{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 3852, "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 } ], "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": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }