717383921146, "step": 6716 }, { "epoch": 1.7437694704049844, "grad_norm": 2.4040071964263916, "learning_rate": 6.913543211754075e-07, "loss": 0.016608765348792076, "step": 6717 }, { "epoch": 1.7440290758047767, "grad_norm": 2.6155917644500732, "learning_rate": 6.909727324975854e-07, "loss": 0.011835113167762756, "step": 6718 }, { "epoch": 1.744288681204569, "grad_norm": 3.7957258224487305, "learning_rate": 6.905915078086392e-07, "loss": 0.053602222353219986, "step": 6719 }, { "epoch": 1.7445482866043613, "grad_norm": 3.363755702972412, "learning_rate": 6.902106471759595e-07, "loss": 0.028038043528795242, "step": 6720 }, { "epoch": 1.7448078920041536, "grad_norm": 1.5834141969680786, "learning_rate": 6.898301506668749e-07, "loss": 0.0021500037983059883, "step": 6721 }, { "epoch": 1.745067497403946, "grad_norm": 1.8473267555236816, "learning_rate": 6.894500183486478e-07, "loss": 0.001781234983354807, "step": 6722 }, { "epoch": 1.7453271028037385, "grad_norm": 2.70328950881958, "learning_rate": 6.890702502884775e-07, "loss": 0.017752215266227722, "step": 6723 }, { "epoch": 1.7455867082035308, "grad_norm": 2.29447603225708, "learning_rate": 6.886908465534975e-07, "loss": 0.003157127182930708, "step": 6724 }, { "epoch": 1.745846313603323, "grad_norm": 3.692763328552246, "learning_rate": 6.883118072107783e-07, "loss": 0.01048562116920948, "step": 6725 }, { "epoch": 1.7461059190031154, "grad_norm": 1.8784483671188354, "learning_rate": 6.879331323273253e-07, "loss": 0.0031387098133563995, "step": 6726 }, { "epoch": 1.7463655244029077, "grad_norm": 0.7619121670722961, "learning_rate": 6.875548219700792e-07, "loss": 0.0016237777890637517, "step": 6727 }, { "epoch": 1.7466251298027, "grad_norm": 2.7396442890167236, "learning_rate": 6.871768762059167e-07, "loss": 0.008459356613457203, "step": 6728 }, { "epoch": 1.7468847352024923, "grad_norm": 4.6321539878845215, "learning_rate": 6.867992951016506e-07, "loss": 0.023340335115790367, "step": 6729 }, { "epoch": 1.7471443406022846, "grad_norm": 2.8483071327209473, "learning_rate": 6.864220787240275e-07, "loss": 0.016224108636379242, "step": 6730 }, { "epoch": 1.747403946002077, "grad_norm": 2.858295202255249, "learning_rate": 6.860452271397317e-07, "loss": 0.05463512986898422, "step": 6731 }, { "epoch": 1.7476635514018692, "grad_norm": 3.2425427436828613, "learning_rate": 6.856687404153814e-07, "loss": 0.038254037499427795, "step": 6732 }, { "epoch": 1.7479231568016615, "grad_norm": 2.664759397506714, "learning_rate": 6.852926186175305e-07, "loss": 0.05570189654827118, "step": 6733 }, { "epoch": 1.7481827622014539, "grad_norm": 3.0217323303222656, "learning_rate": 6.849168618126695e-07, "loss": 0.014187682420015335, "step": 6734 }, { "epoch": 1.7484423676012462, "grad_norm": 4.452895641326904, "learning_rate": 6.845414700672231e-07, "loss": 0.011281531304121017, "step": 6735 }, { "epoch": 1.7487019730010385, "grad_norm": 0.31545165181159973, "learning_rate": 6.841664434475518e-07, "loss": 0.0008371417643502355, "step": 6736 }, { "epoch": 1.7489615784008308, "grad_norm": 2.8283848762512207, "learning_rate": 6.837917820199525e-07, "loss": 0.009044659323990345, "step": 6737 }, { "epoch": 1.749221183800623, "grad_norm": 1.0229309797286987, "learning_rate": 6.834174858506555e-07, "loss": 0.0022338428534567356, "step": 6738 }, { "epoch": 1.7494807892004154, "grad_norm": 4.415688991546631, "learning_rate": 6.830435550058297e-07, "loss": 0.0609736368060112, "step": 6739 }, { "epoch": 1.7497403946002077, "grad_norm": 4.712591648101807, "learning_rate": 6.826699895515751e-07, "loss": 0.011270084418356419, "step": 6740 }, { "epoch": 1.75, "grad_norm": 3.492992401123047, "learning_rate": 6.822967895539311e-07, "loss": 0.03723129630088806, "step": 6741 }, { "epoch": 1.7502596053997923, "grad_norm": 1.4847800731658936, "learning_rate": 6.81923955078871e-07, "loss": 0.0035709331277757883, "step": 6742 }, { "epoch": 1.7505192107995846, "grad_norm": 0.8645675778388977, "learning_rate": 6.815514861923028e-07, "loss": 0.0010410025715827942, "step": 6743 }, { "epoch": 1.750778816199377, "grad_norm": 2.55519437789917, "learning_rate": 6.811793829600707e-07, "loss": 0.01746240258216858, "step": 6744 }, { "epoch": 1.7510384215991692, "grad_norm": 2.7468321323394775, "learning_rate": 6.808076454479534e-07, "loss": 0.020446574315428734, "step": 6745 }, { "epoch": 1.7512980269989615, "grad_norm": 2.5929877758026123, "learning_rate": 6.80436273721667e-07, "loss": 0.04560333490371704, "step": 6746 }, { "epoch": 1.7515576323987538, "grad_norm": 3.3584721088409424, "learning_rate": 6.800652678468604e-07, "loss": 0.025993870571255684, "step": 6747 }, { "epoch": 1.7518172377985461, "grad_norm": 0.9509556889533997, "learning_rate": 6.79694627889119e-07, "loss": 0.0011978764086961746, "step": 6748 }, { "epoch": 1.7520768431983385, "grad_norm": 3.061148166656494, "learning_rate": 6.793243539139636e-07, "loss": 0.06724780797958374, "step": 6749 }, { "epoch": 1.7523364485981308, "grad_norm": 1.2008548974990845, "learning_rate": 6.789544459868508e-07, "loss": 0.011196143925189972, "step": 6750 }, { "epoch": 1.752596053997923, "grad_norm": 3.890047550201416, "learning_rate": 6.785849041731705e-07, "loss": 0.02513910084962845, "step": 6751 }, { "epoch": 1.7528556593977154, "grad_norm": 2.3905513286590576, "learning_rate": 6.782157285382508e-07, "loss": 0.017430434003472328, "step": 6752 }, { "epoch": 1.7531152647975077, "grad_norm": 2.6472291946411133, "learning_rate": 6.778469191473519e-07, "loss": 0.022488906979560852, "step": 6753 }, { "epoch": 1.7533748701973, "grad_norm": 3.108158588409424, "learning_rate": 6.774784760656725e-07, "loss": 0.07025933265686035, "step": 6754 }, { "epoch": 1.7536344755970923, "grad_norm": 1.4974613189697266, "learning_rate": 6.771103993583437e-07, "loss": 0.003743697889149189, "step": 6755 }, { "epoch": 1.7538940809968846, "grad_norm": 1.9389814138412476, "learning_rate": 6.767426890904332e-07, "loss": 0.0051367804408073425, "step": 6756 }, { "epoch": 1.754153686396677, "grad_norm": 4.0564141273498535, "learning_rate": 6.763753453269445e-07, "loss": 0.01871294341981411, "step": 6757 }, { "epoch": 1.7544132917964692, "grad_norm": 3.2839481830596924, "learning_rate": 6.760083681328147e-07, "loss": 0.04902420938014984, "step": 6758 }, { "epoch": 1.7546728971962615, "grad_norm": 3.5834479331970215, "learning_rate": 6.756417575729171e-07, "loss": 0.02656988799571991, "step": 6759 }, { "epoch": 1.754932502596054, "grad_norm": 3.042562961578369, "learning_rate": 6.752755137120606e-07, "loss": 0.018544502556324005, "step": 6760 }, { "epoch": 1.7551921079958464, "grad_norm": 4.1783246994018555, "learning_rate": 6.749096366149881e-07, "loss": 0.026979710906744003, "step": 6761 }, { "epoch": 1.7554517133956387, "grad_norm": 3.133143186569214, "learning_rate": 6.745441263463791e-07, "loss": 0.008267959579825401, "step": 6762 }, { "epoch": 1.755711318795431, "grad_norm": 1.924217939376831, "learning_rate": 6.741789829708467e-07, "loss": 0.0033503833692520857, "step": 6763 }, { "epoch": 1.7559709241952233, "grad_norm": 3.40358829498291, "learning_rate": 6.738142065529396e-07, "loss": 0.009352288208901882, "step": 6764 }, { "epoch": 1.7562305295950156, "grad_norm": 4.098274230957031, "learning_rate": 6.73449797157143e-07, "loss": 0.024984069168567657, "step": 6765 }, { "epoch": 1.7564901349948079, "grad_norm": 4.211545467376709, "learning_rate": 6.730857548478753e-07, "loss": 0.012602311559021473, "step": 6766 }, { "epoch": 1.7567497403946002, "grad_norm": 4.667227745056152, "learning_rate": 6.727220796894913e-07, "loss": 0.03433246165513992, "step": 6767 }, { "epoch": 1.7570093457943925, "grad_norm": 4.044346809387207, "learning_rate": 6.723587717462801e-07, "loss": 0.01889181137084961, "step": 6768 }, { "epoch": 1.7572689511941848, "grad_norm": 2.925035238265991, "learning_rate": 6.719958310824662e-07, "loss": 0.0046855611726641655, "step": 6769 }, { "epoch": 1.757528556593977, "grad_norm": 2.5914084911346436, "learning_rate": 6.7163325776221e-07, "loss": 0.01888330653309822, "step": 6770 }, { "epoch": 1.7577881619937694, "grad_norm": 3.0526297092437744, "learning_rate": 6.712710518496049e-07, "loss": 0.00867446418851614, "step": 6771 }, { "epoch": 1.758047767393562, "grad_norm": 2.388688802719116, "learning_rate": 6.709092134086813e-07, "loss": 0.006699795834720135, "step": 6772 }, { "epoch": 1.7583073727933543, "grad_norm": 4.107931613922119, "learning_rate": 6.705477425034046e-07, "loss": 0.052819326519966125, "step": 6773 }, { "epoch": 1.7585669781931466, "grad_norm": 3.1702828407287598, "learning_rate": 6.701866391976733e-07, "loss": 0.05503985285758972, "step": 6774 }, { "epoch": 1.7588265835929389, "grad_norm": 1.1654568910598755, "learning_rate": 6.698259035553228e-07, "loss": 0.005827699787914753, "step": 6775 }, { "epoch": 1.7590861889927312, "grad_norm": 2.1945223808288574, "learning_rate": 6.694655356401226e-07, "loss": 0.006337322760373354, "step": 6776 }, { "epoch": 1.7593457943925235, "grad_norm": 3.77994966506958, "learning_rate": 6.691055355157783e-07, "loss": 0.05079348385334015, "step": 6777 }, { "epoch": 1.7596053997923158, "grad_norm": 2.6922223567962646, "learning_rate": 6.687459032459289e-07, "loss": 0.011554365046322346, "step": 6778 }, { "epoch": 1.759865005192108, "grad_norm": 1.362868309020996, "learning_rate": 6.683866388941495e-07, "loss": 0.009549522772431374, "step": 6779 }, { "epoch": 1.7601246105919004, "grad_norm": 3.412062406539917, "learning_rate": 6.680277425239496e-07, "loss": 0.015889137983322144, "step": 6780 }, { "epoch": 1.7603842159916927, "grad_norm": 4.681589126586914, "learning_rate": 6.676692141987744e-07, "loss": 0.01610425114631653, "step": 6781 }, { "epoch": 1.760643821391485, "grad_norm": 3.7768595218658447, "learning_rate": 6.673110539820023e-07, "loss": 0.028758203610777855, "step": 6782 }, { "epoch": 1.7609034267912773, "grad_norm": 3.087698459625244, "learning_rate": 6.669532619369493e-07, "loss": 0.010704657062888145, "step": 6783 }, { "epoch": 1.7611630321910696, "grad_norm": 2.9993793964385986, "learning_rate": 6.665958381268634e-07, "loss": 0.05928972736001015, "step": 6784 }, { "epoch": 1.761422637590862, "grad_norm": 4.110957622528076, "learning_rate": 6.662387826149303e-07, "loss": 0.013272776268422604, "step": 6785 }, { "epoch": 1.7616822429906542, "grad_norm": 3.4153854846954346, "learning_rate": 6.658820954642687e-07, "loss": 0.021423380821943283, "step": 6786 }, { "epoch": 1.7619418483904465, "grad_norm": 3.1364738941192627, "learning_rate": 6.655257767379321e-07, "loss": 0.051390573382377625, "step": 6787 }, { "epoch": 1.7622014537902388, "grad_norm": 2.0423243045806885, "learning_rate": 6.651698264989105e-07, "loss": 0.009259672835469246, "step": 6788 }, { "epoch": 1.7624610591900312, "grad_norm": 3.2103307247161865, "learning_rate": 6.648142448101274e-07, "loss": 0.04483429342508316, "step": 6789 }, { "epoch": 1.7627206645898235, "grad_norm": 3.8730409145355225, "learning_rate": 6.644590317344409e-07, "loss": 0.01181235071271658, "step": 6790 }, { "epoch": 1.7629802699896158, "grad_norm": 3.9428601264953613, "learning_rate": 6.641041873346453e-07, "loss": 0.019475899636745453, "step": 6791 }, { "epoch": 1.763239875389408, "grad_norm": 3.3413162231445312, "learning_rate": 6.637497116734686e-07, "loss": 0.0082698380574584, "step": 6792 }, { "epoch": 1.7634994807892004, "grad_norm": 3.1654655933380127, "learning_rate": 6.633956048135741e-07, "loss": 0.02885996177792549, "step": 6793 }, { "epoch": 1.7637590861889927, "grad_norm": 3.4767534732818604, "learning_rate": 6.630418668175598e-07, "loss": 0.023838823661208153, "step": 6794 }, { "epoch": 1.764018691588785, "grad_norm": 1.517066478729248, "learning_rate": 6.626884977479582e-07, "loss": 0.005506153218448162, "step": 6795 }, { "epoch": 1.7642782969885773, "grad_norm": 5.630796909332275, "learning_rate": 6.623354976672376e-07, "loss": 0.005622546188533306, "step": 6796 }, { "epoch": 1.7645379023883696, "grad_norm": 3.6409544944763184, "learning_rate": 6.619828666377998e-07, "loss": 0.019211584702134132, "step": 6797 }, { "epoch": 1.764797507788162, "grad_norm": 3.0502407550811768, "learning_rate": 6.616306047219813e-07, "loss": 0.05767816677689552, "step": 6798 }, { "epoch": 1.7650571131879542, "grad_norm": 4.959212779998779, "learning_rate": 6.612787119820552e-07, "loss": 0.03192659467458725, "step": 6799 }, { "epoch": 1.7653167185877465, "grad_norm": 1.2285993099212646, "learning_rate": 6.609271884802272e-07, "loss": 0.002460741437971592, "step": 6800 }, { "epoch": 1.7655763239875388, "grad_norm": 1.8374109268188477, "learning_rate": 6.605760342786391e-07, "loss": 0.01425892487168312, "step": 6801 }, { "epoch": 1.7658359293873311, "grad_norm": 1.9255341291427612, "learning_rate": 6.602252494393663e-07, "loss": 0.006451833993196487, "step": 6802 }, { "epoch": 1.7660955347871234, "grad_norm": 1.8856101036071777, "learning_rate": 6.598748340244198e-07, "loss": 0.0034472837578505278, "step": 6803 }, { "epoch": 1.7663551401869158, "grad_norm": 1.8284881114959717, "learning_rate": 6.595247880957459e-07, "loss": 0.002392740687355399, "step": 6804 }, { "epoch": 1.766614745586708, "grad_norm": 3.4202721118927, "learning_rate": 6.59175111715223e-07, "loss": 0.054516881704330444, "step": 6805 }, { "epoch": 1.7668743509865004, "grad_norm": 3.746922731399536, "learning_rate": 6.588258049446673e-07, "loss": 0.01300542987883091, "step": 6806 }, { "epoch": 1.7671339563862927, "grad_norm": 1.8446385860443115, "learning_rate": 6.584768678458274e-07, "loss": 0.00903480313718319, "step": 6807 }, { "epoch": 1.7673935617860852, "grad_norm": 3.0905590057373047, "learning_rate": 6.58128300480388e-07, "loss": 0.04762762784957886, "step": 6808 }, { "epoch": 1.7676531671858775, "grad_norm": 4.576015949249268, "learning_rate": 6.577801029099676e-07, "loss": 0.0394175760447979, "step": 6809 }, { "epoch": 1.7679127725856698, "grad_norm": 3.882128953933716, "learning_rate": 6.574322751961193e-07, "loss": 0.007817881181836128, "step": 6810 }, { "epoch": 1.7681723779854621, "grad_norm": 4.1380839347839355, "learning_rate": 6.570848174003316e-07, "loss": 0.012582434341311455, "step": 6811 }, { "epoch": 1.7684319833852544, "grad_norm": 3.745711326599121, "learning_rate": 6.567377295840267e-07, "loss": 0.025542931631207466, "step": 6812 }, { "epoch": 1.7686915887850467, "grad_norm": 3.2241737842559814, "learning_rate": 6.563910118085616e-07, "loss": 0.026292428374290466, "step": 6813 }, { "epoch": 1.768951194184839, "grad_norm": 3.8445985317230225, "learning_rate": 6.560446641352288e-07, "loss": 0.018557999283075333, "step": 6814 }, { "epoch": 1.7692107995846313, "grad_norm": 4.044197082519531, "learning_rate": 6.556986866252535e-07, "loss": 0.025870895013213158, "step": 6815 }, { "epoch": 1.7694704049844237, "grad_norm": 4.214334011077881, "learning_rate": 6.553530793397975e-07, "loss": 0.028057683259248734, "step": 6816 }, { "epoch": 1.769730010384216, "grad_norm": 2.9529690742492676, "learning_rate": 6.550078423399565e-07, "loss": 0.006042844150215387, "step": 6817 }, { "epoch": 1.7699896157840083, "grad_norm": 3.7149341106414795, "learning_rate": 6.546629756867594e-07, "loss": 0.04831494390964508, "step": 6818 }, { "epoch": 1.7702492211838006, "grad_norm": 2.423302173614502, "learning_rate": 6.543184794411715e-07, "loss": 0.007099660113453865, "step": 6819 }, { "epoch": 1.770508826583593, "grad_norm": 3.651637554168701, "learning_rate": 6.539743536640921e-07, "loss": 0.014157862402498722, "step": 6820 }, { "epoch": 1.7707684319833854, "grad_norm": 1.5197432041168213, "learning_rate": 6.536305984163538e-07, "loss": 0.005935035180300474, "step": 6821 }, { "epoch": 1.7710280373831777, "grad_norm": 3.166660785675049, "learning_rate": 6.532872137587254e-07, "loss": 0.010026372969150543, "step": 6822 }, { "epoch": 1.77128764278297, "grad_norm": 3.4891395568847656, "learning_rate": 6.529441997519091e-07, "loss": 0.014865338802337646, "step": 6823 }, { "epoch": 1.7715472481827623, "grad_norm": 2.4026224613189697, "learning_rate": 6.526015564565423e-07, "loss": 0.01116601750254631, "step": 6824 }, { "epoch": 1.7718068535825546, "grad_norm": 3.123279094696045, "learning_rate": 6.522592839331962e-07, "loss": 0.06805633008480072, "step": 6825 }, { "epoch": 1.772066458982347, "grad_norm": 3.7683451175689697, "learning_rate": 6.519173822423768e-07, "loss": 0.022797301411628723, "step": 6826 }, { "epoch": 1.7723260643821392, "grad_norm": 3.2993946075439453, "learning_rate": 6.51575851444525e-07, "loss": 0.030820243060588837, "step": 6827 }, { "epoch": 1.7725856697819315, "grad_norm": 1.3903380632400513, "learning_rate": 6.512346916000144e-07, "loss": 0.003281674813479185, "step": 6828 }, { "epoch": 1.7728452751817239, "grad_norm": 4.629818439483643, "learning_rate": 6.508939027691548e-07, "loss": 0.015687057748436928, "step": 6829 }, { "epoch": 1.7731048805815162, "grad_norm": 4.778449535369873, "learning_rate": 6.505534850121904e-07, "loss": 0.014449586160480976, "step": 6830 }, { "epoch": 1.7733644859813085, "grad_norm": 3.2327213287353516, "learning_rate": 6.502134383892993e-07, "loss": 0.07645508646965027, "step": 6831 }, { "epoch": 1.7736240913811008, "grad_norm": 3.2204084396362305, "learning_rate": 6.498737629605933e-07, "loss": 0.012581576593220234, "step": 6832 }, { "epoch": 1.773883696780893, "grad_norm": 3.203110933303833, "learning_rate": 6.49534458786119e-07, "loss": 0.018096070736646652, "step": 6833 }, { "epoch": 1.7741433021806854, "grad_norm": 2.827314853668213, "learning_rate": 6.491955259258584e-07, "loss": 0.0021219600457698107, "step": 6834 }, { "epoch": 1.7744029075804777, "grad_norm": 2.6873373985290527, "learning_rate": 6.488569644397272e-07, "loss": 0.019309885799884796, "step": 6835 }, { "epoch": 1.77466251298027, "grad_norm": 3.4213855266571045, "learning_rate": 6.485187743875743e-07, "loss": 0.010598575696349144, "step": 6836 }, { "epoch": 1.7749221183800623, "grad_norm": 2.5493507385253906, "learning_rate": 6.481809558291848e-07, "loss": 0.01181956846266985, "step": 6837 }, { "epoch": 1.7751817237798546, "grad_norm": 3.0959134101867676, "learning_rate": 6.478435088242768e-07, "loss": 0.043917372822761536, "step": 6838 }, { "epoch": 1.775441329179647, "grad_norm": 3.4664969444274902, "learning_rate": 6.475064334325039e-07, "loss": 0.06351538747549057, "step": 6839 }, { "epoch": 1.7757009345794392, "grad_norm": 0.8472293615341187, "learning_rate": 6.471697297134526e-07, "loss": 0.003495721844956279, "step": 6840 }, { "epoch": 1.7759605399792315, "grad_norm": 3.3480281829833984, "learning_rate": 6.468333977266446e-07, "loss": 0.03799842298030853, "step": 6841 }, { "epoch": 1.7762201453790238, "grad_norm": 2.980252981185913, "learning_rate": 6.464974375315357e-07, "loss": 0.024563288316130638, "step": 6842 }, { "epoch": 1.7764797507788161, "grad_norm": 2.434126615524292, "learning_rate": 6.461618491875163e-07, "loss": 0.009652648121118546, "step": 6843 }, { "epoch": 1.7767393561786085, "grad_norm": 5.579076290130615, "learning_rate": 6.4582663275391e-07, "loss": 0.006346831098198891, "step": 6844 }, { "epoch": 1.7769989615784008, "grad_norm": 3.9436585903167725, "learning_rate": 6.454917882899764e-07, "loss": 0.03796900063753128, "step": 6845 }, { "epoch": 1.777258566978193, "grad_norm": 2.007281541824341, "learning_rate": 6.451573158549071e-07, "loss": 0.008825259283185005, "step": 6846 }, { "epoch": 1.7775181723779854, "grad_norm": 3.958474636077881, "learning_rate": 6.448232155078303e-07, "loss": 0.007810006849467754, "step": 6847 }, { "epoch": 1.7777777777777777, "grad_norm": 3.008479118347168, "learning_rate": 6.44489487307807e-07, "loss": 0.0139866778627038, "step": 6848 }, { "epoch": 1.77803738317757, "grad_norm": 4.498909950256348, "learning_rate": 6.441561313138321e-07, "loss": 0.03459234535694122, "step": 6849 }, { "epoch": 1.7782969885773623, "grad_norm": 0.706750750541687, "learning_rate": 6.438231475848361e-07, "loss": 0.001112789730541408, "step": 6850 }, { "epoch": 1.7785565939771546, "grad_norm": 3.131253480911255, "learning_rate": 6.434905361796825e-07, "loss": 0.038602739572525024, "step": 6851 }, { "epoch": 1.778816199376947, "grad_norm": 3.945708990097046, "learning_rate": 6.431582971571693e-07, "loss": 0.013256638310849667, "step": 6852 }, { "epoch": 1.7790758047767392, "grad_norm": 3.748558759689331, "learning_rate": 6.428264305760288e-07, "loss": 0.023578163236379623, "step": 6853 }, { "epoch": 1.7793354101765315, "grad_norm": 4.118285179138184, "learning_rate": 6.424949364949279e-07, "loss": 0.008751897141337395, "step": 6854 }, { "epoch": 1.7795950155763238, "grad_norm": 0.9990384578704834, "learning_rate": 6.421638149724661e-07, "loss": 0.0018600879702717066, "step": 6855 }, { "epoch": 1.7798546209761164, "grad_norm": 3.020667314529419, "learning_rate": 6.418330660671794e-07, "loss": 0.05932560935616493, "step": 6856 }, { "epoch": 1.7801142263759087, "grad_norm": 3.1512556076049805, "learning_rate": 6.415026898375355e-07, "loss": 0.0069924467243254185, "step": 6857 }, { "epoch": 1.780373831775701, "grad_norm": 3.4183762073516846, "learning_rate": 6.411726863419386e-07, "loss": 0.012803389690816402, "step": 6858 }, { "epoch": 1.7806334371754933, "grad_norm": 3.8125016689300537, "learning_rate": 6.408430556387243e-07, "loss": 0.005525339860469103, "step": 6859 }, { "epoch": 1.7808930425752856, "grad_norm": 3.3099348545074463, "learning_rate": 6.405137977861643e-07, "loss": 0.008214006200432777, "step": 6860 }, { "epoch": 1.7811526479750779, "grad_norm": 4.221322536468506, "learning_rate": 6.401849128424645e-07, "loss": 0.01157351303845644, "step": 6861 }, { "epoch": 1.7814122533748702, "grad_norm": 1.101500153541565, "learning_rate": 6.398564008657638e-07, "loss": 0.004210884217172861, "step": 6862 }, { "epoch": 1.7816718587746625, "grad_norm": 3.2780375480651855, "learning_rate": 6.395282619141354e-07, "loss": 0.04622558504343033, "step": 6863 }, { "epoch": 1.7819314641744548, "grad_norm": 3.75420880317688, "learning_rate": 6.392004960455867e-07, "loss": 0.0494522824883461, "step": 6864 }, { "epoch": 1.782191069574247, "grad_norm": 6.574037075042725, "learning_rate": 6.388731033180594e-07, "loss": 0.006414007395505905, "step": 6865 }, { "epoch": 1.7824506749740394, "grad_norm": 1.5675638914108276, "learning_rate": 6.385460837894296e-07, "loss": 0.0021829816978424788, "step": 6866 }, { "epoch": 1.7827102803738317, "grad_norm": 3.9700026512145996, "learning_rate": 6.382194375175058e-07, "loss": 0.025734320282936096, "step": 6867 }, { "epoch": 1.7829698857736243, "grad_norm": 2.8655190467834473, "learning_rate": 6.378931645600321e-07, "loss": 0.047233324497938156, "step": 6868 }, { "epoch": 1.7832294911734166, "grad_norm": 3.100987434387207, "learning_rate": 6.375672649746861e-07, "loss": 0.01004604808986187, "step": 6869 }, { "epoch": 1.7834890965732089, "grad_norm": 1.6137924194335938, "learning_rate": 6.372417388190796e-07, "loss": 0.008083546534180641, "step": 6870 }, { "epoch": 1.7837487019730012, "grad_norm": 2.5496866703033447, "learning_rate": 6.369165861507577e-07, "loss": 0.019367549568414688, "step": 6871 }, { "epoch": 1.7840083073727935, "grad_norm": 2.509204626083374, "learning_rate": 6.365918070271999e-07, "loss": 0.012840783223509789, "step": 6872 }, { "epoch": 1.7842679127725858, "grad_norm": 0.8700495958328247, "learning_rate": 6.362674015058207e-07, "loss": 0.0018138159066438675, "step": 6873 }, { "epoch": 1.784527518172378, "grad_norm": 1.7464121580123901, "learning_rate": 6.359433696439665e-07, "loss": 0.008065875619649887, "step": 6874 }, { "epoch": 1.7847871235721704, "grad_norm": 3.547811269760132, "learning_rate": 6.356197114989187e-07, "loss": 0.020610129460692406, "step": 6875 }, { "epoch": 1.7850467289719627, "grad_norm": 3.7834863662719727, "learning_rate": 6.352964271278934e-07, "loss": 0.03408326208591461, "step": 6876 }, { "epoch": 1.785306334371755, "grad_norm": 3.271744966506958, "learning_rate": 6.349735165880393e-07, "loss": 0.03118491917848587, "step": 6877 }, { "epoch": 1.7855659397715473, "grad_norm": 3.249974250793457, "learning_rate": 6.346509799364402e-07, "loss": 0.03283322602510452, "step": 6878 }, { "epoch": 1.7858255451713396, "grad_norm": 3.896493434906006, "learning_rate": 6.343288172301125e-07, "loss": 0.023491034284234047, "step": 6879 }, { "epoch": 1.786085150571132, "grad_norm": 2.5620675086975098, "learning_rate": 6.340070285260071e-07, "loss": 0.032032404094934464, "step": 6880 }, { "epoch": 1.7863447559709242, "grad_norm": 3.69401216506958, "learning_rate": 6.3368561388101e-07, "loss": 0.005160910543054342, "step": 6881 }, { "epoch": 1.7866043613707165, "grad_norm": 8.180468559265137, "learning_rate": 6.333645733519387e-07, "loss": 0.015320900827646255, "step": 6882 }, { "epoch": 1.7868639667705088, "grad_norm": 3.3860373497009277, "learning_rate": 6.330439069955461e-07, "loss": 0.008648926392197609, "step": 6883 }, { "epoch": 1.7871235721703012, "grad_norm": 4.590875148773193, "learning_rate": 6.327236148685192e-07, "loss": 0.021419452503323555, "step": 6884 }, { "epoch": 1.7873831775700935, "grad_norm": 3.7607171535491943, "learning_rate": 6.32403697027478e-07, "loss": 0.0071140918880701065, "step": 6885 }, { "epoch": 1.7876427829698858, "grad_norm": 3.8672165870666504, "learning_rate": 6.320841535289765e-07, "loss": 0.013107357546687126, "step": 6886 }, { "epoch": 1.787902388369678, "grad_norm": 1.6009901762008667, "learning_rate": 6.317649844295029e-07, "loss": 0.006392107345163822, "step": 6887 }, { "epoch": 1.7881619937694704, "grad_norm": 1.7527693510055542, "learning_rate": 6.314461897854787e-07, "loss": 0.0037057185545563698, "step": 6888 }, { "epoch": 1.7884215991692627, "grad_norm": 3.1519412994384766, "learning_rate": 6.311277696532603e-07, "loss": 0.015451785176992416, "step": 6889 }, { "epoch": 1.788681204569055, "grad_norm": 1.7790426015853882, "learning_rate": 6.308097240891359e-07, "loss": 0.0039133112877607346, "step": 6890 }, { "epoch": 1.7889408099688473, "grad_norm": 3.0281896591186523, "learning_rate": 6.304920531493292e-07, "loss": 0.014304382726550102, "step": 6891 }, { "epoch": 1.7892004153686396, "grad_norm": 3.178205728530884, "learning_rate": 6.301747568899975e-07, "loss": 0.036962926387786865, "step": 6892 }, { "epoch": 1.789460020768432, "grad_norm": 3.9853949546813965, "learning_rate": 6.298578353672312e-07, "loss": 0.014416884630918503, "step": 6893 }, { "epoch": 1.7897196261682242, "grad_norm": 3.1139018535614014, "learning_rate": 6.29541288637055e-07, "loss": 0.013952091336250305, "step": 6894 }, { "epoch": 1.7899792315680165, "grad_norm": 3.062504529953003, "learning_rate": 6.292251167554264e-07, "loss": 0.01797386258840561, "step": 6895 }, { "epoch": 1.7902388369678088, "grad_norm": 3.1086537837982178, "learning_rate": 6.289093197782378e-07, "loss": 0.04983610659837723, "step": 6896 }, { "epoch": 1.7904984423676011, "grad_norm": 3.8188247680664062, "learning_rate": 6.285938977613156e-07, "loss": 0.025052186101675034, "step": 6897 }, { "epoch": 1.7907580477673934, "grad_norm": 5.2541728019714355, "learning_rate": 6.282788507604179e-07, "loss": 0.02399366907775402, "step": 6898 }, { "epoch": 1.7910176531671858, "grad_norm": 2.9426076412200928, "learning_rate": 6.279641788312388e-07, "loss": 0.005956175271421671, "step": 6899 }, { "epoch": 1.791277258566978, "grad_norm": 2.1177127361297607, "learning_rate": 6.276498820294043e-07, "loss": 0.005410703830420971, "step": 6900 }, { "epoch": 1.7915368639667704, "grad_norm": 2.2998697757720947, "learning_rate": 6.273359604104756e-07, "loss": 0.00993646401911974, "step": 6901 }, { "epoch": 1.7917964693665627, "grad_norm": 3.4822351932525635, "learning_rate": 6.270224140299466e-07, "loss": 0.013145294040441513, "step": 6902 }, { "epoch": 1.792056074766355, "grad_norm": 2.1313235759735107, "learning_rate": 6.267092429432446e-07, "loss": 0.0045027765445411205, "step": 6903 }, { "epoch": 1.7923156801661475, "grad_norm": 3.927111864089966, "learning_rate": 6.26396447205732e-07, "loss": 0.020879313349723816, "step": 6904 }, { "epoch": 1.7925752855659398, "grad_norm": 0.47978928685188293, "learning_rate": 6.260840268727032e-07, "loss": 0.0009525601053610444, "step": 6905 }, { "epoch": 1.7928348909657321, "grad_norm": 2.105480670928955, "learning_rate": 6.257719819993873e-07, "loss": 0.005806602071970701, "step": 6906 }, { "epoch": 1.7930944963655244, "grad_norm": 2.6695892810821533, "learning_rate": 6.254603126409466e-07, "loss": 0.018391752615571022, "step": 6907 }, { "epoch": 1.7933541017653167, "grad_norm": 2.944913625717163, "learning_rate": 6.251490188524768e-07, "loss": 0.061417706310749054, "step": 6908 }, { "epoch": 1.793613707165109, "grad_norm": 0.9573518633842468, "learning_rate": 6.248381006890083e-07, "loss": 0.0015280753141269088, "step": 6909 }, { "epoch": 1.7938733125649013, "grad_norm": 2.637267827987671, "learning_rate": 6.24527558205504e-07, "loss": 0.0097707724198699, "step": 6910 }, { "epoch": 1.7941329179646937, "grad_norm": 4.580219745635986, "learning_rate": 6.2421739145686e-07, "loss": 0.008801469579339027, "step": 6911 }, { "epoch": 1.794392523364486, "grad_norm": 3.806452751159668, "learning_rate": 6.239076004979075e-07, "loss": 0.01393618993461132, "step": 6912 }, { "epoch": 1.7946521287642783, "grad_norm": 2.1876282691955566, "learning_rate": 6.235981853834102e-07, "loss": 0.01596100814640522, "step": 6913 }, { "epoch": 1.7949117341640706, "grad_norm": 2.4419076442718506, "learning_rate": 6.232891461680653e-07, "loss": 0.010612120851874352, "step": 6914 }, { "epoch": 1.7951713395638629, "grad_norm": 3.373019218444824, "learning_rate": 6.229804829065044e-07, "loss": 0.010738594457507133, "step": 6915 }, { "epoch": 1.7954309449636554, "grad_norm": 3.5234079360961914, "learning_rate": 6.226721956532921e-07, "loss": 0.02449640817940235, "step": 6916 }, { "epoch": 1.7956905503634477, "grad_norm": 3.3970425128936768, "learning_rate": 6.223642844629255e-07, "loss": 0.007116829045116901, "step": 6917 }, { "epoch": 1.79595015576324, "grad_norm": 2.044321060180664, "learning_rate": 6.220567493898376e-07, "loss": 0.006006545387208462, "step": 6918 }, { "epoch": 1.7962097611630323, "grad_norm": 3.2409286499023438, "learning_rate": 6.217495904883928e-07, "loss": 0.00727926567196846, "step": 6919 }, { "epoch": 1.7964693665628246, "grad_norm": 3.3120083808898926, "learning_rate": 6.214428078128902e-07, "loss": 0.012691959738731384, "step": 6920 }, { "epoch": 1.796728971962617, "grad_norm": 3.2974061965942383, "learning_rate": 6.211364014175613e-07, "loss": 0.045014649629592896, "step": 6921 }, { "epoch": 1.7969885773624092, "grad_norm": 0.7195594906806946, "learning_rate": 6.208303713565721e-07, "loss": 0.003152871038764715, "step": 6922 }, { "epoch": 1.7972481827622016, "grad_norm": 2.1363589763641357, "learning_rate": 6.205247176840217e-07, "loss": 0.006353180389851332, "step": 6923 }, { "epoch": 1.7975077881619939, "grad_norm": 3.166658639907837, "learning_rate": 6.202194404539431e-07, "loss": 0.004232748877257109, "step": 6924 }, { "epoch": 1.7977673935617862, "grad_norm": 3.8419857025146484, "learning_rate": 6.199145397203017e-07, "loss": 0.0292468573898077, "step": 6925 }, { "epoch": 1.7980269989615785, "grad_norm": 4.169207572937012, "learning_rate": 6.19610015536997e-07, "loss": 0.013457444496452808, "step": 6926 }, { "epoch": 1.7982866043613708, "grad_norm": 2.6917691230773926, "learning_rate": 6.193058679578619e-07, "loss": 0.009495881386101246, "step": 6927 }, { "epoch": 1.798546209761163, "grad_norm": 3.0139100551605225, "learning_rate": 6.190020970366636e-07, "loss": 0.05606050789356232, "step": 6928 }, { "epoch": 1.7988058151609554, "grad_norm": 3.4073381423950195, "learning_rate": 6.186987028271007e-07, "loss": 0.041320160031318665, "step": 6929 }, { "epoch": 1.7990654205607477, "grad_norm": 3.5999927520751953, "learning_rate": 6.183956853828069e-07, "loss": 0.012004128657281399, "step": 6930 }, { "epoch": 1.79932502596054, "grad_norm": 1.6586089134216309, "learning_rate": 6.180930447573486e-07, "loss": 0.004221175331622362, "step": 6931 }, { "epoch": 1.7995846313603323, "grad_norm": 0.30730631947517395, "learning_rate": 6.177907810042259e-07, "loss": 0.0010660351254045963, "step": 6932 }, { "epoch": 1.7998442367601246, "grad_norm": 5.085046768188477, "learning_rate": 6.174888941768721e-07, "loss": 0.018570035696029663, "step": 6933 }, { "epoch": 1.800103842159917, "grad_norm": 3.3659329414367676, "learning_rate": 6.171873843286534e-07, "loss": 0.04357179254293442, "step": 6934 }, { "epoch": 1.8003634475597092, "grad_norm": 2.9674692153930664, "learning_rate": 6.168862515128703e-07, "loss": 0.017133217304944992, "step": 6935 }, { "epoch": 1.8006230529595015, "grad_norm": 1.596024513244629, "learning_rate": 6.165854957827565e-07, "loss": 0.005086222197860479, "step": 6936 }, { "epoch": 1.8008826583592938, "grad_norm": 1.488836646080017, "learning_rate": 6.16285117191478e-07, "loss": 0.0028102900832891464, "step": 6937 }, { "epoch": 1.8011422637590861, "grad_norm": 2.0749998092651367, "learning_rate": 6.159851157921354e-07, "loss": 0.004816302098333836, "step": 6938 }, { "epoch": 1.8014018691588785, "grad_norm": 2.073747158050537, "learning_rate": 6.156854916377618e-07, "loss": 0.004217231646180153, "step": 6939 }, { "epoch": 1.8016614745586708, "grad_norm": 3.0165793895721436, "learning_rate": 6.153862447813236e-07, "loss": 0.06748174130916595, "step": 6940 }, { "epoch": 1.801921079958463, "grad_norm": 3.1854965686798096, "learning_rate": 6.150873752757217e-07, "loss": 0.00777632649987936, "step": 6941 }, { "epoch": 1.8021806853582554, "grad_norm": 3.4247870445251465, "learning_rate": 6.147888831737882e-07, "loss": 0.009393865242600441, "step": 6942 }, { "epoch": 1.8024402907580477, "grad_norm": 2.2669501304626465, "learning_rate": 6.144907685282909e-07, "loss": 0.013041554018855095, "step": 6943 }, { "epoch": 1.80269989615784, "grad_norm": 0.49743303656578064, "learning_rate": 6.14193031391929e-07, "loss": 0.0007488213595934212, "step": 6944 }, { "epoch": 1.8029595015576323, "grad_norm": 4.02049446105957, "learning_rate": 6.13895671817335e-07, "loss": 0.04386834427714348, "step": 6945 }, { "epoch": 1.8032191069574246, "grad_norm": 4.196509838104248, "learning_rate": 6.135986898570765e-07, "loss": 0.02868589758872986, "step": 6946 }, { "epoch": 1.803478712357217, "grad_norm": 4.387327194213867, "learning_rate": 6.133020855636526e-07, "loss": 0.028170477598905563, "step": 6947 }, { "epoch": 1.8037383177570092, "grad_norm": 3.8475823402404785, "learning_rate": 6.130058589894958e-07, "loss": 0.006615836173295975, "step": 6948 }, { "epoch": 1.8039979231568015, "grad_norm": 2.687760829925537, "learning_rate": 6.127100101869725e-07, "loss": 0.009716751053929329, "step": 6949 }, { "epoch": 1.8042575285565938, "grad_norm": 4.228241920471191, "learning_rate": 6.124145392083817e-07, "loss": 0.015160621143877506, "step": 6950 }, { "epoch": 1.8045171339563861, "grad_norm": 2.8232438564300537, "learning_rate": 6.121194461059566e-07, "loss": 0.005782740656286478, "step": 6951 }, { "epoch": 1.8047767393561787, "grad_norm": 3.43489933013916, "learning_rate": 6.118247309318625e-07, "loss": 0.04322514310479164, "step": 6952 }, { "epoch": 1.805036344755971, "grad_norm": 3.743253469467163, "learning_rate": 6.115303937381976e-07, "loss": 0.02473502792418003, "step": 6953 }, { "epoch": 1.8052959501557633, "grad_norm": 2.161764621734619, "learning_rate": 6.112364345769953e-07, "loss": 0.00968287792056799, "step": 6954 }, { "epoch": 1.8055555555555556, "grad_norm": 3.974393606185913, "learning_rate": 6.1094285350022e-07, "loss": 0.028570009395480156, "step": 6955 }, { "epoch": 1.8058151609553479, "grad_norm": 1.422430157661438, "learning_rate": 6.1064965055977e-07, "loss": 0.009809732437133789, "step": 6956 }, { "epoch": 1.8060747663551402, "grad_norm": 3.2897748947143555, "learning_rate": 6.103568258074773e-07, "loss": 0.006641198415309191, "step": 6957 }, { "epoch": 1.8063343717549325, "grad_norm": 2.991149663925171, "learning_rate": 6.100643792951065e-07, "loss": 0.009586689062416553, "step": 6958 }, { "epoch": 1.8065939771547248, "grad_norm": 2.87155818939209, "learning_rate": 6.097723110743556e-07, "loss": 0.009186727926135063, "step": 6959 }, { "epoch": 1.8068535825545171, "grad_norm": 2.314804792404175, "learning_rate": 6.094806211968551e-07, "loss": 0.008500299416482449, "step": 6960 }, { "epoch": 1.8071131879543094, "grad_norm": 2.7162423133850098, "learning_rate": 6.091893097141695e-07, "loss": 0.013450701721012592, "step": 6961 }, { "epoch": 1.8073727933541017, "grad_norm": 0.9540672302246094, "learning_rate": 6.088983766777959e-07, "loss": 0.0015935588162392378, "step": 6962 }, { "epoch": 1.807632398753894, "grad_norm": 4.305320739746094, "learning_rate": 6.086078221391644e-07, "loss": 0.029418228194117546, "step": 6963 }, { "epoch": 1.8078920041536866, "grad_norm": 4.360906600952148, "learning_rate": 6.083176461496389e-07, "loss": 0.0077460752800107, "step": 6964 }, { "epoch": 1.8081516095534789, "grad_norm": 4.095016002655029, "learning_rate": 6.08027848760515e-07, "loss": 0.02239750698208809, "step": 6965 }, { "epoch": 1.8084112149532712, "grad_norm": 3.3765971660614014, "learning_rate": 6.077384300230233e-07, "loss": 0.0680498257279396, "step": 6966 }, { "epoch": 1.8086708203530635, "grad_norm": 1.865102767944336, "learning_rate": 6.074493899883259e-07, "loss": 0.005740173626691103, "step": 6967 }, { "epoch": 1.8089304257528558, "grad_norm": 4.016392707824707, "learning_rate": 6.07160728707518e-07, "loss": 0.04349514842033386, "step": 6968 }, { "epoch": 1.809190031152648, "grad_norm": 3.2694928646087646, "learning_rate": 6.068724462316291e-07, "loss": 0.01040036603808403, "step": 6969 }, { "epoch": 1.8094496365524404, "grad_norm": 1.3483223915100098, "learning_rate": 6.065845426116207e-07, "loss": 0.002973753260448575, "step": 6970 }, { "epoch": 1.8097092419522327, "grad_norm": 1.447385311126709, "learning_rate": 6.062970178983869e-07, "loss": 0.0013460922054946423, "step": 6971 }, { "epoch": 1.809968847352025, "grad_norm": 2.286693811416626, "learning_rate": 6.060098721427566e-07, "loss": 0.0045221466571092606, "step": 6972 }, { "epoch": 1.8102284527518173, "grad_norm": 0.19315192103385925, "learning_rate": 6.057231053954895e-07, "loss": 0.0006330618634819984, "step": 6973 }, { "epoch": 1.8104880581516096, "grad_norm": 3.2845051288604736, "learning_rate": 6.054367177072803e-07, "loss": 0.06208490952849388, "step": 6974 }, { "epoch": 1.810747663551402, "grad_norm": 3.4805994033813477, "learning_rate": 6.051507091287554e-07, "loss": 0.015216906554996967, "step": 6975 }, { "epoch": 1.8110072689511942, "grad_norm": 3.3725178241729736, "learning_rate": 6.048650797104746e-07, "loss": 0.0562296137213707, "step": 6976 }, { "epoch": 1.8112668743509865, "grad_norm": 3.9820966720581055, "learning_rate": 6.045798295029305e-07, "loss": 0.006802306044846773, "step": 6977 }, { "epoch": 1.8115264797507789, "grad_norm": 4.380641937255859, "learning_rate": 6.042949585565491e-07, "loss": 0.0027612708508968353, "step": 6978 }, { "epoch": 1.8117860851505712, "grad_norm": 2.818065643310547, "learning_rate": 6.040104669216889e-07, "loss": 0.005707257892936468, "step": 6979 }, { "epoch": 1.8120456905503635, "grad_norm": 4.779131889343262, "learning_rate": 6.037263546486413e-07, "loss": 0.01688496582210064, "step": 6980 }, { "epoch": 1.8123052959501558, "grad_norm": 3.0015928745269775, "learning_rate": 6.03442621787631e-07, "loss": 0.011390820145606995, "step": 6981 }, { "epoch": 1.812564901349948, "grad_norm": 1.135155439376831, "learning_rate": 6.031592683888158e-07, "loss": 0.0038340394385159016, "step": 6982 }, { "epoch": 1.8128245067497404, "grad_norm": 2.9016735553741455, "learning_rate": 6.028762945022857e-07, "loss": 0.009713146835565567, "step": 6983 }, { "epoch": 1.8130841121495327, "grad_norm": 3.4050652980804443, "learning_rate": 6.02593700178064e-07, "loss": 0.022347813472151756, "step": 6984 }, { "epoch": 1.813343717549325, "grad_norm": 3.0052382946014404, "learning_rate": 6.023114854661073e-07, "loss": 0.06101396307349205, "step": 6985 }, { "epoch": 1.8136033229491173, "grad_norm": 3.14897084236145, "learning_rate": 6.020296504163043e-07, "loss": 0.04540513828396797, "step": 6986 }, { "epoch": 1.8138629283489096, "grad_norm": 3.3766074180603027, "learning_rate": 6.017481950784767e-07, "loss": 0.008367424830794334, "step": 6987 }, { "epoch": 1.814122533748702, "grad_norm": 2.9910855293273926, "learning_rate": 6.014671195023805e-07, "loss": 0.018971599638462067, "step": 6988 }, { "epoch": 1.8143821391484942, "grad_norm": 3.1565065383911133, "learning_rate": 6.01186423737702e-07, "loss": 0.026090960949659348, "step": 6989 }, { "epoch": 1.8146417445482865, "grad_norm": 2.9151906967163086, "learning_rate": 6.009061078340631e-07, "loss": 0.016435425728559494, "step": 6990 }, { "epoch": 1.8149013499480788, "grad_norm": 2.7145724296569824, "learning_rate": 6.00626171841016e-07, "loss": 0.013151478953659534, "step": 6991 }, { "epoch": 1.8151609553478711, "grad_norm": 3.279041290283203, "learning_rate": 6.003466158080478e-07, "loss": 0.016299303621053696, "step": 6992 }, { "epoch": 1.8154205607476634, "grad_norm": 3.004223108291626, "learning_rate": 6.000674397845777e-07, "loss": 0.008741889148950577, "step": 6993 }, { "epoch": 1.8156801661474558, "grad_norm": 1.2048771381378174, "learning_rate": 5.997886438199571e-07, "loss": 0.0017404037062078714, "step": 6994 }, { "epoch": 1.815939771547248, "grad_norm": 3.027965784072876, "learning_rate": 5.995102279634708e-07, "loss": 0.044673483818769455, "step": 6995 }, { "epoch": 1.8161993769470404, "grad_norm": 1.1865172386169434, "learning_rate": 5.992321922643364e-07, "loss": 0.00305704353377223, "step": 6996 }, { "epoch": 1.8164589823468327, "grad_norm": 2.172804117202759, "learning_rate": 5.98954536771705e-07, "loss": 0.002955416450276971, "step": 6997 }, { "epoch": 1.816718587746625, "grad_norm": 3.4171528816223145, "learning_rate": 5.986772615346586e-07, "loss": 0.002613621763885021, "step": 6998 }, { "epoch": 1.8169781931464173, "grad_norm": 2.931663990020752, "learning_rate": 5.984003666022135e-07, "loss": 0.009670664556324482, "step": 6999 }, { "epoch": 1.8172377985462098, "grad_norm": 4.512659072875977, "learning_rate": 5.981238520233188e-07, "loss": 0.021192722022533417, "step": 7000 }, { "epoch": 1.8174974039460021, "grad_norm": 1.0759708881378174, "learning_rate": 5.978477178468556e-07, "loss": 0.0014512055786326528, "step": 7001 }, { "epoch": 1.8177570093457944, "grad_norm": 1.092248558998108, "learning_rate": 5.975719641216379e-07, "loss": 0.0024137147702276707, "step": 7002 }, { "epoch": 1.8180166147455867, "grad_norm": 2.940964698791504, "learning_rate": 5.972965908964129e-07, "loss": 0.030882954597473145, "step": 7003 }, { "epoch": 1.818276220145379, "grad_norm": 3.916633367538452, "learning_rate": 5.970215982198602e-07, "loss": 0.029748674482107162, "step": 7004 }, { "epoch": 1.8185358255451713, "grad_norm": 3.3833401203155518, "learning_rate": 5.967469861405927e-07, "loss": 0.05832609906792641, "step": 7005 }, { "epoch": 1.8187954309449637, "grad_norm": 2.9385132789611816, "learning_rate": 5.964727547071548e-07, "loss": 0.041485756635665894, "step": 7006 }, { "epoch": 1.819055036344756, "grad_norm": 2.541195869445801, "learning_rate": 5.961989039680243e-07, "loss": 0.0018615820445120335, "step": 7007 }, { "epoch": 1.8193146417445483, "grad_norm": 3.5727481842041016, "learning_rate": 5.959254339716127e-07, "loss": 0.008519445545971394, "step": 7008 }, { "epoch": 1.8195742471443406, "grad_norm": 1.9660803079605103, "learning_rate": 5.956523447662624e-07, "loss": 0.0023153547663241625, "step": 7009 }, { "epoch": 1.8198338525441329, "grad_norm": 1.811285138130188, "learning_rate": 5.953796364002491e-07, "loss": 0.0023013504687696695, "step": 7010 }, { "epoch": 1.8200934579439252, "grad_norm": 2.3768863677978516, "learning_rate": 5.951073089217822e-07, "loss": 0.007364982273429632, "step": 7011 }, { "epoch": 1.8203530633437177, "grad_norm": 4.103195667266846, "learning_rate": 5.948353623790024e-07, "loss": 0.027206487953662872, "step": 7012 }, { "epoch": 1.82061266874351, "grad_norm": 3.2316431999206543, "learning_rate": 5.94563796819984e-07, "loss": 0.013311381451785564, "step": 7013 }, { "epoch": 1.8208722741433023, "grad_norm": 0.40437597036361694, "learning_rate": 5.942926122927334e-07, "loss": 0.0008651587413623929, "step": 7014 }, { "epoch": 1.8211318795430946, "grad_norm": 1.9365547895431519, "learning_rate": 5.940218088451894e-07, "loss": 0.006206659600138664, "step": 7015 }, { "epoch": 1.821391484942887, "grad_norm": 3.27624773979187, "learning_rate": 5.937513865252247e-07, "loss": 0.014456378296017647, "step": 7016 }, { "epoch": 1.8216510903426792, "grad_norm": 4.669627666473389, "learning_rate": 5.934813453806434e-07, "loss": 0.008296409621834755, "step": 7017 }, { "epoch": 1.8219106957424716, "grad_norm": 3.0535428524017334, "learning_rate": 5.932116854591819e-07, "loss": 0.04534134268760681, "step": 7018 }, { "epoch": 1.8221703011422639, "grad_norm": 3.478127956390381, "learning_rate": 5.929424068085109e-07, "loss": 0.020715635269880295, "step": 7019 }, { "epoch": 1.8224299065420562, "grad_norm": 1.765798807144165, "learning_rate": 5.926735094762325e-07, "loss": 0.003978113178163767, "step": 7020 }, { "epoch": 1.8226895119418485, "grad_norm": 3.3386805057525635, "learning_rate": 5.924049935098812e-07, "loss": 0.03287418186664581, "step": 7021 }, { "epoch": 1.8229491173416408, "grad_norm": 3.8893508911132812, "learning_rate": 5.921368589569246e-07, "loss": 0.019073758274316788, "step": 7022 }, { "epoch": 1.823208722741433, "grad_norm": 1.51974356174469, "learning_rate": 5.918691058647626e-07, "loss": 0.00970356073230505, "step": 7023 }, { "epoch": 1.8234683281412254, "grad_norm": 3.7386703491210938, "learning_rate": 5.916017342807287e-07, "loss": 0.013131300918757915, "step": 7024 }, { "epoch": 1.8237279335410177, "grad_norm": 1.5754534006118774, "learning_rate": 5.913347442520865e-07, "loss": 0.003068051300942898, "step": 7025 }, { "epoch": 1.82398753894081, "grad_norm": 3.668041229248047, "learning_rate": 5.910681358260355e-07, "loss": 0.0036926455795764923, "step": 7026 }, { "epoch": 1.8242471443406023, "grad_norm": 4.006649017333984, "learning_rate": 5.908019090497045e-07, "loss": 0.024879589676856995, "step": 7027 }, { "epoch": 1.8245067497403946, "grad_norm": 3.6216254234313965, "learning_rate": 5.905360639701571e-07, "loss": 0.015382388606667519, "step": 7028 }, { "epoch": 1.824766355140187, "grad_norm": 3.6061997413635254, "learning_rate": 5.902706006343887e-07, "loss": 0.03758465126156807, "step": 7029 }, { "epoch": 1.8250259605399792, "grad_norm": 2.9569802284240723, "learning_rate": 5.900055190893262e-07, "loss": 0.024231478571891785, "step": 7030 }, { "epoch": 1.8252855659397715, "grad_norm": 0.2792518734931946, "learning_rate": 5.897408193818311e-07, "loss": 0.0005584136815741658, "step": 7031 }, { "epoch": 1.8255451713395638, "grad_norm": 4.120367050170898, "learning_rate": 5.894765015586959e-07, "loss": 0.0250489953905344, "step": 7032 }, { "epoch": 1.8258047767393562, "grad_norm": 2.5774903297424316, "learning_rate": 5.892125656666454e-07, "loss": 0.001920229522511363, "step": 7033 }, { "epoch": 1.8260643821391485, "grad_norm": 3.126530647277832, "learning_rate": 5.889490117523381e-07, "loss": 0.04384436458349228, "step": 7034 }, { "epoch": 1.8263239875389408, "grad_norm": 3.511042594909668, "learning_rate": 5.886858398623635e-07, "loss": 0.03965388983488083, "step": 7035 }, { "epoch": 1.826583592938733, "grad_norm": 3.467695951461792, "learning_rate": 5.884230500432455e-07, "loss": 0.01975908689200878, "step": 7036 }, { "epoch": 1.8268431983385254, "grad_norm": 3.5129270553588867, "learning_rate": 5.881606423414383e-07, "loss": 0.022930683568120003, "step": 7037 }, { "epoch": 1.8271028037383177, "grad_norm": 1.1024032831192017, "learning_rate": 5.878986168033301e-07, "loss": 0.005938286893069744, "step": 7038 }, { "epoch": 1.82736240913811, "grad_norm": 5.4558587074279785, "learning_rate": 5.876369734752411e-07, "loss": 0.022125594317913055, "step": 7039 }, { "epoch": 1.8276220145379023, "grad_norm": 3.1584270000457764, "learning_rate": 5.873757124034234e-07, "loss": 0.03012877143919468, "step": 7040 }, { "epoch": 1.8278816199376946, "grad_norm": 3.707174777984619, "learning_rate": 5.871148336340623e-07, "loss": 0.014858338981866837, "step": 7041 }, { "epoch": 1.828141225337487, "grad_norm": 3.99273681640625, "learning_rate": 5.868543372132751e-07, "loss": 0.029210854321718216, "step": 7042 }, { "epoch": 1.8284008307372792, "grad_norm": 3.5532126426696777, "learning_rate": 5.865942231871113e-07, "loss": 0.05125981569290161, "step": 7043 }, { "epoch": 1.8286604361370715, "grad_norm": 4.100953102111816, "learning_rate": 5.863344916015538e-07, "loss": 0.007395410444587469, "step": 7044 }, { "epoch": 1.8289200415368638, "grad_norm": 2.6553173065185547, "learning_rate": 5.86075142502517e-07, "loss": 0.005917428992688656, "step": 7045 }, { "epoch": 1.8291796469366561, "grad_norm": 3.2634854316711426, "learning_rate": 5.858161759358472e-07, "loss": 0.04141733795404434, "step": 7046 }, { "epoch": 1.8294392523364484, "grad_norm": 1.8511072397232056, "learning_rate": 5.855575919473248e-07, "loss": 0.008262400515377522, "step": 7047 }, { "epoch": 1.829698857736241, "grad_norm": 3.3841652870178223, "learning_rate": 5.852993905826606e-07, "loss": 0.03803122043609619, "step": 7048 }, { "epoch": 1.8299584631360333, "grad_norm": 0.7139279842376709, "learning_rate": 5.850415718874988e-07, "loss": 0.001121588866226375, "step": 7049 }, { "epoch": 1.8302180685358256, "grad_norm": 3.2755556106567383, "learning_rate": 5.847841359074165e-07, "loss": 0.038230471312999725, "step": 7050 }, { "epoch": 1.8304776739356179, "grad_norm": 3.4116125106811523, "learning_rate": 5.845270826879221e-07, "loss": 0.039587777107954025, "step": 7051 }, { "epoch": 1.8307372793354102, "grad_norm": 4.864570140838623, "learning_rate": 5.842704122744569e-07, "loss": 0.01364913396537304, "step": 7052 }, { "epoch": 1.8309968847352025, "grad_norm": 3.6385414600372314, "learning_rate": 5.840141247123938e-07, "loss": 0.04944679141044617, "step": 7053 }, { "epoch": 1.8312564901349948, "grad_norm": 2.750469446182251, "learning_rate": 5.837582200470388e-07, "loss": 0.03444409742951393, "step": 7054 }, { "epoch": 1.8315160955347871, "grad_norm": 2.834906816482544, "learning_rate": 5.835026983236308e-07, "loss": 0.013838493265211582, "step": 7055 }, { "epoch": 1.8317757009345794, "grad_norm": 3.613842725753784, "learning_rate": 5.832475595873389e-07, "loss": 0.0555809810757637, "step": 7056 }, { "epoch": 1.8320353063343717, "grad_norm": 3.039910078048706, "learning_rate": 5.829928038832668e-07, "loss": 0.027058228850364685, "step": 7057 }, { "epoch": 1.832294911734164, "grad_norm": 1.691103458404541, "learning_rate": 5.827384312564488e-07, "loss": 0.006405962165445089, "step": 7058 }, { "epoch": 1.8325545171339563, "grad_norm": 2.684671640396118, "learning_rate": 5.824844417518527e-07, "loss": 0.01307629980146885, "step": 7059 }, { "epoch": 1.8328141225337489, "grad_norm": 0.189860999584198, "learning_rate": 5.822308354143776e-07, "loss": 0.0004936504410579801, "step": 7060 }, { "epoch": 1.8330737279335412, "grad_norm": 3.4216315746307373, "learning_rate": 5.819776122888553e-07, "loss": 0.016037151217460632, "step": 7061 }, { "epoch": 1.8333333333333335, "grad_norm": 3.4762444496154785, "learning_rate": 5.817247724200503e-07, "loss": 0.029881423339247704, "step": 7062 }, { "epoch": 1.8335929387331258, "grad_norm": 4.297699451446533, "learning_rate": 5.814723158526586e-07, "loss": 0.018072698265314102, "step": 7063 }, { "epoch": 1.833852544132918, "grad_norm": 3.6172127723693848, "learning_rate": 5.812202426313087e-07, "loss": 0.03532277047634125, "step": 7064 }, { "epoch": 1.8341121495327104, "grad_norm": 1.0233930349349976, "learning_rate": 5.809685528005614e-07, "loss": 0.0006762405391782522, "step": 7065 }, { "epoch": 1.8343717549325027, "grad_norm": 6.709595203399658, "learning_rate": 5.807172464049096e-07, "loss": 0.011640362441539764, "step": 7066 }, { "epoch": 1.834631360332295, "grad_norm": 3.4657368659973145, "learning_rate": 5.80466323488779e-07, "loss": 0.04795607179403305, "step": 7067 }, { "epoch": 1.8348909657320873, "grad_norm": 1.063690423965454, "learning_rate": 5.802157840965264e-07, "loss": 0.007218698970973492, "step": 7068 }, { "epoch": 1.8351505711318796, "grad_norm": 3.335324764251709, "learning_rate": 5.799656282724417e-07, "loss": 0.01165764033794403, "step": 7069 }, { "epoch": 1.835410176531672, "grad_norm": 0.7480274438858032, "learning_rate": 5.797158560607467e-07, "loss": 0.0017080811085179448, "step": 7070 }, { "epoch": 1.8356697819314642, "grad_norm": 3.573347806930542, "learning_rate": 5.794664675055956e-07, "loss": 0.005789581686258316, "step": 7071 }, { "epoch": 1.8359293873312565, "grad_norm": 0.16045081615447998, "learning_rate": 5.792174626510742e-07, "loss": 0.00046755996299907565, "step": 7072 }, { "epoch": 1.8361889927310489, "grad_norm": 1.006052017211914, "learning_rate": 5.789688415412015e-07, "loss": 0.0009958737064152956, "step": 7073 }, { "epoch": 1.8364485981308412, "grad_norm": 3.432695150375366, "learning_rate": 5.787206042199273e-07, "loss": 0.023150786757469177, "step": 7074 }, { "epoch": 1.8367082035306335, "grad_norm": 2.0429534912109375, "learning_rate": 5.784727507311344e-07, "loss": 0.0077864257618784904, "step": 7075 }, { "epoch": 1.8369678089304258, "grad_norm": 0.4520798623561859, "learning_rate": 5.78225281118638e-07, "loss": 0.0009367038728669286, "step": 7076 }, { "epoch": 1.837227414330218, "grad_norm": 4.184211254119873, "learning_rate": 5.779781954261845e-07, "loss": 0.013239031657576561, "step": 7077 }, { "epoch": 1.8374870197300104, "grad_norm": 4.0661234855651855, "learning_rate": 5.777314936974541e-07, "loss": 0.029186196625232697, "step": 7078 }, { "epoch": 1.8377466251298027, "grad_norm": 4.711292266845703, "learning_rate": 5.774851759760566e-07, "loss": 0.011131851002573967, "step": 7079 }, { "epoch": 1.838006230529595, "grad_norm": 3.6832830905914307, "learning_rate": 5.77239242305536e-07, "loss": 0.00776339927688241, "step": 7080 }, { "epoch": 1.8382658359293873, "grad_norm": 1.061249017715454, "learning_rate": 5.769936927293679e-07, "loss": 0.0037574907764792442, "step": 7081 }, { "epoch": 1.8385254413291796, "grad_norm": 3.148695945739746, "learning_rate": 5.767485272909598e-07, "loss": 0.006490395404398441, "step": 7082 }, { "epoch": 1.838785046728972, "grad_norm": 2.663745641708374, "learning_rate": 5.765037460336512e-07, "loss": 0.014809470623731613, "step": 7083 }, { "epoch": 1.8390446521287642, "grad_norm": 3.4006471633911133, "learning_rate": 5.762593490007135e-07, "loss": 0.00902776513248682, "step": 7084 }, { "epoch": 1.8393042575285565, "grad_norm": 3.489861249923706, "learning_rate": 5.76015336235351e-07, "loss": 0.05889303237199783, "step": 7085 }, { "epoch": 1.8395638629283488, "grad_norm": 3.7354135513305664, "learning_rate": 5.757717077807e-07, "loss": 0.04708360135555267, "step": 7086 }, { "epoch": 1.8398234683281411, "grad_norm": 3.641718864440918, "learning_rate": 5.75528463679827e-07, "loss": 0.02226543053984642, "step": 7087 }, { "epoch": 1.8400830737279334, "grad_norm": 3.596632242202759, "learning_rate": 5.752856039757334e-07, "loss": 0.024880442768335342, "step": 7088 }, { "epoch": 1.8403426791277258, "grad_norm": 1.2802940607070923, "learning_rate": 5.750431287113501e-07, "loss": 0.005687442142516375, "step": 7089 }, { "epoch": 1.840602284527518, "grad_norm": 3.124325752258301, "learning_rate": 5.74801037929542e-07, "loss": 0.00648588826879859, "step": 7090 }, { "epoch": 1.8408618899273104, "grad_norm": 1.5512921810150146, "learning_rate": 5.745593316731051e-07, "loss": 0.002536448882892728, "step": 7091 }, { "epoch": 1.8411214953271027, "grad_norm": 3.346349000930786, "learning_rate": 5.74318009984767e-07, "loss": 0.01970016397535801, "step": 7092 }, { "epoch": 1.841381100726895, "grad_norm": 5.533229827880859, "learning_rate": 5.740770729071887e-07, "loss": 0.017858559265732765, "step": 7093 }, { "epoch": 1.8416407061266873, "grad_norm": 3.6820521354675293, "learning_rate": 5.738365204829617e-07, "loss": 0.032873738557100296, "step": 7094 }, { "epoch": 1.8419003115264796, "grad_norm": 0.8197044730186462, "learning_rate": 5.735963527546102e-07, "loss": 0.008434636518359184, "step": 7095 }, { "epoch": 1.842159916926272, "grad_norm": 1.965018630027771, "learning_rate": 5.733565697645906e-07, "loss": 0.009002077393233776, "step": 7096 }, { "epoch": 1.8424195223260644, "grad_norm": 3.5178394317626953, "learning_rate": 5.731171715552908e-07, "loss": 0.011603687889873981, "step": 7097 }, { "epoch": 1.8426791277258567, "grad_norm": 1.6715565919876099, "learning_rate": 5.728781581690312e-07, "loss": 0.0016961288638412952, "step": 7098 }, { "epoch": 1.842938733125649, "grad_norm": 3.386895179748535, "learning_rate": 5.726395296480637e-07, "loss": 0.0096864253282547, "step": 7099 }, { "epoch": 1.8431983385254413, "grad_norm": 4.064446926116943, "learning_rate": 5.724012860345723e-07, "loss": 0.014622877351939678, "step": 7100 }, { "epoch": 1.8434579439252337, "grad_norm": 3.6441633701324463, "learning_rate": 5.721634273706735e-07, "loss": 0.004634403623640537, "step": 7101 }, { "epoch": 1.843717549325026, "grad_norm": 1.2781504392623901, "learning_rate": 5.719259536984147e-07, "loss": 0.002729322761297226, "step": 7102 }, { "epoch": 1.8439771547248183, "grad_norm": 3.113659620285034, "learning_rate": 5.716888650597758e-07, "loss": 0.03240089863538742, "step": 7103 }, { "epoch": 1.8442367601246106, "grad_norm": 2.755530834197998, "learning_rate": 5.714521614966689e-07, "loss": 0.006023142486810684, "step": 7104 }, { "epoch": 1.8444963655244029, "grad_norm": 0.9652439951896667, "learning_rate": 5.71215843050938e-07, "loss": 0.005033926106989384, "step": 7105 }, { "epoch": 1.8447559709241952, "grad_norm": 1.383959412574768, "learning_rate": 5.709799097643583e-07, "loss": 0.008798161521553993, "step": 7106 }, { "epoch": 1.8450155763239875, "grad_norm": 3.3037779331207275, "learning_rate": 5.707443616786376e-07, "loss": 0.04370098561048508, "step": 7107 }, { "epoch": 1.8452751817237798, "grad_norm": 3.6492671966552734, "learning_rate": 5.705091988354152e-07, "loss": 0.028737712651491165, "step": 7108 }, { "epoch": 1.8455347871235723, "grad_norm": 3.0923798084259033, "learning_rate": 5.702744212762634e-07, "loss": 0.009643714874982834, "step": 7109 }, { "epoch": 1.8457943925233646, "grad_norm": 2.11399245262146, "learning_rate": 5.700400290426841e-07, "loss": 0.00400822889059782, "step": 7110 }, { "epoch": 1.846053997923157, "grad_norm": 2.928126573562622, "learning_rate": 5.698060221761133e-07, "loss": 0.008346466347575188, "step": 7111 }, { "epoch": 1.8463136033229492, "grad_norm": 4.799402713775635, "learning_rate": 5.695724007179182e-07, "loss": 0.009552659466862679, "step": 7112 }, { "epoch": 1.8465732087227416, "grad_norm": 1.7665077447891235, "learning_rate": 5.693391647093975e-07, "loss": 0.0013998261420056224, "step": 7113 }, { "epoch": 1.8468328141225339, "grad_norm": 1.0418787002563477, "learning_rate": 5.69106314191782e-07, "loss": 0.0014450059970840812, "step": 7114 }, { "epoch": 1.8470924195223262, "grad_norm": 2.312854051589966, "learning_rate": 5.688738492062342e-07, "loss": 0.0035096928477287292, "step": 7115 }, { "epoch": 1.8473520249221185, "grad_norm": 2.9173476696014404, "learning_rate": 5.686417697938483e-07, "loss": 0.010896686464548111, "step": 7116 }, { "epoch": 1.8476116303219108, "grad_norm": 1.8191825151443481, "learning_rate": 5.684100759956516e-07, "loss": 0.006831010803580284, "step": 7117 }, { "epoch": 1.847871235721703, "grad_norm": 3.463315725326538, "learning_rate": 5.681787678526015e-07, "loss": 0.008428117260336876, "step": 7118 }, { "epoch": 1.8481308411214954, "grad_norm": 3.4678637981414795, "learning_rate": 5.679478454055882e-07, "loss": 0.030332889407873154, "step": 7119 }, { "epoch": 1.8483904465212877, "grad_norm": 2.7882330417633057, "learning_rate": 5.677173086954331e-07, "loss": 0.016116581857204437, "step": 7120 }, { "epoch": 1.84865005192108, "grad_norm": 1.8205763101577759, "learning_rate": 5.674871577628907e-07, "loss": 0.00464217271655798, "step": 7121 }, { "epoch": 1.8489096573208723, "grad_norm": 3.180244207382202, "learning_rate": 5.672573926486457e-07, "loss": 0.05642075091600418, "step": 7122 }, { "epoch": 1.8491692627206646, "grad_norm": 3.185000419616699, "learning_rate": 5.670280133933152e-07, "loss": 0.009721651673316956, "step": 7123 }, { "epoch": 1.849428868120457, "grad_norm": 4.647751808166504, "learning_rate": 5.66799020037449e-07, "loss": 0.010541869327425957, "step": 7124 }, { "epoch": 1.8496884735202492, "grad_norm": 4.044257164001465, "learning_rate": 5.66570412621527e-07, "loss": 0.015132857486605644, "step": 7125 }, { "epoch": 1.8499480789200415, "grad_norm": 4.21527624130249, "learning_rate": 5.663421911859622e-07, "loss": 0.010661320760846138, "step": 7126 }, { "epoch": 1.8502076843198338, "grad_norm": 2.743701696395874, "learning_rate": 5.661143557710988e-07, "loss": 0.018814902752637863, "step": 7127 }, { "epoch": 1.8504672897196262, "grad_norm": 3.353222370147705, "learning_rate": 5.658869064172132e-07, "loss": 0.053767502307891846, "step": 7128 }, { "epoch": 1.8507268951194185, "grad_norm": 1.6704447269439697, "learning_rate": 5.656598431645127e-07, "loss": 0.0037753696087747812, "step": 7129 }, { "epoch": 1.8509865005192108, "grad_norm": 4.580047607421875, "learning_rate": 5.654331660531374e-07, "loss": 0.010751841589808464, "step": 7130 }, { "epoch": 1.851246105919003, "grad_norm": 3.6169750690460205, "learning_rate": 5.652068751231583e-07, "loss": 0.007412794977426529, "step": 7131 }, { "epoch": 1.8515057113187954, "grad_norm": 1.3062689304351807, "learning_rate": 5.649809704145786e-07, "loss": 0.003429129021242261, "step": 7132 }, { "epoch": 1.8517653167185877, "grad_norm": 3.168778657913208, "learning_rate": 5.64755451967333e-07, "loss": 0.0148939099162817, "step": 7133 }, { "epoch": 1.85202492211838, "grad_norm": 3.4378480911254883, "learning_rate": 5.645303198212878e-07, "loss": 0.006948903203010559, "step": 7134 }, { "epoch": 1.8522845275181723, "grad_norm": 1.6736140251159668, "learning_rate": 5.643055740162418e-07, "loss": 0.0032363145146518946, "step": 7135 }, { "epoch": 1.8525441329179646, "grad_norm": 3.5367705821990967, "learning_rate": 5.640812145919245e-07, "loss": 0.02695627138018608, "step": 7136 }, { "epoch": 1.852803738317757, "grad_norm": 3.419281482696533, "learning_rate": 5.638572415879972e-07, "loss": 0.013012580573558807, "step": 7137 }, { "epoch": 1.8530633437175492, "grad_norm": 4.267374515533447, "learning_rate": 5.63633655044054e-07, "loss": 0.0024125240743160248, "step": 7138 }, { "epoch": 1.8533229491173415, "grad_norm": 2.5169196128845215, "learning_rate": 5.63410454999619e-07, "loss": 0.016320763155817986, "step": 7139 }, { "epoch": 1.8535825545171338, "grad_norm": 2.9500489234924316, "learning_rate": 5.6318764149415e-07, "loss": 0.05011526495218277, "step": 7140 }, { "epoch": 1.8538421599169261, "grad_norm": 2.428922176361084, "learning_rate": 5.629652145670338e-07, "loss": 0.0035834601148962975, "step": 7141 }, { "epoch": 1.8541017653167184, "grad_norm": 2.6159720420837402, "learning_rate": 5.627431742575914e-07, "loss": 0.008820343762636185, "step": 7142 }, { "epoch": 1.8543613707165107, "grad_norm": 3.2777366638183594, "learning_rate": 5.625215206050745e-07, "loss": 0.009764288552105427, "step": 7143 }, { "epoch": 1.854620976116303, "grad_norm": 4.426848888397217, "learning_rate": 5.623002536486661e-07, "loss": 0.022735562175512314, "step": 7144 }, { "epoch": 1.8548805815160956, "grad_norm": 3.182705879211426, "learning_rate": 5.620793734274812e-07, "loss": 0.02688264474272728, "step": 7145 }, { "epoch": 1.855140186915888, "grad_norm": 3.1705918312072754, "learning_rate": 5.618588799805662e-07, "loss": 0.046316586434841156, "step": 7146 }, { "epoch": 1.8553997923156802, "grad_norm": 3.054004430770874, "learning_rate": 5.616387733468992e-07, "loss": 0.05171571671962738, "step": 7147 }, { "epoch": 1.8556593977154725, "grad_norm": 3.3313586711883545, "learning_rate": 5.614190535653909e-07, "loss": 0.01099427416920662, "step": 7148 }, { "epoch": 1.8559190031152648, "grad_norm": 1.2790848016738892, "learning_rate": 5.611997206748815e-07, "loss": 0.001329638296738267, "step": 7149 }, { "epoch": 1.8561786085150571, "grad_norm": 1.135335922241211, "learning_rate": 5.60980774714145e-07, "loss": 0.00694786012172699, "step": 7150 }, { "epoch": 1.8564382139148494, "grad_norm": 3.0596773624420166, "learning_rate": 5.607622157218851e-07, "loss": 0.006511428393423557, "step": 7151 }, { "epoch": 1.8566978193146417, "grad_norm": 3.2761971950531006, "learning_rate": 5.60544043736739e-07, "loss": 0.030582217499613762, "step": 7152 }, { "epoch": 1.856957424714434, "grad_norm": 2.1002421379089355, "learning_rate": 5.60326258797274e-07, "loss": 0.0036506487522274256, "step": 7153 }, { "epoch": 1.8572170301142263, "grad_norm": 1.804988145828247, "learning_rate": 5.601088609419893e-07, "loss": 0.0013566729612648487, "step": 7154 }, { "epoch": 1.8574766355140186, "grad_norm": 0.5929874181747437, "learning_rate": 5.598918502093165e-07, "loss": 0.0016412114491686225, "step": 7155 }, { "epoch": 1.857736240913811, "grad_norm": 0.332093209028244, "learning_rate": 5.596752266376175e-07, "loss": 0.0007698191329836845, "step": 7156 }, { "epoch": 1.8579958463136035, "grad_norm": 3.389343023300171, "learning_rate": 5.594589902651866e-07, "loss": 0.0586797371506691, "step": 7157 }, { "epoch": 1.8582554517133958, "grad_norm": 3.068917989730835, "learning_rate": 5.592431411302498e-07, "loss": 0.010214882902801037, "step": 7158 }, { "epoch": 1.858515057113188, "grad_norm": 0.8744127154350281, "learning_rate": 5.590276792709639e-07, "loss": 0.0010048249969258904, "step": 7159 }, { "epoch": 1.8587746625129804, "grad_norm": 3.697754144668579, "learning_rate": 5.588126047254173e-07, "loss": 0.017432715743780136, "step": 7160 }, { "epoch": 1.8590342679127727, "grad_norm": 2.818070411682129, "learning_rate": 5.58597917531631e-07, "loss": 0.007921523414552212, "step": 7161 }, { "epoch": 1.859293873312565, "grad_norm": 2.1089565753936768, "learning_rate": 5.583836177275561e-07, "loss": 0.007645646575838327, "step": 7162 }, { "epoch": 1.8595534787123573, "grad_norm": 1.6613259315490723, "learning_rate": 5.581697053510768e-07, "loss": 0.0030720431823283434, "step": 7163 }, { "epoch": 1.8598130841121496, "grad_norm": 3.686690330505371, "learning_rate": 5.579561804400068e-07, "loss": 0.004527957644313574, "step": 7164 }, { "epoch": 1.860072689511942, "grad_norm": 1.8902649879455566, "learning_rate": 5.577430430320932e-07, "loss": 0.006130700930953026, "step": 7165 }, { "epoch": 1.8603322949117342, "grad_norm": 2.9164438247680664, "learning_rate": 5.575302931650135e-07, "loss": 0.014139466919004917, "step": 7166 }, { "epoch": 1.8605919003115265, "grad_norm": 3.5741209983825684, "learning_rate": 5.573179308763772e-07, "loss": 0.04637329652905464, "step": 7167 }, { "epoch": 1.8608515057113189, "grad_norm": 0.7614745497703552, "learning_rate": 5.571059562037244e-07, "loss": 0.002247815951704979, "step": 7168 }, { "epoch": 1.8611111111111112, "grad_norm": 2.2284655570983887, "learning_rate": 5.568943691845285e-07, "loss": 0.004865704104304314, "step": 7169 }, { "epoch": 1.8613707165109035, "grad_norm": 0.23668786883354187, "learning_rate": 5.566831698561925e-07, "loss": 0.0005068673053756356, "step": 7170 }, { "epoch": 1.8616303219106958, "grad_norm": 3.33732008934021, "learning_rate": 5.56472358256052e-07, "loss": 0.05199936777353287, "step": 7171 }, { "epoch": 1.861889927310488, "grad_norm": 2.8600778579711914, "learning_rate": 5.56261934421373e-07, "loss": 0.015387466177344322, "step": 7172 }, { "epoch": 1.8621495327102804, "grad_norm": 3.369030714035034, "learning_rate": 5.560518983893542e-07, "loss": 0.028508726507425308, "step": 7173 }, { "epoch": 1.8624091381100727, "grad_norm": 3.577089309692383, "learning_rate": 5.558422501971251e-07, "loss": 0.03729378432035446, "step": 7174 }, { "epoch": 1.862668743509865, "grad_norm": 1.1888010501861572, "learning_rate": 5.556329898817467e-07, "loss": 0.004399549216032028, "step": 7175 }, { "epoch": 1.8629283489096573, "grad_norm": 1.2537271976470947, "learning_rate": 5.554241174802112e-07, "loss": 0.0018267864361405373, "step": 7176 }, { "epoch": 1.8631879543094496, "grad_norm": 1.7284417152404785, "learning_rate": 5.552156330294427e-07, "loss": 0.002960815094411373, "step": 7177 }, { "epoch": 1.863447559709242, "grad_norm": 2.8542988300323486, "learning_rate": 5.55007536566296e-07, "loss": 0.008320380002260208, "step": 7178 }, { "epoch": 1.8637071651090342, "grad_norm": 2.4446630477905273, "learning_rate": 5.547998281275589e-07, "loss": 0.005740633234381676, "step": 7179 }, { "epoch": 1.8639667705088265, "grad_norm": 3.1427805423736572, "learning_rate": 5.545925077499481e-07, "loss": 0.006156979128718376, "step": 7180 }, { "epoch": 1.8642263759086188, "grad_norm": 5.20102596282959, "learning_rate": 5.543855754701142e-07, "loss": 0.005438312888145447, "step": 7181 }, { "epoch": 1.8644859813084111, "grad_norm": 3.0123918056488037, "learning_rate": 5.541790313246371e-07, "loss": 0.012124699540436268, "step": 7182 }, { "epoch": 1.8647455867082035, "grad_norm": 3.551903247833252, "learning_rate": 5.539728753500302e-07, "loss": 0.01054548192769289, "step": 7183 }, { "epoch": 1.8650051921079958, "grad_norm": 2.740570068359375, "learning_rate": 5.537671075827364e-07, "loss": 0.003923687152564526, "step": 7184 }, { "epoch": 1.865264797507788, "grad_norm": 3.1323344707489014, "learning_rate": 5.535617280591308e-07, "loss": 0.005734456703066826, "step": 7185 }, { "epoch": 1.8655244029075804, "grad_norm": 2.7674126625061035, "learning_rate": 5.5335673681552e-07, "loss": 0.010483269579708576, "step": 7186 }, { "epoch": 1.8657840083073727, "grad_norm": 4.241761207580566, "learning_rate": 5.531521338881417e-07, "loss": 0.028344623744487762, "step": 7187 }, { "epoch": 1.866043613707165, "grad_norm": 3.3853793144226074, "learning_rate": 5.529479193131649e-07, "loss": 0.010326148942112923, "step": 7188 }, { "epoch": 1.8663032191069573, "grad_norm": 2.430368661880493, "learning_rate": 5.527440931266902e-07, "loss": 0.007408294826745987, "step": 7189 }, { "epoch": 1.8665628245067496, "grad_norm": 2.726691722869873, "learning_rate": 5.525406553647495e-07, "loss": 0.00403217226266861, "step": 7190 }, { "epoch": 1.866822429906542, "grad_norm": 2.8402740955352783, "learning_rate": 5.523376060633054e-07, "loss": 0.001946997712366283, "step": 7191 }, { "epoch": 1.8670820353063342, "grad_norm": 2.0517690181732178, "learning_rate": 5.52134945258253e-07, "loss": 0.0024018636904656887, "step": 7192 }, { "epoch": 1.8673416407061267, "grad_norm": 4.063671112060547, "learning_rate": 5.519326729854176e-07, "loss": 0.009747032076120377, "step": 7193 }, { "epoch": 1.867601246105919, "grad_norm": 2.2388319969177246, "learning_rate": 5.517307892805569e-07, "loss": 0.007565214764326811, "step": 7194 }, { "epoch": 1.8678608515057114, "grad_norm": 3.8377561569213867, "learning_rate": 5.515292941793587e-07, "loss": 0.0049830107018351555, "step": 7195 }, { "epoch": 1.8681204569055037, "grad_norm": 0.8282508850097656, "learning_rate": 5.513281877174431e-07, "loss": 0.0009924406185746193, "step": 7196 }, { "epoch": 1.868380062305296, "grad_norm": 2.2598581314086914, "learning_rate": 5.51127469930361e-07, "loss": 0.0042435768991708755, "step": 7197 }, { "epoch": 1.8686396677050883, "grad_norm": 2.0791263580322266, "learning_rate": 5.509271408535949e-07, "loss": 0.004959519952535629, "step": 7198 }, { "epoch": 1.8688992731048806, "grad_norm": 3.4029293060302734, "learning_rate": 5.507272005225578e-07, "loss": 0.0050319042056798935, "step": 7199 }, { "epoch": 1.8691588785046729, "grad_norm": 2.7865939140319824, "learning_rate": 5.505276489725951e-07, "loss": 0.0027828537859022617, "step": 7200 }, { "epoch": 1.8694184839044652, "grad_norm": 2.6338837146759033, "learning_rate": 5.503284862389828e-07, "loss": 0.011571137234568596, "step": 7201 }, { "epoch": 1.8696780893042575, "grad_norm": 2.0863232612609863, "learning_rate": 5.501297123569287e-07, "loss": 0.005394405219703913, "step": 7202 }, { "epoch": 1.8699376947040498, "grad_norm": 3.762585401535034, "learning_rate": 5.499313273615706e-07, "loss": 0.015352385118603706, "step": 7203 }, { "epoch": 1.870197300103842, "grad_norm": 1.2618948221206665, "learning_rate": 5.49733331287979e-07, "loss": 0.002243360038846731, "step": 7204 }, { "epoch": 1.8704569055036346, "grad_norm": 1.4234015941619873, "learning_rate": 5.495357241711554e-07, "loss": 0.002857644110918045, "step": 7205 }, { "epoch": 1.870716510903427, "grad_norm": 5.15160608291626, "learning_rate": 5.493385060460316e-07, "loss": 0.03719927370548248, "step": 7206 }, { "epoch": 1.8709761163032192, "grad_norm": 5.505312442779541, "learning_rate": 5.491416769474714e-07, "loss": 0.01064928900450468, "step": 7207 }, { "epoch": 1.8712357217030116, "grad_norm": 4.743838787078857, "learning_rate": 5.489452369102696e-07, "loss": 0.034599483013153076, "step": 7208 }, { "epoch": 1.8714953271028039, "grad_norm": 4.116814613342285, "learning_rate": 5.487491859691525e-07, "loss": 0.0518580824136734, "step": 7209 }, { "epoch": 1.8717549325025962, "grad_norm": 3.236018657684326, "learning_rate": 5.485535241587777e-07, "loss": 0.01630646362900734, "step": 7210 }, { "epoch": 1.8720145379023885, "grad_norm": 3.2602450847625732, "learning_rate": 5.483582515137331e-07, "loss": 0.01590406522154808, "step": 7211 }, { "epoch": 1.8722741433021808, "grad_norm": 5.187288284301758, "learning_rate": 5.481633680685388e-07, "loss": 0.014265317469835281, "step": 7212 }, { "epoch": 1.872533748701973, "grad_norm": 3.406961679458618, "learning_rate": 5.479688738576457e-07, "loss": 0.024636073037981987, "step": 7213 }, { "epoch": 1.8727933541017654, "grad_norm": 4.208930492401123, "learning_rate": 5.477747689154355e-07, "loss": 0.02095184102654457, "step": 7214 }, { "epoch": 1.8730529595015577, "grad_norm": 0.33554625511169434, "learning_rate": 5.475810532762226e-07, "loss": 0.0004532186721917242, "step": 7215 }, { "epoch": 1.87331256490135, "grad_norm": 3.9780921936035156, "learning_rate": 5.473877269742502e-07, "loss": 0.00929274968802929, "step": 7216 }, { "epoch": 1.8735721703011423, "grad_norm": 0.8608137369155884, "learning_rate": 5.471947900436947e-07, "loss": 0.0013339020079001784, "step": 7217 }, { "epoch": 1.8738317757009346, "grad_norm": 3.3715641498565674, "learning_rate": 5.470022425186627e-07, "loss": 0.008207854814827442, "step": 7218 }, { "epoch": 1.874091381100727, "grad_norm": 3.816756248474121, "learning_rate": 5.468100844331922e-07, "loss": 0.010723773390054703, "step": 7219 }, { "epoch": 1.8743509865005192, "grad_norm": 4.248715877532959, "learning_rate": 5.466183158212525e-07, "loss": 0.015237737447023392, "step": 7220 }, { "epoch": 1.8746105919003115, "grad_norm": 3.657332181930542, "learning_rate": 5.464269367167439e-07, "loss": 0.016989368945360184, "step": 7221 }, { "epoch": 1.8748701973001038, "grad_norm": 3.5508956909179688, "learning_rate": 5.462359471534976e-07, "loss": 0.027230393141508102, "step": 7222 }, { "epoch": 1.8751298026998962, "grad_norm": 3.2702815532684326, "learning_rate": 5.460453471652765e-07, "loss": 0.010301136411726475, "step": 7223 }, { "epoch": 1.8753894080996885, "grad_norm": 3.563422918319702, "learning_rate": 5.45855136785774e-07, "loss": 0.026559986174106598, "step": 7224 }, { "epoch": 1.8756490134994808, "grad_norm": 3.892979860305786, "learning_rate": 5.45665316048615e-07, "loss": 0.02352192997932434, "step": 7225 }, { "epoch": 1.875908618899273, "grad_norm": 2.0905418395996094, "learning_rate": 5.454758849873559e-07, "loss": 0.0027509727515280247, "step": 7226 }, { "epoch": 1.8761682242990654, "grad_norm": 3.4242236614227295, "learning_rate": 5.45286843635483e-07, "loss": 0.039410680532455444, "step": 7227 }, { "epoch": 1.8764278296988577, "grad_norm": 3.1316323280334473, "learning_rate": 5.450981920264152e-07, "loss": 0.05227996036410332, "step": 7228 }, { "epoch": 1.87668743509865, "grad_norm": 0.8102247714996338, "learning_rate": 5.449099301935015e-07, "loss": 0.0009552034898661077, "step": 7229 }, { "epoch": 1.8769470404984423, "grad_norm": 0.7344695329666138, "learning_rate": 5.447220581700222e-07, "loss": 0.0008963648579083383, "step": 7230 }, { "epoch": 1.8772066458982346, "grad_norm": 3.401514768600464, "learning_rate": 5.445345759891889e-07, "loss": 0.005455402657389641, "step": 7231 }, { "epoch": 1.877466251298027, "grad_norm": 4.3959641456604, "learning_rate": 5.44347483684144e-07, "loss": 0.011496535502374172, "step": 7232 }, { "epoch": 1.8777258566978192, "grad_norm": 4.866628170013428, "learning_rate": 5.441607812879614e-07, "loss": 0.004726549610495567, "step": 7233 }, { "epoch": 1.8779854620976115, "grad_norm": 1.5350854396820068, "learning_rate": 5.439744688336453e-07, "loss": 0.003259366611018777, "step": 7234 }, { "epoch": 1.8782450674974038, "grad_norm": 3.604301929473877, "learning_rate": 5.43788546354132e-07, "loss": 0.014456155709922314, "step": 7235 }, { "epoch": 1.8785046728971961, "grad_norm": 3.4382221698760986, "learning_rate": 5.436030138822881e-07, "loss": 0.006947622634470463, "step": 7236 }, { "epoch": 1.8787642782969884, "grad_norm": 1.4954643249511719, "learning_rate": 5.434178714509117e-07, "loss": 0.0019444039789959788, "step": 7237 }, { "epoch": 1.8790238836967808, "grad_norm": 1.1901782751083374, "learning_rate": 5.432331190927316e-07, "loss": 0.005235753022134304, "step": 7238 }, { "epoch": 1.879283489096573, "grad_norm": 3.96826171875, "learning_rate": 5.430487568404078e-07, "loss": 0.016704218462109566, "step": 7239 }, { "epoch": 1.8795430944963654, "grad_norm": 4.567534923553467, "learning_rate": 5.428647847265313e-07, "loss": 0.010883531533181667, "step": 7240 }, { "epoch": 1.879802699896158, "grad_norm": 3.0764214992523193, "learning_rate": 5.426812027836241e-07, "loss": 0.007669569458812475, "step": 7241 }, { "epoch": 1.8800623052959502, "grad_norm": 3.503098726272583, "learning_rate": 5.424980110441392e-07, "loss": 0.005124077666550875, "step": 7242 }, { "epoch": 1.8803219106957425, "grad_norm": 3.903238534927368, "learning_rate": 5.423152095404608e-07, "loss": 0.06315024942159653, "step": 7243 }, { "epoch": 1.8805815160955348, "grad_norm": 4.590921401977539, "learning_rate": 5.421327983049046e-07, "loss": 0.01982627436518669, "step": 7244 }, { "epoch": 1.8808411214953271, "grad_norm": 3.614849090576172, "learning_rate": 5.419507773697157e-07, "loss": 0.009202586486935616, "step": 7245 }, { "epoch": 1.8811007268951194, "grad_norm": 3.4332828521728516, "learning_rate": 5.417691467670719e-07, "loss": 0.013846688903868198, "step": 7246 }, { "epoch": 1.8813603322949117, "grad_norm": 3.8009531497955322, "learning_rate": 5.415879065290811e-07, "loss": 0.02417711727321148, "step": 7247 }, { "epoch": 1.881619937694704, "grad_norm": 3.373260974884033, "learning_rate": 5.414070566877826e-07, "loss": 0.05127542465925217, "step": 7248 }, { "epoch": 1.8818795430944963, "grad_norm": 4.415375232696533, "learning_rate": 5.412265972751465e-07, "loss": 0.036739133298397064, "step": 7249 }, { "epoch": 1.8821391484942886, "grad_norm": 1.5293775796890259, "learning_rate": 5.410465283230735e-07, "loss": 0.0012416953686624765, "step": 7250 }, { "epoch": 1.882398753894081, "grad_norm": 5.034975528717041, "learning_rate": 5.408668498633959e-07, "loss": 0.027757739648222923, "step": 7251 }, { "epoch": 1.8826583592938733, "grad_norm": 2.7355141639709473, "learning_rate": 5.406875619278772e-07, "loss": 0.016620509326457977, "step": 7252 }, { "epoch": 1.8829179646936658, "grad_norm": 4.965260982513428, "learning_rate": 5.405086645482108e-07, "loss": 0.009251350536942482, "step": 7253 }, { "epoch": 1.883177570093458, "grad_norm": 3.565704584121704, "learning_rate": 5.40330157756022e-07, "loss": 0.02084929123520851, "step": 7254 }, { "epoch": 1.8834371754932504, "grad_norm": 5.98294734954834, "learning_rate": 5.401520415828664e-07, "loss": 0.011777237057685852, "step": 7255 }, { "epoch": 1.8836967808930427, "grad_norm": 4.225892066955566, "learning_rate": 5.399743160602311e-07, "loss": 0.01907079480588436, "step": 7256 }, { "epoch": 1.883956386292835, "grad_norm": 2.3408238887786865, "learning_rate": 5.397969812195336e-07, "loss": 0.01151167880743742, "step": 7257 }, { "epoch": 1.8842159916926273, "grad_norm": 3.2871172428131104, "learning_rate": 5.396200370921232e-07, "loss": 0.06408160924911499, "step": 7258 }, { "epoch": 1.8844755970924196, "grad_norm": 4.023031711578369, "learning_rate": 5.394434837092792e-07, "loss": 0.006055739242583513, "step": 7259 }, { "epoch": 1.884735202492212, "grad_norm": 2.772050142288208, "learning_rate": 5.392673211022122e-07, "loss": 0.007118405308574438, "step": 7260 }, { "epoch": 1.8849948078920042, "grad_norm": 1.1042734384536743, "learning_rate": 5.390915493020638e-07, "loss": 0.004739250987768173, "step": 7261 }, { "epoch": 1.8852544132917965, "grad_norm": 5.282007694244385, "learning_rate": 5.389161683399063e-07, "loss": 0.002952813170850277, "step": 7262 }, { "epoch": 1.8855140186915889, "grad_norm": 3.216993808746338, "learning_rate": 5.387411782467429e-07, "loss": 0.007737265899777412, "step": 7263 }, { "epoch": 1.8857736240913812, "grad_norm": 2.613496780395508, "learning_rate": 5.385665790535087e-07, "loss": 0.013674566522240639, "step": 7264 }, { "epoch": 1.8860332294911735, "grad_norm": 3.962094783782959, "learning_rate": 5.383923707910677e-07, "loss": 0.035780660808086395, "step": 7265 }, { "epoch": 1.8862928348909658, "grad_norm": 4.195896148681641, "learning_rate": 5.382185534902165e-07, "loss": 0.00999370776116848, "step": 7266 }, { "epoch": 1.886552440290758, "grad_norm": 3.5218422412872314, "learning_rate": 5.380451271816824e-07, "loss": 0.046357981860637665, "step": 7267 }, { "epoch": 1.8868120456905504, "grad_norm": 2.7192695140838623, "learning_rate": 5.378720918961223e-07, "loss": 0.05301528051495552, "step": 7268 }, { "epoch": 1.8870716510903427, "grad_norm": 3.457899332046509, "learning_rate": 5.376994476641252e-07, "loss": 0.00851336121559143, "step": 7269 }, { "epoch": 1.887331256490135, "grad_norm": 2.3607676029205322, "learning_rate": 5.375271945162114e-07, "loss": 0.004983728285878897, "step": 7270 }, { "epoch": 1.8875908618899273, "grad_norm": 5.748841285705566, "learning_rate": 5.373553324828303e-07, "loss": 0.016796180978417397, "step": 7271 }, { "epoch": 1.8878504672897196, "grad_norm": 0.6426729559898376, "learning_rate": 5.371838615943637e-07, "loss": 0.001826480613090098, "step": 7272 }, { "epoch": 1.888110072689512, "grad_norm": 2.202732563018799, "learning_rate": 5.370127818811234e-07, "loss": 0.003556877840310335, "step": 7273 }, { "epoch": 1.8883696780893042, "grad_norm": 0.28005945682525635, "learning_rate": 5.368420933733525e-07, "loss": 0.0007069173152558506, "step": 7274 }, { "epoch": 1.8886292834890965, "grad_norm": 1.1646060943603516, "learning_rate": 5.366717961012253e-07, "loss": 0.00206271605566144, "step": 7275 }, { "epoch": 1.8888888888888888, "grad_norm": 1.7867320775985718, "learning_rate": 5.365018900948458e-07, "loss": 0.007704711519181728, "step": 7276 }, { "epoch": 1.8891484942886811, "grad_norm": 3.647876024246216, "learning_rate": 5.363323753842498e-07, "loss": 0.03651953116059303, "step": 7277 }, { "epoch": 1.8894080996884735, "grad_norm": 4.194567680358887, "learning_rate": 5.361632519994034e-07, "loss": 0.004274024628102779, "step": 7278 }, { "epoch": 1.8896677050882658, "grad_norm": 0.6671724319458008, "learning_rate": 5.359945199702042e-07, "loss": 0.005262645427137613, "step": 7279 }, { "epoch": 1.889927310488058, "grad_norm": 3.396265983581543, "learning_rate": 5.358261793264796e-07, "loss": 0.05592833086848259, "step": 7280 }, { "epoch": 1.8901869158878504, "grad_norm": 3.596566915512085, "learning_rate": 5.356582300979887e-07, "loss": 0.007165045477449894, "step": 7281 }, { "epoch": 1.8904465212876427, "grad_norm": 3.2963225841522217, "learning_rate": 5.354906723144211e-07, "loss": 0.037583403289318085, "step": 7282 }, { "epoch": 1.890706126687435, "grad_norm": 0.8235634565353394, "learning_rate": 5.353235060053971e-07, "loss": 0.0014504549326375127, "step": 7283 }, { "epoch": 1.8909657320872273, "grad_norm": 3.9416277408599854, "learning_rate": 5.351567312004675e-07, "loss": 0.0155373839661479, "step": 7284 }, { "epoch": 1.8912253374870196, "grad_norm": 1.685674786567688, "learning_rate": 5.349903479291153e-07, "loss": 0.0017687587533146143, "step": 7285 }, { "epoch": 1.891484942886812, "grad_norm": 3.3965375423431396, "learning_rate": 5.348243562207519e-07, "loss": 0.008781928569078445, "step": 7286 }, { "epoch": 1.8917445482866042, "grad_norm": 4.351215362548828, "learning_rate": 5.34658756104722e-07, "loss": 0.020458150655031204, "step": 7287 }, { "epoch": 1.8920041536863965, "grad_norm": 3.919637680053711, "learning_rate": 5.344935476102992e-07, "loss": 0.027189107611775398, "step": 7288 }, { "epoch": 1.892263759086189, "grad_norm": 4.263038158416748, "learning_rate": 5.34328730766689e-07, "loss": 0.015373742207884789, "step": 7289 }, { "epoch": 1.8925233644859814, "grad_norm": 0.3996925950050354, "learning_rate": 5.341643056030269e-07, "loss": 0.001280707074329257, "step": 7290 }, { "epoch": 1.8927829698857737, "grad_norm": 0.6620967984199524, "learning_rate": 5.340002721483798e-07, "loss": 0.0010656446684151888, "step": 7291 }, { "epoch": 1.893042575285566, "grad_norm": 1.6675306558609009, "learning_rate": 5.338366304317445e-07, "loss": 0.004965006373822689, "step": 7292 }, { "epoch": 1.8933021806853583, "grad_norm": 3.8704562187194824, "learning_rate": 5.336733804820501e-07, "loss": 0.014905963093042374, "step": 7293 }, { "epoch": 1.8935617860851506, "grad_norm": 3.771791934967041, "learning_rate": 5.335105223281544e-07, "loss": 0.012255201116204262, "step": 7294 }, { "epoch": 1.8938213914849429, "grad_norm": 3.661487579345703, "learning_rate": 5.333480559988477e-07, "loss": 0.035550639033317566, "step": 7295 }, { "epoch": 1.8940809968847352, "grad_norm": 0.37157633900642395, "learning_rate": 5.331859815228501e-07, "loss": 0.0008983806474134326, "step": 7296 }, { "epoch": 1.8943406022845275, "grad_norm": 3.3810174465179443, "learning_rate": 5.330242989288127e-07, "loss": 0.0224231518805027, "step": 7297 }, { "epoch": 1.8946002076843198, "grad_norm": 2.8371894359588623, "learning_rate": 5.328630082453173e-07, "loss": 0.01134933065623045, "step": 7298 }, { "epoch": 1.894859813084112, "grad_norm": 2.0708727836608887, "learning_rate": 5.327021095008762e-07, "loss": 0.008093970827758312, "step": 7299 }, { "epoch": 1.8951194184839044, "grad_norm": 2.6017870903015137, "learning_rate": 5.325416027239327e-07, "loss": 0.0033627236261963844, "step": 7300 }, { "epoch": 1.895379023883697, "grad_norm": 3.3077187538146973, "learning_rate": 5.323814879428608e-07, "loss": 0.005075275432318449, "step": 7301 }, { "epoch": 1.8956386292834893, "grad_norm": 3.547349214553833, "learning_rate": 5.322217651859652e-07, "loss": 0.011944230645895004, "step": 7302 }, { "epoch": 1.8958982346832816, "grad_norm": 3.0603857040405273, "learning_rate": 5.320624344814812e-07, "loss": 0.00893373042345047, "step": 7303 }, { "epoch": 1.8961578400830739, "grad_norm": 2.998150110244751, "learning_rate": 5.319034958575744e-07, "loss": 0.024934053421020508, "step": 7304 }, { "epoch": 1.8964174454828662, "grad_norm": 1.6358262300491333, "learning_rate": 5.317449493423419e-07, "loss": 0.006830524653196335, "step": 7305 }, { "epoch": 1.8966770508826585, "grad_norm": 2.599655866622925, "learning_rate": 5.315867949638114e-07, "loss": 0.007113161496818066, "step": 7306 }, { "epoch": 1.8969366562824508, "grad_norm": 2.6768031120300293, "learning_rate": 5.314290327499404e-07, "loss": 0.012122849002480507, "step": 7307 }, { "epoch": 1.897196261682243, "grad_norm": 3.3730947971343994, "learning_rate": 5.312716627286179e-07, "loss": 0.04862159490585327, "step": 7308 }, { "epoch": 1.8974558670820354, "grad_norm": 3.7619946002960205, "learning_rate": 5.311146849276632e-07, "loss": 0.02854781597852707, "step": 7309 }, { "epoch": 1.8977154724818277, "grad_norm": 5.3588457107543945, "learning_rate": 5.309580993748266e-07, "loss": 0.006535135675221682, "step": 7310 }, { "epoch": 1.89797507788162, "grad_norm": 4.015567779541016, "learning_rate": 5.308019060977887e-07, "loss": 0.034720852971076965, "step": 7311 }, { "epoch": 1.8982346832814123, "grad_norm": 2.126422643661499, "learning_rate": 5.306461051241607e-07, "loss": 0.01383209228515625, "step": 7312 }, { "epoch": 1.8984942886812046, "grad_norm": 2.940131664276123, "learning_rate": 5.304906964814849e-07, "loss": 0.06385219842195511, "step": 7313 }, { "epoch": 1.898753894080997, "grad_norm": 3.4686801433563232, "learning_rate": 5.303356801972342e-07, "loss": 0.05083303898572922, "step": 7314 }, { "epoch": 1.8990134994807892, "grad_norm": 1.2934755086898804, "learning_rate": 5.301810562988113e-07, "loss": 0.009885527193546295, "step": 7315 }, { "epoch": 1.8992731048805815, "grad_norm": 3.265181064605713, "learning_rate": 5.300268248135509e-07, "loss": 0.044082529842853546, "step": 7316 }, { "epoch": 1.8995327102803738, "grad_norm": 2.445885419845581, "learning_rate": 5.29872985768717e-07, "loss": 0.004708084277808666, "step": 7317 }, { "epoch": 1.8997923156801662, "grad_norm": 0.5616148710250854, "learning_rate": 5.297195391915051e-07, "loss": 0.0009042596211656928, "step": 7318 }, { "epoch": 1.9000519210799585, "grad_norm": 2.5711944103240967, "learning_rate": 5.295664851090409e-07, "loss": 0.0028767837211489677, "step": 7319 }, { "epoch": 1.9003115264797508, "grad_norm": 3.9505455493927, "learning_rate": 5.29413823548381e-07, "loss": 0.016484634950757027, "step": 7320 }, { "epoch": 1.900571131879543, "grad_norm": 3.203468084335327, "learning_rate": 5.292615545365124e-07, "loss": 0.015005475841462612, "step": 7321 }, { "epoch": 1.9008307372793354, "grad_norm": 1.9154140949249268, "learning_rate": 5.29109678100353e-07, "loss": 0.006114573683589697, "step": 7322 }, { "epoch": 1.9010903426791277, "grad_norm": 3.230736017227173, "learning_rate": 5.289581942667505e-07, "loss": 0.026531070470809937, "step": 7323 }, { "epoch": 1.90134994807892, "grad_norm": 3.725271463394165, "learning_rate": 5.288071030624846e-07, "loss": 0.010862372815608978, "step": 7324 }, { "epoch": 1.9016095534787123, "grad_norm": 3.099076509475708, "learning_rate": 5.286564045142638e-07, "loss": 0.057212598621845245, "step": 7325 }, { "epoch": 1.9018691588785046, "grad_norm": 4.626403331756592, "learning_rate": 5.28506098648729e-07, "loss": 0.015801869332790375, "step": 7326 }, { "epoch": 1.902128764278297, "grad_norm": 4.723788738250732, "learning_rate": 5.283561854924504e-07, "loss": 0.04521571099758148, "step": 7327 }, { "epoch": 1.9023883696780892, "grad_norm": 0.83754563331604, "learning_rate": 5.282066650719291e-07, "loss": 0.001832381240092218, "step": 7328 }, { "epoch": 1.9026479750778815, "grad_norm": 2.479773759841919, "learning_rate": 5.280575374135975e-07, "loss": 0.005530287511646748, "step": 7329 }, { "epoch": 1.9029075804776738, "grad_norm": 3.6503489017486572, "learning_rate": 5.279088025438173e-07, "loss": 0.01958397775888443, "step": 7330 }, { "epoch": 1.9031671858774661, "grad_norm": 1.6931699514389038, "learning_rate": 5.277604604888816e-07, "loss": 0.0014858600916340947, "step": 7331 }, { "epoch": 1.9034267912772584, "grad_norm": 3.2970833778381348, "learning_rate": 5.27612511275014e-07, "loss": 0.0071150814183056355, "step": 7332 }, { "epoch": 1.9036863966770508, "grad_norm": 3.604883909225464, "learning_rate": 5.274649549283685e-07, "loss": 0.014169922098517418, "step": 7333 }, { "epoch": 1.903946002076843, "grad_norm": 1.9839894771575928, "learning_rate": 5.273177914750298e-07, "loss": 0.006038830615580082, "step": 7334 }, { "epoch": 1.9042056074766354, "grad_norm": 3.4350249767303467, "learning_rate": 5.271710209410127e-07, "loss": 0.0029337371233850718, "step": 7335 }, { "epoch": 1.9044652128764277, "grad_norm": 3.7255659103393555, "learning_rate": 5.270246433522629e-07, "loss": 0.04348261281847954, "step": 7336 }, { "epoch": 1.9047248182762202, "grad_norm": 2.5492608547210693, "learning_rate": 5.268786587346573e-07, "loss": 0.011341666802763939, "step": 7337 }, { "epoch": 1.9049844236760125, "grad_norm": 2.0188193321228027, "learning_rate": 5.267330671140016e-07, "loss": 0.004260779824107885, "step": 7338 }, { "epoch": 1.9052440290758048, "grad_norm": 2.0649852752685547, "learning_rate": 5.265878685160339e-07, "loss": 0.005818515550345182, "step": 7339 }, { "epoch": 1.9055036344755971, "grad_norm": 2.1068637371063232, "learning_rate": 5.264430629664214e-07, "loss": 0.006110815797001123, "step": 7340 }, { "epoch": 1.9057632398753894, "grad_norm": 2.4533910751342773, "learning_rate": 5.262986504907629e-07, "loss": 0.006498333066701889, "step": 7341 }, { "epoch": 1.9060228452751817, "grad_norm": 3.775411605834961, "learning_rate": 5.261546311145868e-07, "loss": 0.04178459197282791, "step": 7342 }, { "epoch": 1.906282450674974, "grad_norm": 3.2785980701446533, "learning_rate": 5.260110048633527e-07, "loss": 0.004815869033336639, "step": 7343 }, { "epoch": 1.9065420560747663, "grad_norm": 0.3220641613006592, "learning_rate": 5.258677717624502e-07, "loss": 0.000725969672203064, "step": 7344 }, { "epoch": 1.9068016614745587, "grad_norm": 2.956380844116211, "learning_rate": 5.257249318372001e-07, "loss": 0.010019868612289429, "step": 7345 }, { "epoch": 1.907061266874351, "grad_norm": 0.9587599039077759, "learning_rate": 5.255824851128525e-07, "loss": 0.004926767200231552, "step": 7346 }, { "epoch": 1.9073208722741433, "grad_norm": 3.3018641471862793, "learning_rate": 5.254404316145894e-07, "loss": 0.015121713280677795, "step": 7347 }, { "epoch": 1.9075804776739356, "grad_norm": 4.210992813110352, "learning_rate": 5.252987713675222e-07, "loss": 0.0351373627781868, "step": 7348 }, { "epoch": 1.907840083073728, "grad_norm": 3.3469202518463135, "learning_rate": 5.251575043966932e-07, "loss": 0.040311142802238464, "step": 7349 }, { "epoch": 1.9080996884735204, "grad_norm": 4.3810505867004395, "learning_rate": 5.250166307270754e-07, "loss": 0.025353116914629936, "step": 7350 }, { "epoch": 1.9083592938733127, "grad_norm": 3.818844795227051, "learning_rate": 5.24876150383572e-07, "loss": 0.04590035229921341, "step": 7351 }, { "epoch": 1.908618899273105, "grad_norm": 3.422590970993042, "learning_rate": 5.247360633910168e-07, "loss": 0.04668577015399933, "step": 7352 }, { "epoch": 1.9088785046728973, "grad_norm": 3.4895293712615967, "learning_rate": 5.245963697741734e-07, "loss": 0.015951432287693024, "step": 7353 }, { "epoch": 1.9091381100726896, "grad_norm": 1.2808440923690796, "learning_rate": 5.244570695577371e-07, "loss": 0.005524486768990755, "step": 7354 }, { "epoch": 1.909397715472482, "grad_norm": 2.3871185779571533, "learning_rate": 5.243181627663325e-07, "loss": 0.006104829255491495, "step": 7355 }, { "epoch": 1.9096573208722742, "grad_norm": 3.714897871017456, "learning_rate": 5.241796494245157e-07, "loss": 0.006446209270507097, "step": 7356 }, { "epoch": 1.9099169262720666, "grad_norm": 0.6087303161621094, "learning_rate": 5.240415295567719e-07, "loss": 0.0009732777252793312, "step": 7357 }, { "epoch": 1.9101765316718589, "grad_norm": 3.61208176612854, "learning_rate": 5.239038031875184e-07, "loss": 0.049889713525772095, "step": 7358 }, { "epoch": 1.9104361370716512, "grad_norm": 0.9707542061805725, "learning_rate": 5.237664703411014e-07, "loss": 0.004920959938317537, "step": 7359 }, { "epoch": 1.9106957424714435, "grad_norm": 0.4957505166530609, "learning_rate": 5.236295310417988e-07, "loss": 0.0010198852978646755, "step": 7360 }, { "epoch": 1.9109553478712358, "grad_norm": 3.295208692550659, "learning_rate": 5.234929853138177e-07, "loss": 0.003958222921937704, "step": 7361 }, { "epoch": 1.911214953271028, "grad_norm": 3.9541213512420654, "learning_rate": 5.233568331812964e-07, "loss": 0.006789405830204487, "step": 7362 }, { "epoch": 1.9114745586708204, "grad_norm": 4.438393592834473, "learning_rate": 5.232210746683039e-07, "loss": 0.021040940657258034, "step": 7363 }, { "epoch": 1.9117341640706127, "grad_norm": 0.14898468554019928, "learning_rate": 5.230857097988388e-07, "loss": 0.0003817762481048703, "step": 7364 }, { "epoch": 1.911993769470405, "grad_norm": 1.602406620979309, "learning_rate": 5.229507385968308e-07, "loss": 0.002584646688774228, "step": 7365 }, { "epoch": 1.9122533748701973, "grad_norm": 2.964902400970459, "learning_rate": 5.228161610861392e-07, "loss": 0.0031783687882125378, "step": 7366 }, { "epoch": 1.9125129802699896, "grad_norm": 3.311774492263794, "learning_rate": 5.226819772905546e-07, "loss": 0.04207735136151314, "step": 7367 }, { "epoch": 1.912772585669782, "grad_norm": 7.059079647064209, "learning_rate": 5.22548187233798e-07, "loss": 0.021877991035580635, "step": 7368 }, { "epoch": 1.9130321910695742, "grad_norm": 2.8070106506347656, "learning_rate": 5.224147909395197e-07, "loss": 0.005849101580679417, "step": 7369 }, { "epoch": 1.9132917964693665, "grad_norm": 2.639572858810425, "learning_rate": 5.222817884313015e-07, "loss": 0.0023643458262085915, "step": 7370 }, { "epoch": 1.9135514018691588, "grad_norm": 3.0686798095703125, "learning_rate": 5.22149179732655e-07, "loss": 0.006326314527541399, "step": 7371 }, { "epoch": 1.9138110072689511, "grad_norm": 4.193237781524658, "learning_rate": 5.220169648670227e-07, "loss": 0.051908574998378754, "step": 7372 }, { "epoch": 1.9140706126687435, "grad_norm": 0.662465512752533, "learning_rate": 5.21885143857777e-07, "loss": 0.0006088418886065483, "step": 7373 }, { "epoch": 1.9143302180685358, "grad_norm": 4.596851348876953, "learning_rate": 5.217537167282206e-07, "loss": 0.03164263069629669, "step": 7374 }, { "epoch": 1.914589823468328, "grad_norm": 0.9720121622085571, "learning_rate": 5.21622683501587e-07, "loss": 0.005001503974199295, "step": 7375 }, { "epoch": 1.9148494288681204, "grad_norm": 2.793678045272827, "learning_rate": 5.214920442010399e-07, "loss": 0.006431759800761938, "step": 7376 }, { "epoch": 1.9151090342679127, "grad_norm": 3.2892911434173584, "learning_rate": 5.213617988496733e-07, "loss": 0.03660944849252701, "step": 7377 }, { "epoch": 1.915368639667705, "grad_norm": 4.248115539550781, "learning_rate": 5.212319474705119e-07, "loss": 0.01921408623456955, "step": 7378 }, { "epoch": 1.9156282450674973, "grad_norm": 4.0808939933776855, "learning_rate": 5.211024900865099e-07, "loss": 0.011941619217395782, "step": 7379 }, { "epoch": 1.9158878504672896, "grad_norm": 0.8923670053482056, "learning_rate": 5.209734267205526e-07, "loss": 0.003274268703535199, "step": 7380 }, { "epoch": 1.916147455867082, "grad_norm": 0.9343645572662354, "learning_rate": 5.208447573954557e-07, "loss": 0.000996266957372427, "step": 7381 }, { "epoch": 1.9164070612668742, "grad_norm": 1.1037530899047852, "learning_rate": 5.207164821339647e-07, "loss": 0.00143895810469985, "step": 7382 }, { "epoch": 1.9166666666666665, "grad_norm": 3.5924577713012695, "learning_rate": 5.205886009587559e-07, "loss": 0.051975347101688385, "step": 7383 }, { "epoch": 1.9169262720664588, "grad_norm": 0.771475076675415, "learning_rate": 5.204611138924358e-07, "loss": 0.005632089450955391, "step": 7384 }, { "epoch": 1.9171858774662511, "grad_norm": 4.763246536254883, "learning_rate": 5.203340209575409e-07, "loss": 0.01910097897052765, "step": 7385 }, { "epoch": 1.9174454828660437, "grad_norm": 3.6220269203186035, "learning_rate": 5.202073221765384e-07, "loss": 0.048830170184373856, "step": 7386 }, { "epoch": 1.917705088265836, "grad_norm": 5.188525676727295, "learning_rate": 5.20081017571826e-07, "loss": 0.018410373479127884, "step": 7387 }, { "epoch": 1.9179646936656283, "grad_norm": 3.7594196796417236, "learning_rate": 5.199551071657315e-07, "loss": 0.006070416420698166, "step": 7388 }, { "epoch": 1.9182242990654206, "grad_norm": 2.516655206680298, "learning_rate": 5.198295909805126e-07, "loss": 0.005552830174565315, "step": 7389 }, { "epoch": 1.9184839044652129, "grad_norm": 2.2369208335876465, "learning_rate": 5.197044690383577e-07, "loss": 0.0022831042297184467, "step": 7390 }, { "epoch": 1.9187435098650052, "grad_norm": 2.8165855407714844, "learning_rate": 5.195797413613861e-07, "loss": 0.006876862607896328, "step": 7391 }, { "epoch": 1.9190031152647975, "grad_norm": 3.9640631675720215, "learning_rate": 5.194554079716461e-07, "loss": 0.011928298510611057, "step": 7392 }, { "epoch": 1.9192627206645898, "grad_norm": 3.2773067951202393, "learning_rate": 5.19331468891117e-07, "loss": 0.006251771003007889, "step": 7393 }, { "epoch": 1.9195223260643821, "grad_norm": 3.2849128246307373, "learning_rate": 5.192079241417092e-07, "loss": 0.046006932854652405, "step": 7394 }, { "epoch": 1.9197819314641744, "grad_norm": 3.0055723190307617, "learning_rate": 5.190847737452617e-07, "loss": 0.03940066695213318, "step": 7395 }, { "epoch": 1.9200415368639667, "grad_norm": 3.0563912391662598, "learning_rate": 5.189620177235454e-07, "loss": 0.04095057025551796, "step": 7396 }, { "epoch": 1.920301142263759, "grad_norm": 3.9765279293060303, "learning_rate": 5.1883965609826e-07, "loss": 0.008205227553844452, "step": 7397 }, { "epoch": 1.9205607476635516, "grad_norm": 3.8130369186401367, "learning_rate": 5.187176888910366e-07, "loss": 0.0036794054321944714, "step": 7398 }, { "epoch": 1.9208203530633439, "grad_norm": 1.4061260223388672, "learning_rate": 5.185961161234364e-07, "loss": 0.011448485776782036, "step": 7399 }, { "epoch": 1.9210799584631362, "grad_norm": 3.091383934020996, "learning_rate": 5.184749378169505e-07, "loss": 0.032873380929231644, "step": 7400 }, { "epoch": 1.9213395638629285, "grad_norm": 4.012529373168945, "learning_rate": 5.183541539930002e-07, "loss": 0.019971713423728943, "step": 7401 }, { "epoch": 1.9215991692627208, "grad_norm": 0.7654219269752502, "learning_rate": 5.182337646729378e-07, "loss": 0.001430743606761098, "step": 7402 }, { "epoch": 1.921858774662513, "grad_norm": 4.256779193878174, "learning_rate": 5.18113769878045e-07, "loss": 0.008581621572375298, "step": 7403 }, { "epoch": 1.9221183800623054, "grad_norm": 3.1133153438568115, "learning_rate": 5.179941696295345e-07, "loss": 0.00972704403102398, "step": 7404 }, { "epoch": 1.9223779854620977, "grad_norm": 3.646655559539795, "learning_rate": 5.178749639485482e-07, "loss": 0.019611120223999023, "step": 7405 }, { "epoch": 1.92263759086189, "grad_norm": 2.778761625289917, "learning_rate": 5.177561528561597e-07, "loss": 0.003341378876939416, "step": 7406 }, { "epoch": 1.9228971962616823, "grad_norm": 3.7893214225769043, "learning_rate": 5.176377363733718e-07, "loss": 0.018231607973575592, "step": 7407 }, { "epoch": 1.9231568016614746, "grad_norm": 4.281126022338867, "learning_rate": 5.175197145211176e-07, "loss": 0.020477458834648132, "step": 7408 }, { "epoch": 1.923416407061267, "grad_norm": 0.5098006725311279, "learning_rate": 5.174020873202611e-07, "loss": 0.0010418423917144537, "step": 7409 }, { "epoch": 1.9236760124610592, "grad_norm": 5.450545310974121, "learning_rate": 5.172848547915956e-07, "loss": 0.01903531141579151, "step": 7410 }, { "epoch": 1.9239356178608515, "grad_norm": 3.0570523738861084, "learning_rate": 5.171680169558456e-07, "loss": 0.010006336495280266, "step": 7411 }, { "epoch": 1.9241952232606438, "grad_norm": 2.4167022705078125, "learning_rate": 5.170515738336647e-07, "loss": 0.01151976827532053, "step": 7412 }, { "epoch": 1.9244548286604362, "grad_norm": 2.664854049682617, "learning_rate": 5.169355254456381e-07, "loss": 0.003931707702577114, "step": 7413 }, { "epoch": 1.9247144340602285, "grad_norm": 1.8898561000823975, "learning_rate": 5.168198718122804e-07, "loss": 0.005349913612008095, "step": 7414 }, { "epoch": 1.9249740394600208, "grad_norm": 2.7902889251708984, "learning_rate": 5.167046129540361e-07, "loss": 0.03385451063513756, "step": 7415 }, { "epoch": 1.925233644859813, "grad_norm": 4.04743766784668, "learning_rate": 5.165897488912805e-07, "loss": 0.01196955144405365, "step": 7416 }, { "epoch": 1.9254932502596054, "grad_norm": 2.1401846408843994, "learning_rate": 5.164752796443191e-07, "loss": 0.009422939270734787, "step": 7417 }, { "epoch": 1.9257528556593977, "grad_norm": 2.6773979663848877, "learning_rate": 5.163612052333871e-07, "loss": 0.006044348236173391, "step": 7418 }, { "epoch": 1.92601246105919, "grad_norm": 0.6640409827232361, "learning_rate": 5.162475256786508e-07, "loss": 0.003879061434417963, "step": 7419 }, { "epoch": 1.9262720664589823, "grad_norm": 3.7669124603271484, "learning_rate": 5.161342410002055e-07, "loss": 0.051194850355386734, "step": 7420 }, { "epoch": 1.9265316718587746, "grad_norm": 4.207799434661865, "learning_rate": 5.160213512180778e-07, "loss": 0.01653052493929863, "step": 7421 }, { "epoch": 1.926791277258567, "grad_norm": 4.1290693283081055, "learning_rate": 5.159088563522239e-07, "loss": 0.032128699123859406, "step": 7422 }, { "epoch": 1.9270508826583592, "grad_norm": 1.1297845840454102, "learning_rate": 5.157967564225303e-07, "loss": 0.0019908370450139046, "step": 7423 }, { "epoch": 1.9273104880581515, "grad_norm": 2.210944175720215, "learning_rate": 5.156850514488136e-07, "loss": 0.008894028142094612, "step": 7424 }, { "epoch": 1.9275700934579438, "grad_norm": 3.9457006454467773, "learning_rate": 5.155737414508208e-07, "loss": 0.012354056350886822, "step": 7425 }, { "epoch": 1.9278296988577361, "grad_norm": 4.684595108032227, "learning_rate": 5.154628264482289e-07, "loss": 0.023439856246113777, "step": 7426 }, { "epoch": 1.9280893042575284, "grad_norm": 3.2306768894195557, "learning_rate": 5.153523064606452e-07, "loss": 0.039573341608047485, "step": 7427 }, { "epoch": 1.9283489096573208, "grad_norm": 3.9095070362091064, "learning_rate": 5.15242181507607e-07, "loss": 0.0251602903008461, "step": 7428 }, { "epoch": 1.928608515057113, "grad_norm": 4.432699203491211, "learning_rate": 5.151324516085819e-07, "loss": 0.017959730699658394, "step": 7429 }, { "epoch": 1.9288681204569054, "grad_norm": 4.414709091186523, "learning_rate": 5.15023116782968e-07, "loss": 0.04231696203351021, "step": 7430 }, { "epoch": 1.9291277258566977, "grad_norm": 2.10012149810791, "learning_rate": 5.149141770500926e-07, "loss": 0.014068751595914364, "step": 7431 }, { "epoch": 1.92938733125649, "grad_norm": 4.500402450561523, "learning_rate": 5.148056324292142e-07, "loss": 0.007685068063437939, "step": 7432 }, { "epoch": 1.9296469366562823, "grad_norm": 3.098372220993042, "learning_rate": 5.146974829395207e-07, "loss": 0.006029227748513222, "step": 7433 }, { "epoch": 1.9299065420560748, "grad_norm": 4.325308322906494, "learning_rate": 5.145897286001304e-07, "loss": 0.013994935899972916, "step": 7434 }, { "epoch": 1.9301661474558671, "grad_norm": 3.721496820449829, "learning_rate": 5.144823694300921e-07, "loss": 0.041057806462049484, "step": 7435 }, { "epoch": 1.9304257528556594, "grad_norm": 0.8684055209159851, "learning_rate": 5.143754054483844e-07, "loss": 0.0043010832741856575, "step": 7436 }, { "epoch": 1.9306853582554517, "grad_norm": 2.9042437076568604, "learning_rate": 5.14268836673916e-07, "loss": 0.00682584336027503, "step": 7437 }, { "epoch": 1.930944963655244, "grad_norm": 0.5182605981826782, "learning_rate": 5.141626631255257e-07, "loss": 0.0009404345182701945, "step": 7438 }, { "epoch": 1.9312045690550363, "grad_norm": 2.5756919384002686, "learning_rate": 5.140568848219826e-07, "loss": 0.004116247408092022, "step": 7439 }, { "epoch": 1.9314641744548287, "grad_norm": 3.1372427940368652, "learning_rate": 5.139515017819862e-07, "loss": 0.011149942874908447, "step": 7440 }, { "epoch": 1.931723779854621, "grad_norm": 4.065782070159912, "learning_rate": 5.138465140241657e-07, "loss": 0.007686255965381861, "step": 7441 }, { "epoch": 1.9319833852544133, "grad_norm": 5.031244277954102, "learning_rate": 5.137419215670798e-07, "loss": 0.007249088492244482, "step": 7442 }, { "epoch": 1.9322429906542056, "grad_norm": 3.5215718746185303, "learning_rate": 5.136377244292192e-07, "loss": 0.03974473103880882, "step": 7443 }, { "epoch": 1.9325025960539979, "grad_norm": 2.1945910453796387, "learning_rate": 5.135339226290027e-07, "loss": 0.009214668534696102, "step": 7444 }, { "epoch": 1.9327622014537902, "grad_norm": 3.407639503479004, "learning_rate": 5.134305161847805e-07, "loss": 0.013553671538829803, "step": 7445 }, { "epoch": 1.9330218068535827, "grad_norm": 1.5758694410324097, "learning_rate": 5.133275051148322e-07, "loss": 0.0037204334512352943, "step": 7446 }, { "epoch": 1.933281412253375, "grad_norm": 4.605482578277588, "learning_rate": 5.132248894373682e-07, "loss": 0.012653294019401073, "step": 7447 }, { "epoch": 1.9335410176531673, "grad_norm": 4.355397701263428, "learning_rate": 5.131226691705285e-07, "loss": 0.02047853171825409, "step": 7448 }, { "epoch": 1.9338006230529596, "grad_norm": 4.304808139801025, "learning_rate": 5.130208443323829e-07, "loss": 0.053284354507923126, "step": 7449 }, { "epoch": 1.934060228452752, "grad_norm": 0.6071298718452454, "learning_rate": 5.129194149409321e-07, "loss": 0.0017295723082497716, "step": 7450 }, { "epoch": 1.9343198338525442, "grad_norm": 3.7049927711486816, "learning_rate": 5.128183810141066e-07, "loss": 0.042199477553367615, "step": 7451 }, { "epoch": 1.9345794392523366, "grad_norm": 1.7468910217285156, "learning_rate": 5.127177425697662e-07, "loss": 0.005469500552862883, "step": 7452 }, { "epoch": 1.9348390446521289, "grad_norm": 3.68658447265625, "learning_rate": 5.126174996257024e-07, "loss": 0.0279570072889328, "step": 7453 }, { "epoch": 1.9350986500519212, "grad_norm": 2.747297763824463, "learning_rate": 5.125176521996353e-07, "loss": 0.011467298492789268, "step": 7454 }, { "epoch": 1.9353582554517135, "grad_norm": 4.458613872528076, "learning_rate": 5.124182003092153e-07, "loss": 0.02629726380109787, "step": 7455 }, { "epoch": 1.9356178608515058, "grad_norm": 3.593946695327759, "learning_rate": 5.123191439720241e-07, "loss": 0.01697412133216858, "step": 7456 }, { "epoch": 1.935877466251298, "grad_norm": 1.8886265754699707, "learning_rate": 5.122204832055723e-07, "loss": 0.009073678404092789, "step": 7457 }, { "epoch": 1.9361370716510904, "grad_norm": 4.148972511291504, "learning_rate": 5.121222180273004e-07, "loss": 0.05574100464582443, "step": 7458 }, { "epoch": 1.9363966770508827, "grad_norm": 2.952674388885498, "learning_rate": 5.120243484545798e-07, "loss": 0.009048601612448692, "step": 7459 }, { "epoch": 1.936656282450675, "grad_norm": 1.7899762392044067, "learning_rate": 5.119268745047115e-07, "loss": 0.0015201595379039645, "step": 7460 }, { "epoch": 1.9369158878504673, "grad_norm": 4.3761091232299805, "learning_rate": 5.118297961949268e-07, "loss": 0.013211953453719616, "step": 7461 }, { "epoch": 1.9371754932502596, "grad_norm": 2.7409801483154297, "learning_rate": 5.117331135423866e-07, "loss": 0.011543329805135727, "step": 7462 }, { "epoch": 1.937435098650052, "grad_norm": 4.281731128692627, "learning_rate": 5.116368265641828e-07, "loss": 0.035110317170619965, "step": 7463 }, { "epoch": 1.9376947040498442, "grad_norm": 3.8161888122558594, "learning_rate": 5.115409352773361e-07, "loss": 0.029499679803848267, "step": 7464 }, { "epoch": 1.9379543094496365, "grad_norm": 1.4395610094070435, "learning_rate": 5.11445439698798e-07, "loss": 0.00231447396799922, "step": 7465 }, { "epoch": 1.9382139148494288, "grad_norm": 3.9906585216522217, "learning_rate": 5.113503398454503e-07, "loss": 0.03255588561296463, "step": 7466 }, { "epoch": 1.9384735202492211, "grad_norm": 3.986720561981201, "learning_rate": 5.11255635734104e-07, "loss": 0.03633507713675499, "step": 7467 }, { "epoch": 1.9387331256490135, "grad_norm": 0.8778521418571472, "learning_rate": 5.11161327381501e-07, "loss": 0.002381341066211462, "step": 7468 }, { "epoch": 1.9389927310488058, "grad_norm": 3.8043389320373535, "learning_rate": 5.110674148043129e-07, "loss": 0.02994507923722267, "step": 7469 }, { "epoch": 1.939252336448598, "grad_norm": 1.2418330907821655, "learning_rate": 5.109738980191408e-07, "loss": 0.007993156090378761, "step": 7470 }, { "epoch": 1.9395119418483904, "grad_norm": 3.4288320541381836, "learning_rate": 5.108807770425167e-07, "loss": 0.01357682328671217, "step": 7471 }, { "epoch": 1.9397715472481827, "grad_norm": 1.9772356748580933, "learning_rate": 5.107880518909023e-07, "loss": 0.0027578547596931458, "step": 7472 }, { "epoch": 1.940031152647975, "grad_norm": 4.07466983795166, "learning_rate": 5.106957225806889e-07, "loss": 0.02554759383201599, "step": 7473 }, { "epoch": 1.9402907580477673, "grad_norm": 5.934898853302002, "learning_rate": 5.106037891281988e-07, "loss": 0.007156727835536003, "step": 7474 }, { "epoch": 1.9405503634475596, "grad_norm": 3.28840970993042, "learning_rate": 5.105122515496832e-07, "loss": 0.0028577172197401524, "step": 7475 }, { "epoch": 1.940809968847352, "grad_norm": 0.6693241000175476, "learning_rate": 5.10421109861324e-07, "loss": 0.0032522394321858883, "step": 7476 }, { "epoch": 1.9410695742471442, "grad_norm": 4.665596008300781, "learning_rate": 5.103303640792333e-07, "loss": 0.01341063529253006, "step": 7477 }, { "epoch": 1.9413291796469365, "grad_norm": 4.602598190307617, "learning_rate": 5.102400142194524e-07, "loss": 0.028071872889995575, "step": 7478 }, { "epoch": 1.9415887850467288, "grad_norm": 0.29839804768562317, "learning_rate": 5.101500602979533e-07, "loss": 0.0007038483745418489, "step": 7479 }, { "epoch": 1.9418483904465211, "grad_norm": 1.5973501205444336, "learning_rate": 5.100605023306377e-07, "loss": 0.004098794423043728, "step": 7480 }, { "epoch": 1.9421079958463134, "grad_norm": 2.0127761363983154, "learning_rate": 5.099713403333375e-07, "loss": 0.0023641111329197884, "step": 7481 }, { "epoch": 1.942367601246106, "grad_norm": 4.343639850616455, "learning_rate": 5.098825743218146e-07, "loss": 0.011279802769422531, "step": 7482 }, { "epoch": 1.9426272066458983, "grad_norm": 3.2618141174316406, "learning_rate": 5.097942043117604e-07, "loss": 0.011605750769376755, "step": 7483 }, { "epoch": 1.9428868120456906, "grad_norm": 3.2192211151123047, "learning_rate": 5.097062303187973e-07, "loss": 0.012626316398382187, "step": 7484 }, { "epoch": 1.9431464174454829, "grad_norm": 2.3280632495880127, "learning_rate": 5.096186523584765e-07, "loss": 0.008124733343720436, "step": 7485 }, { "epoch": 1.9434060228452752, "grad_norm": 2.1719887256622314, "learning_rate": 5.095314704462803e-07, "loss": 0.00631493516266346, "step": 7486 }, { "epoch": 1.9436656282450675, "grad_norm": 3.4556467533111572, "learning_rate": 5.0944468459762e-07, "loss": 0.027271810919046402, "step": 7487 }, { "epoch": 1.9439252336448598, "grad_norm": 3.3849189281463623, "learning_rate": 5.093582948278375e-07, "loss": 0.01098781730979681, "step": 7488 }, { "epoch": 1.9441848390446521, "grad_norm": 3.73421049118042, "learning_rate": 5.092723011522047e-07, "loss": 0.04907284304499626, "step": 7489 }, { "epoch": 1.9444444444444444, "grad_norm": 2.17192006111145, "learning_rate": 5.091867035859232e-07, "loss": 0.005497557111084461, "step": 7490 }, { "epoch": 1.9447040498442367, "grad_norm": 2.961153984069824, "learning_rate": 5.091015021441246e-07, "loss": 0.008774995803833008, "step": 7491 }, { "epoch": 1.944963655244029, "grad_norm": 3.2748584747314453, "learning_rate": 5.090166968418707e-07, "loss": 0.007201547268778086, "step": 7492 }, { "epoch": 1.9452232606438213, "grad_norm": 0.3106982707977295, "learning_rate": 5.08932287694153e-07, "loss": 0.0006892210803925991, "step": 7493 }, { "epoch": 1.9454828660436139, "grad_norm": 2.203852653503418, "learning_rate": 5.088482747158933e-07, "loss": 0.005676380358636379, "step": 7494 }, { "epoch": 1.9457424714434062, "grad_norm": 5.56152868270874, "learning_rate": 5.087646579219432e-07, "loss": 0.012758525088429451, "step": 7495 }, { "epoch": 1.9460020768431985, "grad_norm": 0.5434372425079346, "learning_rate": 5.086814373270836e-07, "loss": 0.000675056828185916, "step": 7496 }, { "epoch": 1.9462616822429908, "grad_norm": 1.0454511642456055, "learning_rate": 5.085986129460268e-07, "loss": 0.003975324798375368, "step": 7497 }, { "epoch": 1.946521287642783, "grad_norm": 0.35569262504577637, "learning_rate": 5.085161847934141e-07, "loss": 0.0007565533742308617, "step": 7498 }, { "epoch": 1.9467808930425754, "grad_norm": 3.13509202003479, "learning_rate": 5.084341528838165e-07, "loss": 0.011128056794404984, "step": 7499 }, { "epoch": 1.9470404984423677, "grad_norm": 1.5081311464309692, "learning_rate": 5.083525172317356e-07, "loss": 0.00766620272770524, "step": 7500 }, { "epoch": 1.94730010384216, "grad_norm": 3.664608955383301, "learning_rate": 5.082712778516029e-07, "loss": 0.00441700266674161, "step": 7501 }, { "epoch": 1.9475597092419523, "grad_norm": 3.588001251220703, "learning_rate": 5.081904347577796e-07, "loss": 0.016787555068731308, "step": 7502 }, { "epoch": 1.9478193146417446, "grad_norm": 4.257080078125, "learning_rate": 5.081099879645564e-07, "loss": 0.009976796805858612, "step": 7503 }, { "epoch": 1.948078920041537, "grad_norm": 1.1797499656677246, "learning_rate": 5.080299374861551e-07, "loss": 0.0026661118026822805, "step": 7504 }, { "epoch": 1.9483385254413292, "grad_norm": 3.330235719680786, "learning_rate": 5.079502833367265e-07, "loss": 0.01935042068362236, "step": 7505 }, { "epoch": 1.9485981308411215, "grad_norm": 3.387782335281372, "learning_rate": 5.078710255303515e-07, "loss": 0.026852089911699295, "step": 7506 }, { "epoch": 1.9488577362409139, "grad_norm": 3.3577044010162354, "learning_rate": 5.077921640810416e-07, "loss": 0.034325968474149704, "step": 7507 }, { "epoch": 1.9491173416407062, "grad_norm": 4.5701069831848145, "learning_rate": 5.077136990027372e-07, "loss": 0.013953939080238342, "step": 7508 }, { "epoch": 1.9493769470404985, "grad_norm": 0.905861496925354, "learning_rate": 5.076356303093091e-07, "loss": 0.0018929196521639824, "step": 7509 }, { "epoch": 1.9496365524402908, "grad_norm": 1.1398825645446777, "learning_rate": 5.075579580145585e-07, "loss": 0.00138233439065516, "step": 7510 }, { "epoch": 1.949896157840083, "grad_norm": 3.5242855548858643, "learning_rate": 5.074806821322155e-07, "loss": 0.006147813051939011, "step": 7511 }, { "epoch": 1.9501557632398754, "grad_norm": 4.0042595863342285, "learning_rate": 5.074038026759412e-07, "loss": 0.04490246996283531, "step": 7512 }, { "epoch": 1.9504153686396677, "grad_norm": 1.7341364622116089, "learning_rate": 5.07327319659326e-07, "loss": 0.0021707930136471987, "step": 7513 }, { "epoch": 1.95067497403946, "grad_norm": 0.45338279008865356, "learning_rate": 5.072512330958902e-07, "loss": 0.0007628168677911162, "step": 7514 }, { "epoch": 1.9509345794392523, "grad_norm": 3.7563297748565674, "learning_rate": 5.071755429990844e-07, "loss": 0.04332790896296501, "step": 7515 }, { "epoch": 1.9511941848390446, "grad_norm": 1.7553690671920776, "learning_rate": 5.071002493822885e-07, "loss": 0.003236586693674326, "step": 7516 }, { "epoch": 1.951453790238837, "grad_norm": 1.0203025341033936, "learning_rate": 5.070253522588128e-07, "loss": 0.001894861925393343, "step": 7517 }, { "epoch": 1.9517133956386292, "grad_norm": 3.5152370929718018, "learning_rate": 5.06950851641898e-07, "loss": 0.010165700688958168, "step": 7518 }, { "epoch": 1.9519730010384215, "grad_norm": 1.2684009075164795, "learning_rate": 5.068767475447131e-07, "loss": 0.004762305412441492, "step": 7519 }, { "epoch": 1.9522326064382138, "grad_norm": 4.670705318450928, "learning_rate": 5.068030399803587e-07, "loss": 0.02366233430802822, "step": 7520 }, { "epoch": 1.9524922118380061, "grad_norm": 6.225174903869629, "learning_rate": 5.067297289618643e-07, "loss": 0.016368623822927475, "step": 7521 }, { "epoch": 1.9527518172377984, "grad_norm": 4.085736274719238, "learning_rate": 5.0665681450219e-07, "loss": 0.0556163564324379, "step": 7522 }, { "epoch": 1.9530114226375908, "grad_norm": 4.6119208335876465, "learning_rate": 5.065842966142251e-07, "loss": 0.01756507344543934, "step": 7523 }, { "epoch": 1.953271028037383, "grad_norm": 3.1908905506134033, "learning_rate": 5.065121753107889e-07, "loss": 0.014208132401108742, "step": 7524 }, { "epoch": 1.9535306334371754, "grad_norm": 4.011569023132324, "learning_rate": 5.064404506046314e-07, "loss": 0.03811315447092056, "step": 7525 }, { "epoch": 1.9537902388369677, "grad_norm": 9.108850479125977, "learning_rate": 5.063691225084313e-07, "loss": 0.008588157594203949, "step": 7526 }, { "epoch": 1.95404984423676, "grad_norm": 1.8120609521865845, "learning_rate": 5.062981910347979e-07, "loss": 0.007962450385093689, "step": 7527 }, { "epoch": 1.9543094496365523, "grad_norm": 2.613875150680542, "learning_rate": 5.062276561962705e-07, "loss": 0.011151318438351154, "step": 7528 }, { "epoch": 1.9545690550363446, "grad_norm": 4.200826644897461, "learning_rate": 5.06157518005318e-07, "loss": 0.02059153839945793, "step": 7529 }, { "epoch": 1.9548286604361371, "grad_norm": 0.08951236307621002, "learning_rate": 5.060877764743391e-07, "loss": 0.0003905462799593806, "step": 7530 }, { "epoch": 1.9550882658359294, "grad_norm": 4.167377471923828, "learning_rate": 5.060184316156627e-07, "loss": 0.019765488803386688, "step": 7531 }, { "epoch": 1.9553478712357217, "grad_norm": 2.4153852462768555, "learning_rate": 5.059494834415468e-07, "loss": 0.0019260382978245616, "step": 7532 }, { "epoch": 1.955607476635514, "grad_norm": 1.3837751150131226, "learning_rate": 5.058809319641804e-07, "loss": 0.006355760619044304, "step": 7533 }, { "epoch": 1.9558670820353063, "grad_norm": 4.878229141235352, "learning_rate": 5.05812777195682e-07, "loss": 0.009930131025612354, "step": 7534 }, { "epoch": 1.9561266874350987, "grad_norm": 2.136509418487549, "learning_rate": 5.057450191480994e-07, "loss": 0.003984416369348764, "step": 7535 }, { "epoch": 1.956386292834891, "grad_norm": 3.2099227905273438, "learning_rate": 5.056776578334107e-07, "loss": 0.010264036245644093, "step": 7536 }, { "epoch": 1.9566458982346833, "grad_norm": 0.8820711970329285, "learning_rate": 5.056106932635236e-07, "loss": 0.001354666892439127, "step": 7537 }, { "epoch": 1.9569055036344756, "grad_norm": 3.650312662124634, "learning_rate": 5.055441254502768e-07, "loss": 0.018610211089253426, "step": 7538 }, { "epoch": 1.9571651090342679, "grad_norm": 2.952615261077881, "learning_rate": 5.054779544054372e-07, "loss": 0.002554355887696147, "step": 7539 }, { "epoch": 1.9574247144340602, "grad_norm": 1.8532259464263916, "learning_rate": 5.054121801407024e-07, "loss": 0.006598471663892269, "step": 7540 }, { "epoch": 1.9576843198338525, "grad_norm": 4.2524943351745605, "learning_rate": 5.053468026677003e-07, "loss": 0.021761540323495865, "step": 7541 }, { "epoch": 1.957943925233645, "grad_norm": 6.51881742477417, "learning_rate": 5.052818219979871e-07, "loss": 0.010047923773527145, "step": 7542 }, { "epoch": 1.9582035306334373, "grad_norm": 4.619829177856445, "learning_rate": 5.05217238143051e-07, "loss": 0.031220847740769386, "step": 7543 }, { "epoch": 1.9584631360332296, "grad_norm": 0.9818623661994934, "learning_rate": 5.051530511143083e-07, "loss": 0.003794305492192507, "step": 7544 }, { "epoch": 1.958722741433022, "grad_norm": 4.041805744171143, "learning_rate": 5.050892609231061e-07, "loss": 0.03933563828468323, "step": 7545 }, { "epoch": 1.9589823468328142, "grad_norm": 2.2889914512634277, "learning_rate": 5.050258675807208e-07, "loss": 0.007701908238232136, "step": 7546 }, { "epoch": 1.9592419522326066, "grad_norm": 0.1781739741563797, "learning_rate": 5.049628710983592e-07, "loss": 0.0004211645573377609, "step": 7547 }, { "epoch": 1.9595015576323989, "grad_norm": 2.0316059589385986, "learning_rate": 5.049002714871572e-07, "loss": 0.001676631043665111, "step": 7548 }, { "epoch": 1.9597611630321912, "grad_norm": 2.5499308109283447, "learning_rate": 5.048380687581814e-07, "loss": 0.006991156376898289, "step": 7549 }, { "epoch": 1.9600207684319835, "grad_norm": 3.3568544387817383, "learning_rate": 5.047762629224276e-07, "loss": 0.038648415356874466, "step": 7550 }, { "epoch": 1.9602803738317758, "grad_norm": 1.2762439250946045, "learning_rate": 5.047148539908217e-07, "loss": 0.0014600250869989395, "step": 7551 }, { "epoch": 1.960539979231568, "grad_norm": 0.6989502310752869, "learning_rate": 5.046538419742194e-07, "loss": 0.005842781160026789, "step": 7552 }, { "epoch": 1.9607995846313604, "grad_norm": 1.9857368469238281, "learning_rate": 5.045932268834061e-07, "loss": 0.0029631969518959522, "step": 7553 }, { "epoch": 1.9610591900311527, "grad_norm": 3.266078233718872, "learning_rate": 5.045330087290976e-07, "loss": 0.04761793464422226, "step": 7554 }, { "epoch": 1.961318795430945, "grad_norm": 3.487441062927246, "learning_rate": 5.044731875219385e-07, "loss": 0.03181172162294388, "step": 7555 }, { "epoch": 1.9615784008307373, "grad_norm": 4.920881271362305, "learning_rate": 5.044137632725041e-07, "loss": 0.006416680756956339, "step": 7556 }, { "epoch": 1.9618380062305296, "grad_norm": 1.180408000946045, "learning_rate": 5.043547359912995e-07, "loss": 0.0015004808083176613, "step": 7557 }, { "epoch": 1.962097611630322, "grad_norm": 3.9420275688171387, "learning_rate": 5.042961056887591e-07, "loss": 0.014973174780607224, "step": 7558 }, { "epoch": 1.9623572170301142, "grad_norm": 3.742079257965088, "learning_rate": 5.042378723752473e-07, "loss": 0.05472417175769806, "step": 7559 }, { "epoch": 1.9626168224299065, "grad_norm": 2.916116952896118, "learning_rate": 5.041800360610583e-07, "loss": 0.008225452154874802, "step": 7560 }, { "epoch": 1.9628764278296988, "grad_norm": 2.605210065841675, "learning_rate": 5.041225967564167e-07, "loss": 0.0024243025109171867, "step": 7561 }, { "epoch": 1.9631360332294912, "grad_norm": 2.4785242080688477, "learning_rate": 5.040655544714762e-07, "loss": 0.0168341975659132, "step": 7562 }, { "epoch": 1.9633956386292835, "grad_norm": 3.618093252182007, "learning_rate": 5.040089092163204e-07, "loss": 0.01795738935470581, "step": 7563 }, { "epoch": 1.9636552440290758, "grad_norm": 3.365795850753784, "learning_rate": 5.039526610009634e-07, "loss": 0.021300990134477615, "step": 7564 }, { "epoch": 1.963914849428868, "grad_norm": 4.217456340789795, "learning_rate": 5.038968098353482e-07, "loss": 0.03910613805055618, "step": 7565 }, { "epoch": 1.9641744548286604, "grad_norm": 3.7801835536956787, "learning_rate": 5.038413557293478e-07, "loss": 0.026543647050857544, "step": 7566 }, { "epoch": 1.9644340602284527, "grad_norm": 3.9806060791015625, "learning_rate": 5.037862986927657e-07, "loss": 0.010401435196399689, "step": 7567 }, { "epoch": 1.964693665628245, "grad_norm": 3.853823184967041, "learning_rate": 5.037316387353347e-07, "loss": 0.04802954941987991, "step": 7568 }, { "epoch": 1.9649532710280373, "grad_norm": 0.8213972449302673, "learning_rate": 5.036773758667168e-07, "loss": 0.0026966293808072805, "step": 7569 }, { "epoch": 1.9652128764278296, "grad_norm": 4.615248203277588, "learning_rate": 5.036235100965053e-07, "loss": 0.02670138329267502, "step": 7570 }, { "epoch": 1.965472481827622, "grad_norm": 3.7526040077209473, "learning_rate": 5.035700414342218e-07, "loss": 0.03774642199277878, "step": 7571 }, { "epoch": 1.9657320872274142, "grad_norm": 4.200477600097656, "learning_rate": 5.035169698893185e-07, "loss": 0.03301376476883888, "step": 7572 }, { "epoch": 1.9659916926272065, "grad_norm": 0.758489191532135, "learning_rate": 5.034642954711773e-07, "loss": 0.0009838612750172615, "step": 7573 }, { "epoch": 1.9662512980269988, "grad_norm": 3.319847583770752, "learning_rate": 5.034120181891097e-07, "loss": 0.023433906957507133, "step": 7574 }, { "epoch": 1.9665109034267911, "grad_norm": 3.496439218521118, "learning_rate": 5.033601380523576e-07, "loss": 0.049383945763111115, "step": 7575 }, { "epoch": 1.9667705088265834, "grad_norm": 0.6874250173568726, "learning_rate": 5.033086550700915e-07, "loss": 0.007158142980188131, "step": 7576 }, { "epoch": 1.9670301142263757, "grad_norm": 3.796433448791504, "learning_rate": 5.03257569251413e-07, "loss": 0.0368887335062027, "step": 7577 }, { "epoch": 1.9672897196261683, "grad_norm": 0.9295914173126221, "learning_rate": 5.032068806053528e-07, "loss": 0.0014976980164647102, "step": 7578 }, { "epoch": 1.9675493250259606, "grad_norm": 4.443777561187744, "learning_rate": 5.031565891408709e-07, "loss": 0.019771091639995575, "step": 7579 }, { "epoch": 1.967808930425753, "grad_norm": 3.547762393951416, "learning_rate": 5.031066948668585e-07, "loss": 0.008037898689508438, "step": 7580 }, { "epoch": 1.9680685358255452, "grad_norm": 2.2532715797424316, "learning_rate": 5.030571977921354e-07, "loss": 0.006354399956762791, "step": 7581 }, { "epoch": 1.9683281412253375, "grad_norm": 1.2829958200454712, "learning_rate": 5.030080979254515e-07, "loss": 0.011142079718410969, "step": 7582 }, { "epoch": 1.9685877466251298, "grad_norm": 2.488168716430664, "learning_rate": 5.029593952754865e-07, "loss": 0.007397132460027933, "step": 7583 }, { "epoch": 1.9688473520249221, "grad_norm": 4.16187858581543, "learning_rate": 5.029110898508502e-07, "loss": 0.011476762592792511, "step": 7584 }, { "epoch": 1.9691069574247144, "grad_norm": 4.407505989074707, "learning_rate": 5.028631816600815e-07, "loss": 0.019968898966908455, "step": 7585 }, { "epoch": 1.9693665628245067, "grad_norm": 4.481733322143555, "learning_rate": 5.028156707116497e-07, "loss": 0.004932123702019453, "step": 7586 }, { "epoch": 1.969626168224299, "grad_norm": 0.09768381714820862, "learning_rate": 5.027685570139535e-07, "loss": 0.000331950286636129, "step": 7587 }, { "epoch": 1.9698857736240913, "grad_norm": 4.714725971221924, "learning_rate": 5.02721840575322e-07, "loss": 0.014514814130961895, "step": 7588 }, { "epoch": 1.9701453790238836, "grad_norm": 3.3303182125091553, "learning_rate": 5.02675521404013e-07, "loss": 0.008730028755962849, "step": 7589 }, { "epoch": 1.9704049844236762, "grad_norm": 5.003809452056885, "learning_rate": 5.026295995082151e-07, "loss": 0.026247456669807434, "step": 7590 }, { "epoch": 1.9706645898234685, "grad_norm": 3.4611518383026123, "learning_rate": 5.025840748960458e-07, "loss": 0.010819965973496437, "step": 7591 }, { "epoch": 1.9709241952232608, "grad_norm": 1.3188283443450928, "learning_rate": 5.025389475755532e-07, "loss": 0.00529808783903718, "step": 7592 }, { "epoch": 1.971183800623053, "grad_norm": 3.4689624309539795, "learning_rate": 5.024942175547146e-07, "loss": 0.02649364061653614, "step": 7593 }, { "epoch": 1.9714434060228454, "grad_norm": 1.9628334045410156, "learning_rate": 5.024498848414372e-07, "loss": 0.0026526316069066525, "step": 7594 }, { "epoch": 1.9717030114226377, "grad_norm": 3.8687312602996826, "learning_rate": 5.024059494435582e-07, "loss": 0.02918553724884987, "step": 7595 }, { "epoch": 1.97196261682243, "grad_norm": 3.50976824760437, "learning_rate": 5.023624113688442e-07, "loss": 0.026419682428240776, "step": 7596 }, { "epoch": 1.9722222222222223, "grad_norm": 4.1032328605651855, "learning_rate": 5.023192706249917e-07, "loss": 0.03173719346523285, "step": 7597 }, { "epoch": 1.9724818276220146, "grad_norm": 2.4240331649780273, "learning_rate": 5.022765272196273e-07, "loss": 0.0062370640225708485, "step": 7598 }, { "epoch": 1.972741433021807, "grad_norm": 3.6597840785980225, "learning_rate": 5.022341811603066e-07, "loss": 0.0323762483894825, "step": 7599 }, { "epoch": 1.9730010384215992, "grad_norm": 3.3954007625579834, "learning_rate": 5.021922324545157e-07, "loss": 0.002220367779955268, "step": 7600 }, { "epoch": 1.9732606438213915, "grad_norm": 4.570201396942139, "learning_rate": 5.021506811096703e-07, "loss": 0.011202114634215832, "step": 7601 }, { "epoch": 1.9735202492211839, "grad_norm": 3.770310640335083, "learning_rate": 5.021095271331153e-07, "loss": 0.010909839533269405, "step": 7602 }, { "epoch": 1.9737798546209762, "grad_norm": 2.530940055847168, "learning_rate": 5.020687705321261e-07, "loss": 0.0031513399444520473, "step": 7603 }, { "epoch": 1.9740394600207685, "grad_norm": 3.9444808959960938, "learning_rate": 5.020284113139077e-07, "loss": 0.024930063635110855, "step": 7604 }, { "epoch": 1.9742990654205608, "grad_norm": 3.702787160873413, "learning_rate": 5.019884494855942e-07, "loss": 0.014311658218502998, "step": 7605 }, { "epoch": 1.974558670820353, "grad_norm": 2.620448112487793, "learning_rate": 5.019488850542504e-07, "loss": 0.0058686137199401855, "step": 7606 }, { "epoch": 1.9748182762201454, "grad_norm": 3.098017454147339, "learning_rate": 5.0190971802687e-07, "loss": 0.016466284170746803, "step": 7607 }, { "epoch": 1.9750778816199377, "grad_norm": 4.059018611907959, "learning_rate": 5.018709484103771e-07, "loss": 0.015336194075644016, "step": 7608 }, { "epoch": 1.97533748701973, "grad_norm": 4.18746280670166, "learning_rate": 5.018325762116251e-07, "loss": 0.01970347762107849, "step": 7609 }, { "epoch": 1.9755970924195223, "grad_norm": 2.17427134513855, "learning_rate": 5.017946014373977e-07, "loss": 0.0016759589780122042, "step": 7610 }, { "epoch": 1.9758566978193146, "grad_norm": 1.1246789693832397, "learning_rate": 5.017570240944074e-07, "loss": 0.009074024856090546, "step": 7611 }, { "epoch": 1.976116303219107, "grad_norm": 4.671160697937012, "learning_rate": 5.017198441892972e-07, "loss": 0.007997333072125912, "step": 7612 }, { "epoch": 1.9763759086188992, "grad_norm": 1.8251498937606812, "learning_rate": 5.016830617286399e-07, "loss": 0.0019229406025260687, "step": 7613 }, { "epoch": 1.9766355140186915, "grad_norm": 1.278839349746704, "learning_rate": 5.016466767189378e-07, "loss": 0.0035036911722272635, "step": 7614 }, { "epoch": 1.9768951194184838, "grad_norm": 3.687342405319214, "learning_rate": 5.016106891666227e-07, "loss": 0.010088279843330383, "step": 7615 }, { "epoch": 1.9771547248182761, "grad_norm": 0.33641183376312256, "learning_rate": 5.015750990780564e-07, "loss": 0.001096275052987039, "step": 7616 }, { "epoch": 1.9774143302180685, "grad_norm": 3.591958522796631, "learning_rate": 5.015399064595305e-07, "loss": 0.009475704282522202, "step": 7617 }, { "epoch": 1.9776739356178608, "grad_norm": 3.698127269744873, "learning_rate": 5.01505111317266e-07, "loss": 0.02231019362807274, "step": 7618 }, { "epoch": 1.977933541017653, "grad_norm": 2.20625638961792, "learning_rate": 5.014707136574146e-07, "loss": 0.002608305774629116, "step": 7619 }, { "epoch": 1.9781931464174454, "grad_norm": 4.167311191558838, "learning_rate": 5.014367134860563e-07, "loss": 0.01620514690876007, "step": 7620 }, { "epoch": 1.9784527518172377, "grad_norm": 1.61149263381958, "learning_rate": 5.014031108092017e-07, "loss": 0.0027399975806474686, "step": 7621 }, { "epoch": 1.97871235721703, "grad_norm": 2.0788350105285645, "learning_rate": 5.01369905632791e-07, "loss": 0.007507216185331345, "step": 7622 }, { "epoch": 1.9789719626168223, "grad_norm": 4.120263576507568, "learning_rate": 5.013370979626944e-07, "loss": 0.014323829673230648, "step": 7623 }, { "epoch": 1.9792315680166146, "grad_norm": 0.13775740563869476, "learning_rate": 5.013046878047113e-07, "loss": 0.000368311011698097, "step": 7624 }, { "epoch": 1.979491173416407, "grad_norm": 1.833883285522461, "learning_rate": 5.01272675164571e-07, "loss": 0.0076026166789233685, "step": 7625 }, { "epoch": 1.9797507788161994, "grad_norm": 3.8753488063812256, "learning_rate": 5.012410600479326e-07, "loss": 0.0184082742780447, "step": 7626 }, { "epoch": 1.9800103842159917, "grad_norm": 3.9783644676208496, "learning_rate": 5.012098424603852e-07, "loss": 0.031811490654945374, "step": 7627 }, { "epoch": 1.980269989615784, "grad_norm": 1.241299033164978, "learning_rate": 5.011790224074471e-07, "loss": 0.002162822289392352, "step": 7628 }, { "epoch": 1.9805295950155763, "grad_norm": 1.9057338237762451, "learning_rate": 5.011485998945668e-07, "loss": 0.0030032088980078697, "step": 7629 }, { "epoch": 1.9807892004153687, "grad_norm": 0.11300015449523926, "learning_rate": 5.01118574927122e-07, "loss": 0.0004182655247859657, "step": 7630 }, { "epoch": 1.981048805815161, "grad_norm": 4.571908950805664, "learning_rate": 5.010889475104205e-07, "loss": 0.033155255019664764, "step": 7631 }, { "epoch": 1.9813084112149533, "grad_norm": 2.441946506500244, "learning_rate": 5.010597176497002e-07, "loss": 0.003771853167563677, "step": 7632 }, { "epoch": 1.9815680166147456, "grad_norm": 3.1432554721832275, "learning_rate": 5.010308853501275e-07, "loss": 0.0028556501492857933, "step": 7633 }, { "epoch": 1.9818276220145379, "grad_norm": 4.375800132751465, "learning_rate": 5.010024506168001e-07, "loss": 0.04348035156726837, "step": 7634 }, { "epoch": 1.9820872274143302, "grad_norm": 4.383745193481445, "learning_rate": 5.009744134547438e-07, "loss": 0.007314388640224934, "step": 7635 }, { "epoch": 1.9823468328141225, "grad_norm": 4.425081729888916, "learning_rate": 5.009467738689157e-07, "loss": 0.039034515619277954, "step": 7636 }, { "epoch": 1.9826064382139148, "grad_norm": 3.5888781547546387, "learning_rate": 5.009195318642012e-07, "loss": 0.02779734879732132, "step": 7637 }, { "epoch": 1.9828660436137073, "grad_norm": 3.924286127090454, "learning_rate": 5.008926874454166e-07, "loss": 0.029707927256822586, "step": 7638 }, { "epoch": 1.9831256490134996, "grad_norm": 4.822940349578857, "learning_rate": 5.008662406173071e-07, "loss": 0.024246670305728912, "step": 7639 }, { "epoch": 1.983385254413292, "grad_norm": 1.5827624797821045, "learning_rate": 5.008401913845478e-07, "loss": 0.00948254857212305, "step": 7640 }, { "epoch": 1.9836448598130842, "grad_norm": 1.1708109378814697, "learning_rate": 5.008145397517439e-07, "loss": 0.0037285704165697098, "step": 7641 }, { "epoch": 1.9839044652128766, "grad_norm": 2.5591750144958496, "learning_rate": 5.007892857234296e-07, "loss": 0.003684740513563156, "step": 7642 }, { "epoch": 1.9841640706126689, "grad_norm": 3.0102972984313965, "learning_rate": 5.007644293040695e-07, "loss": 0.009382588788866997, "step": 7643 }, { "epoch": 1.9844236760124612, "grad_norm": 0.2982277572154999, "learning_rate": 5.007399704980578e-07, "loss": 0.0004247108590789139, "step": 7644 }, { "epoch": 1.9846832814122535, "grad_norm": 4.642423629760742, "learning_rate": 5.007159093097181e-07, "loss": 0.004301370587199926, "step": 7645 }, { "epoch": 1.9849428868120458, "grad_norm": 2.7809689044952393, "learning_rate": 5.006922457433039e-07, "loss": 0.009050899185240269, "step": 7646 }, { "epoch": 1.985202492211838, "grad_norm": 2.0816471576690674, "learning_rate": 5.006689798029981e-07, "loss": 0.006726243533194065, "step": 7647 }, { "epoch": 1.9854620976116304, "grad_norm": 0.960081934928894, "learning_rate": 5.00646111492914e-07, "loss": 0.002242127899080515, "step": 7648 }, { "epoch": 1.9857217030114227, "grad_norm": 2.0472097396850586, "learning_rate": 5.006236408170939e-07, "loss": 0.003253812436014414, "step": 7649 }, { "epoch": 1.985981308411215, "grad_norm": 4.109157085418701, "learning_rate": 5.006015677795102e-07, "loss": 0.009092522785067558, "step": 7650 }, { "epoch": 1.9862409138110073, "grad_norm": 1.5865919589996338, "learning_rate": 5.005798923840652e-07, "loss": 0.005628820974379778, "step": 7651 }, { "epoch": 1.9865005192107996, "grad_norm": 3.942011833190918, "learning_rate": 5.005586146345901e-07, "loss": 0.021580703556537628, "step": 7652 }, { "epoch": 1.986760124610592, "grad_norm": 0.6142634749412537, "learning_rate": 5.005377345348463e-07, "loss": 0.0010989129077643156, "step": 7653 }, { "epoch": 1.9870197300103842, "grad_norm": 3.9426114559173584, "learning_rate": 5.005172520885257e-07, "loss": 0.03994600847363472, "step": 7654 }, { "epoch": 1.9872793354101765, "grad_norm": 3.074249267578125, "learning_rate": 5.004971672992484e-07, "loss": 0.014167876914143562, "step": 7655 }, { "epoch": 1.9875389408099688, "grad_norm": 3.097294569015503, "learning_rate": 5.004774801705649e-07, "loss": 0.00492623308673501, "step": 7656 }, { "epoch": 1.9877985462097612, "grad_norm": 2.9102766513824463, "learning_rate": 5.00458190705956e-07, "loss": 0.002009684918448329, "step": 7657 }, { "epoch": 1.9880581516095535, "grad_norm": 0.8005115389823914, "learning_rate": 5.00439298908831e-07, "loss": 0.005415376741439104, "step": 7658 }, { "epoch": 1.9883177570093458, "grad_norm": 3.966430425643921, "learning_rate": 5.004208047825299e-07, "loss": 0.003539395285770297, "step": 7659 }, { "epoch": 1.988577362409138, "grad_norm": 1.3358259201049805, "learning_rate": 5.004027083303222e-07, "loss": 0.003964547999203205, "step": 7660 }, { "epoch": 1.9888369678089304, "grad_norm": 0.9281079769134521, "learning_rate": 5.003850095554065e-07, "loss": 0.0018031361978501081, "step": 7661 }, { "epoch": 1.9890965732087227, "grad_norm": 2.059988260269165, "learning_rate": 5.00367708460912e-07, "loss": 0.009577656164765358, "step": 7662 }, { "epoch": 1.989356178608515, "grad_norm": 4.241560459136963, "learning_rate": 5.003508050498967e-07, "loss": 0.03465881198644638, "step": 7663 }, { "epoch": 1.9896157840083073, "grad_norm": 5.014860153198242, "learning_rate": 5.00334299325349e-07, "loss": 0.001644836156629026, "step": 7664 }, { "epoch": 1.9898753894080996, "grad_norm": 3.7811665534973145, "learning_rate": 5.003181912901865e-07, "loss": 0.008876295760273933, "step": 7665 }, { "epoch": 1.990134994807892, "grad_norm": 1.3504682779312134, "learning_rate": 5.003024809472572e-07, "loss": 0.001871571526862681, "step": 7666 }, { "epoch": 1.9903946002076842, "grad_norm": 3.6948635578155518, "learning_rate": 5.002871682993378e-07, "loss": 0.00808645412325859, "step": 7667 }, { "epoch": 1.9906542056074765, "grad_norm": 1.159423828125, "learning_rate": 5.002722533491354e-07, "loss": 0.0018908327911049128, "step": 7668 }, { "epoch": 1.9909138110072688, "grad_norm": 2.2320759296417236, "learning_rate": 5.002577360992869e-07, "loss": 0.004744209349155426, "step": 7669 }, { "epoch": 1.9911734164070611, "grad_norm": 2.228510856628418, "learning_rate": 5.002436165523583e-07, "loss": 0.005249138921499252, "step": 7670 }, { "epoch": 1.9914330218068534, "grad_norm": 2.218503713607788, "learning_rate": 5.002298947108454e-07, "loss": 0.003351774299517274, "step": 7671 }, { "epoch": 1.9916926272066457, "grad_norm": 4.210080623626709, "learning_rate": 5.002165705771746e-07, "loss": 0.023450084030628204, "step": 7672 }, { "epoch": 1.991952232606438, "grad_norm": 3.766568660736084, "learning_rate": 5.002036441537007e-07, "loss": 0.039119794964790344, "step": 7673 }, { "epoch": 1.9922118380062306, "grad_norm": 2.7124974727630615, "learning_rate": 5.00191115442709e-07, "loss": 0.005609865300357342, "step": 7674 }, { "epoch": 1.992471443406023, "grad_norm": 3.0480222702026367, "learning_rate": 5.001789844464145e-07, "loss": 0.006263800896704197, "step": 7675 }, { "epoch": 1.9927310488058152, "grad_norm": 4.154672622680664, "learning_rate": 5.001672511669613e-07, "loss": 0.011678007431328297, "step": 7676 }, { "epoch": 1.9929906542056075, "grad_norm": 3.481156349182129, "learning_rate": 5.001559156064238e-07, "loss": 0.010045747272670269, "step": 7677 }, { "epoch": 1.9932502596053998, "grad_norm": 1.577587604522705, "learning_rate": 5.00144977766806e-07, "loss": 0.007719546090811491, "step": 7678 }, { "epoch": 1.9935098650051921, "grad_norm": 3.146655321121216, "learning_rate": 5.001344376500408e-07, "loss": 0.012590918689966202, "step": 7679 }, { "epoch": 1.9937694704049844, "grad_norm": 2.511948585510254, "learning_rate": 5.001242952579924e-07, "loss": 0.003179384395480156, "step": 7680 }, { "epoch": 1.9940290758047767, "grad_norm": 2.9970626831054688, "learning_rate": 5.001145505924532e-07, "loss": 0.007250980474054813, "step": 7681 }, { "epoch": 1.994288681204569, "grad_norm": 4.067587375640869, "learning_rate": 5.001052036551458e-07, "loss": 0.04737027734518051, "step": 7682 }, { "epoch": 1.9945482866043613, "grad_norm": 2.0208866596221924, "learning_rate": 5.000962544477227e-07, "loss": 0.005356420297175646, "step": 7683 }, { "epoch": 1.9948078920041536, "grad_norm": 4.105834484100342, "learning_rate": 5.000877029717655e-07, "loss": 0.014595864340662956, "step": 7684 }, { "epoch": 1.995067497403946, "grad_norm": 7.904292583465576, "learning_rate": 5.000795492287867e-07, "loss": 0.011930219829082489, "step": 7685 }, { "epoch": 1.9953271028037385, "grad_norm": 2.9657955169677734, "learning_rate": 5.000717932202271e-07, "loss": 0.0033392985351383686, "step": 7686 }, { "epoch": 1.9955867082035308, "grad_norm": 4.2459187507629395, "learning_rate": 5.000644349474577e-07, "loss": 0.02439943514764309, "step": 7687 }, { "epoch": 1.995846313603323, "grad_norm": 4.422745704650879, "learning_rate": 5.000574744117799e-07, "loss": 0.007104120682924986, "step": 7688 }, { "epoch": 1.9961059190031154, "grad_norm": 0.8379095792770386, "learning_rate": 5.000509116144236e-07, "loss": 0.0012893825769424438, "step": 7689 }, { "epoch": 1.9963655244029077, "grad_norm": 3.9828217029571533, "learning_rate": 5.000447465565491e-07, "loss": 0.017411155626177788, "step": 7690 }, { "epoch": 1.9966251298027, "grad_norm": 3.88564395904541, "learning_rate": 5.000389792392462e-07, "loss": 0.04283437877893448, "step": 7691 }, { "epoch": 1.9968847352024923, "grad_norm": 3.2411437034606934, "learning_rate": 5.000336096635346e-07, "loss": 0.008005509153008461, "step": 7692 }, { "epoch": 1.9971443406022846, "grad_norm": 3.0741894245147705, "learning_rate": 5.000286378303634e-07, "loss": 0.005651957355439663, "step": 7693 }, { "epoch": 1.997403946002077, "grad_norm": 4.128803730010986, "learning_rate": 5.000240637406114e-07, "loss": 0.05321303755044937, "step": 7694 }, { "epoch": 1.9976635514018692, "grad_norm": 1.2954665422439575, "learning_rate": 5.000198873950875e-07, "loss": 0.0017337705940008163, "step": 7695 }, { "epoch": 1.9979231568016615, "grad_norm": 2.5457770824432373, "learning_rate": 5.000161087945296e-07, "loss": 0.005197938531637192, "step": 7696 }, { "epoch": 1.9981827622014539, "grad_norm": 3.575122117996216, "learning_rate": 5.000127279396061e-07, "loss": 0.04494870454072952, "step": 7697 }, { "epoch": 1.9984423676012462, "grad_norm": 0.7769529223442078, "learning_rate": 5.000097448309143e-07, "loss": 0.011516678147017956, "step": 7698 }, { "epoch": 1.9987019730010385, "grad_norm": 0.8410825729370117, "learning_rate": 5.000071594689815e-07, "loss": 0.0039059282280504704, "step": 7699 }, { "epoch": 1.9989615784008308, "grad_norm": 3.560511827468872, "learning_rate": 5.000049718542651e-07, "loss": 0.030136100947856903, "step": 7700 }, { "epoch": 1.999221183800623, "grad_norm": 3.240534543991089, "learning_rate": 5.000031819871517e-07, "loss": 0.008124545216560364, "step": 7701 }, { "epoch": 1.9994807892004154, "grad_norm": 5.1699957847595215, "learning_rate": 5.000017898679574e-07, "loss": 0.013893505558371544, "step": 7702 }, { "epoch": 1.9997403946002077, "grad_norm": 2.076751470565796, "learning_rate": 5.000007954969287e-07, "loss": 0.008646814152598381, "step": 7703 }, { "epoch": 2.0, "grad_norm": 0.35851481556892395, "learning_rate": 5.000001988742409e-07, "loss": 0.0007540645892731845, "step": 7704 } ], "logging_steps": 1, "max_steps": 7704, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }