{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 2.0, "eval_steps": 500, "global_step": 7704, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00025960539979231567, "grad_norm": 26.493032455444336, "learning_rate": 0.0, "loss": 0.30881375074386597, "step": 1 }, { "epoch": 0.0005192107995846313, "grad_norm": 33.08597183227539, "learning_rate": 2.1551724137931035e-08, "loss": 0.3634331226348877, "step": 2 }, { "epoch": 0.000778816199376947, "grad_norm": 36.19953918457031, "learning_rate": 4.310344827586207e-08, "loss": 0.37143877148628235, "step": 3 }, { "epoch": 0.0010384215991692627, "grad_norm": 26.900283813476562, "learning_rate": 6.465517241379311e-08, "loss": 0.30546796321868896, "step": 4 }, { "epoch": 0.0012980269989615785, "grad_norm": 35.835968017578125, "learning_rate": 8.620689655172414e-08, "loss": 0.38070282340049744, "step": 5 }, { "epoch": 0.001557632398753894, "grad_norm": 30.058963775634766, "learning_rate": 1.0775862068965518e-07, "loss": 0.34866654872894287, "step": 6 }, { "epoch": 0.0018172377985462098, "grad_norm": 38.288238525390625, "learning_rate": 1.2931034482758622e-07, "loss": 0.39639484882354736, "step": 7 }, { "epoch": 0.0020768431983385254, "grad_norm": 24.725461959838867, "learning_rate": 1.5086206896551727e-07, "loss": 0.29217034578323364, "step": 8 }, { "epoch": 0.002336448598130841, "grad_norm": 24.87229347229004, "learning_rate": 1.7241379310344828e-07, "loss": 0.2957771420478821, "step": 9 }, { "epoch": 0.002596053997923157, "grad_norm": 23.54578971862793, "learning_rate": 1.9396551724137932e-07, "loss": 0.3178044557571411, "step": 10 }, { "epoch": 0.0028556593977154725, "grad_norm": 23.147645950317383, "learning_rate": 2.1551724137931036e-07, "loss": 0.33592915534973145, "step": 11 }, { "epoch": 0.003115264797507788, "grad_norm": 25.183185577392578, "learning_rate": 2.3706896551724138e-07, "loss": 0.34157222509384155, "step": 12 }, { "epoch": 0.0033748701973001036, "grad_norm": 20.62165069580078, "learning_rate": 2.5862068965517245e-07, "loss": 0.29687637090682983, "step": 13 }, { "epoch": 0.0036344755970924196, "grad_norm": 17.411287307739258, "learning_rate": 2.801724137931035e-07, "loss": 0.2664971351623535, "step": 14 }, { "epoch": 0.003894080996884735, "grad_norm": 14.045934677124023, "learning_rate": 3.0172413793103453e-07, "loss": 0.32109129428863525, "step": 15 }, { "epoch": 0.004153686396677051, "grad_norm": 16.288734436035156, "learning_rate": 3.2327586206896557e-07, "loss": 0.2534257769584656, "step": 16 }, { "epoch": 0.004413291796469367, "grad_norm": 25.617290496826172, "learning_rate": 3.4482758620689656e-07, "loss": 0.33992597460746765, "step": 17 }, { "epoch": 0.004672897196261682, "grad_norm": 25.657901763916016, "learning_rate": 3.663793103448276e-07, "loss": 0.33357465267181396, "step": 18 }, { "epoch": 0.004932502596053998, "grad_norm": 29.105175018310547, "learning_rate": 3.8793103448275865e-07, "loss": 0.33323654532432556, "step": 19 }, { "epoch": 0.005192107995846314, "grad_norm": 15.23852252960205, "learning_rate": 4.094827586206897e-07, "loss": 0.2474779188632965, "step": 20 }, { "epoch": 0.005451713395638629, "grad_norm": 12.253477096557617, "learning_rate": 4.3103448275862073e-07, "loss": 0.23435838520526886, "step": 21 }, { "epoch": 0.005711318795430945, "grad_norm": 17.401926040649414, "learning_rate": 4.5258620689655177e-07, "loss": 0.2564501464366913, "step": 22 }, { "epoch": 0.005970924195223261, "grad_norm": 24.22475814819336, "learning_rate": 4.7413793103448276e-07, "loss": 0.30346331000328064, "step": 23 }, { "epoch": 0.006230529595015576, "grad_norm": 11.473700523376465, "learning_rate": 4.956896551724138e-07, "loss": 0.2441878765821457, "step": 24 }, { "epoch": 0.006490134994807892, "grad_norm": 13.34585189819336, "learning_rate": 5.172413793103449e-07, "loss": 0.2795218229293823, "step": 25 }, { "epoch": 0.006749740394600207, "grad_norm": 10.188233375549316, "learning_rate": 5.387931034482759e-07, "loss": 0.2651021480560303, "step": 26 }, { "epoch": 0.007009345794392523, "grad_norm": 11.684386253356934, "learning_rate": 5.60344827586207e-07, "loss": 0.2458612322807312, "step": 27 }, { "epoch": 0.007268951194184839, "grad_norm": 6.333677768707275, "learning_rate": 5.81896551724138e-07, "loss": 0.2382061779499054, "step": 28 }, { "epoch": 0.007528556593977154, "grad_norm": 7.899011611938477, "learning_rate": 6.034482758620691e-07, "loss": 0.2804311513900757, "step": 29 }, { "epoch": 0.00778816199376947, "grad_norm": 12.176069259643555, "learning_rate": 6.25e-07, "loss": 0.28721535205841064, "step": 30 }, { "epoch": 0.008047767393561786, "grad_norm": 14.507472038269043, "learning_rate": 6.465517241379311e-07, "loss": 0.29816022515296936, "step": 31 }, { "epoch": 0.008307372793354102, "grad_norm": 14.998342514038086, "learning_rate": 6.68103448275862e-07, "loss": 0.24752050638198853, "step": 32 }, { "epoch": 0.008566978193146417, "grad_norm": 13.286417007446289, "learning_rate": 6.896551724137931e-07, "loss": 0.22487664222717285, "step": 33 }, { "epoch": 0.008826583592938734, "grad_norm": 12.095267295837402, "learning_rate": 7.112068965517242e-07, "loss": 0.21072927117347717, "step": 34 }, { "epoch": 0.009086188992731049, "grad_norm": 11.955726623535156, "learning_rate": 7.327586206896552e-07, "loss": 0.2464357614517212, "step": 35 }, { "epoch": 0.009345794392523364, "grad_norm": 7.97135591506958, "learning_rate": 7.543103448275863e-07, "loss": 0.2713427245616913, "step": 36 }, { "epoch": 0.00960539979231568, "grad_norm": 7.150877952575684, "learning_rate": 7.758620689655173e-07, "loss": 0.26975858211517334, "step": 37 }, { "epoch": 0.009865005192107996, "grad_norm": 6.836562633514404, "learning_rate": 7.974137931034484e-07, "loss": 0.2421075850725174, "step": 38 }, { "epoch": 0.010124610591900311, "grad_norm": 11.348214149475098, "learning_rate": 8.189655172413794e-07, "loss": 0.277168333530426, "step": 39 }, { "epoch": 0.010384215991692628, "grad_norm": 8.353156089782715, "learning_rate": 8.405172413793105e-07, "loss": 0.20149089395999908, "step": 40 }, { "epoch": 0.010643821391484943, "grad_norm": 18.90726089477539, "learning_rate": 8.620689655172415e-07, "loss": 0.24545857310295105, "step": 41 }, { "epoch": 0.010903426791277258, "grad_norm": 9.116273880004883, "learning_rate": 8.836206896551724e-07, "loss": 0.23614321649074554, "step": 42 }, { "epoch": 0.011163032191069575, "grad_norm": 14.686635971069336, "learning_rate": 9.051724137931035e-07, "loss": 0.2769784927368164, "step": 43 }, { "epoch": 0.01142263759086189, "grad_norm": 5.20591926574707, "learning_rate": 9.267241379310346e-07, "loss": 0.21059194207191467, "step": 44 }, { "epoch": 0.011682242990654205, "grad_norm": 6.22241735458374, "learning_rate": 9.482758620689655e-07, "loss": 0.1997520923614502, "step": 45 }, { "epoch": 0.011941848390446522, "grad_norm": 4.934725284576416, "learning_rate": 9.698275862068966e-07, "loss": 0.21309278905391693, "step": 46 }, { "epoch": 0.012201453790238837, "grad_norm": 14.327861785888672, "learning_rate": 9.913793103448276e-07, "loss": 0.2608850300312042, "step": 47 }, { "epoch": 0.012461059190031152, "grad_norm": 9.705903053283691, "learning_rate": 1.0129310344827586e-06, "loss": 0.2267390638589859, "step": 48 }, { "epoch": 0.01272066458982347, "grad_norm": 9.720457077026367, "learning_rate": 1.0344827586206898e-06, "loss": 0.24654261767864227, "step": 49 }, { "epoch": 0.012980269989615784, "grad_norm": 6.311122417449951, "learning_rate": 1.0560344827586208e-06, "loss": 0.19172322750091553, "step": 50 }, { "epoch": 0.0132398753894081, "grad_norm": 6.511159420013428, "learning_rate": 1.0775862068965518e-06, "loss": 0.196778804063797, "step": 51 }, { "epoch": 0.013499480789200415, "grad_norm": 5.09488582611084, "learning_rate": 1.0991379310344828e-06, "loss": 0.1842140257358551, "step": 52 }, { "epoch": 0.013759086188992731, "grad_norm": 7.950708389282227, "learning_rate": 1.120689655172414e-06, "loss": 0.1980053037405014, "step": 53 }, { "epoch": 0.014018691588785047, "grad_norm": 10.244484901428223, "learning_rate": 1.142241379310345e-06, "loss": 0.24652954936027527, "step": 54 }, { "epoch": 0.014278296988577362, "grad_norm": 12.767365455627441, "learning_rate": 1.163793103448276e-06, "loss": 0.24263547360897064, "step": 55 }, { "epoch": 0.014537902388369679, "grad_norm": 5.574886322021484, "learning_rate": 1.185344827586207e-06, "loss": 0.21138687431812286, "step": 56 }, { "epoch": 0.014797507788161994, "grad_norm": 7.189504623413086, "learning_rate": 1.2068965517241381e-06, "loss": 0.2548171281814575, "step": 57 }, { "epoch": 0.015057113187954309, "grad_norm": 10.101699829101562, "learning_rate": 1.2284482758620691e-06, "loss": 0.2684274911880493, "step": 58 }, { "epoch": 0.015316718587746626, "grad_norm": 6.288949489593506, "learning_rate": 1.25e-06, "loss": 0.17961880564689636, "step": 59 }, { "epoch": 0.01557632398753894, "grad_norm": 10.175830841064453, "learning_rate": 1.271551724137931e-06, "loss": 0.21740397810935974, "step": 60 }, { "epoch": 0.015835929387331256, "grad_norm": 10.871426582336426, "learning_rate": 1.2931034482758623e-06, "loss": 0.23830696940422058, "step": 61 }, { "epoch": 0.016095534787123573, "grad_norm": 3.6401515007019043, "learning_rate": 1.3146551724137933e-06, "loss": 0.17544765770435333, "step": 62 }, { "epoch": 0.016355140186915886, "grad_norm": 4.350811958312988, "learning_rate": 1.336206896551724e-06, "loss": 0.20826268196105957, "step": 63 }, { "epoch": 0.016614745586708203, "grad_norm": 7.23914909362793, "learning_rate": 1.3577586206896553e-06, "loss": 0.23236125707626343, "step": 64 }, { "epoch": 0.01687435098650052, "grad_norm": 9.075438499450684, "learning_rate": 1.3793103448275862e-06, "loss": 0.23794904351234436, "step": 65 }, { "epoch": 0.017133956386292833, "grad_norm": 7.008607387542725, "learning_rate": 1.4008620689655172e-06, "loss": 0.18524351716041565, "step": 66 }, { "epoch": 0.01739356178608515, "grad_norm": 5.1376214027404785, "learning_rate": 1.4224137931034484e-06, "loss": 0.19980892539024353, "step": 67 }, { "epoch": 0.017653167185877467, "grad_norm": 4.3825154304504395, "learning_rate": 1.4439655172413794e-06, "loss": 0.2508780360221863, "step": 68 }, { "epoch": 0.01791277258566978, "grad_norm": 3.412696599960327, "learning_rate": 1.4655172413793104e-06, "loss": 0.1794978678226471, "step": 69 }, { "epoch": 0.018172377985462097, "grad_norm": 5.152624607086182, "learning_rate": 1.4870689655172416e-06, "loss": 0.18680207431316376, "step": 70 }, { "epoch": 0.018431983385254414, "grad_norm": 5.56689977645874, "learning_rate": 1.5086206896551726e-06, "loss": 0.17662358283996582, "step": 71 }, { "epoch": 0.018691588785046728, "grad_norm": 2.959162473678589, "learning_rate": 1.5301724137931034e-06, "loss": 0.1827951967716217, "step": 72 }, { "epoch": 0.018951194184839044, "grad_norm": 7.402022838592529, "learning_rate": 1.5517241379310346e-06, "loss": 0.21049058437347412, "step": 73 }, { "epoch": 0.01921079958463136, "grad_norm": 6.92476749420166, "learning_rate": 1.5732758620689656e-06, "loss": 0.22278863191604614, "step": 74 }, { "epoch": 0.019470404984423675, "grad_norm": 3.4119603633880615, "learning_rate": 1.5948275862068968e-06, "loss": 0.17251047492027283, "step": 75 }, { "epoch": 0.01973001038421599, "grad_norm": 3.5850000381469727, "learning_rate": 1.6163793103448278e-06, "loss": 0.18299147486686707, "step": 76 }, { "epoch": 0.01998961578400831, "grad_norm": 7.329703330993652, "learning_rate": 1.6379310344827587e-06, "loss": 0.2308977246284485, "step": 77 }, { "epoch": 0.020249221183800622, "grad_norm": 5.087424278259277, "learning_rate": 1.65948275862069e-06, "loss": 0.18340125679969788, "step": 78 }, { "epoch": 0.02050882658359294, "grad_norm": 3.5880308151245117, "learning_rate": 1.681034482758621e-06, "loss": 0.22179841995239258, "step": 79 }, { "epoch": 0.020768431983385256, "grad_norm": 7.20912504196167, "learning_rate": 1.7025862068965517e-06, "loss": 0.22962023317813873, "step": 80 }, { "epoch": 0.02102803738317757, "grad_norm": 6.147843837738037, "learning_rate": 1.724137931034483e-06, "loss": 0.17625057697296143, "step": 81 }, { "epoch": 0.021287642782969886, "grad_norm": 3.6094930171966553, "learning_rate": 1.745689655172414e-06, "loss": 0.17558670043945312, "step": 82 }, { "epoch": 0.021547248182762203, "grad_norm": 6.853363037109375, "learning_rate": 1.7672413793103449e-06, "loss": 0.22858379781246185, "step": 83 }, { "epoch": 0.021806853582554516, "grad_norm": 5.862078666687012, "learning_rate": 1.788793103448276e-06, "loss": 0.2117193192243576, "step": 84 }, { "epoch": 0.022066458982346833, "grad_norm": 4.279626369476318, "learning_rate": 1.810344827586207e-06, "loss": 0.21613912284374237, "step": 85 }, { "epoch": 0.02232606438213915, "grad_norm": 5.088603496551514, "learning_rate": 1.831896551724138e-06, "loss": 0.22461816668510437, "step": 86 }, { "epoch": 0.022585669781931463, "grad_norm": 5.279255390167236, "learning_rate": 1.8534482758620693e-06, "loss": 0.19398245215415955, "step": 87 }, { "epoch": 0.02284527518172378, "grad_norm": 5.398169994354248, "learning_rate": 1.8750000000000003e-06, "loss": 0.2248784601688385, "step": 88 }, { "epoch": 0.023104880581516097, "grad_norm": 3.2151293754577637, "learning_rate": 1.896551724137931e-06, "loss": 0.17585435509681702, "step": 89 }, { "epoch": 0.02336448598130841, "grad_norm": 4.543318748474121, "learning_rate": 1.9181034482758622e-06, "loss": 0.1920384168624878, "step": 90 }, { "epoch": 0.023624091381100727, "grad_norm": 7.454247951507568, "learning_rate": 1.9396551724137932e-06, "loss": 0.20781096816062927, "step": 91 }, { "epoch": 0.023883696780893044, "grad_norm": 3.640583038330078, "learning_rate": 1.9612068965517242e-06, "loss": 0.20872822403907776, "step": 92 }, { "epoch": 0.024143302180685357, "grad_norm": 3.3529255390167236, "learning_rate": 1.982758620689655e-06, "loss": 0.16610077023506165, "step": 93 }, { "epoch": 0.024402907580477674, "grad_norm": 12.344358444213867, "learning_rate": 2.004310344827586e-06, "loss": 0.18674594163894653, "step": 94 }, { "epoch": 0.02466251298026999, "grad_norm": 2.99529767036438, "learning_rate": 2.025862068965517e-06, "loss": 0.17005591094493866, "step": 95 }, { "epoch": 0.024922118380062305, "grad_norm": 8.363637924194336, "learning_rate": 2.0474137931034486e-06, "loss": 0.17141121625900269, "step": 96 }, { "epoch": 0.02518172377985462, "grad_norm": 9.401901245117188, "learning_rate": 2.0689655172413796e-06, "loss": 0.18854010105133057, "step": 97 }, { "epoch": 0.02544132917964694, "grad_norm": 5.83868408203125, "learning_rate": 2.0905172413793106e-06, "loss": 0.20968753099441528, "step": 98 }, { "epoch": 0.02570093457943925, "grad_norm": 7.986198425292969, "learning_rate": 2.1120689655172416e-06, "loss": 0.17470046877861023, "step": 99 }, { "epoch": 0.02596053997923157, "grad_norm": 8.175525665283203, "learning_rate": 2.1336206896551725e-06, "loss": 0.1747843325138092, "step": 100 }, { "epoch": 0.026220145379023882, "grad_norm": 4.830190181732178, "learning_rate": 2.1551724137931035e-06, "loss": 0.17286519706249237, "step": 101 }, { "epoch": 0.0264797507788162, "grad_norm": 7.515536308288574, "learning_rate": 2.1767241379310345e-06, "loss": 0.20090913772583008, "step": 102 }, { "epoch": 0.026739356178608516, "grad_norm": 6.979458332061768, "learning_rate": 2.1982758620689655e-06, "loss": 0.1923002153635025, "step": 103 }, { "epoch": 0.02699896157840083, "grad_norm": 4.671876430511475, "learning_rate": 2.219827586206897e-06, "loss": 0.16815975308418274, "step": 104 }, { "epoch": 0.027258566978193146, "grad_norm": 8.002513885498047, "learning_rate": 2.241379310344828e-06, "loss": 0.2158738076686859, "step": 105 }, { "epoch": 0.027518172377985463, "grad_norm": 9.73463249206543, "learning_rate": 2.262931034482759e-06, "loss": 0.2026848942041397, "step": 106 }, { "epoch": 0.027777777777777776, "grad_norm": 3.476544141769409, "learning_rate": 2.28448275862069e-06, "loss": 0.16349993646144867, "step": 107 }, { "epoch": 0.028037383177570093, "grad_norm": 3.524379253387451, "learning_rate": 2.306034482758621e-06, "loss": 0.1560702621936798, "step": 108 }, { "epoch": 0.02829698857736241, "grad_norm": 4.540988922119141, "learning_rate": 2.327586206896552e-06, "loss": 0.168179452419281, "step": 109 }, { "epoch": 0.028556593977154723, "grad_norm": 4.535221576690674, "learning_rate": 2.349137931034483e-06, "loss": 0.23165851831436157, "step": 110 }, { "epoch": 0.02881619937694704, "grad_norm": 7.782125473022461, "learning_rate": 2.370689655172414e-06, "loss": 0.23734615743160248, "step": 111 }, { "epoch": 0.029075804776739357, "grad_norm": 4.362497329711914, "learning_rate": 2.392241379310345e-06, "loss": 0.20048323273658752, "step": 112 }, { "epoch": 0.02933541017653167, "grad_norm": 4.9733781814575195, "learning_rate": 2.4137931034482762e-06, "loss": 0.21510255336761475, "step": 113 }, { "epoch": 0.029595015576323987, "grad_norm": 2.249379873275757, "learning_rate": 2.4353448275862072e-06, "loss": 0.17332340776920319, "step": 114 }, { "epoch": 0.029854620976116304, "grad_norm": 2.5672945976257324, "learning_rate": 2.4568965517241382e-06, "loss": 0.18745915591716766, "step": 115 }, { "epoch": 0.030114226375908618, "grad_norm": 2.9369590282440186, "learning_rate": 2.4784482758620692e-06, "loss": 0.2024538815021515, "step": 116 }, { "epoch": 0.030373831775700934, "grad_norm": 3.334728956222534, "learning_rate": 2.5e-06, "loss": 0.16009104251861572, "step": 117 }, { "epoch": 0.03063343717549325, "grad_norm": 3.590211868286133, "learning_rate": 2.521551724137931e-06, "loss": 0.21161338686943054, "step": 118 }, { "epoch": 0.030893042575285565, "grad_norm": 4.292784214019775, "learning_rate": 2.543103448275862e-06, "loss": 0.19962286949157715, "step": 119 }, { "epoch": 0.03115264797507788, "grad_norm": 5.388044834136963, "learning_rate": 2.5646551724137936e-06, "loss": 0.15654881298542023, "step": 120 }, { "epoch": 0.0314122533748702, "grad_norm": 3.964723825454712, "learning_rate": 2.5862068965517246e-06, "loss": 0.22150859236717224, "step": 121 }, { "epoch": 0.03167185877466251, "grad_norm": 6.57093620300293, "learning_rate": 2.6077586206896556e-06, "loss": 0.18162310123443604, "step": 122 }, { "epoch": 0.031931464174454825, "grad_norm": 6.371257305145264, "learning_rate": 2.6293103448275866e-06, "loss": 0.16692183911800385, "step": 123 }, { "epoch": 0.032191069574247146, "grad_norm": 2.5801219940185547, "learning_rate": 2.6508620689655175e-06, "loss": 0.17248718440532684, "step": 124 }, { "epoch": 0.03245067497403946, "grad_norm": 5.74022912979126, "learning_rate": 2.672413793103448e-06, "loss": 0.22750148177146912, "step": 125 }, { "epoch": 0.03271028037383177, "grad_norm": 8.321017265319824, "learning_rate": 2.69396551724138e-06, "loss": 0.16305723786354065, "step": 126 }, { "epoch": 0.03296988577362409, "grad_norm": 3.894865036010742, "learning_rate": 2.7155172413793105e-06, "loss": 0.19587905704975128, "step": 127 }, { "epoch": 0.033229491173416406, "grad_norm": 6.304594039916992, "learning_rate": 2.7370689655172415e-06, "loss": 0.17772074043750763, "step": 128 }, { "epoch": 0.03348909657320872, "grad_norm": 6.391052722930908, "learning_rate": 2.7586206896551725e-06, "loss": 0.19482538104057312, "step": 129 }, { "epoch": 0.03374870197300104, "grad_norm": 2.6514196395874023, "learning_rate": 2.7801724137931035e-06, "loss": 0.17343385517597198, "step": 130 }, { "epoch": 0.03400830737279335, "grad_norm": 2.9819414615631104, "learning_rate": 2.8017241379310345e-06, "loss": 0.1600264012813568, "step": 131 }, { "epoch": 0.03426791277258567, "grad_norm": 3.645534038543701, "learning_rate": 2.823275862068966e-06, "loss": 0.1939956694841385, "step": 132 }, { "epoch": 0.03452751817237799, "grad_norm": 2.9662857055664062, "learning_rate": 2.844827586206897e-06, "loss": 0.18078738451004028, "step": 133 }, { "epoch": 0.0347871235721703, "grad_norm": 2.4623539447784424, "learning_rate": 2.866379310344828e-06, "loss": 0.16157013177871704, "step": 134 }, { "epoch": 0.035046728971962614, "grad_norm": 6.632139205932617, "learning_rate": 2.887931034482759e-06, "loss": 0.17674389481544495, "step": 135 }, { "epoch": 0.035306334371754934, "grad_norm": 5.614828109741211, "learning_rate": 2.90948275862069e-06, "loss": 0.20509997010231018, "step": 136 }, { "epoch": 0.03556593977154725, "grad_norm": 4.133851051330566, "learning_rate": 2.931034482758621e-06, "loss": 0.20856188237667084, "step": 137 }, { "epoch": 0.03582554517133956, "grad_norm": 1.9442671537399292, "learning_rate": 2.9525862068965522e-06, "loss": 0.16597259044647217, "step": 138 }, { "epoch": 0.03608515057113188, "grad_norm": 4.191993713378906, "learning_rate": 2.9741379310344832e-06, "loss": 0.1723909229040146, "step": 139 }, { "epoch": 0.036344755970924195, "grad_norm": 5.281088352203369, "learning_rate": 2.9956896551724142e-06, "loss": 0.20163168013095856, "step": 140 }, { "epoch": 0.03660436137071651, "grad_norm": 6.760980129241943, "learning_rate": 3.017241379310345e-06, "loss": 0.23958587646484375, "step": 141 }, { "epoch": 0.03686396677050883, "grad_norm": 3.5350446701049805, "learning_rate": 3.0387931034482758e-06, "loss": 0.18042194843292236, "step": 142 }, { "epoch": 0.03712357217030114, "grad_norm": 4.04268217086792, "learning_rate": 3.0603448275862068e-06, "loss": 0.16705310344696045, "step": 143 }, { "epoch": 0.037383177570093455, "grad_norm": 5.176905632019043, "learning_rate": 3.081896551724138e-06, "loss": 0.19060397148132324, "step": 144 }, { "epoch": 0.037642782969885775, "grad_norm": 6.104772567749023, "learning_rate": 3.103448275862069e-06, "loss": 0.1683473289012909, "step": 145 }, { "epoch": 0.03790238836967809, "grad_norm": 2.602423667907715, "learning_rate": 3.125e-06, "loss": 0.16104543209075928, "step": 146 }, { "epoch": 0.0381619937694704, "grad_norm": 7.914226055145264, "learning_rate": 3.146551724137931e-06, "loss": 0.2229699194431305, "step": 147 }, { "epoch": 0.03842159916926272, "grad_norm": 2.9940102100372314, "learning_rate": 3.168103448275862e-06, "loss": 0.21602454781532288, "step": 148 }, { "epoch": 0.038681204569055036, "grad_norm": 5.908970355987549, "learning_rate": 3.1896551724137935e-06, "loss": 0.15488679707050323, "step": 149 }, { "epoch": 0.03894080996884735, "grad_norm": 3.6412980556488037, "learning_rate": 3.2112068965517245e-06, "loss": 0.1718800961971283, "step": 150 }, { "epoch": 0.03920041536863967, "grad_norm": 5.638095378875732, "learning_rate": 3.2327586206896555e-06, "loss": 0.1771140992641449, "step": 151 }, { "epoch": 0.03946002076843198, "grad_norm": 5.147150993347168, "learning_rate": 3.2543103448275865e-06, "loss": 0.21349367499351501, "step": 152 }, { "epoch": 0.0397196261682243, "grad_norm": 6.994207859039307, "learning_rate": 3.2758620689655175e-06, "loss": 0.1628064215183258, "step": 153 }, { "epoch": 0.03997923156801662, "grad_norm": 3.5203306674957275, "learning_rate": 3.2974137931034485e-06, "loss": 0.1514279544353485, "step": 154 }, { "epoch": 0.04023883696780893, "grad_norm": 6.057727813720703, "learning_rate": 3.31896551724138e-06, "loss": 0.16256004571914673, "step": 155 }, { "epoch": 0.040498442367601244, "grad_norm": 1.9922481775283813, "learning_rate": 3.340517241379311e-06, "loss": 0.16324645280838013, "step": 156 }, { "epoch": 0.040758047767393564, "grad_norm": 5.4350666999816895, "learning_rate": 3.362068965517242e-06, "loss": 0.18523657321929932, "step": 157 }, { "epoch": 0.04101765316718588, "grad_norm": 7.959524631500244, "learning_rate": 3.383620689655173e-06, "loss": 0.2029598355293274, "step": 158 }, { "epoch": 0.04127725856697819, "grad_norm": 4.2450480461120605, "learning_rate": 3.4051724137931034e-06, "loss": 0.1590932011604309, "step": 159 }, { "epoch": 0.04153686396677051, "grad_norm": 9.66146469116211, "learning_rate": 3.4267241379310344e-06, "loss": 0.17766794562339783, "step": 160 }, { "epoch": 0.041796469366562825, "grad_norm": 2.830456256866455, "learning_rate": 3.448275862068966e-06, "loss": 0.14888204634189606, "step": 161 }, { "epoch": 0.04205607476635514, "grad_norm": 8.868423461914062, "learning_rate": 3.469827586206897e-06, "loss": 0.20059752464294434, "step": 162 }, { "epoch": 0.04231568016614746, "grad_norm": 9.148317337036133, "learning_rate": 3.491379310344828e-06, "loss": 0.1764640212059021, "step": 163 }, { "epoch": 0.04257528556593977, "grad_norm": 1.766520619392395, "learning_rate": 3.512931034482759e-06, "loss": 0.1651400774717331, "step": 164 }, { "epoch": 0.042834890965732085, "grad_norm": 13.98501205444336, "learning_rate": 3.5344827586206898e-06, "loss": 0.24191120266914368, "step": 165 }, { "epoch": 0.043094496365524405, "grad_norm": 2.593844413757324, "learning_rate": 3.5560344827586208e-06, "loss": 0.18691739439964294, "step": 166 }, { "epoch": 0.04335410176531672, "grad_norm": 6.090912342071533, "learning_rate": 3.577586206896552e-06, "loss": 0.18826225399971008, "step": 167 }, { "epoch": 0.04361370716510903, "grad_norm": 6.419212818145752, "learning_rate": 3.599137931034483e-06, "loss": 0.20015421509742737, "step": 168 }, { "epoch": 0.04387331256490135, "grad_norm": 5.59576940536499, "learning_rate": 3.620689655172414e-06, "loss": 0.21465960144996643, "step": 169 }, { "epoch": 0.044132917964693666, "grad_norm": 5.701845169067383, "learning_rate": 3.642241379310345e-06, "loss": 0.22056584060192108, "step": 170 }, { "epoch": 0.04439252336448598, "grad_norm": 4.359399795532227, "learning_rate": 3.663793103448276e-06, "loss": 0.16945557296276093, "step": 171 }, { "epoch": 0.0446521287642783, "grad_norm": 8.722148895263672, "learning_rate": 3.685344827586207e-06, "loss": 0.23970764875411987, "step": 172 }, { "epoch": 0.04491173416407061, "grad_norm": 4.0208330154418945, "learning_rate": 3.7068965517241385e-06, "loss": 0.17236192524433136, "step": 173 }, { "epoch": 0.045171339563862926, "grad_norm": 10.16695499420166, "learning_rate": 3.7284482758620695e-06, "loss": 0.178751140832901, "step": 174 }, { "epoch": 0.04543094496365525, "grad_norm": 6.479930400848389, "learning_rate": 3.7500000000000005e-06, "loss": 0.17964544892311096, "step": 175 }, { "epoch": 0.04569055036344756, "grad_norm": 4.562987804412842, "learning_rate": 3.771551724137931e-06, "loss": 0.15014541149139404, "step": 176 }, { "epoch": 0.045950155763239874, "grad_norm": 8.790274620056152, "learning_rate": 3.793103448275862e-06, "loss": 0.21400389075279236, "step": 177 }, { "epoch": 0.046209761163032194, "grad_norm": 3.584289073944092, "learning_rate": 3.8146551724137935e-06, "loss": 0.15767940878868103, "step": 178 }, { "epoch": 0.04646936656282451, "grad_norm": 3.841867685317993, "learning_rate": 3.8362068965517245e-06, "loss": 0.17452183365821838, "step": 179 }, { "epoch": 0.04672897196261682, "grad_norm": 5.340777397155762, "learning_rate": 3.857758620689656e-06, "loss": 0.1904277205467224, "step": 180 }, { "epoch": 0.04698857736240914, "grad_norm": 6.333534240722656, "learning_rate": 3.8793103448275865e-06, "loss": 0.2292891889810562, "step": 181 }, { "epoch": 0.047248182762201454, "grad_norm": 2.588327646255493, "learning_rate": 3.900862068965518e-06, "loss": 0.16129066050052643, "step": 182 }, { "epoch": 0.04750778816199377, "grad_norm": 5.2229228019714355, "learning_rate": 3.9224137931034484e-06, "loss": 0.18930277228355408, "step": 183 }, { "epoch": 0.04776739356178609, "grad_norm": 5.499538898468018, "learning_rate": 3.94396551724138e-06, "loss": 0.1590391993522644, "step": 184 }, { "epoch": 0.0480269989615784, "grad_norm": 3.071500301361084, "learning_rate": 3.96551724137931e-06, "loss": 0.18527376651763916, "step": 185 }, { "epoch": 0.048286604361370715, "grad_norm": 5.6555047035217285, "learning_rate": 3.987068965517242e-06, "loss": 0.15650302171707153, "step": 186 }, { "epoch": 0.048546209761163035, "grad_norm": 3.650153875350952, "learning_rate": 4.008620689655172e-06, "loss": 0.1465233415365219, "step": 187 }, { "epoch": 0.04880581516095535, "grad_norm": 4.10042142868042, "learning_rate": 4.030172413793104e-06, "loss": 0.16121768951416016, "step": 188 }, { "epoch": 0.04906542056074766, "grad_norm": 3.1300861835479736, "learning_rate": 4.051724137931034e-06, "loss": 0.15226449072360992, "step": 189 }, { "epoch": 0.04932502596053998, "grad_norm": 2.457634449005127, "learning_rate": 4.073275862068966e-06, "loss": 0.1605234146118164, "step": 190 }, { "epoch": 0.049584631360332296, "grad_norm": 2.3399291038513184, "learning_rate": 4.094827586206897e-06, "loss": 0.1424296796321869, "step": 191 }, { "epoch": 0.04984423676012461, "grad_norm": 6.644623279571533, "learning_rate": 4.116379310344828e-06, "loss": 0.20630499720573425, "step": 192 }, { "epoch": 0.05010384215991693, "grad_norm": 5.842675685882568, "learning_rate": 4.137931034482759e-06, "loss": 0.1799265593290329, "step": 193 }, { "epoch": 0.05036344755970924, "grad_norm": 2.728978157043457, "learning_rate": 4.15948275862069e-06, "loss": 0.17563268542289734, "step": 194 }, { "epoch": 0.050623052959501556, "grad_norm": 2.7977640628814697, "learning_rate": 4.181034482758621e-06, "loss": 0.16256183385849, "step": 195 }, { "epoch": 0.05088265835929388, "grad_norm": 3.889425754547119, "learning_rate": 4.2025862068965526e-06, "loss": 0.19589287042617798, "step": 196 }, { "epoch": 0.05114226375908619, "grad_norm": 3.5422937870025635, "learning_rate": 4.224137931034483e-06, "loss": 0.17377299070358276, "step": 197 }, { "epoch": 0.0514018691588785, "grad_norm": 1.9122322797775269, "learning_rate": 4.2456896551724145e-06, "loss": 0.15801164507865906, "step": 198 }, { "epoch": 0.051661474558670824, "grad_norm": 2.813732147216797, "learning_rate": 4.267241379310345e-06, "loss": 0.18961195647716522, "step": 199 }, { "epoch": 0.05192107995846314, "grad_norm": 3.6777029037475586, "learning_rate": 4.288793103448276e-06, "loss": 0.21118327975273132, "step": 200 }, { "epoch": 0.05218068535825545, "grad_norm": 2.3781869411468506, "learning_rate": 4.310344827586207e-06, "loss": 0.18215464055538177, "step": 201 }, { "epoch": 0.052440290758047764, "grad_norm": 2.799623727798462, "learning_rate": 4.3318965517241385e-06, "loss": 0.17887458205223083, "step": 202 }, { "epoch": 0.052699896157840084, "grad_norm": 2.048494338989258, "learning_rate": 4.353448275862069e-06, "loss": 0.17874237895011902, "step": 203 }, { "epoch": 0.0529595015576324, "grad_norm": 3.4896762371063232, "learning_rate": 4.3750000000000005e-06, "loss": 0.16831515729427338, "step": 204 }, { "epoch": 0.05321910695742471, "grad_norm": 3.402509927749634, "learning_rate": 4.396551724137931e-06, "loss": 0.2068026065826416, "step": 205 }, { "epoch": 0.05347871235721703, "grad_norm": 4.084611892700195, "learning_rate": 4.4181034482758624e-06, "loss": 0.15354010462760925, "step": 206 }, { "epoch": 0.053738317757009345, "grad_norm": 2.806406259536743, "learning_rate": 4.439655172413794e-06, "loss": 0.20589181780815125, "step": 207 }, { "epoch": 0.05399792315680166, "grad_norm": 1.9871960878372192, "learning_rate": 4.461206896551724e-06, "loss": 0.18320158123970032, "step": 208 }, { "epoch": 0.05425752855659398, "grad_norm": 7.132951736450195, "learning_rate": 4.482758620689656e-06, "loss": 0.1941656470298767, "step": 209 }, { "epoch": 0.05451713395638629, "grad_norm": 1.6523672342300415, "learning_rate": 4.504310344827586e-06, "loss": 0.15862713754177094, "step": 210 }, { "epoch": 0.054776739356178605, "grad_norm": 9.519768714904785, "learning_rate": 4.525862068965518e-06, "loss": 0.22071096301078796, "step": 211 }, { "epoch": 0.055036344755970926, "grad_norm": 3.391305923461914, "learning_rate": 4.547413793103448e-06, "loss": 0.18698054552078247, "step": 212 }, { "epoch": 0.05529595015576324, "grad_norm": 3.0141208171844482, "learning_rate": 4.56896551724138e-06, "loss": 0.18300002813339233, "step": 213 }, { "epoch": 0.05555555555555555, "grad_norm": 6.484405994415283, "learning_rate": 4.590517241379311e-06, "loss": 0.15994206070899963, "step": 214 }, { "epoch": 0.05581516095534787, "grad_norm": 4.151312828063965, "learning_rate": 4.612068965517242e-06, "loss": 0.17187371850013733, "step": 215 }, { "epoch": 0.056074766355140186, "grad_norm": 3.968634605407715, "learning_rate": 4.633620689655173e-06, "loss": 0.18580839037895203, "step": 216 }, { "epoch": 0.0563343717549325, "grad_norm": 2.813121795654297, "learning_rate": 4.655172413793104e-06, "loss": 0.15126726031303406, "step": 217 }, { "epoch": 0.05659397715472482, "grad_norm": 3.7420594692230225, "learning_rate": 4.676724137931034e-06, "loss": 0.15505164861679077, "step": 218 }, { "epoch": 0.05685358255451713, "grad_norm": 3.699432611465454, "learning_rate": 4.698275862068966e-06, "loss": 0.24055396020412445, "step": 219 }, { "epoch": 0.05711318795430945, "grad_norm": 3.461106061935425, "learning_rate": 4.719827586206897e-06, "loss": 0.1761658489704132, "step": 220 }, { "epoch": 0.05737279335410177, "grad_norm": 2.480583906173706, "learning_rate": 4.741379310344828e-06, "loss": 0.14994078874588013, "step": 221 }, { "epoch": 0.05763239875389408, "grad_norm": 1.6046788692474365, "learning_rate": 4.762931034482759e-06, "loss": 0.15188482403755188, "step": 222 }, { "epoch": 0.057892004153686394, "grad_norm": 3.2408876419067383, "learning_rate": 4.78448275862069e-06, "loss": 0.17881394922733307, "step": 223 }, { "epoch": 0.058151609553478714, "grad_norm": 4.270179748535156, "learning_rate": 4.806034482758621e-06, "loss": 0.1687372326850891, "step": 224 }, { "epoch": 0.05841121495327103, "grad_norm": 2.798147678375244, "learning_rate": 4.8275862068965525e-06, "loss": 0.16570065915584564, "step": 225 }, { "epoch": 0.05867082035306334, "grad_norm": 6.086611747741699, "learning_rate": 4.849137931034483e-06, "loss": 0.212249755859375, "step": 226 }, { "epoch": 0.05893042575285566, "grad_norm": 2.8216068744659424, "learning_rate": 4.8706896551724145e-06, "loss": 0.18051661550998688, "step": 227 }, { "epoch": 0.059190031152647975, "grad_norm": 5.830135822296143, "learning_rate": 4.892241379310345e-06, "loss": 0.18400394916534424, "step": 228 }, { "epoch": 0.05944963655244029, "grad_norm": 2.9246201515197754, "learning_rate": 4.9137931034482765e-06, "loss": 0.17532242834568024, "step": 229 }, { "epoch": 0.05970924195223261, "grad_norm": 3.2492096424102783, "learning_rate": 4.935344827586207e-06, "loss": 0.14892646670341492, "step": 230 }, { "epoch": 0.05996884735202492, "grad_norm": 6.0275397300720215, "learning_rate": 4.9568965517241384e-06, "loss": 0.14693573117256165, "step": 231 }, { "epoch": 0.060228452751817235, "grad_norm": 3.676851987838745, "learning_rate": 4.97844827586207e-06, "loss": 0.18537473678588867, "step": 232 }, { "epoch": 0.060488058151609556, "grad_norm": 9.805270195007324, "learning_rate": 5e-06, "loss": 0.19504375755786896, "step": 233 }, { "epoch": 0.06074766355140187, "grad_norm": 5.534234046936035, "learning_rate": 4.99999980112576e-06, "loss": 0.20208905637264252, "step": 234 }, { "epoch": 0.06100726895119418, "grad_norm": 5.432924270629883, "learning_rate": 4.999999204503072e-06, "loss": 0.1807359755039215, "step": 235 }, { "epoch": 0.0612668743509865, "grad_norm": 9.194381713867188, "learning_rate": 4.999998210132044e-06, "loss": 0.1791306734085083, "step": 236 }, { "epoch": 0.061526479750778816, "grad_norm": 6.60745906829834, "learning_rate": 4.999996818012849e-06, "loss": 0.19117704033851624, "step": 237 }, { "epoch": 0.06178608515057113, "grad_norm": 7.350625514984131, "learning_rate": 4.999995028145735e-06, "loss": 0.17696911096572876, "step": 238 }, { "epoch": 0.06204569055036345, "grad_norm": 8.633665084838867, "learning_rate": 4.999992840531019e-06, "loss": 0.23586943745613098, "step": 239 }, { "epoch": 0.06230529595015576, "grad_norm": 2.9915926456451416, "learning_rate": 4.999990255169086e-06, "loss": 0.18584705889225006, "step": 240 }, { "epoch": 0.06256490134994808, "grad_norm": 4.896827697753906, "learning_rate": 4.999987272060394e-06, "loss": 0.1856250762939453, "step": 241 }, { "epoch": 0.0628245067497404, "grad_norm": 5.643903732299805, "learning_rate": 4.999983891205471e-06, "loss": 0.1635006219148636, "step": 242 }, { "epoch": 0.0630841121495327, "grad_norm": 1.528049111366272, "learning_rate": 4.999980112604912e-06, "loss": 0.18491607904434204, "step": 243 }, { "epoch": 0.06334371754932502, "grad_norm": 5.3573079109191895, "learning_rate": 4.999975936259389e-06, "loss": 0.19272570312023163, "step": 244 }, { "epoch": 0.06360332294911734, "grad_norm": 3.2686572074890137, "learning_rate": 4.9999713621696375e-06, "loss": 0.19585445523262024, "step": 245 }, { "epoch": 0.06386292834890965, "grad_norm": 2.5445687770843506, "learning_rate": 4.999966390336466e-06, "loss": 0.16618525981903076, "step": 246 }, { "epoch": 0.06412253374870197, "grad_norm": 2.673389196395874, "learning_rate": 4.999961020760754e-06, "loss": 0.1856500208377838, "step": 247 }, { "epoch": 0.06438213914849429, "grad_norm": 5.08747673034668, "learning_rate": 4.999955253443451e-06, "loss": 0.17744220793247223, "step": 248 }, { "epoch": 0.0646417445482866, "grad_norm": 3.4631764888763428, "learning_rate": 4.999949088385577e-06, "loss": 0.21400876343250275, "step": 249 }, { "epoch": 0.06490134994807892, "grad_norm": 4.717981815338135, "learning_rate": 4.9999425255882205e-06, "loss": 0.20574966073036194, "step": 250 }, { "epoch": 0.06516095534787124, "grad_norm": 3.1709601879119873, "learning_rate": 4.999935565052543e-06, "loss": 0.1828537881374359, "step": 251 }, { "epoch": 0.06542056074766354, "grad_norm": 3.3796730041503906, "learning_rate": 4.999928206779774e-06, "loss": 0.1930818110704422, "step": 252 }, { "epoch": 0.06568016614745587, "grad_norm": 3.8881592750549316, "learning_rate": 4.999920450771214e-06, "loss": 0.1966610550880432, "step": 253 }, { "epoch": 0.06593977154724819, "grad_norm": 4.788054943084717, "learning_rate": 4.999912297028235e-06, "loss": 0.22244784235954285, "step": 254 }, { "epoch": 0.06619937694704049, "grad_norm": 2.9976963996887207, "learning_rate": 4.999903745552278e-06, "loss": 0.18524152040481567, "step": 255 }, { "epoch": 0.06645898234683281, "grad_norm": 4.986732482910156, "learning_rate": 4.9998947963448554e-06, "loss": 0.2150808572769165, "step": 256 }, { "epoch": 0.06671858774662513, "grad_norm": 2.5336241722106934, "learning_rate": 4.999885449407547e-06, "loss": 0.18663188815116882, "step": 257 }, { "epoch": 0.06697819314641744, "grad_norm": 5.567897319793701, "learning_rate": 4.999875704742009e-06, "loss": 0.1946118175983429, "step": 258 }, { "epoch": 0.06723779854620976, "grad_norm": 3.377563238143921, "learning_rate": 4.99986556234996e-06, "loss": 0.1868019700050354, "step": 259 }, { "epoch": 0.06749740394600208, "grad_norm": 9.775577545166016, "learning_rate": 4.999855022233195e-06, "loss": 0.18961545825004578, "step": 260 }, { "epoch": 0.06775700934579439, "grad_norm": 4.779046535491943, "learning_rate": 4.999844084393577e-06, "loss": 0.19130223989486694, "step": 261 }, { "epoch": 0.0680166147455867, "grad_norm": 14.29584789276123, "learning_rate": 4.99983274883304e-06, "loss": 0.24854326248168945, "step": 262 }, { "epoch": 0.06827622014537903, "grad_norm": 3.9892120361328125, "learning_rate": 4.999821015553586e-06, "loss": 0.15667863190174103, "step": 263 }, { "epoch": 0.06853582554517133, "grad_norm": 2.926456928253174, "learning_rate": 4.999808884557291e-06, "loss": 0.16000252962112427, "step": 264 }, { "epoch": 0.06879543094496365, "grad_norm": 5.123176097869873, "learning_rate": 4.9997963558463e-06, "loss": 0.18751877546310425, "step": 265 }, { "epoch": 0.06905503634475597, "grad_norm": 6.435518264770508, "learning_rate": 4.999783429422826e-06, "loss": 0.15903380513191223, "step": 266 }, { "epoch": 0.06931464174454828, "grad_norm": 2.6448402404785156, "learning_rate": 4.999770105289155e-06, "loss": 0.19634830951690674, "step": 267 }, { "epoch": 0.0695742471443406, "grad_norm": 4.890232086181641, "learning_rate": 4.999756383447642e-06, "loss": 0.19890347123146057, "step": 268 }, { "epoch": 0.06983385254413292, "grad_norm": 4.381546974182129, "learning_rate": 4.999742263900714e-06, "loss": 0.21375583112239838, "step": 269 }, { "epoch": 0.07009345794392523, "grad_norm": 3.737780809402466, "learning_rate": 4.9997277466508654e-06, "loss": 0.1474035680294037, "step": 270 }, { "epoch": 0.07035306334371755, "grad_norm": 5.606150150299072, "learning_rate": 4.9997128317006625e-06, "loss": 0.19879716634750366, "step": 271 }, { "epoch": 0.07061266874350987, "grad_norm": 3.247302532196045, "learning_rate": 4.999697519052743e-06, "loss": 0.16521042585372925, "step": 272 }, { "epoch": 0.07087227414330217, "grad_norm": 2.6324973106384277, "learning_rate": 4.999681808709814e-06, "loss": 0.20160385966300964, "step": 273 }, { "epoch": 0.0711318795430945, "grad_norm": 2.0625979900360107, "learning_rate": 4.999665700674651e-06, "loss": 0.19540265202522278, "step": 274 }, { "epoch": 0.07139148494288682, "grad_norm": 2.85563063621521, "learning_rate": 4.999649194950103e-06, "loss": 0.198085218667984, "step": 275 }, { "epoch": 0.07165109034267912, "grad_norm": 3.3202311992645264, "learning_rate": 4.999632291539088e-06, "loss": 0.17118288576602936, "step": 276 }, { "epoch": 0.07191069574247144, "grad_norm": 3.6365158557891846, "learning_rate": 4.999614990444594e-06, "loss": 0.1703692078590393, "step": 277 }, { "epoch": 0.07217030114226376, "grad_norm": 4.556649684906006, "learning_rate": 4.999597291669679e-06, "loss": 0.20715852081775665, "step": 278 }, { "epoch": 0.07242990654205607, "grad_norm": 1.2504528760910034, "learning_rate": 4.99957919521747e-06, "loss": 0.15755081176757812, "step": 279 }, { "epoch": 0.07268951194184839, "grad_norm": 3.571620225906372, "learning_rate": 4.9995607010911696e-06, "loss": 0.1979547142982483, "step": 280 }, { "epoch": 0.07294911734164071, "grad_norm": 3.478311538696289, "learning_rate": 4.999541809294045e-06, "loss": 0.15023788809776306, "step": 281 }, { "epoch": 0.07320872274143302, "grad_norm": 4.343663692474365, "learning_rate": 4.999522519829435e-06, "loss": 0.15106618404388428, "step": 282 }, { "epoch": 0.07346832814122534, "grad_norm": 1.2190685272216797, "learning_rate": 4.999502832700752e-06, "loss": 0.15331235527992249, "step": 283 }, { "epoch": 0.07372793354101766, "grad_norm": 2.6507456302642822, "learning_rate": 4.999482747911475e-06, "loss": 0.16486616432666779, "step": 284 }, { "epoch": 0.07398753894080996, "grad_norm": 5.13680362701416, "learning_rate": 4.999462265465154e-06, "loss": 0.14376884698867798, "step": 285 }, { "epoch": 0.07424714434060228, "grad_norm": 1.5335215330123901, "learning_rate": 4.99944138536541e-06, "loss": 0.14699000120162964, "step": 286 }, { "epoch": 0.0745067497403946, "grad_norm": 1.0839756727218628, "learning_rate": 4.999420107615935e-06, "loss": 0.1473913937807083, "step": 287 }, { "epoch": 0.07476635514018691, "grad_norm": 2.0339598655700684, "learning_rate": 4.9993984322204905e-06, "loss": 0.15934163331985474, "step": 288 }, { "epoch": 0.07502596053997923, "grad_norm": 2.0913054943084717, "learning_rate": 4.999376359182906e-06, "loss": 0.1635827124118805, "step": 289 }, { "epoch": 0.07528556593977155, "grad_norm": 1.318588137626648, "learning_rate": 4.999353888507086e-06, "loss": 0.15233764052391052, "step": 290 }, { "epoch": 0.07554517133956386, "grad_norm": 2.212083101272583, "learning_rate": 4.999331020197002e-06, "loss": 0.16547361016273499, "step": 291 }, { "epoch": 0.07580477673935618, "grad_norm": 1.264451026916504, "learning_rate": 4.999307754256696e-06, "loss": 0.19689923524856567, "step": 292 }, { "epoch": 0.0760643821391485, "grad_norm": 1.383230209350586, "learning_rate": 4.999284090690282e-06, "loss": 0.14297375082969666, "step": 293 }, { "epoch": 0.0763239875389408, "grad_norm": 2.333632707595825, "learning_rate": 4.999260029501942e-06, "loss": 0.21532827615737915, "step": 294 }, { "epoch": 0.07658359293873312, "grad_norm": 4.145650386810303, "learning_rate": 4.999235570695931e-06, "loss": 0.1544724553823471, "step": 295 }, { "epoch": 0.07684319833852545, "grad_norm": 4.033267498016357, "learning_rate": 4.9992107142765715e-06, "loss": 0.1488204002380371, "step": 296 }, { "epoch": 0.07710280373831775, "grad_norm": 5.0765204429626465, "learning_rate": 4.999185460248257e-06, "loss": 0.17012086510658264, "step": 297 }, { "epoch": 0.07736240913811007, "grad_norm": 2.6621272563934326, "learning_rate": 4.999159808615453e-06, "loss": 0.14829808473587036, "step": 298 }, { "epoch": 0.07762201453790239, "grad_norm": 7.6853227615356445, "learning_rate": 4.999133759382693e-06, "loss": 0.19593581557273865, "step": 299 }, { "epoch": 0.0778816199376947, "grad_norm": 2.3959169387817383, "learning_rate": 4.999107312554583e-06, "loss": 0.14504459500312805, "step": 300 }, { "epoch": 0.07814122533748702, "grad_norm": 6.704405784606934, "learning_rate": 4.9990804681358e-06, "loss": 0.16790315508842468, "step": 301 }, { "epoch": 0.07840083073727934, "grad_norm": 6.347667217254639, "learning_rate": 4.999053226131084e-06, "loss": 0.20930039882659912, "step": 302 }, { "epoch": 0.07866043613707165, "grad_norm": 1.507530689239502, "learning_rate": 4.9990255865452565e-06, "loss": 0.15549427270889282, "step": 303 }, { "epoch": 0.07892004153686397, "grad_norm": 5.461438179016113, "learning_rate": 4.9989975493832005e-06, "loss": 0.15556904673576355, "step": 304 }, { "epoch": 0.07917964693665629, "grad_norm": 2.9057157039642334, "learning_rate": 4.998969114649873e-06, "loss": 0.20091670751571655, "step": 305 }, { "epoch": 0.0794392523364486, "grad_norm": 2.173197031021118, "learning_rate": 4.9989402823503006e-06, "loss": 0.18469281494617462, "step": 306 }, { "epoch": 0.07969885773624091, "grad_norm": 1.5458030700683594, "learning_rate": 4.9989110524895795e-06, "loss": 0.1784062385559082, "step": 307 }, { "epoch": 0.07995846313603323, "grad_norm": 5.17405891418457, "learning_rate": 4.998881425072879e-06, "loss": 0.1837070882320404, "step": 308 }, { "epoch": 0.08021806853582554, "grad_norm": 2.0893588066101074, "learning_rate": 4.998851400105433e-06, "loss": 0.14344362914562225, "step": 309 }, { "epoch": 0.08047767393561786, "grad_norm": 3.994081974029541, "learning_rate": 4.998820977592553e-06, "loss": 0.2002202421426773, "step": 310 }, { "epoch": 0.08073727933541018, "grad_norm": 4.819305419921875, "learning_rate": 4.998790157539615e-06, "loss": 0.1825810819864273, "step": 311 }, { "epoch": 0.08099688473520249, "grad_norm": 1.66962730884552, "learning_rate": 4.998758939952068e-06, "loss": 0.21133461594581604, "step": 312 }, { "epoch": 0.08125649013499481, "grad_norm": 4.628571510314941, "learning_rate": 4.998727324835429e-06, "loss": 0.17945489287376404, "step": 313 }, { "epoch": 0.08151609553478713, "grad_norm": 6.923367023468018, "learning_rate": 4.998695312195289e-06, "loss": 0.1773732304573059, "step": 314 }, { "epoch": 0.08177570093457943, "grad_norm": 6.832416534423828, "learning_rate": 4.998662902037306e-06, "loss": 0.16500814259052277, "step": 315 }, { "epoch": 0.08203530633437175, "grad_norm": 4.444311141967773, "learning_rate": 4.998630094367209e-06, "loss": 0.15766559541225433, "step": 316 }, { "epoch": 0.08229491173416408, "grad_norm": 0.9057044386863708, "learning_rate": 4.998596889190799e-06, "loss": 0.14444893598556519, "step": 317 }, { "epoch": 0.08255451713395638, "grad_norm": 2.753089189529419, "learning_rate": 4.998563286513944e-06, "loss": 0.15843352675437927, "step": 318 }, { "epoch": 0.0828141225337487, "grad_norm": 5.184789657592773, "learning_rate": 4.998529286342585e-06, "loss": 0.1912737935781479, "step": 319 }, { "epoch": 0.08307372793354102, "grad_norm": 4.277538776397705, "learning_rate": 4.998494888682734e-06, "loss": 0.21995413303375244, "step": 320 }, { "epoch": 0.08333333333333333, "grad_norm": 2.652477502822876, "learning_rate": 4.99846009354047e-06, "loss": 0.1678324192762375, "step": 321 }, { "epoch": 0.08359293873312565, "grad_norm": 2.3373851776123047, "learning_rate": 4.998424900921944e-06, "loss": 0.18271851539611816, "step": 322 }, { "epoch": 0.08385254413291797, "grad_norm": 2.909444808959961, "learning_rate": 4.998389310833378e-06, "loss": 0.1743454933166504, "step": 323 }, { "epoch": 0.08411214953271028, "grad_norm": 3.9864895343780518, "learning_rate": 4.998353323281063e-06, "loss": 0.15420842170715332, "step": 324 }, { "epoch": 0.0843717549325026, "grad_norm": 2.4789798259735107, "learning_rate": 4.998316938271361e-06, "loss": 0.1647869050502777, "step": 325 }, { "epoch": 0.08463136033229492, "grad_norm": 2.913198471069336, "learning_rate": 4.998280155810704e-06, "loss": 0.16932925581932068, "step": 326 }, { "epoch": 0.08489096573208722, "grad_norm": 4.138684272766113, "learning_rate": 4.998242975905593e-06, "loss": 0.185330331325531, "step": 327 }, { "epoch": 0.08515057113187954, "grad_norm": 2.9485185146331787, "learning_rate": 4.998205398562603e-06, "loss": 0.17558008432388306, "step": 328 }, { "epoch": 0.08541017653167186, "grad_norm": 3.4920153617858887, "learning_rate": 4.998167423788375e-06, "loss": 0.1557396650314331, "step": 329 }, { "epoch": 0.08566978193146417, "grad_norm": 4.903194904327393, "learning_rate": 4.998129051589624e-06, "loss": 0.19835640490055084, "step": 330 }, { "epoch": 0.08592938733125649, "grad_norm": 2.7904317378997803, "learning_rate": 4.998090281973131e-06, "loss": 0.15449321269989014, "step": 331 }, { "epoch": 0.08618899273104881, "grad_norm": 1.1501896381378174, "learning_rate": 4.99805111494575e-06, "loss": 0.16146257519721985, "step": 332 }, { "epoch": 0.08644859813084112, "grad_norm": 5.549028396606445, "learning_rate": 4.9980115505144065e-06, "loss": 0.18605072796344757, "step": 333 }, { "epoch": 0.08670820353063344, "grad_norm": 2.2069547176361084, "learning_rate": 4.997971588686093e-06, "loss": 0.15798325836658478, "step": 334 }, { "epoch": 0.08696780893042576, "grad_norm": 3.661423683166504, "learning_rate": 4.9979312294678745e-06, "loss": 0.19068759679794312, "step": 335 }, { "epoch": 0.08722741433021806, "grad_norm": 1.8779270648956299, "learning_rate": 4.997890472866885e-06, "loss": 0.17053472995758057, "step": 336 }, { "epoch": 0.08748701973001038, "grad_norm": 2.628756046295166, "learning_rate": 4.99784931889033e-06, "loss": 0.18781046569347382, "step": 337 }, { "epoch": 0.0877466251298027, "grad_norm": 3.347198486328125, "learning_rate": 4.997807767545485e-06, "loss": 0.1540624350309372, "step": 338 }, { "epoch": 0.08800623052959501, "grad_norm": 1.0390416383743286, "learning_rate": 4.9977658188396935e-06, "loss": 0.152253657579422, "step": 339 }, { "epoch": 0.08826583592938733, "grad_norm": 2.653606653213501, "learning_rate": 4.997723472780373e-06, "loss": 0.14635790884494781, "step": 340 }, { "epoch": 0.08852544132917965, "grad_norm": 2.031266450881958, "learning_rate": 4.997680729375008e-06, "loss": 0.14509323239326477, "step": 341 }, { "epoch": 0.08878504672897196, "grad_norm": 3.666862964630127, "learning_rate": 4.997637588631156e-06, "loss": 0.20078957080841064, "step": 342 }, { "epoch": 0.08904465212876428, "grad_norm": 5.557047367095947, "learning_rate": 4.997594050556443e-06, "loss": 0.1525324583053589, "step": 343 }, { "epoch": 0.0893042575285566, "grad_norm": 2.99737811088562, "learning_rate": 4.9975501151585635e-06, "loss": 0.18028578162193298, "step": 344 }, { "epoch": 0.0895638629283489, "grad_norm": 0.9941532015800476, "learning_rate": 4.997505782445286e-06, "loss": 0.15595948696136475, "step": 345 }, { "epoch": 0.08982346832814123, "grad_norm": 3.6784744262695312, "learning_rate": 4.997461052424447e-06, "loss": 0.17797742784023285, "step": 346 }, { "epoch": 0.09008307372793355, "grad_norm": 3.9965896606445312, "learning_rate": 4.997415925103955e-06, "loss": 0.18175366520881653, "step": 347 }, { "epoch": 0.09034267912772585, "grad_norm": 3.598421812057495, "learning_rate": 4.997370400491786e-06, "loss": 0.16459819674491882, "step": 348 }, { "epoch": 0.09060228452751817, "grad_norm": 2.795027017593384, "learning_rate": 4.997324478595987e-06, "loss": 0.19047008454799652, "step": 349 }, { "epoch": 0.0908618899273105, "grad_norm": 1.8659802675247192, "learning_rate": 4.997278159424678e-06, "loss": 0.16556808352470398, "step": 350 }, { "epoch": 0.0911214953271028, "grad_norm": 3.209381103515625, "learning_rate": 4.997231442986047e-06, "loss": 0.19286660850048065, "step": 351 }, { "epoch": 0.09138110072689512, "grad_norm": 2.8543200492858887, "learning_rate": 4.997184329288351e-06, "loss": 0.19286467134952545, "step": 352 }, { "epoch": 0.09164070612668744, "grad_norm": 3.153524875640869, "learning_rate": 4.99713681833992e-06, "loss": 0.18380849063396454, "step": 353 }, { "epoch": 0.09190031152647975, "grad_norm": 3.3384103775024414, "learning_rate": 4.99708891014915e-06, "loss": 0.18959854543209076, "step": 354 }, { "epoch": 0.09215991692627207, "grad_norm": 1.7476879358291626, "learning_rate": 4.997040604724514e-06, "loss": 0.15234260261058807, "step": 355 }, { "epoch": 0.09241952232606439, "grad_norm": 5.859508037567139, "learning_rate": 4.9969919020745485e-06, "loss": 0.14875568449497223, "step": 356 }, { "epoch": 0.0926791277258567, "grad_norm": 1.8558025360107422, "learning_rate": 4.996942802207866e-06, "loss": 0.14554736018180847, "step": 357 }, { "epoch": 0.09293873312564901, "grad_norm": 5.259397506713867, "learning_rate": 4.996893305133142e-06, "loss": 0.1971375048160553, "step": 358 }, { "epoch": 0.09319833852544133, "grad_norm": 2.338233232498169, "learning_rate": 4.996843410859129e-06, "loss": 0.14351218938827515, "step": 359 }, { "epoch": 0.09345794392523364, "grad_norm": 2.2254245281219482, "learning_rate": 4.996793119394648e-06, "loss": 0.1455444097518921, "step": 360 }, { "epoch": 0.09371754932502596, "grad_norm": 4.406637191772461, "learning_rate": 4.996742430748587e-06, "loss": 0.15604691207408905, "step": 361 }, { "epoch": 0.09397715472481828, "grad_norm": 2.33426570892334, "learning_rate": 4.996691344929909e-06, "loss": 0.20729988813400269, "step": 362 }, { "epoch": 0.09423676012461059, "grad_norm": 4.3667731285095215, "learning_rate": 4.996639861947643e-06, "loss": 0.1839103400707245, "step": 363 }, { "epoch": 0.09449636552440291, "grad_norm": 3.8674817085266113, "learning_rate": 4.99658798181089e-06, "loss": 0.17285628616809845, "step": 364 }, { "epoch": 0.09475597092419523, "grad_norm": 2.70582914352417, "learning_rate": 4.996535704528823e-06, "loss": 0.15863148868083954, "step": 365 }, { "epoch": 0.09501557632398754, "grad_norm": 1.8883378505706787, "learning_rate": 4.996483030110682e-06, "loss": 0.1828474998474121, "step": 366 }, { "epoch": 0.09527518172377986, "grad_norm": 1.572060227394104, "learning_rate": 4.996429958565779e-06, "loss": 0.1401478797197342, "step": 367 }, { "epoch": 0.09553478712357218, "grad_norm": 6.749952793121338, "learning_rate": 4.996376489903495e-06, "loss": 0.164112389087677, "step": 368 }, { "epoch": 0.09579439252336448, "grad_norm": 4.129411697387695, "learning_rate": 4.996322624133284e-06, "loss": 0.15077120065689087, "step": 369 }, { "epoch": 0.0960539979231568, "grad_norm": 4.958012104034424, "learning_rate": 4.996268361264666e-06, "loss": 0.14366358518600464, "step": 370 }, { "epoch": 0.09631360332294912, "grad_norm": 4.273137092590332, "learning_rate": 4.996213701307235e-06, "loss": 0.16222725808620453, "step": 371 }, { "epoch": 0.09657320872274143, "grad_norm": 7.518589019775391, "learning_rate": 4.9961586442706525e-06, "loss": 0.2153668999671936, "step": 372 }, { "epoch": 0.09683281412253375, "grad_norm": 6.282024383544922, "learning_rate": 4.9961031901646524e-06, "loss": 0.1970100700855255, "step": 373 }, { "epoch": 0.09709241952232607, "grad_norm": 2.578824281692505, "learning_rate": 4.996047338999037e-06, "loss": 0.1865856945514679, "step": 374 }, { "epoch": 0.09735202492211838, "grad_norm": 2.5622966289520264, "learning_rate": 4.99599109078368e-06, "loss": 0.15749093890190125, "step": 375 }, { "epoch": 0.0976116303219107, "grad_norm": 2.964451551437378, "learning_rate": 4.995934445528524e-06, "loss": 0.15549103915691376, "step": 376 }, { "epoch": 0.09787123572170302, "grad_norm": 4.499732494354248, "learning_rate": 4.995877403243584e-06, "loss": 0.18742817640304565, "step": 377 }, { "epoch": 0.09813084112149532, "grad_norm": 3.3507068157196045, "learning_rate": 4.995819963938942e-06, "loss": 0.1529548466205597, "step": 378 }, { "epoch": 0.09839044652128764, "grad_norm": 4.914780616760254, "learning_rate": 4.995762127624753e-06, "loss": 0.15118087828159332, "step": 379 }, { "epoch": 0.09865005192107996, "grad_norm": 0.7882607579231262, "learning_rate": 4.995703894311242e-06, "loss": 0.14934226870536804, "step": 380 }, { "epoch": 0.09890965732087227, "grad_norm": 2.692214250564575, "learning_rate": 4.995645264008701e-06, "loss": 0.19091889262199402, "step": 381 }, { "epoch": 0.09916926272066459, "grad_norm": 3.7163450717926025, "learning_rate": 4.995586236727496e-06, "loss": 0.17284610867500305, "step": 382 }, { "epoch": 0.09942886812045691, "grad_norm": 2.0848114490509033, "learning_rate": 4.995526812478061e-06, "loss": 0.21864858269691467, "step": 383 }, { "epoch": 0.09968847352024922, "grad_norm": 3.384596824645996, "learning_rate": 4.995466991270903e-06, "loss": 0.1496678590774536, "step": 384 }, { "epoch": 0.09994807892004154, "grad_norm": 1.3882570266723633, "learning_rate": 4.9954067731165945e-06, "loss": 0.15405291318893433, "step": 385 }, { "epoch": 0.10020768431983386, "grad_norm": 2.69234299659729, "learning_rate": 4.995346158025781e-06, "loss": 0.1447930485010147, "step": 386 }, { "epoch": 0.10046728971962617, "grad_norm": 1.127137541770935, "learning_rate": 4.995285146009179e-06, "loss": 0.15312492847442627, "step": 387 }, { "epoch": 0.10072689511941849, "grad_norm": 2.01901912689209, "learning_rate": 4.995223737077573e-06, "loss": 0.2046917974948883, "step": 388 }, { "epoch": 0.1009865005192108, "grad_norm": 4.624377250671387, "learning_rate": 4.9951619312418196e-06, "loss": 0.1519964188337326, "step": 389 }, { "epoch": 0.10124610591900311, "grad_norm": 1.370639443397522, "learning_rate": 4.995099728512843e-06, "loss": 0.1609353870153427, "step": 390 }, { "epoch": 0.10150571131879543, "grad_norm": 4.443675994873047, "learning_rate": 4.995037128901642e-06, "loss": 0.15773311257362366, "step": 391 }, { "epoch": 0.10176531671858775, "grad_norm": 1.157870888710022, "learning_rate": 4.99497413241928e-06, "loss": 0.15258081257343292, "step": 392 }, { "epoch": 0.10202492211838006, "grad_norm": 1.9757283926010132, "learning_rate": 4.994910739076895e-06, "loss": 0.20346295833587646, "step": 393 }, { "epoch": 0.10228452751817238, "grad_norm": 1.818998098373413, "learning_rate": 4.994846948885693e-06, "loss": 0.1633879393339157, "step": 394 }, { "epoch": 0.1025441329179647, "grad_norm": 3.4261796474456787, "learning_rate": 4.99478276185695e-06, "loss": 0.17383749783039093, "step": 395 }, { "epoch": 0.102803738317757, "grad_norm": 1.5702756643295288, "learning_rate": 4.994718178002013e-06, "loss": 0.13778595626354218, "step": 396 }, { "epoch": 0.10306334371754933, "grad_norm": 1.6733174324035645, "learning_rate": 4.9946531973323005e-06, "loss": 0.14944422245025635, "step": 397 }, { "epoch": 0.10332294911734165, "grad_norm": 1.1642560958862305, "learning_rate": 4.9945878198592975e-06, "loss": 0.14754937589168549, "step": 398 }, { "epoch": 0.10358255451713395, "grad_norm": 1.7975839376449585, "learning_rate": 4.994522045594563e-06, "loss": 0.16382966935634613, "step": 399 }, { "epoch": 0.10384215991692627, "grad_norm": 1.709892749786377, "learning_rate": 4.9944558745497236e-06, "loss": 0.15582896769046783, "step": 400 }, { "epoch": 0.1041017653167186, "grad_norm": 3.4715511798858643, "learning_rate": 4.9943893067364765e-06, "loss": 0.16403912007808685, "step": 401 }, { "epoch": 0.1043613707165109, "grad_norm": 2.988701581954956, "learning_rate": 4.99432234216659e-06, "loss": 0.16873395442962646, "step": 402 }, { "epoch": 0.10462097611630322, "grad_norm": 1.1232988834381104, "learning_rate": 4.994254980851901e-06, "loss": 0.1445494294166565, "step": 403 }, { "epoch": 0.10488058151609553, "grad_norm": 2.652386426925659, "learning_rate": 4.994187222804319e-06, "loss": 0.1634778082370758, "step": 404 }, { "epoch": 0.10514018691588785, "grad_norm": 3.4670910835266113, "learning_rate": 4.99411906803582e-06, "loss": 0.1862848997116089, "step": 405 }, { "epoch": 0.10539979231568017, "grad_norm": 3.443472146987915, "learning_rate": 4.994050516558454e-06, "loss": 0.15790852904319763, "step": 406 }, { "epoch": 0.10565939771547248, "grad_norm": 2.8006937503814697, "learning_rate": 4.993981568384339e-06, "loss": 0.16157102584838867, "step": 407 }, { "epoch": 0.1059190031152648, "grad_norm": 4.2308268547058105, "learning_rate": 4.993912223525661e-06, "loss": 0.18012985587120056, "step": 408 }, { "epoch": 0.10617860851505712, "grad_norm": 3.2461657524108887, "learning_rate": 4.993842481994683e-06, "loss": 0.17713908851146698, "step": 409 }, { "epoch": 0.10643821391484942, "grad_norm": 1.8459417819976807, "learning_rate": 4.99377234380373e-06, "loss": 0.17620046436786652, "step": 410 }, { "epoch": 0.10669781931464174, "grad_norm": 2.3345680236816406, "learning_rate": 4.9937018089652024e-06, "loss": 0.20170894265174866, "step": 411 }, { "epoch": 0.10695742471443406, "grad_norm": 3.0434205532073975, "learning_rate": 4.99363087749157e-06, "loss": 0.17652148008346558, "step": 412 }, { "epoch": 0.10721703011422637, "grad_norm": 4.065282821655273, "learning_rate": 4.993559549395369e-06, "loss": 0.2153504341840744, "step": 413 }, { "epoch": 0.10747663551401869, "grad_norm": 1.9912054538726807, "learning_rate": 4.993487824689211e-06, "loss": 0.15625865757465363, "step": 414 }, { "epoch": 0.10773624091381101, "grad_norm": 1.1919394731521606, "learning_rate": 4.993415703385776e-06, "loss": 0.1392943561077118, "step": 415 }, { "epoch": 0.10799584631360332, "grad_norm": 2.2062466144561768, "learning_rate": 4.99334318549781e-06, "loss": 0.14548742771148682, "step": 416 }, { "epoch": 0.10825545171339564, "grad_norm": 2.3815255165100098, "learning_rate": 4.993270271038136e-06, "loss": 0.14378434419631958, "step": 417 }, { "epoch": 0.10851505711318796, "grad_norm": 1.2458542585372925, "learning_rate": 4.993196960019642e-06, "loss": 0.1491820514202118, "step": 418 }, { "epoch": 0.10877466251298026, "grad_norm": 4.515323638916016, "learning_rate": 4.993123252455288e-06, "loss": 0.16596074402332306, "step": 419 }, { "epoch": 0.10903426791277258, "grad_norm": 1.742958426475525, "learning_rate": 4.993049148358103e-06, "loss": 0.1438429057598114, "step": 420 }, { "epoch": 0.1092938733125649, "grad_norm": 3.972852945327759, "learning_rate": 4.992974647741188e-06, "loss": 0.14795330166816711, "step": 421 }, { "epoch": 0.10955347871235721, "grad_norm": 4.0465497970581055, "learning_rate": 4.992899750617712e-06, "loss": 0.2065192461013794, "step": 422 }, { "epoch": 0.10981308411214953, "grad_norm": 3.036853551864624, "learning_rate": 4.9928244570009165e-06, "loss": 0.1451602280139923, "step": 423 }, { "epoch": 0.11007268951194185, "grad_norm": 3.7677183151245117, "learning_rate": 4.99274876690411e-06, "loss": 0.1482478380203247, "step": 424 }, { "epoch": 0.11033229491173416, "grad_norm": 1.1145589351654053, "learning_rate": 4.992672680340675e-06, "loss": 0.147009015083313, "step": 425 }, { "epoch": 0.11059190031152648, "grad_norm": 10.920927047729492, "learning_rate": 4.9925961973240595e-06, "loss": 0.17535415291786194, "step": 426 }, { "epoch": 0.1108515057113188, "grad_norm": 1.9013742208480835, "learning_rate": 4.992519317867784e-06, "loss": 0.15261100232601166, "step": 427 }, { "epoch": 0.1111111111111111, "grad_norm": 4.2049994468688965, "learning_rate": 4.992442041985442e-06, "loss": 0.1673760563135147, "step": 428 }, { "epoch": 0.11137071651090343, "grad_norm": 4.973546981811523, "learning_rate": 4.9923643696906915e-06, "loss": 0.14553481340408325, "step": 429 }, { "epoch": 0.11163032191069575, "grad_norm": 1.2730872631072998, "learning_rate": 4.992286300997263e-06, "loss": 0.15103399753570557, "step": 430 }, { "epoch": 0.11188992731048805, "grad_norm": 6.4402055740356445, "learning_rate": 4.992207835918959e-06, "loss": 0.19575029611587524, "step": 431 }, { "epoch": 0.11214953271028037, "grad_norm": 4.984111785888672, "learning_rate": 4.99212897446965e-06, "loss": 0.1541980803012848, "step": 432 }, { "epoch": 0.11240913811007269, "grad_norm": 2.1429238319396973, "learning_rate": 4.992049716663274e-06, "loss": 0.14749819040298462, "step": 433 }, { "epoch": 0.112668743509865, "grad_norm": 2.894331693649292, "learning_rate": 4.991970062513845e-06, "loss": 0.15096339583396912, "step": 434 }, { "epoch": 0.11292834890965732, "grad_norm": 5.763020038604736, "learning_rate": 4.991890012035444e-06, "loss": 0.2005748152732849, "step": 435 }, { "epoch": 0.11318795430944964, "grad_norm": 2.5792500972747803, "learning_rate": 4.991809565242221e-06, "loss": 0.1774093508720398, "step": 436 }, { "epoch": 0.11344755970924195, "grad_norm": 7.7461748123168945, "learning_rate": 4.991728722148398e-06, "loss": 0.15947116911411285, "step": 437 }, { "epoch": 0.11370716510903427, "grad_norm": 4.183899879455566, "learning_rate": 4.991647482768265e-06, "loss": 0.18557488918304443, "step": 438 }, { "epoch": 0.11396677050882659, "grad_norm": 1.576807975769043, "learning_rate": 4.991565847116184e-06, "loss": 0.182738795876503, "step": 439 }, { "epoch": 0.1142263759086189, "grad_norm": 3.6562235355377197, "learning_rate": 4.991483815206586e-06, "loss": 0.16196967661380768, "step": 440 }, { "epoch": 0.11448598130841121, "grad_norm": 3.448214292526245, "learning_rate": 4.991401387053973e-06, "loss": 0.15095674991607666, "step": 441 }, { "epoch": 0.11474558670820353, "grad_norm": 3.524411678314209, "learning_rate": 4.991318562672917e-06, "loss": 0.16248801350593567, "step": 442 }, { "epoch": 0.11500519210799584, "grad_norm": 1.732315182685852, "learning_rate": 4.991235342078058e-06, "loss": 0.15949997305870056, "step": 443 }, { "epoch": 0.11526479750778816, "grad_norm": 1.651301622390747, "learning_rate": 4.9911517252841066e-06, "loss": 0.13929246366024017, "step": 444 }, { "epoch": 0.11552440290758048, "grad_norm": 3.7732813358306885, "learning_rate": 4.991067712305848e-06, "loss": 0.21940088272094727, "step": 445 }, { "epoch": 0.11578400830737279, "grad_norm": 5.614531993865967, "learning_rate": 4.9909833031581294e-06, "loss": 0.16275757551193237, "step": 446 }, { "epoch": 0.11604361370716511, "grad_norm": 3.5158534049987793, "learning_rate": 4.990898497855876e-06, "loss": 0.17715850472450256, "step": 447 }, { "epoch": 0.11630321910695743, "grad_norm": 2.8858039379119873, "learning_rate": 4.990813296414077e-06, "loss": 0.1977224200963974, "step": 448 }, { "epoch": 0.11656282450674973, "grad_norm": 3.3986105918884277, "learning_rate": 4.990727698847796e-06, "loss": 0.1831243336200714, "step": 449 }, { "epoch": 0.11682242990654206, "grad_norm": 3.881941318511963, "learning_rate": 4.990641705172163e-06, "loss": 0.1738479733467102, "step": 450 }, { "epoch": 0.11708203530633438, "grad_norm": 2.7864794731140137, "learning_rate": 4.99055531540238e-06, "loss": 0.17881640791893005, "step": 451 }, { "epoch": 0.11734164070612668, "grad_norm": 4.712965488433838, "learning_rate": 4.99046852955372e-06, "loss": 0.1906452476978302, "step": 452 }, { "epoch": 0.117601246105919, "grad_norm": 2.7893917560577393, "learning_rate": 4.990381347641524e-06, "loss": 0.14130036532878876, "step": 453 }, { "epoch": 0.11786085150571132, "grad_norm": 1.4497795104980469, "learning_rate": 4.9902937696812034e-06, "loss": 0.13050776720046997, "step": 454 }, { "epoch": 0.11812045690550363, "grad_norm": 2.6096394062042236, "learning_rate": 4.99020579568824e-06, "loss": 0.16796143352985382, "step": 455 }, { "epoch": 0.11838006230529595, "grad_norm": 2.225696325302124, "learning_rate": 4.990117425678186e-06, "loss": 0.1511036902666092, "step": 456 }, { "epoch": 0.11863966770508827, "grad_norm": 2.1474204063415527, "learning_rate": 4.990028659666663e-06, "loss": 0.14518392086029053, "step": 457 }, { "epoch": 0.11889927310488058, "grad_norm": 1.677028775215149, "learning_rate": 4.989939497669363e-06, "loss": 0.14543962478637695, "step": 458 }, { "epoch": 0.1191588785046729, "grad_norm": 2.9730310440063477, "learning_rate": 4.989849939702047e-06, "loss": 0.17253482341766357, "step": 459 }, { "epoch": 0.11941848390446522, "grad_norm": 1.7401862144470215, "learning_rate": 4.989759985780548e-06, "loss": 0.14678889513015747, "step": 460 }, { "epoch": 0.11967808930425752, "grad_norm": 2.3588504791259766, "learning_rate": 4.989669635920767e-06, "loss": 0.15637540817260742, "step": 461 }, { "epoch": 0.11993769470404984, "grad_norm": 2.538734197616577, "learning_rate": 4.989578890138676e-06, "loss": 0.1282447725534439, "step": 462 }, { "epoch": 0.12019730010384216, "grad_norm": 0.958745539188385, "learning_rate": 4.989487748450318e-06, "loss": 0.1409270465373993, "step": 463 }, { "epoch": 0.12045690550363447, "grad_norm": 4.851242542266846, "learning_rate": 4.9893962108718015e-06, "loss": 0.22463083267211914, "step": 464 }, { "epoch": 0.12071651090342679, "grad_norm": 5.209906101226807, "learning_rate": 4.989304277419311e-06, "loss": 0.19802328944206238, "step": 465 }, { "epoch": 0.12097611630321911, "grad_norm": 5.107081890106201, "learning_rate": 4.989211948109099e-06, "loss": 0.20082537829875946, "step": 466 }, { "epoch": 0.12123572170301142, "grad_norm": 2.0984628200531006, "learning_rate": 4.989119222957483e-06, "loss": 0.18066588044166565, "step": 467 }, { "epoch": 0.12149532710280374, "grad_norm": 3.6756696701049805, "learning_rate": 4.98902610198086e-06, "loss": 0.19615745544433594, "step": 468 }, { "epoch": 0.12175493250259606, "grad_norm": 6.849240779876709, "learning_rate": 4.988932585195688e-06, "loss": 0.21537843346595764, "step": 469 }, { "epoch": 0.12201453790238836, "grad_norm": 1.2190890312194824, "learning_rate": 4.988838672618499e-06, "loss": 0.1429261863231659, "step": 470 }, { "epoch": 0.12227414330218069, "grad_norm": 1.693670630455017, "learning_rate": 4.988744364265896e-06, "loss": 0.18044167757034302, "step": 471 }, { "epoch": 0.122533748701973, "grad_norm": 6.607396125793457, "learning_rate": 4.988649660154551e-06, "loss": 0.17147275805473328, "step": 472 }, { "epoch": 0.12279335410176531, "grad_norm": 5.676530838012695, "learning_rate": 4.988554560301203e-06, "loss": 0.1684752255678177, "step": 473 }, { "epoch": 0.12305295950155763, "grad_norm": 3.2958221435546875, "learning_rate": 4.988459064722664e-06, "loss": 0.1231340616941452, "step": 474 }, { "epoch": 0.12331256490134995, "grad_norm": 6.7702317237854, "learning_rate": 4.988363173435818e-06, "loss": 0.18092258274555206, "step": 475 }, { "epoch": 0.12357217030114226, "grad_norm": 5.693624973297119, "learning_rate": 4.9882668864576135e-06, "loss": 0.18729931116104126, "step": 476 }, { "epoch": 0.12383177570093458, "grad_norm": 1.84119713306427, "learning_rate": 4.988170203805074e-06, "loss": 0.15363255143165588, "step": 477 }, { "epoch": 0.1240913811007269, "grad_norm": 6.85548210144043, "learning_rate": 4.9880731254952885e-06, "loss": 0.14941227436065674, "step": 478 }, { "epoch": 0.1243509865005192, "grad_norm": 4.975020408630371, "learning_rate": 4.987975651545421e-06, "loss": 0.13664516806602478, "step": 479 }, { "epoch": 0.12461059190031153, "grad_norm": 2.7991867065429688, "learning_rate": 4.9878777819727e-06, "loss": 0.16558687388896942, "step": 480 }, { "epoch": 0.12487019730010385, "grad_norm": 4.567789077758789, "learning_rate": 4.9877795167944285e-06, "loss": 0.1585719883441925, "step": 481 }, { "epoch": 0.12512980269989615, "grad_norm": 4.2380757331848145, "learning_rate": 4.987680856027976e-06, "loss": 0.16859620809555054, "step": 482 }, { "epoch": 0.12538940809968846, "grad_norm": 3.5048882961273193, "learning_rate": 4.987581799690785e-06, "loss": 0.16769979894161224, "step": 483 }, { "epoch": 0.1256490134994808, "grad_norm": 1.514815092086792, "learning_rate": 4.987482347800365e-06, "loss": 0.13417431712150574, "step": 484 }, { "epoch": 0.1259086188992731, "grad_norm": 6.471580505371094, "learning_rate": 4.987382500374298e-06, "loss": 0.21183665096759796, "step": 485 }, { "epoch": 0.1261682242990654, "grad_norm": 5.621553897857666, "learning_rate": 4.987282257430234e-06, "loss": 0.2015511393547058, "step": 486 }, { "epoch": 0.12642782969885774, "grad_norm": 1.335161805152893, "learning_rate": 4.987181618985893e-06, "loss": 0.1706988513469696, "step": 487 }, { "epoch": 0.12668743509865005, "grad_norm": 5.279466152191162, "learning_rate": 4.987080585059068e-06, "loss": 0.14900296926498413, "step": 488 }, { "epoch": 0.12694704049844235, "grad_norm": 3.2065892219543457, "learning_rate": 4.9869791556676175e-06, "loss": 0.15903116762638092, "step": 489 }, { "epoch": 0.1272066458982347, "grad_norm": 3.374295711517334, "learning_rate": 4.986877330829472e-06, "loss": 0.14163348078727722, "step": 490 }, { "epoch": 0.127466251298027, "grad_norm": 2.311927080154419, "learning_rate": 4.986775110562632e-06, "loss": 0.15052694082260132, "step": 491 }, { "epoch": 0.1277258566978193, "grad_norm": 3.51096510887146, "learning_rate": 4.9866724948851685e-06, "loss": 0.18820136785507202, "step": 492 }, { "epoch": 0.12798546209761164, "grad_norm": 1.6219485998153687, "learning_rate": 4.98656948381522e-06, "loss": 0.15082715451717377, "step": 493 }, { "epoch": 0.12824506749740394, "grad_norm": 3.5319550037384033, "learning_rate": 4.986466077370998e-06, "loss": 0.15039047598838806, "step": 494 }, { "epoch": 0.12850467289719625, "grad_norm": 4.1491475105285645, "learning_rate": 4.9863622755707815e-06, "loss": 0.1917169690132141, "step": 495 }, { "epoch": 0.12876427829698858, "grad_norm": 0.9212135672569275, "learning_rate": 4.986258078432921e-06, "loss": 0.1505240947008133, "step": 496 }, { "epoch": 0.1290238836967809, "grad_norm": 2.7041194438934326, "learning_rate": 4.986153485975835e-06, "loss": 0.2048410326242447, "step": 497 }, { "epoch": 0.1292834890965732, "grad_norm": 1.5821806192398071, "learning_rate": 4.986048498218014e-06, "loss": 0.17316670715808868, "step": 498 }, { "epoch": 0.12954309449636553, "grad_norm": 1.3393187522888184, "learning_rate": 4.985943115178018e-06, "loss": 0.20265641808509827, "step": 499 }, { "epoch": 0.12980269989615784, "grad_norm": 4.544067859649658, "learning_rate": 4.985837336874475e-06, "loss": 0.15340915322303772, "step": 500 }, { "epoch": 0.13006230529595014, "grad_norm": 2.302405595779419, "learning_rate": 4.985731163326084e-06, "loss": 0.18747897446155548, "step": 501 }, { "epoch": 0.13032191069574248, "grad_norm": 1.5345863103866577, "learning_rate": 4.985624594551616e-06, "loss": 0.1539071649312973, "step": 502 }, { "epoch": 0.13058151609553478, "grad_norm": 1.2481849193572998, "learning_rate": 4.985517630569908e-06, "loss": 0.13044705986976624, "step": 503 }, { "epoch": 0.1308411214953271, "grad_norm": 1.5456883907318115, "learning_rate": 4.98541027139987e-06, "loss": 0.15013496577739716, "step": 504 }, { "epoch": 0.13110072689511942, "grad_norm": 2.2959017753601074, "learning_rate": 4.98530251706048e-06, "loss": 0.17088960111141205, "step": 505 }, { "epoch": 0.13136033229491173, "grad_norm": 1.0989983081817627, "learning_rate": 4.9851943675707864e-06, "loss": 0.15267784893512726, "step": 506 }, { "epoch": 0.13161993769470404, "grad_norm": 2.475637435913086, "learning_rate": 4.985085822949908e-06, "loss": 0.14876261353492737, "step": 507 }, { "epoch": 0.13187954309449637, "grad_norm": 1.5138416290283203, "learning_rate": 4.984976883217032e-06, "loss": 0.1682138442993164, "step": 508 }, { "epoch": 0.13213914849428868, "grad_norm": 4.290208339691162, "learning_rate": 4.984867548391418e-06, "loss": 0.14483986794948578, "step": 509 }, { "epoch": 0.13239875389408098, "grad_norm": 4.200908660888672, "learning_rate": 4.984757818492393e-06, "loss": 0.14678660035133362, "step": 510 }, { "epoch": 0.13265835929387332, "grad_norm": 2.4624037742614746, "learning_rate": 4.984647693539355e-06, "loss": 0.13627436757087708, "step": 511 }, { "epoch": 0.13291796469366562, "grad_norm": 3.9858081340789795, "learning_rate": 4.984537173551772e-06, "loss": 0.14454571902751923, "step": 512 }, { "epoch": 0.13317757009345793, "grad_norm": 3.572359561920166, "learning_rate": 4.98442625854918e-06, "loss": 0.1764489710330963, "step": 513 }, { "epoch": 0.13343717549325027, "grad_norm": 2.5586657524108887, "learning_rate": 4.984314948551187e-06, "loss": 0.16124680638313293, "step": 514 }, { "epoch": 0.13369678089304257, "grad_norm": 5.691349029541016, "learning_rate": 4.98420324357747e-06, "loss": 0.17647451162338257, "step": 515 }, { "epoch": 0.13395638629283488, "grad_norm": 3.0229525566101074, "learning_rate": 4.984091143647777e-06, "loss": 0.19242620468139648, "step": 516 }, { "epoch": 0.1342159916926272, "grad_norm": 2.90967059135437, "learning_rate": 4.983978648781923e-06, "loss": 0.15815161168575287, "step": 517 }, { "epoch": 0.13447559709241952, "grad_norm": 4.489492416381836, "learning_rate": 4.983865758999795e-06, "loss": 0.17274656891822815, "step": 518 }, { "epoch": 0.13473520249221183, "grad_norm": 1.8925111293792725, "learning_rate": 4.98375247432135e-06, "loss": 0.13492321968078613, "step": 519 }, { "epoch": 0.13499480789200416, "grad_norm": 1.233404278755188, "learning_rate": 4.983638794766613e-06, "loss": 0.16171064972877502, "step": 520 }, { "epoch": 0.13525441329179647, "grad_norm": 3.676962375640869, "learning_rate": 4.9835247203556815e-06, "loss": 0.17504611611366272, "step": 521 }, { "epoch": 0.13551401869158877, "grad_norm": 2.432905673980713, "learning_rate": 4.9834102511087205e-06, "loss": 0.139023095369339, "step": 522 }, { "epoch": 0.1357736240913811, "grad_norm": 1.453686237335205, "learning_rate": 4.983295387045965e-06, "loss": 0.19279426336288452, "step": 523 }, { "epoch": 0.1360332294911734, "grad_norm": 1.6402262449264526, "learning_rate": 4.98318012818772e-06, "loss": 0.1336382031440735, "step": 524 }, { "epoch": 0.13629283489096572, "grad_norm": 2.051002025604248, "learning_rate": 4.983064474554362e-06, "loss": 0.1411638855934143, "step": 525 }, { "epoch": 0.13655244029075805, "grad_norm": 4.269655227661133, "learning_rate": 4.9829484261663355e-06, "loss": 0.19634854793548584, "step": 526 }, { "epoch": 0.13681204569055036, "grad_norm": 3.1028196811676025, "learning_rate": 4.982831983044155e-06, "loss": 0.14280305802822113, "step": 527 }, { "epoch": 0.13707165109034267, "grad_norm": 2.3650784492492676, "learning_rate": 4.982715145208405e-06, "loss": 0.14571471512317657, "step": 528 }, { "epoch": 0.137331256490135, "grad_norm": 3.0714311599731445, "learning_rate": 4.982597912679739e-06, "loss": 0.14162388443946838, "step": 529 }, { "epoch": 0.1375908618899273, "grad_norm": 3.0857532024383545, "learning_rate": 4.9824802854788825e-06, "loss": 0.19133388996124268, "step": 530 }, { "epoch": 0.1378504672897196, "grad_norm": 1.3616688251495361, "learning_rate": 4.982362263626629e-06, "loss": 0.12950947880744934, "step": 531 }, { "epoch": 0.13811007268951195, "grad_norm": 0.9546668529510498, "learning_rate": 4.9822438471438404e-06, "loss": 0.1368541568517685, "step": 532 }, { "epoch": 0.13836967808930425, "grad_norm": 0.9758488535881042, "learning_rate": 4.982125036051452e-06, "loss": 0.1332048624753952, "step": 533 }, { "epoch": 0.13862928348909656, "grad_norm": 1.991361141204834, "learning_rate": 4.982005830370466e-06, "loss": 0.17747114598751068, "step": 534 }, { "epoch": 0.1388888888888889, "grad_norm": 1.992957592010498, "learning_rate": 4.981886230121956e-06, "loss": 0.17552542686462402, "step": 535 }, { "epoch": 0.1391484942886812, "grad_norm": 1.734963297843933, "learning_rate": 4.981766235327062e-06, "loss": 0.15166205167770386, "step": 536 }, { "epoch": 0.1394080996884735, "grad_norm": 2.1418042182922363, "learning_rate": 4.981645846007e-06, "loss": 0.1404399573802948, "step": 537 }, { "epoch": 0.13966770508826584, "grad_norm": 1.4380006790161133, "learning_rate": 4.98152506218305e-06, "loss": 0.18046779930591583, "step": 538 }, { "epoch": 0.13992731048805815, "grad_norm": 2.145642042160034, "learning_rate": 4.981403883876564e-06, "loss": 0.17688386142253876, "step": 539 }, { "epoch": 0.14018691588785046, "grad_norm": 3.7084076404571533, "learning_rate": 4.981282311108964e-06, "loss": 0.14304335415363312, "step": 540 }, { "epoch": 0.1404465212876428, "grad_norm": 1.0964720249176025, "learning_rate": 4.98116034390174e-06, "loss": 0.1793137490749359, "step": 541 }, { "epoch": 0.1407061266874351, "grad_norm": 2.360703468322754, "learning_rate": 4.981037982276456e-06, "loss": 0.1567421406507492, "step": 542 }, { "epoch": 0.1409657320872274, "grad_norm": 3.052215337753296, "learning_rate": 4.980915226254739e-06, "loss": 0.14438071846961975, "step": 543 }, { "epoch": 0.14122533748701974, "grad_norm": 1.0365489721298218, "learning_rate": 4.980792075858291e-06, "loss": 0.142978236079216, "step": 544 }, { "epoch": 0.14148494288681204, "grad_norm": 2.1871399879455566, "learning_rate": 4.9806685311088834e-06, "loss": 0.1524900645017624, "step": 545 }, { "epoch": 0.14174454828660435, "grad_norm": 1.607697606086731, "learning_rate": 4.980544592028354e-06, "loss": 0.13501092791557312, "step": 546 }, { "epoch": 0.14200415368639668, "grad_norm": 1.2485724687576294, "learning_rate": 4.980420258638615e-06, "loss": 0.13799138367176056, "step": 547 }, { "epoch": 0.142263759086189, "grad_norm": 4.6421637535095215, "learning_rate": 4.980295530961643e-06, "loss": 0.15860742330551147, "step": 548 }, { "epoch": 0.1425233644859813, "grad_norm": 1.5751545429229736, "learning_rate": 4.980170409019488e-06, "loss": 0.1434209942817688, "step": 549 }, { "epoch": 0.14278296988577363, "grad_norm": 2.4185171127319336, "learning_rate": 4.980044892834269e-06, "loss": 0.13701488077640533, "step": 550 }, { "epoch": 0.14304257528556594, "grad_norm": 2.743389368057251, "learning_rate": 4.979918982428174e-06, "loss": 0.16662216186523438, "step": 551 }, { "epoch": 0.14330218068535824, "grad_norm": 3.772907018661499, "learning_rate": 4.9797926778234615e-06, "loss": 0.14486731588840485, "step": 552 }, { "epoch": 0.14356178608515058, "grad_norm": 1.991045355796814, "learning_rate": 4.9796659790424595e-06, "loss": 0.13899707794189453, "step": 553 }, { "epoch": 0.14382139148494288, "grad_norm": 1.1297516822814941, "learning_rate": 4.979538886107565e-06, "loss": 0.13563813269138336, "step": 554 }, { "epoch": 0.1440809968847352, "grad_norm": 2.1427018642425537, "learning_rate": 4.979411399041244e-06, "loss": 0.14226707816123962, "step": 555 }, { "epoch": 0.14434060228452752, "grad_norm": 1.070486068725586, "learning_rate": 4.979283517866036e-06, "loss": 0.14166074991226196, "step": 556 }, { "epoch": 0.14460020768431983, "grad_norm": 2.357175588607788, "learning_rate": 4.979155242604545e-06, "loss": 0.1905166506767273, "step": 557 }, { "epoch": 0.14485981308411214, "grad_norm": 1.3562159538269043, "learning_rate": 4.979026573279447e-06, "loss": 0.168259397149086, "step": 558 }, { "epoch": 0.14511941848390447, "grad_norm": 2.2339985370635986, "learning_rate": 4.97889750991349e-06, "loss": 0.13558799028396606, "step": 559 }, { "epoch": 0.14537902388369678, "grad_norm": 1.9686832427978516, "learning_rate": 4.978768052529488e-06, "loss": 0.1659511923789978, "step": 560 }, { "epoch": 0.14563862928348908, "grad_norm": 1.8885436058044434, "learning_rate": 4.978638201150327e-06, "loss": 0.1546463668346405, "step": 561 }, { "epoch": 0.14589823468328142, "grad_norm": 2.5185599327087402, "learning_rate": 4.9785079557989605e-06, "loss": 0.15894436836242676, "step": 562 }, { "epoch": 0.14615784008307373, "grad_norm": 1.9309266805648804, "learning_rate": 4.978377316498414e-06, "loss": 0.17436295747756958, "step": 563 }, { "epoch": 0.14641744548286603, "grad_norm": 1.9802175760269165, "learning_rate": 4.97824628327178e-06, "loss": 0.13707497715950012, "step": 564 }, { "epoch": 0.14667705088265837, "grad_norm": 1.262899398803711, "learning_rate": 4.978114856142224e-06, "loss": 0.14910410344600677, "step": 565 }, { "epoch": 0.14693665628245067, "grad_norm": 2.155357837677002, "learning_rate": 4.977983035132977e-06, "loss": 0.1611238420009613, "step": 566 }, { "epoch": 0.14719626168224298, "grad_norm": 1.7412850856781006, "learning_rate": 4.977850820267346e-06, "loss": 0.1590101718902588, "step": 567 }, { "epoch": 0.1474558670820353, "grad_norm": 3.2746710777282715, "learning_rate": 4.9777182115686995e-06, "loss": 0.14744940400123596, "step": 568 }, { "epoch": 0.14771547248182762, "grad_norm": 1.4990284442901611, "learning_rate": 4.977585209060481e-06, "loss": 0.16039317846298218, "step": 569 }, { "epoch": 0.14797507788161993, "grad_norm": 2.627997636795044, "learning_rate": 4.977451812766203e-06, "loss": 0.1917704939842224, "step": 570 }, { "epoch": 0.14823468328141226, "grad_norm": 1.630319595336914, "learning_rate": 4.977318022709445e-06, "loss": 0.19095705449581146, "step": 571 }, { "epoch": 0.14849428868120457, "grad_norm": 2.1005563735961914, "learning_rate": 4.977183838913862e-06, "loss": 0.14820599555969238, "step": 572 }, { "epoch": 0.14875389408099687, "grad_norm": 2.36879563331604, "learning_rate": 4.97704926140317e-06, "loss": 0.14981324970722198, "step": 573 }, { "epoch": 0.1490134994807892, "grad_norm": 3.742894172668457, "learning_rate": 4.9769142902011615e-06, "loss": 0.16630589962005615, "step": 574 }, { "epoch": 0.14927310488058151, "grad_norm": 4.092385292053223, "learning_rate": 4.9767789253316964e-06, "loss": 0.1831657886505127, "step": 575 }, { "epoch": 0.14953271028037382, "grad_norm": 1.261873722076416, "learning_rate": 4.976643166818704e-06, "loss": 0.1420586109161377, "step": 576 }, { "epoch": 0.14979231568016615, "grad_norm": 4.7921271324157715, "learning_rate": 4.976507014686183e-06, "loss": 0.19693724811077118, "step": 577 }, { "epoch": 0.15005192107995846, "grad_norm": 1.4069263935089111, "learning_rate": 4.976370468958201e-06, "loss": 0.14249220490455627, "step": 578 }, { "epoch": 0.15031152647975077, "grad_norm": 1.4785072803497314, "learning_rate": 4.976233529658898e-06, "loss": 0.14328446984291077, "step": 579 }, { "epoch": 0.1505711318795431, "grad_norm": 2.0501253604888916, "learning_rate": 4.9760961968124815e-06, "loss": 0.155215322971344, "step": 580 }, { "epoch": 0.1508307372793354, "grad_norm": 2.2228729724884033, "learning_rate": 4.975958470443228e-06, "loss": 0.13664713501930237, "step": 581 }, { "epoch": 0.15109034267912771, "grad_norm": 0.8777739405632019, "learning_rate": 4.9758203505754845e-06, "loss": 0.16640238463878632, "step": 582 }, { "epoch": 0.15134994807892005, "grad_norm": 3.5218498706817627, "learning_rate": 4.975681837233668e-06, "loss": 0.13468728959560394, "step": 583 }, { "epoch": 0.15160955347871236, "grad_norm": 5.687356948852539, "learning_rate": 4.975542930442263e-06, "loss": 0.1702696979045868, "step": 584 }, { "epoch": 0.15186915887850466, "grad_norm": 3.59616756439209, "learning_rate": 4.975403630225827e-06, "loss": 0.18906691670417786, "step": 585 }, { "epoch": 0.152128764278297, "grad_norm": 3.0998053550720215, "learning_rate": 4.975263936608984e-06, "loss": 0.1221320852637291, "step": 586 }, { "epoch": 0.1523883696780893, "grad_norm": 3.3339226245880127, "learning_rate": 4.975123849616429e-06, "loss": 0.1476742923259735, "step": 587 }, { "epoch": 0.1526479750778816, "grad_norm": 1.7601169347763062, "learning_rate": 4.974983369272925e-06, "loss": 0.14323046803474426, "step": 588 }, { "epoch": 0.15290758047767394, "grad_norm": 7.065764427185059, "learning_rate": 4.9748424956033076e-06, "loss": 0.18637219071388245, "step": 589 }, { "epoch": 0.15316718587746625, "grad_norm": 3.8112173080444336, "learning_rate": 4.9747012286324784e-06, "loss": 0.13489459455013275, "step": 590 }, { "epoch": 0.15342679127725856, "grad_norm": 0.8728705644607544, "learning_rate": 4.9745595683854115e-06, "loss": 0.14231091737747192, "step": 591 }, { "epoch": 0.1536863966770509, "grad_norm": 4.241068363189697, "learning_rate": 4.974417514887148e-06, "loss": 0.170458123087883, "step": 592 }, { "epoch": 0.1539460020768432, "grad_norm": 5.4670305252075195, "learning_rate": 4.9742750681628e-06, "loss": 0.17236308753490448, "step": 593 }, { "epoch": 0.1542056074766355, "grad_norm": 0.874913215637207, "learning_rate": 4.97413222823755e-06, "loss": 0.14191952347755432, "step": 594 }, { "epoch": 0.15446521287642784, "grad_norm": 4.892989158630371, "learning_rate": 4.973988995136648e-06, "loss": 0.17631220817565918, "step": 595 }, { "epoch": 0.15472481827622014, "grad_norm": 2.8158822059631348, "learning_rate": 4.9738453688854145e-06, "loss": 0.1430496871471405, "step": 596 }, { "epoch": 0.15498442367601245, "grad_norm": 2.447906255722046, "learning_rate": 4.973701349509238e-06, "loss": 0.16812047362327576, "step": 597 }, { "epoch": 0.15524402907580478, "grad_norm": 1.4402189254760742, "learning_rate": 4.973556937033579e-06, "loss": 0.13005127012729645, "step": 598 }, { "epoch": 0.1555036344755971, "grad_norm": 3.285464286804199, "learning_rate": 4.973412131483966e-06, "loss": 0.14287739992141724, "step": 599 }, { "epoch": 0.1557632398753894, "grad_norm": 2.689537525177002, "learning_rate": 4.973266932885999e-06, "loss": 0.13861629366874695, "step": 600 }, { "epoch": 0.15602284527518173, "grad_norm": 1.6409128904342651, "learning_rate": 4.973121341265343e-06, "loss": 0.16324928402900696, "step": 601 }, { "epoch": 0.15628245067497404, "grad_norm": 1.6250011920928955, "learning_rate": 4.9729753566477376e-06, "loss": 0.17544081807136536, "step": 602 }, { "epoch": 0.15654205607476634, "grad_norm": 2.431417465209961, "learning_rate": 4.972828979058988e-06, "loss": 0.18938396871089935, "step": 603 }, { "epoch": 0.15680166147455868, "grad_norm": 3.566960573196411, "learning_rate": 4.972682208524971e-06, "loss": 0.15683212876319885, "step": 604 }, { "epoch": 0.15706126687435099, "grad_norm": 1.143561601638794, "learning_rate": 4.972535045071632e-06, "loss": 0.13867947459220886, "step": 605 }, { "epoch": 0.1573208722741433, "grad_norm": 3.293412923812866, "learning_rate": 4.972387488724986e-06, "loss": 0.14926454424858093, "step": 606 }, { "epoch": 0.15758047767393563, "grad_norm": 4.07432746887207, "learning_rate": 4.972239539511119e-06, "loss": 0.16576983034610748, "step": 607 }, { "epoch": 0.15784008307372793, "grad_norm": 1.5417015552520752, "learning_rate": 4.9720911974561835e-06, "loss": 0.14239105582237244, "step": 608 }, { "epoch": 0.15809968847352024, "grad_norm": 4.042348384857178, "learning_rate": 4.971942462586403e-06, "loss": 0.1712881326675415, "step": 609 }, { "epoch": 0.15835929387331257, "grad_norm": 4.7444610595703125, "learning_rate": 4.971793334928071e-06, "loss": 0.1395210176706314, "step": 610 }, { "epoch": 0.15861889927310488, "grad_norm": 3.56238055229187, "learning_rate": 4.9716438145075505e-06, "loss": 0.183371439576149, "step": 611 }, { "epoch": 0.1588785046728972, "grad_norm": 3.6256444454193115, "learning_rate": 4.971493901351272e-06, "loss": 0.16141541302204132, "step": 612 }, { "epoch": 0.15913811007268952, "grad_norm": 2.9284780025482178, "learning_rate": 4.971343595485736e-06, "loss": 0.20248635113239288, "step": 613 }, { "epoch": 0.15939771547248183, "grad_norm": 1.89718759059906, "learning_rate": 4.971192896937516e-06, "loss": 0.16097882390022278, "step": 614 }, { "epoch": 0.15965732087227413, "grad_norm": 0.7767889499664307, "learning_rate": 4.97104180573325e-06, "loss": 0.1404905617237091, "step": 615 }, { "epoch": 0.15991692627206647, "grad_norm": 3.5821008682250977, "learning_rate": 4.970890321899647e-06, "loss": 0.16317123174667358, "step": 616 }, { "epoch": 0.16017653167185877, "grad_norm": 1.529128909111023, "learning_rate": 4.970738445463488e-06, "loss": 0.1740923821926117, "step": 617 }, { "epoch": 0.16043613707165108, "grad_norm": 1.9485045671463013, "learning_rate": 4.9705861764516195e-06, "loss": 0.16981011629104614, "step": 618 }, { "epoch": 0.16069574247144341, "grad_norm": 2.801853656768799, "learning_rate": 4.97043351489096e-06, "loss": 0.13524675369262695, "step": 619 }, { "epoch": 0.16095534787123572, "grad_norm": 2.244046926498413, "learning_rate": 4.970280460808495e-06, "loss": 0.1787937581539154, "step": 620 }, { "epoch": 0.16121495327102803, "grad_norm": 4.769516468048096, "learning_rate": 4.970127014231283e-06, "loss": 0.14051033556461334, "step": 621 }, { "epoch": 0.16147455867082036, "grad_norm": 3.0804946422576904, "learning_rate": 4.969973175186449e-06, "loss": 0.16236986219882965, "step": 622 }, { "epoch": 0.16173416407061267, "grad_norm": 5.572113990783691, "learning_rate": 4.969818943701189e-06, "loss": 0.18247359991073608, "step": 623 }, { "epoch": 0.16199376947040497, "grad_norm": 3.563438892364502, "learning_rate": 4.969664319802766e-06, "loss": 0.14019417762756348, "step": 624 }, { "epoch": 0.1622533748701973, "grad_norm": 3.1188080310821533, "learning_rate": 4.9695093035185154e-06, "loss": 0.1639358252286911, "step": 625 }, { "epoch": 0.16251298026998962, "grad_norm": 2.4696128368377686, "learning_rate": 4.96935389487584e-06, "loss": 0.1264353096485138, "step": 626 }, { "epoch": 0.16277258566978192, "grad_norm": 3.7759876251220703, "learning_rate": 4.969198093902212e-06, "loss": 0.1665867269039154, "step": 627 }, { "epoch": 0.16303219106957426, "grad_norm": 1.0335386991500854, "learning_rate": 4.969041900625174e-06, "loss": 0.14665067195892334, "step": 628 }, { "epoch": 0.16329179646936656, "grad_norm": 2.2939391136169434, "learning_rate": 4.968885315072338e-06, "loss": 0.1531972587108612, "step": 629 }, { "epoch": 0.16355140186915887, "grad_norm": 2.1533381938934326, "learning_rate": 4.968728337271383e-06, "loss": 0.16868479549884796, "step": 630 }, { "epoch": 0.1638110072689512, "grad_norm": 1.6050641536712646, "learning_rate": 4.968570967250061e-06, "loss": 0.13505469262599945, "step": 631 }, { "epoch": 0.1640706126687435, "grad_norm": 1.278440237045288, "learning_rate": 4.9684132050361896e-06, "loss": 0.1414593607187271, "step": 632 }, { "epoch": 0.16433021806853582, "grad_norm": 2.4325149059295654, "learning_rate": 4.968255050657658e-06, "loss": 0.15594032406806946, "step": 633 }, { "epoch": 0.16458982346832815, "grad_norm": 1.7544941902160645, "learning_rate": 4.968096504142426e-06, "loss": 0.168398916721344, "step": 634 }, { "epoch": 0.16484942886812046, "grad_norm": 3.994680881500244, "learning_rate": 4.96793756551852e-06, "loss": 0.16368339955806732, "step": 635 }, { "epoch": 0.16510903426791276, "grad_norm": 3.076707601547241, "learning_rate": 4.967778234814035e-06, "loss": 0.1412467360496521, "step": 636 }, { "epoch": 0.1653686396677051, "grad_norm": 1.7792186737060547, "learning_rate": 4.96761851205714e-06, "loss": 0.17613843083381653, "step": 637 }, { "epoch": 0.1656282450674974, "grad_norm": 2.6651995182037354, "learning_rate": 4.9674583972760676e-06, "loss": 0.1746181696653366, "step": 638 }, { "epoch": 0.1658878504672897, "grad_norm": 2.1281895637512207, "learning_rate": 4.967297890499124e-06, "loss": 0.18735434114933014, "step": 639 }, { "epoch": 0.16614745586708204, "grad_norm": 3.004265069961548, "learning_rate": 4.967136991754683e-06, "loss": 0.13024160265922546, "step": 640 }, { "epoch": 0.16640706126687435, "grad_norm": 3.7307493686676025, "learning_rate": 4.966975701071188e-06, "loss": 0.14347386360168457, "step": 641 }, { "epoch": 0.16666666666666666, "grad_norm": 1.8170634508132935, "learning_rate": 4.96681401847715e-06, "loss": 0.16254568099975586, "step": 642 }, { "epoch": 0.166926272066459, "grad_norm": 2.640381097793579, "learning_rate": 4.966651944001153e-06, "loss": 0.18376685678958893, "step": 643 }, { "epoch": 0.1671858774662513, "grad_norm": 5.059758186340332, "learning_rate": 4.966489477671846e-06, "loss": 0.1443699300289154, "step": 644 }, { "epoch": 0.1674454828660436, "grad_norm": 2.7722458839416504, "learning_rate": 4.966326619517951e-06, "loss": 0.16779932379722595, "step": 645 }, { "epoch": 0.16770508826583594, "grad_norm": 4.485286712646484, "learning_rate": 4.966163369568256e-06, "loss": 0.13241076469421387, "step": 646 }, { "epoch": 0.16796469366562825, "grad_norm": 6.449831485748291, "learning_rate": 4.965999727851621e-06, "loss": 0.18787233531475067, "step": 647 }, { "epoch": 0.16822429906542055, "grad_norm": 1.1325451135635376, "learning_rate": 4.965835694396973e-06, "loss": 0.14463919401168823, "step": 648 }, { "epoch": 0.16848390446521289, "grad_norm": 1.938537836074829, "learning_rate": 4.965671269233312e-06, "loss": 0.17008303105831146, "step": 649 }, { "epoch": 0.1687435098650052, "grad_norm": 1.705251693725586, "learning_rate": 4.9655064523897015e-06, "loss": 0.15220513939857483, "step": 650 }, { "epoch": 0.1690031152647975, "grad_norm": 2.9700865745544434, "learning_rate": 4.965341243895279e-06, "loss": 0.20417314767837524, "step": 651 }, { "epoch": 0.16926272066458983, "grad_norm": 1.2573093175888062, "learning_rate": 4.9651756437792485e-06, "loss": 0.1807861328125, "step": 652 }, { "epoch": 0.16952232606438214, "grad_norm": 1.1604633331298828, "learning_rate": 4.965009652070885e-06, "loss": 0.18498916923999786, "step": 653 }, { "epoch": 0.16978193146417445, "grad_norm": 0.6475570797920227, "learning_rate": 4.964843268799533e-06, "loss": 0.13570433855056763, "step": 654 }, { "epoch": 0.17004153686396678, "grad_norm": 2.3335258960723877, "learning_rate": 4.964676493994604e-06, "loss": 0.1663290560245514, "step": 655 }, { "epoch": 0.1703011422637591, "grad_norm": 1.8894842863082886, "learning_rate": 4.964509327685579e-06, "loss": 0.13018761575222015, "step": 656 }, { "epoch": 0.1705607476635514, "grad_norm": 1.2265080213546753, "learning_rate": 4.964341769902012e-06, "loss": 0.1796884834766388, "step": 657 }, { "epoch": 0.17082035306334373, "grad_norm": 1.4832634925842285, "learning_rate": 4.964173820673521e-06, "loss": 0.12758612632751465, "step": 658 }, { "epoch": 0.17107995846313603, "grad_norm": 2.1791820526123047, "learning_rate": 4.964005480029797e-06, "loss": 0.13354802131652832, "step": 659 }, { "epoch": 0.17133956386292834, "grad_norm": 1.3497661352157593, "learning_rate": 4.963836748000597e-06, "loss": 0.1462550163269043, "step": 660 }, { "epoch": 0.17159916926272067, "grad_norm": 1.9515339136123657, "learning_rate": 4.963667624615751e-06, "loss": 0.15267646312713623, "step": 661 }, { "epoch": 0.17185877466251298, "grad_norm": 1.1643575429916382, "learning_rate": 4.9634981099051545e-06, "loss": 0.14950358867645264, "step": 662 }, { "epoch": 0.1721183800623053, "grad_norm": 2.621387243270874, "learning_rate": 4.963328203898775e-06, "loss": 0.16176219284534454, "step": 663 }, { "epoch": 0.17237798546209762, "grad_norm": 3.1328020095825195, "learning_rate": 4.963157906626648e-06, "loss": 0.16163554787635803, "step": 664 }, { "epoch": 0.17263759086188993, "grad_norm": 1.6968307495117188, "learning_rate": 4.9629872181188774e-06, "loss": 0.15151429176330566, "step": 665 }, { "epoch": 0.17289719626168223, "grad_norm": 2.043006658554077, "learning_rate": 4.962816138405637e-06, "loss": 0.14885354042053223, "step": 666 }, { "epoch": 0.17315680166147457, "grad_norm": 2.9228100776672363, "learning_rate": 4.9626446675171704e-06, "loss": 0.1378011852502823, "step": 667 }, { "epoch": 0.17341640706126688, "grad_norm": 3.0615415573120117, "learning_rate": 4.962472805483789e-06, "loss": 0.14774253964424133, "step": 668 }, { "epoch": 0.17367601246105918, "grad_norm": 1.881534457206726, "learning_rate": 4.962300552335875e-06, "loss": 0.15725667774677277, "step": 669 }, { "epoch": 0.17393561786085152, "grad_norm": 5.556959629058838, "learning_rate": 4.962127908103878e-06, "loss": 0.18054935336112976, "step": 670 }, { "epoch": 0.17419522326064382, "grad_norm": 1.433811902999878, "learning_rate": 4.961954872818318e-06, "loss": 0.15963834524154663, "step": 671 }, { "epoch": 0.17445482866043613, "grad_norm": 2.4027116298675537, "learning_rate": 4.961781446509784e-06, "loss": 0.18362829089164734, "step": 672 }, { "epoch": 0.17471443406022846, "grad_norm": 1.117735505104065, "learning_rate": 4.961607629208933e-06, "loss": 0.12363755702972412, "step": 673 }, { "epoch": 0.17497403946002077, "grad_norm": 2.7063803672790527, "learning_rate": 4.961433420946492e-06, "loss": 0.1929568499326706, "step": 674 }, { "epoch": 0.17523364485981308, "grad_norm": 2.0945677757263184, "learning_rate": 4.961258821753257e-06, "loss": 0.14318422973155975, "step": 675 }, { "epoch": 0.1754932502596054, "grad_norm": 3.8866469860076904, "learning_rate": 4.961083831660095e-06, "loss": 0.1547536998987198, "step": 676 }, { "epoch": 0.17575285565939772, "grad_norm": 3.319244623184204, "learning_rate": 4.960908450697937e-06, "loss": 0.18098962306976318, "step": 677 }, { "epoch": 0.17601246105919002, "grad_norm": 2.2167835235595703, "learning_rate": 4.960732678897788e-06, "loss": 0.15388521552085876, "step": 678 }, { "epoch": 0.17627206645898236, "grad_norm": 8.007393836975098, "learning_rate": 4.960556516290721e-06, "loss": 0.18519365787506104, "step": 679 }, { "epoch": 0.17653167185877466, "grad_norm": 1.8834011554718018, "learning_rate": 4.960379962907878e-06, "loss": 0.1504184901714325, "step": 680 }, { "epoch": 0.17679127725856697, "grad_norm": 2.939762830734253, "learning_rate": 4.960203018780467e-06, "loss": 0.14833174645900726, "step": 681 }, { "epoch": 0.1770508826583593, "grad_norm": 6.142423629760742, "learning_rate": 4.9600256839397695e-06, "loss": 0.17820106446743011, "step": 682 }, { "epoch": 0.1773104880581516, "grad_norm": 3.815699815750122, "learning_rate": 4.959847958417134e-06, "loss": 0.17030736804008484, "step": 683 }, { "epoch": 0.17757009345794392, "grad_norm": 4.002564907073975, "learning_rate": 4.959669842243979e-06, "loss": 0.17746210098266602, "step": 684 }, { "epoch": 0.17782969885773625, "grad_norm": 5.115752696990967, "learning_rate": 4.95949133545179e-06, "loss": 0.20600220561027527, "step": 685 }, { "epoch": 0.17808930425752856, "grad_norm": 4.070754528045654, "learning_rate": 4.959312438072123e-06, "loss": 0.1593736857175827, "step": 686 }, { "epoch": 0.17834890965732086, "grad_norm": 5.685986518859863, "learning_rate": 4.959133150136605e-06, "loss": 0.19910694658756256, "step": 687 }, { "epoch": 0.1786085150571132, "grad_norm": 3.743213415145874, "learning_rate": 4.958953471676927e-06, "loss": 0.16497129201889038, "step": 688 }, { "epoch": 0.1788681204569055, "grad_norm": 5.547244548797607, "learning_rate": 4.958773402724854e-06, "loss": 0.17044761776924133, "step": 689 }, { "epoch": 0.1791277258566978, "grad_norm": 5.325279235839844, "learning_rate": 4.958592943312218e-06, "loss": 0.1555238664150238, "step": 690 }, { "epoch": 0.17938733125649015, "grad_norm": 6.8043694496154785, "learning_rate": 4.958412093470919e-06, "loss": 0.20715372264385223, "step": 691 }, { "epoch": 0.17964693665628245, "grad_norm": 2.029676914215088, "learning_rate": 4.958230853232928e-06, "loss": 0.17712348699569702, "step": 692 }, { "epoch": 0.17990654205607476, "grad_norm": 4.2031450271606445, "learning_rate": 4.958049222630285e-06, "loss": 0.1894916445016861, "step": 693 }, { "epoch": 0.1801661474558671, "grad_norm": 11.198884963989258, "learning_rate": 4.957867201695096e-06, "loss": 0.1970914602279663, "step": 694 }, { "epoch": 0.1804257528556594, "grad_norm": 5.055878162384033, "learning_rate": 4.957684790459539e-06, "loss": 0.15540093183517456, "step": 695 }, { "epoch": 0.1806853582554517, "grad_norm": 1.530955195426941, "learning_rate": 4.957501988955861e-06, "loss": 0.187856525182724, "step": 696 }, { "epoch": 0.18094496365524404, "grad_norm": 2.822929859161377, "learning_rate": 4.9573187972163765e-06, "loss": 0.143517404794693, "step": 697 }, { "epoch": 0.18120456905503635, "grad_norm": 5.150654315948486, "learning_rate": 4.9571352152734696e-06, "loss": 0.16692832112312317, "step": 698 }, { "epoch": 0.18146417445482865, "grad_norm": 4.626124382019043, "learning_rate": 4.956951243159593e-06, "loss": 0.15490129590034485, "step": 699 }, { "epoch": 0.181723779854621, "grad_norm": 3.6985654830932617, "learning_rate": 4.956766880907269e-06, "loss": 0.15215985476970673, "step": 700 }, { "epoch": 0.1819833852544133, "grad_norm": 1.3592275381088257, "learning_rate": 4.956582128549089e-06, "loss": 0.13631075620651245, "step": 701 }, { "epoch": 0.1822429906542056, "grad_norm": 2.453643560409546, "learning_rate": 4.9563969861177126e-06, "loss": 0.17221879959106445, "step": 702 }, { "epoch": 0.18250259605399793, "grad_norm": 3.524130344390869, "learning_rate": 4.9562114536458686e-06, "loss": 0.1861657351255417, "step": 703 }, { "epoch": 0.18276220145379024, "grad_norm": 4.461541175842285, "learning_rate": 4.956025531166356e-06, "loss": 0.15439657866954803, "step": 704 }, { "epoch": 0.18302180685358255, "grad_norm": 2.049950361251831, "learning_rate": 4.955839218712039e-06, "loss": 0.16491739451885223, "step": 705 }, { "epoch": 0.18328141225337488, "grad_norm": 2.0396053791046143, "learning_rate": 4.955652516315856e-06, "loss": 0.17260399460792542, "step": 706 }, { "epoch": 0.1835410176531672, "grad_norm": 1.0969191789627075, "learning_rate": 4.955465424010812e-06, "loss": 0.14096906781196594, "step": 707 }, { "epoch": 0.1838006230529595, "grad_norm": 3.187723398208618, "learning_rate": 4.955277941829978e-06, "loss": 0.13972122967243195, "step": 708 }, { "epoch": 0.18406022845275183, "grad_norm": 3.594277858734131, "learning_rate": 4.955090069806499e-06, "loss": 0.16320841014385223, "step": 709 }, { "epoch": 0.18431983385254413, "grad_norm": 2.8075778484344482, "learning_rate": 4.954901807973586e-06, "loss": 0.12594783306121826, "step": 710 }, { "epoch": 0.18457943925233644, "grad_norm": 1.7943955659866333, "learning_rate": 4.954713156364517e-06, "loss": 0.140581414103508, "step": 711 }, { "epoch": 0.18483904465212878, "grad_norm": 2.9889841079711914, "learning_rate": 4.954524115012645e-06, "loss": 0.1463010609149933, "step": 712 }, { "epoch": 0.18509865005192108, "grad_norm": 3.565938949584961, "learning_rate": 4.9543346839513856e-06, "loss": 0.14806053042411804, "step": 713 }, { "epoch": 0.1853582554517134, "grad_norm": 2.056234121322632, "learning_rate": 4.954144863214226e-06, "loss": 0.14708393812179565, "step": 714 }, { "epoch": 0.18561786085150572, "grad_norm": 1.2089844942092896, "learning_rate": 4.953954652834724e-06, "loss": 0.1475439965724945, "step": 715 }, { "epoch": 0.18587746625129803, "grad_norm": 1.6442136764526367, "learning_rate": 4.953764052846502e-06, "loss": 0.15212489664554596, "step": 716 }, { "epoch": 0.18613707165109034, "grad_norm": 3.171586275100708, "learning_rate": 4.953573063283257e-06, "loss": 0.14387434720993042, "step": 717 }, { "epoch": 0.18639667705088267, "grad_norm": 2.3936800956726074, "learning_rate": 4.953381684178748e-06, "loss": 0.13548313081264496, "step": 718 }, { "epoch": 0.18665628245067498, "grad_norm": 2.9637253284454346, "learning_rate": 4.9531899155668076e-06, "loss": 0.16230641305446625, "step": 719 }, { "epoch": 0.18691588785046728, "grad_norm": 2.802398920059204, "learning_rate": 4.952997757481338e-06, "loss": 0.1713087260723114, "step": 720 }, { "epoch": 0.18717549325025962, "grad_norm": 1.9056416749954224, "learning_rate": 4.952805209956306e-06, "loss": 0.19507166743278503, "step": 721 }, { "epoch": 0.18743509865005192, "grad_norm": 2.507052421569824, "learning_rate": 4.95261227302575e-06, "loss": 0.1915706992149353, "step": 722 }, { "epoch": 0.18769470404984423, "grad_norm": 2.952115535736084, "learning_rate": 4.952418946723778e-06, "loss": 0.15594014525413513, "step": 723 }, { "epoch": 0.18795430944963656, "grad_norm": 1.0056740045547485, "learning_rate": 4.952225231084565e-06, "loss": 0.14498737454414368, "step": 724 }, { "epoch": 0.18821391484942887, "grad_norm": 2.732292652130127, "learning_rate": 4.9520311261423556e-06, "loss": 0.169405996799469, "step": 725 }, { "epoch": 0.18847352024922118, "grad_norm": 3.4126155376434326, "learning_rate": 4.951836631931461e-06, "loss": 0.15460290014743805, "step": 726 }, { "epoch": 0.1887331256490135, "grad_norm": 3.1835720539093018, "learning_rate": 4.951641748486268e-06, "loss": 0.13483497500419617, "step": 727 }, { "epoch": 0.18899273104880582, "grad_norm": 1.991335153579712, "learning_rate": 4.951446475841223e-06, "loss": 0.16579541563987732, "step": 728 }, { "epoch": 0.18925233644859812, "grad_norm": 2.1125965118408203, "learning_rate": 4.951250814030847e-06, "loss": 0.19295397400856018, "step": 729 }, { "epoch": 0.18951194184839046, "grad_norm": 2.6507396697998047, "learning_rate": 4.951054763089731e-06, "loss": 0.13889482617378235, "step": 730 }, { "epoch": 0.18977154724818276, "grad_norm": 3.7656304836273193, "learning_rate": 4.95085832305253e-06, "loss": 0.1956333965063095, "step": 731 }, { "epoch": 0.19003115264797507, "grad_norm": 2.4912517070770264, "learning_rate": 4.950661493953969e-06, "loss": 0.1774275302886963, "step": 732 }, { "epoch": 0.1902907580477674, "grad_norm": 2.697248935699463, "learning_rate": 4.950464275828845e-06, "loss": 0.1345430612564087, "step": 733 }, { "epoch": 0.1905503634475597, "grad_norm": 3.007168769836426, "learning_rate": 4.9502666687120215e-06, "loss": 0.14123135805130005, "step": 734 }, { "epoch": 0.19080996884735202, "grad_norm": 2.748274564743042, "learning_rate": 4.95006867263843e-06, "loss": 0.14475218951702118, "step": 735 }, { "epoch": 0.19106957424714435, "grad_norm": 3.295030117034912, "learning_rate": 4.949870287643072e-06, "loss": 0.1532915085554123, "step": 736 }, { "epoch": 0.19132917964693666, "grad_norm": 1.6636075973510742, "learning_rate": 4.949671513761018e-06, "loss": 0.14637303352355957, "step": 737 }, { "epoch": 0.19158878504672897, "grad_norm": 1.745643138885498, "learning_rate": 4.949472351027405e-06, "loss": 0.16271524131298065, "step": 738 }, { "epoch": 0.1918483904465213, "grad_norm": 1.89266037940979, "learning_rate": 4.949272799477442e-06, "loss": 0.15129487216472626, "step": 739 }, { "epoch": 0.1921079958463136, "grad_norm": 1.7598381042480469, "learning_rate": 4.949072859146406e-06, "loss": 0.13269385695457458, "step": 740 }, { "epoch": 0.1923676012461059, "grad_norm": 2.909017324447632, "learning_rate": 4.948872530069639e-06, "loss": 0.15349891781806946, "step": 741 }, { "epoch": 0.19262720664589825, "grad_norm": 1.2617806196212769, "learning_rate": 4.948671812282557e-06, "loss": 0.1558079868555069, "step": 742 }, { "epoch": 0.19288681204569055, "grad_norm": 1.645241141319275, "learning_rate": 4.948470705820642e-06, "loss": 0.14354169368743896, "step": 743 }, { "epoch": 0.19314641744548286, "grad_norm": 1.1495155096054077, "learning_rate": 4.948269210719443e-06, "loss": 0.14629808068275452, "step": 744 }, { "epoch": 0.1934060228452752, "grad_norm": 2.537968158721924, "learning_rate": 4.948067327014583e-06, "loss": 0.13258527219295502, "step": 745 }, { "epoch": 0.1936656282450675, "grad_norm": 7.789155960083008, "learning_rate": 4.947865054741748e-06, "loss": 0.14940127730369568, "step": 746 }, { "epoch": 0.1939252336448598, "grad_norm": 3.060382604598999, "learning_rate": 4.947662393936695e-06, "loss": 0.19654959440231323, "step": 747 }, { "epoch": 0.19418483904465214, "grad_norm": 5.403552532196045, "learning_rate": 4.947459344635251e-06, "loss": 0.1604142189025879, "step": 748 }, { "epoch": 0.19444444444444445, "grad_norm": 3.683692693710327, "learning_rate": 4.9472559068733104e-06, "loss": 0.1624574512243271, "step": 749 }, { "epoch": 0.19470404984423675, "grad_norm": 3.245325803756714, "learning_rate": 4.9470520806868356e-06, "loss": 0.15699315071105957, "step": 750 }, { "epoch": 0.1949636552440291, "grad_norm": 3.0018246173858643, "learning_rate": 4.946847866111859e-06, "loss": 0.1416621059179306, "step": 751 }, { "epoch": 0.1952232606438214, "grad_norm": 1.6931846141815186, "learning_rate": 4.9466432631844805e-06, "loss": 0.13667243719100952, "step": 752 }, { "epoch": 0.1954828660436137, "grad_norm": 4.497149467468262, "learning_rate": 4.946438271940869e-06, "loss": 0.15599729120731354, "step": 753 }, { "epoch": 0.19574247144340604, "grad_norm": 3.7798948287963867, "learning_rate": 4.946232892417264e-06, "loss": 0.13993939757347107, "step": 754 }, { "epoch": 0.19600207684319834, "grad_norm": 2.2770519256591797, "learning_rate": 4.94602712464997e-06, "loss": 0.2002446949481964, "step": 755 }, { "epoch": 0.19626168224299065, "grad_norm": 4.7828474044799805, "learning_rate": 4.945820968675363e-06, "loss": 0.17854604125022888, "step": 756 }, { "epoch": 0.19652128764278298, "grad_norm": 5.145047664642334, "learning_rate": 4.945614424529886e-06, "loss": 0.14924287796020508, "step": 757 }, { "epoch": 0.1967808930425753, "grad_norm": 4.630312919616699, "learning_rate": 4.945407492250052e-06, "loss": 0.19336295127868652, "step": 758 }, { "epoch": 0.1970404984423676, "grad_norm": 1.0283145904541016, "learning_rate": 4.945200171872442e-06, "loss": 0.12273910641670227, "step": 759 }, { "epoch": 0.19730010384215993, "grad_norm": 1.7363466024398804, "learning_rate": 4.944992463433704e-06, "loss": 0.14490409195423126, "step": 760 }, { "epoch": 0.19755970924195224, "grad_norm": 7.316575527191162, "learning_rate": 4.944784366970558e-06, "loss": 0.15837591886520386, "step": 761 }, { "epoch": 0.19781931464174454, "grad_norm": 3.1281673908233643, "learning_rate": 4.944575882519789e-06, "loss": 0.13781790435314178, "step": 762 }, { "epoch": 0.19807892004153688, "grad_norm": 1.2333273887634277, "learning_rate": 4.944367010118254e-06, "loss": 0.16850176453590393, "step": 763 }, { "epoch": 0.19833852544132918, "grad_norm": 3.0381503105163574, "learning_rate": 4.944157749802876e-06, "loss": 0.17324228584766388, "step": 764 }, { "epoch": 0.1985981308411215, "grad_norm": 1.7431823015213013, "learning_rate": 4.9439481016106465e-06, "loss": 0.1641484647989273, "step": 765 }, { "epoch": 0.19885773624091382, "grad_norm": 4.638725757598877, "learning_rate": 4.943738065578628e-06, "loss": 0.20277249813079834, "step": 766 }, { "epoch": 0.19911734164070613, "grad_norm": 2.68569016456604, "learning_rate": 4.943527641743949e-06, "loss": 0.16428686678409576, "step": 767 }, { "epoch": 0.19937694704049844, "grad_norm": 1.376389980316162, "learning_rate": 4.9433168301438084e-06, "loss": 0.15708574652671814, "step": 768 }, { "epoch": 0.19963655244029077, "grad_norm": 2.895153284072876, "learning_rate": 4.943105630815472e-06, "loss": 0.1698860079050064, "step": 769 }, { "epoch": 0.19989615784008308, "grad_norm": 4.280593395233154, "learning_rate": 4.942894043796276e-06, "loss": 0.15120628476142883, "step": 770 }, { "epoch": 0.20015576323987538, "grad_norm": 1.1333955526351929, "learning_rate": 4.942682069123623e-06, "loss": 0.13323985040187836, "step": 771 }, { "epoch": 0.20041536863966772, "grad_norm": 1.7880268096923828, "learning_rate": 4.942469706834987e-06, "loss": 0.1309472769498825, "step": 772 }, { "epoch": 0.20067497403946002, "grad_norm": 1.2510486841201782, "learning_rate": 4.942256956967907e-06, "loss": 0.1636420488357544, "step": 773 }, { "epoch": 0.20093457943925233, "grad_norm": 4.209747791290283, "learning_rate": 4.942043819559993e-06, "loss": 0.17814870178699493, "step": 774 }, { "epoch": 0.20119418483904467, "grad_norm": 3.910644054412842, "learning_rate": 4.941830294648924e-06, "loss": 0.14358699321746826, "step": 775 }, { "epoch": 0.20145379023883697, "grad_norm": 2.7922580242156982, "learning_rate": 4.941616382272444e-06, "loss": 0.1323028951883316, "step": 776 }, { "epoch": 0.20171339563862928, "grad_norm": 1.5550202131271362, "learning_rate": 4.94140208246837e-06, "loss": 0.14958223700523376, "step": 777 }, { "epoch": 0.2019730010384216, "grad_norm": 2.153350591659546, "learning_rate": 4.941187395274584e-06, "loss": 0.1515781134366989, "step": 778 }, { "epoch": 0.20223260643821392, "grad_norm": 1.086204171180725, "learning_rate": 4.940972320729036e-06, "loss": 0.13639438152313232, "step": 779 }, { "epoch": 0.20249221183800623, "grad_norm": 1.0451949834823608, "learning_rate": 4.940756858869751e-06, "loss": 0.13205945491790771, "step": 780 }, { "epoch": 0.20275181723779856, "grad_norm": 2.122274875640869, "learning_rate": 4.940541009734814e-06, "loss": 0.18265405297279358, "step": 781 }, { "epoch": 0.20301142263759087, "grad_norm": 1.1470533609390259, "learning_rate": 4.940324773362383e-06, "loss": 0.14647865295410156, "step": 782 }, { "epoch": 0.20327102803738317, "grad_norm": 2.1747052669525146, "learning_rate": 4.940108149790684e-06, "loss": 0.17479760944843292, "step": 783 }, { "epoch": 0.2035306334371755, "grad_norm": 2.4125683307647705, "learning_rate": 4.939891139058011e-06, "loss": 0.1327369511127472, "step": 784 }, { "epoch": 0.2037902388369678, "grad_norm": 2.161757230758667, "learning_rate": 4.939673741202726e-06, "loss": 0.1786506175994873, "step": 785 }, { "epoch": 0.20404984423676012, "grad_norm": 1.0412883758544922, "learning_rate": 4.939455956263262e-06, "loss": 0.12333005666732788, "step": 786 }, { "epoch": 0.20430944963655245, "grad_norm": 3.0833349227905273, "learning_rate": 4.9392377842781156e-06, "loss": 0.15023472905158997, "step": 787 }, { "epoch": 0.20456905503634476, "grad_norm": 1.6082741022109985, "learning_rate": 4.939019225285856e-06, "loss": 0.1608155220746994, "step": 788 }, { "epoch": 0.20482866043613707, "grad_norm": 1.0133849382400513, "learning_rate": 4.938800279325119e-06, "loss": 0.14084790647029877, "step": 789 }, { "epoch": 0.2050882658359294, "grad_norm": 1.4704010486602783, "learning_rate": 4.93858094643461e-06, "loss": 0.142598956823349, "step": 790 }, { "epoch": 0.2053478712357217, "grad_norm": 2.9298925399780273, "learning_rate": 4.938361226653101e-06, "loss": 0.1498049944639206, "step": 791 }, { "epoch": 0.205607476635514, "grad_norm": 2.8293235301971436, "learning_rate": 4.938141120019435e-06, "loss": 0.15308767557144165, "step": 792 }, { "epoch": 0.20586708203530635, "grad_norm": 3.2531356811523438, "learning_rate": 4.9379206265725196e-06, "loss": 0.13574251532554626, "step": 793 }, { "epoch": 0.20612668743509865, "grad_norm": 5.71563720703125, "learning_rate": 4.937699746351334e-06, "loss": 0.13789597153663635, "step": 794 }, { "epoch": 0.20638629283489096, "grad_norm": 2.2611587047576904, "learning_rate": 4.937478479394925e-06, "loss": 0.15034985542297363, "step": 795 }, { "epoch": 0.2066458982346833, "grad_norm": 2.2849838733673096, "learning_rate": 4.937256825742409e-06, "loss": 0.13956809043884277, "step": 796 }, { "epoch": 0.2069055036344756, "grad_norm": 3.1259899139404297, "learning_rate": 4.937034785432967e-06, "loss": 0.161848783493042, "step": 797 }, { "epoch": 0.2071651090342679, "grad_norm": 2.6949429512023926, "learning_rate": 4.936812358505851e-06, "loss": 0.15118756890296936, "step": 798 }, { "epoch": 0.20742471443406024, "grad_norm": 1.6194558143615723, "learning_rate": 4.936589545000381e-06, "loss": 0.14215701818466187, "step": 799 }, { "epoch": 0.20768431983385255, "grad_norm": 1.3541743755340576, "learning_rate": 4.936366344955947e-06, "loss": 0.16043809056282043, "step": 800 }, { "epoch": 0.20794392523364486, "grad_norm": 2.751011371612549, "learning_rate": 4.936142758412004e-06, "loss": 0.134968101978302, "step": 801 }, { "epoch": 0.2082035306334372, "grad_norm": 1.0176416635513306, "learning_rate": 4.935918785408076e-06, "loss": 0.14677457511425018, "step": 802 }, { "epoch": 0.2084631360332295, "grad_norm": 2.1661949157714844, "learning_rate": 4.935694425983759e-06, "loss": 0.17071965336799622, "step": 803 }, { "epoch": 0.2087227414330218, "grad_norm": 4.019558429718018, "learning_rate": 4.935469680178713e-06, "loss": 0.1980905830860138, "step": 804 }, { "epoch": 0.2089823468328141, "grad_norm": 2.9907102584838867, "learning_rate": 4.9352445480326675e-06, "loss": 0.17206215858459473, "step": 805 }, { "epoch": 0.20924195223260644, "grad_norm": 2.367276906967163, "learning_rate": 4.935019029585421e-06, "loss": 0.1685027778148651, "step": 806 }, { "epoch": 0.20950155763239875, "grad_norm": 2.7409794330596924, "learning_rate": 4.934793124876842e-06, "loss": 0.17652791738510132, "step": 807 }, { "epoch": 0.20976116303219106, "grad_norm": 2.1391351222991943, "learning_rate": 4.934566833946863e-06, "loss": 0.13999679684638977, "step": 808 }, { "epoch": 0.2100207684319834, "grad_norm": 2.6468729972839355, "learning_rate": 4.934340156835488e-06, "loss": 0.1463983654975891, "step": 809 }, { "epoch": 0.2102803738317757, "grad_norm": 3.0219082832336426, "learning_rate": 4.934113093582787e-06, "loss": 0.1479005515575409, "step": 810 }, { "epoch": 0.210539979231568, "grad_norm": 1.198811411857605, "learning_rate": 4.933885644228902e-06, "loss": 0.11725498735904694, "step": 811 }, { "epoch": 0.21079958463136034, "grad_norm": 4.074282646179199, "learning_rate": 4.933657808814038e-06, "loss": 0.17392607033252716, "step": 812 }, { "epoch": 0.21105919003115264, "grad_norm": 3.268251657485962, "learning_rate": 4.933429587378473e-06, "loss": 0.14221017062664032, "step": 813 }, { "epoch": 0.21131879543094495, "grad_norm": 1.5106528997421265, "learning_rate": 4.933200979962551e-06, "loss": 0.1352197378873825, "step": 814 }, { "epoch": 0.21157840083073728, "grad_norm": 2.8233625888824463, "learning_rate": 4.932971986606685e-06, "loss": 0.13207153975963593, "step": 815 }, { "epoch": 0.2118380062305296, "grad_norm": 1.354764699935913, "learning_rate": 4.9327426073513555e-06, "loss": 0.12942273914813995, "step": 816 }, { "epoch": 0.2120976116303219, "grad_norm": 1.7366182804107666, "learning_rate": 4.9325128422371095e-06, "loss": 0.12352126091718674, "step": 817 }, { "epoch": 0.21235721703011423, "grad_norm": 0.9631526470184326, "learning_rate": 4.932282691304567e-06, "loss": 0.12478172779083252, "step": 818 }, { "epoch": 0.21261682242990654, "grad_norm": 1.5539448261260986, "learning_rate": 4.932052154594413e-06, "loss": 0.1782074272632599, "step": 819 }, { "epoch": 0.21287642782969884, "grad_norm": 2.1883864402770996, "learning_rate": 4.931821232147399e-06, "loss": 0.18060094118118286, "step": 820 }, { "epoch": 0.21313603322949118, "grad_norm": 1.563141942024231, "learning_rate": 4.931589924004349e-06, "loss": 0.17181339859962463, "step": 821 }, { "epoch": 0.21339563862928349, "grad_norm": 3.1819047927856445, "learning_rate": 4.931358230206152e-06, "loss": 0.12615704536437988, "step": 822 }, { "epoch": 0.2136552440290758, "grad_norm": 2.034116744995117, "learning_rate": 4.931126150793767e-06, "loss": 0.16377493739128113, "step": 823 }, { "epoch": 0.21391484942886813, "grad_norm": 0.880215048789978, "learning_rate": 4.930893685808219e-06, "loss": 0.13217875361442566, "step": 824 }, { "epoch": 0.21417445482866043, "grad_norm": 1.1776665449142456, "learning_rate": 4.930660835290603e-06, "loss": 0.14380526542663574, "step": 825 }, { "epoch": 0.21443406022845274, "grad_norm": 2.8170173168182373, "learning_rate": 4.930427599282082e-06, "loss": 0.1760464906692505, "step": 826 }, { "epoch": 0.21469366562824507, "grad_norm": 1.222659707069397, "learning_rate": 4.930193977823887e-06, "loss": 0.13633209466934204, "step": 827 }, { "epoch": 0.21495327102803738, "grad_norm": 1.053088903427124, "learning_rate": 4.929959970957316e-06, "loss": 0.15893366932868958, "step": 828 }, { "epoch": 0.21521287642782969, "grad_norm": 3.020416259765625, "learning_rate": 4.929725578723738e-06, "loss": 0.15191788971424103, "step": 829 }, { "epoch": 0.21547248182762202, "grad_norm": 1.052577018737793, "learning_rate": 4.929490801164585e-06, "loss": 0.15263579785823822, "step": 830 }, { "epoch": 0.21573208722741433, "grad_norm": 2.1771090030670166, "learning_rate": 4.929255638321363e-06, "loss": 0.13782045245170593, "step": 831 }, { "epoch": 0.21599169262720663, "grad_norm": 0.7113531231880188, "learning_rate": 4.9290200902356424e-06, "loss": 0.129910409450531, "step": 832 }, { "epoch": 0.21625129802699897, "grad_norm": 0.9578180909156799, "learning_rate": 4.928784156949062e-06, "loss": 0.13688024878501892, "step": 833 }, { "epoch": 0.21651090342679127, "grad_norm": 1.2829830646514893, "learning_rate": 4.928547838503331e-06, "loss": 0.16487625241279602, "step": 834 }, { "epoch": 0.21677050882658358, "grad_norm": 1.031772494316101, "learning_rate": 4.928311134940224e-06, "loss": 0.13467241823673248, "step": 835 }, { "epoch": 0.21703011422637591, "grad_norm": 2.513284206390381, "learning_rate": 4.928074046301586e-06, "loss": 0.16769133508205414, "step": 836 }, { "epoch": 0.21728971962616822, "grad_norm": 1.7433719635009766, "learning_rate": 4.9278365726293274e-06, "loss": 0.16080935299396515, "step": 837 }, { "epoch": 0.21754932502596053, "grad_norm": 2.525892734527588, "learning_rate": 4.927598713965428e-06, "loss": 0.12803992629051208, "step": 838 }, { "epoch": 0.21780893042575286, "grad_norm": 2.007791042327881, "learning_rate": 4.927360470351937e-06, "loss": 0.16998663544654846, "step": 839 }, { "epoch": 0.21806853582554517, "grad_norm": 1.062831163406372, "learning_rate": 4.927121841830969e-06, "loss": 0.13085174560546875, "step": 840 }, { "epoch": 0.21832814122533747, "grad_norm": 1.1744836568832397, "learning_rate": 4.926882828444709e-06, "loss": 0.12744295597076416, "step": 841 }, { "epoch": 0.2185877466251298, "grad_norm": 1.1278977394104004, "learning_rate": 4.92664343023541e-06, "loss": 0.135942742228508, "step": 842 }, { "epoch": 0.21884735202492211, "grad_norm": 2.6182684898376465, "learning_rate": 4.926403647245391e-06, "loss": 0.15433835983276367, "step": 843 }, { "epoch": 0.21910695742471442, "grad_norm": 1.2703752517700195, "learning_rate": 4.926163479517039e-06, "loss": 0.13619937002658844, "step": 844 }, { "epoch": 0.21936656282450676, "grad_norm": 3.076341390609741, "learning_rate": 4.9259229270928115e-06, "loss": 0.15933530032634735, "step": 845 }, { "epoch": 0.21962616822429906, "grad_norm": 1.6837691068649292, "learning_rate": 4.925681990015233e-06, "loss": 0.1551382839679718, "step": 846 }, { "epoch": 0.21988577362409137, "grad_norm": 1.6569993495941162, "learning_rate": 4.9254406683268955e-06, "loss": 0.17663908004760742, "step": 847 }, { "epoch": 0.2201453790238837, "grad_norm": 4.082376956939697, "learning_rate": 4.925198962070459e-06, "loss": 0.16840018332004547, "step": 848 }, { "epoch": 0.220404984423676, "grad_norm": 2.3027331829071045, "learning_rate": 4.92495687128865e-06, "loss": 0.1469523012638092, "step": 849 }, { "epoch": 0.22066458982346832, "grad_norm": 3.0805916786193848, "learning_rate": 4.924714396024267e-06, "loss": 0.17529305815696716, "step": 850 }, { "epoch": 0.22092419522326065, "grad_norm": 3.5256221294403076, "learning_rate": 4.9244715363201735e-06, "loss": 0.16387152671813965, "step": 851 }, { "epoch": 0.22118380062305296, "grad_norm": 1.2077308893203735, "learning_rate": 4.924228292219301e-06, "loss": 0.1420653760433197, "step": 852 }, { "epoch": 0.22144340602284526, "grad_norm": 0.7885842323303223, "learning_rate": 4.923984663764649e-06, "loss": 0.12602418661117554, "step": 853 }, { "epoch": 0.2217030114226376, "grad_norm": 1.743175745010376, "learning_rate": 4.923740650999287e-06, "loss": 0.14573659002780914, "step": 854 }, { "epoch": 0.2219626168224299, "grad_norm": 0.885120153427124, "learning_rate": 4.92349625396635e-06, "loss": 0.13144522905349731, "step": 855 }, { "epoch": 0.2222222222222222, "grad_norm": 2.2932660579681396, "learning_rate": 4.923251472709041e-06, "loss": 0.14660102128982544, "step": 856 }, { "epoch": 0.22248182762201454, "grad_norm": 3.1675875186920166, "learning_rate": 4.923006307270632e-06, "loss": 0.17930305004119873, "step": 857 }, { "epoch": 0.22274143302180685, "grad_norm": 1.94259774684906, "learning_rate": 4.922760757694464e-06, "loss": 0.13780882954597473, "step": 858 }, { "epoch": 0.22300103842159916, "grad_norm": 0.956764280796051, "learning_rate": 4.922514824023944e-06, "loss": 0.13013482093811035, "step": 859 }, { "epoch": 0.2232606438213915, "grad_norm": 3.1232378482818604, "learning_rate": 4.922268506302547e-06, "loss": 0.15050853788852692, "step": 860 }, { "epoch": 0.2235202492211838, "grad_norm": 1.1388620138168335, "learning_rate": 4.922021804573815e-06, "loss": 0.18373316526412964, "step": 861 }, { "epoch": 0.2237798546209761, "grad_norm": 1.4069851636886597, "learning_rate": 4.921774718881362e-06, "loss": 0.13625383377075195, "step": 862 }, { "epoch": 0.22403946002076844, "grad_norm": 3.1972811222076416, "learning_rate": 4.921527249268866e-06, "loss": 0.19337062537670135, "step": 863 }, { "epoch": 0.22429906542056074, "grad_norm": 1.2054671049118042, "learning_rate": 4.921279395780073e-06, "loss": 0.15030166506767273, "step": 864 }, { "epoch": 0.22455867082035305, "grad_norm": 1.4665591716766357, "learning_rate": 4.921031158458799e-06, "loss": 0.13229897618293762, "step": 865 }, { "epoch": 0.22481827622014539, "grad_norm": 1.9485846757888794, "learning_rate": 4.920782537348926e-06, "loss": 0.1461373269557953, "step": 866 }, { "epoch": 0.2250778816199377, "grad_norm": 1.2959620952606201, "learning_rate": 4.9205335324944045e-06, "loss": 0.13254734873771667, "step": 867 }, { "epoch": 0.22533748701973, "grad_norm": 3.251371383666992, "learning_rate": 4.9202841439392536e-06, "loss": 0.1399160623550415, "step": 868 }, { "epoch": 0.22559709241952233, "grad_norm": 1.6632837057113647, "learning_rate": 4.920034371727559e-06, "loss": 0.1486300826072693, "step": 869 }, { "epoch": 0.22585669781931464, "grad_norm": 2.170768976211548, "learning_rate": 4.919784215903474e-06, "loss": 0.15243345499038696, "step": 870 }, { "epoch": 0.22611630321910695, "grad_norm": 1.3836679458618164, "learning_rate": 4.9195336765112215e-06, "loss": 0.16246472299098969, "step": 871 }, { "epoch": 0.22637590861889928, "grad_norm": 2.3222930431365967, "learning_rate": 4.919282753595091e-06, "loss": 0.1276114135980606, "step": 872 }, { "epoch": 0.2266355140186916, "grad_norm": 2.9078612327575684, "learning_rate": 4.919031447199439e-06, "loss": 0.17222553491592407, "step": 873 }, { "epoch": 0.2268951194184839, "grad_norm": 1.0388494729995728, "learning_rate": 4.918779757368692e-06, "loss": 0.1433994323015213, "step": 874 }, { "epoch": 0.22715472481827623, "grad_norm": 2.6683058738708496, "learning_rate": 4.918527684147342e-06, "loss": 0.12779635190963745, "step": 875 }, { "epoch": 0.22741433021806853, "grad_norm": 1.026923418045044, "learning_rate": 4.918275227579951e-06, "loss": 0.13191060721874237, "step": 876 }, { "epoch": 0.22767393561786084, "grad_norm": 4.8789262771606445, "learning_rate": 4.918022387711145e-06, "loss": 0.17949515581130981, "step": 877 }, { "epoch": 0.22793354101765317, "grad_norm": 1.3580267429351807, "learning_rate": 4.917769164585623e-06, "loss": 0.1257268190383911, "step": 878 }, { "epoch": 0.22819314641744548, "grad_norm": 4.779221534729004, "learning_rate": 4.917515558248148e-06, "loss": 0.15443959832191467, "step": 879 }, { "epoch": 0.2284527518172378, "grad_norm": 4.534980773925781, "learning_rate": 4.9172615687435525e-06, "loss": 0.14021822810173035, "step": 880 }, { "epoch": 0.22871235721703012, "grad_norm": 0.7515465617179871, "learning_rate": 4.917007196116734e-06, "loss": 0.13096609711647034, "step": 881 }, { "epoch": 0.22897196261682243, "grad_norm": 4.941282749176025, "learning_rate": 4.916752440412661e-06, "loss": 0.17599119246006012, "step": 882 }, { "epoch": 0.22923156801661473, "grad_norm": 2.23367977142334, "learning_rate": 4.91649730167637e-06, "loss": 0.13309407234191895, "step": 883 }, { "epoch": 0.22949117341640707, "grad_norm": 2.7227916717529297, "learning_rate": 4.916241779952961e-06, "loss": 0.1737423539161682, "step": 884 }, { "epoch": 0.22975077881619937, "grad_norm": 0.6840508580207825, "learning_rate": 4.915985875287607e-06, "loss": 0.12455843389034271, "step": 885 }, { "epoch": 0.23001038421599168, "grad_norm": 1.826210856437683, "learning_rate": 4.915729587725544e-06, "loss": 0.14070409536361694, "step": 886 }, { "epoch": 0.23026998961578402, "grad_norm": 3.7775182723999023, "learning_rate": 4.915472917312079e-06, "loss": 0.15471510589122772, "step": 887 }, { "epoch": 0.23052959501557632, "grad_norm": 1.2683265209197998, "learning_rate": 4.915215864092584e-06, "loss": 0.12647467851638794, "step": 888 }, { "epoch": 0.23078920041536863, "grad_norm": 2.3637678623199463, "learning_rate": 4.914958428112502e-06, "loss": 0.1729821264743805, "step": 889 }, { "epoch": 0.23104880581516096, "grad_norm": 2.444972276687622, "learning_rate": 4.91470060941734e-06, "loss": 0.1395980566740036, "step": 890 }, { "epoch": 0.23130841121495327, "grad_norm": 1.66452956199646, "learning_rate": 4.914442408052676e-06, "loss": 0.12722274661064148, "step": 891 }, { "epoch": 0.23156801661474558, "grad_norm": 1.5472853183746338, "learning_rate": 4.9141838240641535e-06, "loss": 0.16724801063537598, "step": 892 }, { "epoch": 0.2318276220145379, "grad_norm": 2.275840997695923, "learning_rate": 4.913924857497483e-06, "loss": 0.12949049472808838, "step": 893 }, { "epoch": 0.23208722741433022, "grad_norm": 3.399348020553589, "learning_rate": 4.913665508398447e-06, "loss": 0.13829216361045837, "step": 894 }, { "epoch": 0.23234683281412252, "grad_norm": 2.055492639541626, "learning_rate": 4.913405776812888e-06, "loss": 0.1450934112071991, "step": 895 }, { "epoch": 0.23260643821391486, "grad_norm": 1.1657028198242188, "learning_rate": 4.913145662786725e-06, "loss": 0.12407904863357544, "step": 896 }, { "epoch": 0.23286604361370716, "grad_norm": 5.063736438751221, "learning_rate": 4.912885166365938e-06, "loss": 0.14044928550720215, "step": 897 }, { "epoch": 0.23312564901349947, "grad_norm": 2.085477590560913, "learning_rate": 4.912624287596577e-06, "loss": 0.16759014129638672, "step": 898 }, { "epoch": 0.2333852544132918, "grad_norm": 0.8822512626647949, "learning_rate": 4.91236302652476e-06, "loss": 0.13948357105255127, "step": 899 }, { "epoch": 0.2336448598130841, "grad_norm": 1.205288052558899, "learning_rate": 4.9121013831966704e-06, "loss": 0.16997981071472168, "step": 900 }, { "epoch": 0.23390446521287642, "grad_norm": 3.784809112548828, "learning_rate": 4.911839357658562e-06, "loss": 0.13051365315914154, "step": 901 }, { "epoch": 0.23416407061266875, "grad_norm": 3.8970136642456055, "learning_rate": 4.911576949956755e-06, "loss": 0.15087798237800598, "step": 902 }, { "epoch": 0.23442367601246106, "grad_norm": 1.336790919303894, "learning_rate": 4.9113141601376365e-06, "loss": 0.1624382585287094, "step": 903 }, { "epoch": 0.23468328141225336, "grad_norm": 3.401092767715454, "learning_rate": 4.911050988247663e-06, "loss": 0.1382785439491272, "step": 904 }, { "epoch": 0.2349428868120457, "grad_norm": 4.282845497131348, "learning_rate": 4.910787434333355e-06, "loss": 0.1376722753047943, "step": 905 }, { "epoch": 0.235202492211838, "grad_norm": 1.4380797147750854, "learning_rate": 4.910523498441304e-06, "loss": 0.13472113013267517, "step": 906 }, { "epoch": 0.2354620976116303, "grad_norm": 1.7230374813079834, "learning_rate": 4.91025918061817e-06, "loss": 0.12573915719985962, "step": 907 }, { "epoch": 0.23572170301142265, "grad_norm": 2.9950573444366455, "learning_rate": 4.909994480910674e-06, "loss": 0.14768072962760925, "step": 908 }, { "epoch": 0.23598130841121495, "grad_norm": 2.798447847366333, "learning_rate": 4.909729399365613e-06, "loss": 0.1291234791278839, "step": 909 }, { "epoch": 0.23624091381100726, "grad_norm": 2.645216464996338, "learning_rate": 4.909463936029843e-06, "loss": 0.1334693729877472, "step": 910 }, { "epoch": 0.2365005192107996, "grad_norm": 2.6524717807769775, "learning_rate": 4.909198090950296e-06, "loss": 0.18006695806980133, "step": 911 }, { "epoch": 0.2367601246105919, "grad_norm": 0.9864537119865417, "learning_rate": 4.908931864173965e-06, "loss": 0.13194143772125244, "step": 912 }, { "epoch": 0.2370197300103842, "grad_norm": 4.1365251541137695, "learning_rate": 4.908665255747914e-06, "loss": 0.1539589911699295, "step": 913 }, { "epoch": 0.23727933541017654, "grad_norm": 3.1950905323028564, "learning_rate": 4.908398265719273e-06, "loss": 0.12890255451202393, "step": 914 }, { "epoch": 0.23753894080996885, "grad_norm": 0.8684861660003662, "learning_rate": 4.908130894135238e-06, "loss": 0.14223253726959229, "step": 915 }, { "epoch": 0.23779854620976115, "grad_norm": 1.0421682596206665, "learning_rate": 4.907863141043076e-06, "loss": 0.12673285603523254, "step": 916 }, { "epoch": 0.2380581516095535, "grad_norm": 2.4523608684539795, "learning_rate": 4.90759500649012e-06, "loss": 0.15065450966358185, "step": 917 }, { "epoch": 0.2383177570093458, "grad_norm": 2.0621721744537354, "learning_rate": 4.9073264905237685e-06, "loss": 0.18158645927906036, "step": 918 }, { "epoch": 0.2385773624091381, "grad_norm": 1.3106149435043335, "learning_rate": 4.907057593191489e-06, "loss": 0.14705893397331238, "step": 919 }, { "epoch": 0.23883696780893043, "grad_norm": 0.8858506083488464, "learning_rate": 4.906788314540819e-06, "loss": 0.137366384267807, "step": 920 }, { "epoch": 0.23909657320872274, "grad_norm": 1.7078686952590942, "learning_rate": 4.906518654619358e-06, "loss": 0.13221199810504913, "step": 921 }, { "epoch": 0.23935617860851505, "grad_norm": 3.964555501937866, "learning_rate": 4.906248613474776e-06, "loss": 0.19142529368400574, "step": 922 }, { "epoch": 0.23961578400830738, "grad_norm": 2.426499605178833, "learning_rate": 4.905978191154811e-06, "loss": 0.18427523970603943, "step": 923 }, { "epoch": 0.2398753894080997, "grad_norm": 4.131654739379883, "learning_rate": 4.905707387707267e-06, "loss": 0.16290602087974548, "step": 924 }, { "epoch": 0.240134994807892, "grad_norm": 0.947967529296875, "learning_rate": 4.905436203180016e-06, "loss": 0.1482866108417511, "step": 925 }, { "epoch": 0.24039460020768433, "grad_norm": 0.8546525239944458, "learning_rate": 4.9051646376209976e-06, "loss": 0.1303931325674057, "step": 926 }, { "epoch": 0.24065420560747663, "grad_norm": 1.7171611785888672, "learning_rate": 4.904892691078218e-06, "loss": 0.1410500705242157, "step": 927 }, { "epoch": 0.24091381100726894, "grad_norm": 3.5752172470092773, "learning_rate": 4.904620363599752e-06, "loss": 0.15024134516716003, "step": 928 }, { "epoch": 0.24117341640706128, "grad_norm": 1.8849412202835083, "learning_rate": 4.904347655233738e-06, "loss": 0.15764538943767548, "step": 929 }, { "epoch": 0.24143302180685358, "grad_norm": 2.5461714267730713, "learning_rate": 4.904074566028388e-06, "loss": 0.13346630334854126, "step": 930 }, { "epoch": 0.2416926272066459, "grad_norm": 5.884583473205566, "learning_rate": 4.903801096031976e-06, "loss": 0.19388219714164734, "step": 931 }, { "epoch": 0.24195223260643822, "grad_norm": 0.6363177299499512, "learning_rate": 4.903527245292847e-06, "loss": 0.14172567427158356, "step": 932 }, { "epoch": 0.24221183800623053, "grad_norm": 1.8276339769363403, "learning_rate": 4.903253013859408e-06, "loss": 0.13635757565498352, "step": 933 }, { "epoch": 0.24247144340602284, "grad_norm": 4.251108169555664, "learning_rate": 4.9029784017801404e-06, "loss": 0.13425984978675842, "step": 934 }, { "epoch": 0.24273104880581517, "grad_norm": 2.2281086444854736, "learning_rate": 4.902703409103588e-06, "loss": 0.13126906752586365, "step": 935 }, { "epoch": 0.24299065420560748, "grad_norm": 1.1254390478134155, "learning_rate": 4.902428035878363e-06, "loss": 0.14061516523361206, "step": 936 }, { "epoch": 0.24325025960539978, "grad_norm": 1.3535698652267456, "learning_rate": 4.902152282153145e-06, "loss": 0.16369131207466125, "step": 937 }, { "epoch": 0.24350986500519212, "grad_norm": 6.398259162902832, "learning_rate": 4.901876147976682e-06, "loss": 0.17183136940002441, "step": 938 }, { "epoch": 0.24376947040498442, "grad_norm": 4.247622013092041, "learning_rate": 4.901599633397787e-06, "loss": 0.12428100407123566, "step": 939 }, { "epoch": 0.24402907580477673, "grad_norm": 1.9345061779022217, "learning_rate": 4.901322738465341e-06, "loss": 0.18326683342456818, "step": 940 }, { "epoch": 0.24428868120456906, "grad_norm": 6.201003551483154, "learning_rate": 4.901045463228295e-06, "loss": 0.18952417373657227, "step": 941 }, { "epoch": 0.24454828660436137, "grad_norm": 2.5773684978485107, "learning_rate": 4.900767807735664e-06, "loss": 0.15872198343276978, "step": 942 }, { "epoch": 0.24480789200415368, "grad_norm": 1.2644466161727905, "learning_rate": 4.90048977203653e-06, "loss": 0.1457417607307434, "step": 943 }, { "epoch": 0.245067497403946, "grad_norm": 3.121623992919922, "learning_rate": 4.900211356180044e-06, "loss": 0.13326393067836761, "step": 944 }, { "epoch": 0.24532710280373832, "grad_norm": 2.3301076889038086, "learning_rate": 4.899932560215423e-06, "loss": 0.15900510549545288, "step": 945 }, { "epoch": 0.24558670820353062, "grad_norm": 2.8172590732574463, "learning_rate": 4.899653384191952e-06, "loss": 0.1349811851978302, "step": 946 }, { "epoch": 0.24584631360332296, "grad_norm": 1.9744046926498413, "learning_rate": 4.899373828158984e-06, "loss": 0.17423765361309052, "step": 947 }, { "epoch": 0.24610591900311526, "grad_norm": 0.9694259762763977, "learning_rate": 4.899093892165938e-06, "loss": 0.1469397246837616, "step": 948 }, { "epoch": 0.24636552440290757, "grad_norm": 1.5956592559814453, "learning_rate": 4.898813576262299e-06, "loss": 0.13289673626422882, "step": 949 }, { "epoch": 0.2466251298026999, "grad_norm": 3.8576912879943848, "learning_rate": 4.8985328804976205e-06, "loss": 0.16016530990600586, "step": 950 }, { "epoch": 0.2468847352024922, "grad_norm": 0.7392709255218506, "learning_rate": 4.898251804921523e-06, "loss": 0.12125159800052643, "step": 951 }, { "epoch": 0.24714434060228452, "grad_norm": 1.8918119668960571, "learning_rate": 4.897970349583697e-06, "loss": 0.1349000185728073, "step": 952 }, { "epoch": 0.24740394600207685, "grad_norm": 1.1558736562728882, "learning_rate": 4.8976885145338935e-06, "loss": 0.1477629691362381, "step": 953 }, { "epoch": 0.24766355140186916, "grad_norm": 2.4298367500305176, "learning_rate": 4.897406299821936e-06, "loss": 0.1693316251039505, "step": 954 }, { "epoch": 0.24792315680166147, "grad_norm": 3.045726776123047, "learning_rate": 4.897123705497714e-06, "loss": 0.19152489304542542, "step": 955 }, { "epoch": 0.2481827622014538, "grad_norm": 3.532133102416992, "learning_rate": 4.896840731611185e-06, "loss": 0.1598229706287384, "step": 956 }, { "epoch": 0.2484423676012461, "grad_norm": 1.9878307580947876, "learning_rate": 4.896557378212369e-06, "loss": 0.13632220029830933, "step": 957 }, { "epoch": 0.2487019730010384, "grad_norm": 2.6632683277130127, "learning_rate": 4.8962736453513595e-06, "loss": 0.1580597311258316, "step": 958 }, { "epoch": 0.24896157840083075, "grad_norm": 1.6472822427749634, "learning_rate": 4.895989533078311e-06, "loss": 0.1487291157245636, "step": 959 }, { "epoch": 0.24922118380062305, "grad_norm": 2.6488897800445557, "learning_rate": 4.895705041443452e-06, "loss": 0.13974641263484955, "step": 960 }, { "epoch": 0.24948078920041536, "grad_norm": 3.060864210128784, "learning_rate": 4.89542017049707e-06, "loss": 0.14443600177764893, "step": 961 }, { "epoch": 0.2497403946002077, "grad_norm": 1.4706839323043823, "learning_rate": 4.895134920289526e-06, "loss": 0.12696698307991028, "step": 962 }, { "epoch": 0.25, "grad_norm": 1.0703004598617554, "learning_rate": 4.894849290871245e-06, "loss": 0.14224982261657715, "step": 963 }, { "epoch": 0.2502596053997923, "grad_norm": 3.258899688720703, "learning_rate": 4.89456328229272e-06, "loss": 0.14140334725379944, "step": 964 }, { "epoch": 0.2505192107995846, "grad_norm": 3.0790810585021973, "learning_rate": 4.894276894604511e-06, "loss": 0.12772399187088013, "step": 965 }, { "epoch": 0.2507788161993769, "grad_norm": 4.610372066497803, "learning_rate": 4.893990127857244e-06, "loss": 0.1484367996454239, "step": 966 }, { "epoch": 0.2510384215991693, "grad_norm": 4.4339399337768555, "learning_rate": 4.893702982101614e-06, "loss": 0.15044137835502625, "step": 967 }, { "epoch": 0.2512980269989616, "grad_norm": 3.7851858139038086, "learning_rate": 4.89341545738838e-06, "loss": 0.14734897017478943, "step": 968 }, { "epoch": 0.2515576323987539, "grad_norm": 6.376601219177246, "learning_rate": 4.893127553768372e-06, "loss": 0.17343442142009735, "step": 969 }, { "epoch": 0.2518172377985462, "grad_norm": 1.33059561252594, "learning_rate": 4.892839271292483e-06, "loss": 0.16908106207847595, "step": 970 }, { "epoch": 0.2520768431983385, "grad_norm": 3.682486057281494, "learning_rate": 4.892550610011675e-06, "loss": 0.1200704425573349, "step": 971 }, { "epoch": 0.2523364485981308, "grad_norm": 5.837192535400391, "learning_rate": 4.8922615699769775e-06, "loss": 0.14675123989582062, "step": 972 }, { "epoch": 0.2525960539979232, "grad_norm": 4.217096328735352, "learning_rate": 4.891972151239486e-06, "loss": 0.13892647624015808, "step": 973 }, { "epoch": 0.2528556593977155, "grad_norm": 1.0223653316497803, "learning_rate": 4.891682353850362e-06, "loss": 0.13013261556625366, "step": 974 }, { "epoch": 0.2531152647975078, "grad_norm": 2.939969062805176, "learning_rate": 4.8913921778608355e-06, "loss": 0.12992927432060242, "step": 975 }, { "epoch": 0.2533748701973001, "grad_norm": 3.281804084777832, "learning_rate": 4.891101623322205e-06, "loss": 0.15317898988723755, "step": 976 }, { "epoch": 0.2536344755970924, "grad_norm": 4.6689677238464355, "learning_rate": 4.890810690285831e-06, "loss": 0.1335676908493042, "step": 977 }, { "epoch": 0.2538940809968847, "grad_norm": 1.568006157875061, "learning_rate": 4.8905193788031454e-06, "loss": 0.14437952637672424, "step": 978 }, { "epoch": 0.25415368639667707, "grad_norm": 1.531802773475647, "learning_rate": 4.890227688925645e-06, "loss": 0.1454084813594818, "step": 979 }, { "epoch": 0.2544132917964694, "grad_norm": 4.68159294128418, "learning_rate": 4.8899356207048945e-06, "loss": 0.17101795971393585, "step": 980 }, { "epoch": 0.2546728971962617, "grad_norm": 3.0705649852752686, "learning_rate": 4.889643174192523e-06, "loss": 0.14051896333694458, "step": 981 }, { "epoch": 0.254932502596054, "grad_norm": 3.997494697570801, "learning_rate": 4.8893503494402305e-06, "loss": 0.16514870524406433, "step": 982 }, { "epoch": 0.2551921079958463, "grad_norm": 1.1786080598831177, "learning_rate": 4.889057146499781e-06, "loss": 0.1746668517589569, "step": 983 }, { "epoch": 0.2554517133956386, "grad_norm": 1.895906925201416, "learning_rate": 4.888763565423004e-06, "loss": 0.15786397457122803, "step": 984 }, { "epoch": 0.25571131879543096, "grad_norm": 4.5167622566223145, "learning_rate": 4.888469606261803e-06, "loss": 0.13091643154621124, "step": 985 }, { "epoch": 0.25597092419522327, "grad_norm": 4.636639595031738, "learning_rate": 4.888175269068138e-06, "loss": 0.14173372089862823, "step": 986 }, { "epoch": 0.2562305295950156, "grad_norm": 1.9731069803237915, "learning_rate": 4.887880553894044e-06, "loss": 0.11856777220964432, "step": 987 }, { "epoch": 0.2564901349948079, "grad_norm": 2.9433577060699463, "learning_rate": 4.887585460791618e-06, "loss": 0.14887341856956482, "step": 988 }, { "epoch": 0.2567497403946002, "grad_norm": 2.170095443725586, "learning_rate": 4.887289989813029e-06, "loss": 0.11642491817474365, "step": 989 }, { "epoch": 0.2570093457943925, "grad_norm": 3.516932487487793, "learning_rate": 4.886994141010505e-06, "loss": 0.16193488240242004, "step": 990 }, { "epoch": 0.25726895119418486, "grad_norm": 4.515347957611084, "learning_rate": 4.886697914436348e-06, "loss": 0.17492187023162842, "step": 991 }, { "epoch": 0.25752855659397716, "grad_norm": 1.0487316846847534, "learning_rate": 4.8864013101429235e-06, "loss": 0.1344137340784073, "step": 992 }, { "epoch": 0.25778816199376947, "grad_norm": 2.931760549545288, "learning_rate": 4.886104328182665e-06, "loss": 0.13998162746429443, "step": 993 }, { "epoch": 0.2580477673935618, "grad_norm": 3.614565372467041, "learning_rate": 4.885806968608072e-06, "loss": 0.15486428141593933, "step": 994 }, { "epoch": 0.2583073727933541, "grad_norm": 2.7558722496032715, "learning_rate": 4.8855092314717095e-06, "loss": 0.14246024191379547, "step": 995 }, { "epoch": 0.2585669781931464, "grad_norm": 1.1163569688796997, "learning_rate": 4.885211116826212e-06, "loss": 0.13246525824069977, "step": 996 }, { "epoch": 0.25882658359293875, "grad_norm": 3.448315143585205, "learning_rate": 4.884912624724279e-06, "loss": 0.15283770859241486, "step": 997 }, { "epoch": 0.25908618899273106, "grad_norm": 1.9732745885849, "learning_rate": 4.884613755218677e-06, "loss": 0.16031642258167267, "step": 998 }, { "epoch": 0.25934579439252337, "grad_norm": 1.6486350297927856, "learning_rate": 4.884314508362239e-06, "loss": 0.13062684237957, "step": 999 }, { "epoch": 0.25960539979231567, "grad_norm": 1.8700430393218994, "learning_rate": 4.884014884207865e-06, "loss": 0.1454603672027588, "step": 1000 }, { "epoch": 0.259865005192108, "grad_norm": 3.0139989852905273, "learning_rate": 4.8837148828085225e-06, "loss": 0.17627154290676117, "step": 1001 }, { "epoch": 0.2601246105919003, "grad_norm": 1.0117521286010742, "learning_rate": 4.883414504217244e-06, "loss": 0.12800610065460205, "step": 1002 }, { "epoch": 0.26038421599169265, "grad_norm": 3.554802656173706, "learning_rate": 4.88311374848713e-06, "loss": 0.1784728467464447, "step": 1003 }, { "epoch": 0.26064382139148495, "grad_norm": 0.6720311045646667, "learning_rate": 4.882812615671347e-06, "loss": 0.13948297500610352, "step": 1004 }, { "epoch": 0.26090342679127726, "grad_norm": 1.3294035196304321, "learning_rate": 4.882511105823128e-06, "loss": 0.14296185970306396, "step": 1005 }, { "epoch": 0.26116303219106957, "grad_norm": 2.3032114505767822, "learning_rate": 4.882209218995775e-06, "loss": 0.12046447396278381, "step": 1006 }, { "epoch": 0.2614226375908619, "grad_norm": 1.5368077754974365, "learning_rate": 4.881906955242652e-06, "loss": 0.1370471715927124, "step": 1007 }, { "epoch": 0.2616822429906542, "grad_norm": 2.0015740394592285, "learning_rate": 4.881604314617194e-06, "loss": 0.13082095980644226, "step": 1008 }, { "epoch": 0.26194184839044654, "grad_norm": 1.797667145729065, "learning_rate": 4.8813012971729e-06, "loss": 0.16042670607566833, "step": 1009 }, { "epoch": 0.26220145379023885, "grad_norm": 2.816323757171631, "learning_rate": 4.880997902963337e-06, "loss": 0.15638317167758942, "step": 1010 }, { "epoch": 0.26246105919003115, "grad_norm": 1.4898452758789062, "learning_rate": 4.880694132042138e-06, "loss": 0.15313643217086792, "step": 1011 }, { "epoch": 0.26272066458982346, "grad_norm": 1.4620025157928467, "learning_rate": 4.8803899844630034e-06, "loss": 0.13157975673675537, "step": 1012 }, { "epoch": 0.26298026998961577, "grad_norm": 0.877397358417511, "learning_rate": 4.880085460279699e-06, "loss": 0.1272658109664917, "step": 1013 }, { "epoch": 0.2632398753894081, "grad_norm": 2.3226449489593506, "learning_rate": 4.879780559546058e-06, "loss": 0.1217062920331955, "step": 1014 }, { "epoch": 0.26349948078920044, "grad_norm": 1.0606647729873657, "learning_rate": 4.879475282315978e-06, "loss": 0.12137018144130707, "step": 1015 }, { "epoch": 0.26375908618899274, "grad_norm": 1.05094575881958, "learning_rate": 4.879169628643428e-06, "loss": 0.13727599382400513, "step": 1016 }, { "epoch": 0.26401869158878505, "grad_norm": 2.321270704269409, "learning_rate": 4.878863598582439e-06, "loss": 0.1511802077293396, "step": 1017 }, { "epoch": 0.26427829698857735, "grad_norm": 1.508430004119873, "learning_rate": 4.87855719218711e-06, "loss": 0.12537333369255066, "step": 1018 }, { "epoch": 0.26453790238836966, "grad_norm": 1.432470440864563, "learning_rate": 4.878250409511608e-06, "loss": 0.131652370095253, "step": 1019 }, { "epoch": 0.26479750778816197, "grad_norm": 1.1156948804855347, "learning_rate": 4.877943250610163e-06, "loss": 0.1356620341539383, "step": 1020 }, { "epoch": 0.26505711318795433, "grad_norm": 2.337869167327881, "learning_rate": 4.8776357155370745e-06, "loss": 0.1730906069278717, "step": 1021 }, { "epoch": 0.26531671858774664, "grad_norm": 3.2590770721435547, "learning_rate": 4.877327804346709e-06, "loss": 0.15579888224601746, "step": 1022 }, { "epoch": 0.26557632398753894, "grad_norm": 4.275121688842773, "learning_rate": 4.877019517093497e-06, "loss": 0.12488903850317001, "step": 1023 }, { "epoch": 0.26583592938733125, "grad_norm": 0.8079078197479248, "learning_rate": 4.876710853831935e-06, "loss": 0.12158416211605072, "step": 1024 }, { "epoch": 0.26609553478712356, "grad_norm": 1.9373412132263184, "learning_rate": 4.8764018146165905e-06, "loss": 0.12502890825271606, "step": 1025 }, { "epoch": 0.26635514018691586, "grad_norm": 1.5628888607025146, "learning_rate": 4.876092399502093e-06, "loss": 0.13030323386192322, "step": 1026 }, { "epoch": 0.2666147455867082, "grad_norm": 1.3708178997039795, "learning_rate": 4.87578260854314e-06, "loss": 0.12337571382522583, "step": 1027 }, { "epoch": 0.26687435098650053, "grad_norm": 1.850164771080017, "learning_rate": 4.875472441794496e-06, "loss": 0.14997535943984985, "step": 1028 }, { "epoch": 0.26713395638629284, "grad_norm": 3.0181617736816406, "learning_rate": 4.875161899310992e-06, "loss": 0.14467865228652954, "step": 1029 }, { "epoch": 0.26739356178608514, "grad_norm": 2.6419477462768555, "learning_rate": 4.874850981147523e-06, "loss": 0.1555178165435791, "step": 1030 }, { "epoch": 0.26765316718587745, "grad_norm": 1.7641966342926025, "learning_rate": 4.874539687359054e-06, "loss": 0.17524635791778564, "step": 1031 }, { "epoch": 0.26791277258566976, "grad_norm": 1.2717543840408325, "learning_rate": 4.874228018000613e-06, "loss": 0.1717900037765503, "step": 1032 }, { "epoch": 0.2681723779854621, "grad_norm": 2.0474376678466797, "learning_rate": 4.8739159731272975e-06, "loss": 0.1277804970741272, "step": 1033 }, { "epoch": 0.2684319833852544, "grad_norm": 1.8371354341506958, "learning_rate": 4.873603552794268e-06, "loss": 0.15111416578292847, "step": 1034 }, { "epoch": 0.26869158878504673, "grad_norm": 0.5717582702636719, "learning_rate": 4.873290757056756e-06, "loss": 0.1319199800491333, "step": 1035 }, { "epoch": 0.26895119418483904, "grad_norm": 0.8306218981742859, "learning_rate": 4.872977585970054e-06, "loss": 0.12952099740505219, "step": 1036 }, { "epoch": 0.26921079958463134, "grad_norm": 1.2921875715255737, "learning_rate": 4.872664039589525e-06, "loss": 0.1609450876712799, "step": 1037 }, { "epoch": 0.26947040498442365, "grad_norm": 2.8182201385498047, "learning_rate": 4.872350117970596e-06, "loss": 0.16015924513339996, "step": 1038 }, { "epoch": 0.269730010384216, "grad_norm": 1.1104657649993896, "learning_rate": 4.8720358211687615e-06, "loss": 0.1513601839542389, "step": 1039 }, { "epoch": 0.2699896157840083, "grad_norm": 1.964600920677185, "learning_rate": 4.871721149239583e-06, "loss": 0.16780072450637817, "step": 1040 }, { "epoch": 0.2702492211838006, "grad_norm": 1.1526204347610474, "learning_rate": 4.8714061022386845e-06, "loss": 0.15263211727142334, "step": 1041 }, { "epoch": 0.27050882658359293, "grad_norm": 2.6984384059906006, "learning_rate": 4.8710906802217624e-06, "loss": 0.1834818720817566, "step": 1042 }, { "epoch": 0.27076843198338524, "grad_norm": 1.8203703165054321, "learning_rate": 4.8707748832445745e-06, "loss": 0.16760298609733582, "step": 1043 }, { "epoch": 0.27102803738317754, "grad_norm": 0.8651371002197266, "learning_rate": 4.870458711362946e-06, "loss": 0.13794095814228058, "step": 1044 }, { "epoch": 0.2712876427829699, "grad_norm": 1.6664544343948364, "learning_rate": 4.8701421646327694e-06, "loss": 0.1454606056213379, "step": 1045 }, { "epoch": 0.2715472481827622, "grad_norm": 2.2198703289031982, "learning_rate": 4.869825243110003e-06, "loss": 0.17911559343338013, "step": 1046 }, { "epoch": 0.2718068535825545, "grad_norm": 2.714691638946533, "learning_rate": 4.869507946850671e-06, "loss": 0.1656208038330078, "step": 1047 }, { "epoch": 0.2720664589823468, "grad_norm": 1.6317884922027588, "learning_rate": 4.869190275910865e-06, "loss": 0.16456016898155212, "step": 1048 }, { "epoch": 0.27232606438213913, "grad_norm": 1.735112190246582, "learning_rate": 4.868872230346741e-06, "loss": 0.14672791957855225, "step": 1049 }, { "epoch": 0.27258566978193144, "grad_norm": 1.0715079307556152, "learning_rate": 4.8685538102145215e-06, "loss": 0.14672714471817017, "step": 1050 }, { "epoch": 0.2728452751817238, "grad_norm": 1.7029563188552856, "learning_rate": 4.868235015570497e-06, "loss": 0.13202176988124847, "step": 1051 }, { "epoch": 0.2731048805815161, "grad_norm": 0.9946675300598145, "learning_rate": 4.867915846471024e-06, "loss": 0.12608695030212402, "step": 1052 }, { "epoch": 0.2733644859813084, "grad_norm": 2.6714329719543457, "learning_rate": 4.867596302972523e-06, "loss": 0.14030727744102478, "step": 1053 }, { "epoch": 0.2736240913811007, "grad_norm": 3.3419651985168457, "learning_rate": 4.867276385131482e-06, "loss": 0.15163199603557587, "step": 1054 }, { "epoch": 0.273883696780893, "grad_norm": 1.602829098701477, "learning_rate": 4.866956093004454e-06, "loss": 0.12998315691947937, "step": 1055 }, { "epoch": 0.27414330218068533, "grad_norm": 2.5811173915863037, "learning_rate": 4.866635426648062e-06, "loss": 0.17972788214683533, "step": 1056 }, { "epoch": 0.2744029075804777, "grad_norm": 2.089737892150879, "learning_rate": 4.866314386118991e-06, "loss": 0.15266868472099304, "step": 1057 }, { "epoch": 0.27466251298027, "grad_norm": 0.9972239136695862, "learning_rate": 4.865992971473993e-06, "loss": 0.16502803564071655, "step": 1058 }, { "epoch": 0.2749221183800623, "grad_norm": 1.7597382068634033, "learning_rate": 4.865671182769888e-06, "loss": 0.14402055740356445, "step": 1059 }, { "epoch": 0.2751817237798546, "grad_norm": 4.550919532775879, "learning_rate": 4.865349020063561e-06, "loss": 0.17748430371284485, "step": 1060 }, { "epoch": 0.2754413291796469, "grad_norm": 2.4850552082061768, "learning_rate": 4.865026483411961e-06, "loss": 0.16254256665706635, "step": 1061 }, { "epoch": 0.2757009345794392, "grad_norm": 2.917238712310791, "learning_rate": 4.864703572872107e-06, "loss": 0.16035591065883636, "step": 1062 }, { "epoch": 0.2759605399792316, "grad_norm": 2.3184897899627686, "learning_rate": 4.864380288501082e-06, "loss": 0.14280733466148376, "step": 1063 }, { "epoch": 0.2762201453790239, "grad_norm": 1.0891625881195068, "learning_rate": 4.864056630356034e-06, "loss": 0.1281307339668274, "step": 1064 }, { "epoch": 0.2764797507788162, "grad_norm": 3.182417154312134, "learning_rate": 4.86373259849418e-06, "loss": 0.15381504595279694, "step": 1065 }, { "epoch": 0.2767393561786085, "grad_norm": 3.094916343688965, "learning_rate": 4.8634081929728e-06, "loss": 0.12803512811660767, "step": 1066 }, { "epoch": 0.2769989615784008, "grad_norm": 1.607938289642334, "learning_rate": 4.863083413849243e-06, "loss": 0.12891148030757904, "step": 1067 }, { "epoch": 0.2772585669781931, "grad_norm": 1.61530339717865, "learning_rate": 4.862758261180921e-06, "loss": 0.1645042896270752, "step": 1068 }, { "epoch": 0.2775181723779855, "grad_norm": 1.1736423969268799, "learning_rate": 4.8624327350253145e-06, "loss": 0.1582232415676117, "step": 1069 }, { "epoch": 0.2777777777777778, "grad_norm": 1.4690752029418945, "learning_rate": 4.862106835439968e-06, "loss": 0.1449793577194214, "step": 1070 }, { "epoch": 0.2780373831775701, "grad_norm": 0.7574937343597412, "learning_rate": 4.861780562482495e-06, "loss": 0.13167649507522583, "step": 1071 }, { "epoch": 0.2782969885773624, "grad_norm": 1.384006381034851, "learning_rate": 4.861453916210571e-06, "loss": 0.18537598848342896, "step": 1072 }, { "epoch": 0.2785565939771547, "grad_norm": 1.9968029260635376, "learning_rate": 4.861126896681941e-06, "loss": 0.14764976501464844, "step": 1073 }, { "epoch": 0.278816199376947, "grad_norm": 1.1662977933883667, "learning_rate": 4.8607995039544135e-06, "loss": 0.12029270827770233, "step": 1074 }, { "epoch": 0.2790758047767394, "grad_norm": 1.0474225282669067, "learning_rate": 4.8604717380858655e-06, "loss": 0.12556761503219604, "step": 1075 }, { "epoch": 0.2793354101765317, "grad_norm": 2.754443407058716, "learning_rate": 4.8601435991342375e-06, "loss": 0.1439093053340912, "step": 1076 }, { "epoch": 0.279595015576324, "grad_norm": 1.0186623334884644, "learning_rate": 4.859815087157536e-06, "loss": 0.16477754712104797, "step": 1077 }, { "epoch": 0.2798546209761163, "grad_norm": 2.1511807441711426, "learning_rate": 4.859486202213836e-06, "loss": 0.17657685279846191, "step": 1078 }, { "epoch": 0.2801142263759086, "grad_norm": 4.311154842376709, "learning_rate": 4.859156944361276e-06, "loss": 0.16694073379039764, "step": 1079 }, { "epoch": 0.2803738317757009, "grad_norm": 1.740757942199707, "learning_rate": 4.858827313658062e-06, "loss": 0.1312982439994812, "step": 1080 }, { "epoch": 0.28063343717549327, "grad_norm": 5.505327224731445, "learning_rate": 4.858497310162465e-06, "loss": 0.14076386392116547, "step": 1081 }, { "epoch": 0.2808930425752856, "grad_norm": 1.907550573348999, "learning_rate": 4.858166933932821e-06, "loss": 0.13174933195114136, "step": 1082 }, { "epoch": 0.2811526479750779, "grad_norm": 1.187332034111023, "learning_rate": 4.857836185027534e-06, "loss": 0.14443150162696838, "step": 1083 }, { "epoch": 0.2814122533748702, "grad_norm": 3.1695778369903564, "learning_rate": 4.857505063505073e-06, "loss": 0.17912966012954712, "step": 1084 }, { "epoch": 0.2816718587746625, "grad_norm": 2.6176295280456543, "learning_rate": 4.857173569423971e-06, "loss": 0.16333088278770447, "step": 1085 }, { "epoch": 0.2819314641744548, "grad_norm": 2.6130757331848145, "learning_rate": 4.856841702842831e-06, "loss": 0.1516454815864563, "step": 1086 }, { "epoch": 0.28219106957424717, "grad_norm": 3.3494012355804443, "learning_rate": 4.856509463820318e-06, "loss": 0.18382225930690765, "step": 1087 }, { "epoch": 0.2824506749740395, "grad_norm": 0.5623272061347961, "learning_rate": 4.856176852415164e-06, "loss": 0.11910340189933777, "step": 1088 }, { "epoch": 0.2827102803738318, "grad_norm": 1.8591208457946777, "learning_rate": 4.855843868686168e-06, "loss": 0.1503314971923828, "step": 1089 }, { "epoch": 0.2829698857736241, "grad_norm": 2.2235968112945557, "learning_rate": 4.855510512692194e-06, "loss": 0.14510558545589447, "step": 1090 }, { "epoch": 0.2832294911734164, "grad_norm": 2.349909543991089, "learning_rate": 4.8551767844921695e-06, "loss": 0.15894022583961487, "step": 1091 }, { "epoch": 0.2834890965732087, "grad_norm": 2.0551648139953613, "learning_rate": 4.854842684145093e-06, "loss": 0.12010103464126587, "step": 1092 }, { "epoch": 0.28374870197300106, "grad_norm": 3.7804040908813477, "learning_rate": 4.854508211710025e-06, "loss": 0.1365443915128708, "step": 1093 }, { "epoch": 0.28400830737279337, "grad_norm": 1.3109321594238281, "learning_rate": 4.85417336724609e-06, "loss": 0.1288025677204132, "step": 1094 }, { "epoch": 0.2842679127725857, "grad_norm": 1.055371642112732, "learning_rate": 4.853838150812485e-06, "loss": 0.12599201500415802, "step": 1095 }, { "epoch": 0.284527518172378, "grad_norm": 5.294534206390381, "learning_rate": 4.853502562468464e-06, "loss": 0.1468266248703003, "step": 1096 }, { "epoch": 0.2847871235721703, "grad_norm": 4.361739158630371, "learning_rate": 4.853166602273356e-06, "loss": 0.16064178943634033, "step": 1097 }, { "epoch": 0.2850467289719626, "grad_norm": 2.8608052730560303, "learning_rate": 4.852830270286548e-06, "loss": 0.157931387424469, "step": 1098 }, { "epoch": 0.28530633437175495, "grad_norm": 2.0804555416107178, "learning_rate": 4.852493566567496e-06, "loss": 0.15141871571540833, "step": 1099 }, { "epoch": 0.28556593977154726, "grad_norm": 4.997710227966309, "learning_rate": 4.852156491175724e-06, "loss": 0.13887448608875275, "step": 1100 }, { "epoch": 0.28582554517133957, "grad_norm": 2.7231149673461914, "learning_rate": 4.851819044170816e-06, "loss": 0.1491992473602295, "step": 1101 }, { "epoch": 0.2860851505711319, "grad_norm": 1.8617726564407349, "learning_rate": 4.851481225612426e-06, "loss": 0.14678579568862915, "step": 1102 }, { "epoch": 0.2863447559709242, "grad_norm": 2.328329563140869, "learning_rate": 4.851143035560274e-06, "loss": 0.13291382789611816, "step": 1103 }, { "epoch": 0.2866043613707165, "grad_norm": 4.074077129364014, "learning_rate": 4.850804474074143e-06, "loss": 0.1415977030992508, "step": 1104 }, { "epoch": 0.28686396677050885, "grad_norm": 1.5129317045211792, "learning_rate": 4.8504655412138816e-06, "loss": 0.1411786526441574, "step": 1105 }, { "epoch": 0.28712357217030116, "grad_norm": 3.395947217941284, "learning_rate": 4.850126237039408e-06, "loss": 0.15446233749389648, "step": 1106 }, { "epoch": 0.28738317757009346, "grad_norm": 1.6939716339111328, "learning_rate": 4.849786561610702e-06, "loss": 0.1318541318178177, "step": 1107 }, { "epoch": 0.28764278296988577, "grad_norm": 2.9871466159820557, "learning_rate": 4.84944651498781e-06, "loss": 0.1289360076189041, "step": 1108 }, { "epoch": 0.2879023883696781, "grad_norm": 2.5281386375427246, "learning_rate": 4.849106097230845e-06, "loss": 0.141859233379364, "step": 1109 }, { "epoch": 0.2881619937694704, "grad_norm": 1.621991515159607, "learning_rate": 4.848765308399987e-06, "loss": 0.14831802248954773, "step": 1110 }, { "epoch": 0.28842159916926274, "grad_norm": 2.9067001342773438, "learning_rate": 4.848424148555476e-06, "loss": 0.15290480852127075, "step": 1111 }, { "epoch": 0.28868120456905505, "grad_norm": 1.5392061471939087, "learning_rate": 4.848082617757624e-06, "loss": 0.12727530300617218, "step": 1112 }, { "epoch": 0.28894080996884736, "grad_norm": 5.28272819519043, "learning_rate": 4.847740716066804e-06, "loss": 0.1536540985107422, "step": 1113 }, { "epoch": 0.28920041536863966, "grad_norm": 1.8905482292175293, "learning_rate": 4.847398443543459e-06, "loss": 0.1338718980550766, "step": 1114 }, { "epoch": 0.28946002076843197, "grad_norm": 1.5165537595748901, "learning_rate": 4.847055800248092e-06, "loss": 0.16741767525672913, "step": 1115 }, { "epoch": 0.2897196261682243, "grad_norm": 2.3058884143829346, "learning_rate": 4.846712786241275e-06, "loss": 0.1279565840959549, "step": 1116 }, { "epoch": 0.28997923156801664, "grad_norm": 2.521000623703003, "learning_rate": 4.8463694015836465e-06, "loss": 0.14765724539756775, "step": 1117 }, { "epoch": 0.29023883696780894, "grad_norm": 1.863874912261963, "learning_rate": 4.846025646335909e-06, "loss": 0.13840261101722717, "step": 1118 }, { "epoch": 0.29049844236760125, "grad_norm": 0.5972314476966858, "learning_rate": 4.845681520558829e-06, "loss": 0.12029758095741272, "step": 1119 }, { "epoch": 0.29075804776739356, "grad_norm": 1.1364840269088745, "learning_rate": 4.845337024313242e-06, "loss": 0.1386958509683609, "step": 1120 }, { "epoch": 0.29101765316718586, "grad_norm": 1.265966773033142, "learning_rate": 4.844992157660044e-06, "loss": 0.13253158330917358, "step": 1121 }, { "epoch": 0.29127725856697817, "grad_norm": 1.310542345046997, "learning_rate": 4.844646920660203e-06, "loss": 0.15137071907520294, "step": 1122 }, { "epoch": 0.29153686396677053, "grad_norm": 0.7703648209571838, "learning_rate": 4.844301313374747e-06, "loss": 0.1334477961063385, "step": 1123 }, { "epoch": 0.29179646936656284, "grad_norm": 1.5851274728775024, "learning_rate": 4.843955335864772e-06, "loss": 0.12221211194992065, "step": 1124 }, { "epoch": 0.29205607476635514, "grad_norm": 1.0167049169540405, "learning_rate": 4.843608988191439e-06, "loss": 0.14000239968299866, "step": 1125 }, { "epoch": 0.29231568016614745, "grad_norm": 2.809708595275879, "learning_rate": 4.843262270415974e-06, "loss": 0.1482102870941162, "step": 1126 }, { "epoch": 0.29257528556593976, "grad_norm": 2.1340677738189697, "learning_rate": 4.842915182599669e-06, "loss": 0.12140713632106781, "step": 1127 }, { "epoch": 0.29283489096573206, "grad_norm": 1.242988109588623, "learning_rate": 4.842567724803881e-06, "loss": 0.13322770595550537, "step": 1128 }, { "epoch": 0.2930944963655244, "grad_norm": 1.435341477394104, "learning_rate": 4.842219897090033e-06, "loss": 0.11854631453752518, "step": 1129 }, { "epoch": 0.29335410176531673, "grad_norm": 2.5639898777008057, "learning_rate": 4.8418716995196125e-06, "loss": 0.15417636930942535, "step": 1130 }, { "epoch": 0.29361370716510904, "grad_norm": 1.607115387916565, "learning_rate": 4.841523132154173e-06, "loss": 0.1470354199409485, "step": 1131 }, { "epoch": 0.29387331256490135, "grad_norm": 1.4092921018600464, "learning_rate": 4.841174195055333e-06, "loss": 0.1176551803946495, "step": 1132 }, { "epoch": 0.29413291796469365, "grad_norm": 3.955483913421631, "learning_rate": 4.8408248882847775e-06, "loss": 0.15616807341575623, "step": 1133 }, { "epoch": 0.29439252336448596, "grad_norm": 1.7640281915664673, "learning_rate": 4.8404752119042556e-06, "loss": 0.1301402747631073, "step": 1134 }, { "epoch": 0.2946521287642783, "grad_norm": 1.5485137701034546, "learning_rate": 4.8401251659755805e-06, "loss": 0.16826346516609192, "step": 1135 }, { "epoch": 0.2949117341640706, "grad_norm": 3.8202056884765625, "learning_rate": 4.839774750560635e-06, "loss": 0.1710115671157837, "step": 1136 }, { "epoch": 0.29517133956386293, "grad_norm": 3.951657772064209, "learning_rate": 4.839423965721362e-06, "loss": 0.13740235567092896, "step": 1137 }, { "epoch": 0.29543094496365524, "grad_norm": 1.4207240343093872, "learning_rate": 4.839072811519773e-06, "loss": 0.15737183392047882, "step": 1138 }, { "epoch": 0.29569055036344755, "grad_norm": 1.3279924392700195, "learning_rate": 4.838721288017945e-06, "loss": 0.1265566051006317, "step": 1139 }, { "epoch": 0.29595015576323985, "grad_norm": 2.808276653289795, "learning_rate": 4.838369395278019e-06, "loss": 0.12932536005973816, "step": 1140 }, { "epoch": 0.2962097611630322, "grad_norm": 3.309812545776367, "learning_rate": 4.838017133362201e-06, "loss": 0.11815972626209259, "step": 1141 }, { "epoch": 0.2964693665628245, "grad_norm": 0.638448178768158, "learning_rate": 4.8376645023327625e-06, "loss": 0.12999215722084045, "step": 1142 }, { "epoch": 0.2967289719626168, "grad_norm": 1.3180761337280273, "learning_rate": 4.837311502252042e-06, "loss": 0.1237274706363678, "step": 1143 }, { "epoch": 0.29698857736240913, "grad_norm": 1.388680338859558, "learning_rate": 4.836958133182441e-06, "loss": 0.16089676320552826, "step": 1144 }, { "epoch": 0.29724818276220144, "grad_norm": 2.917825698852539, "learning_rate": 4.836604395186426e-06, "loss": 0.17182905972003937, "step": 1145 }, { "epoch": 0.29750778816199375, "grad_norm": 2.403254747390747, "learning_rate": 4.836250288326532e-06, "loss": 0.14222347736358643, "step": 1146 }, { "epoch": 0.2977673935617861, "grad_norm": 3.3748221397399902, "learning_rate": 4.835895812665356e-06, "loss": 0.17639638483524323, "step": 1147 }, { "epoch": 0.2980269989615784, "grad_norm": 1.7689467668533325, "learning_rate": 4.835540968265559e-06, "loss": 0.12121307849884033, "step": 1148 }, { "epoch": 0.2982866043613707, "grad_norm": 1.4856823682785034, "learning_rate": 4.8351857551898734e-06, "loss": 0.12804405391216278, "step": 1149 }, { "epoch": 0.29854620976116303, "grad_norm": 2.6868271827697754, "learning_rate": 4.83483017350109e-06, "loss": 0.17864419519901276, "step": 1150 }, { "epoch": 0.29880581516095533, "grad_norm": 5.677618026733398, "learning_rate": 4.834474223262068e-06, "loss": 0.16736054420471191, "step": 1151 }, { "epoch": 0.29906542056074764, "grad_norm": 2.6890625953674316, "learning_rate": 4.834117904535732e-06, "loss": 0.17327070236206055, "step": 1152 }, { "epoch": 0.29932502596054, "grad_norm": 6.931493282318115, "learning_rate": 4.83376121738507e-06, "loss": 0.1426052451133728, "step": 1153 }, { "epoch": 0.2995846313603323, "grad_norm": 4.9485087394714355, "learning_rate": 4.833404161873137e-06, "loss": 0.13315634429454803, "step": 1154 }, { "epoch": 0.2998442367601246, "grad_norm": 2.7914998531341553, "learning_rate": 4.8330467380630515e-06, "loss": 0.14174216985702515, "step": 1155 }, { "epoch": 0.3001038421599169, "grad_norm": 2.6091527938842773, "learning_rate": 4.8326889460179984e-06, "loss": 0.1556772142648697, "step": 1156 }, { "epoch": 0.30036344755970923, "grad_norm": 6.432187080383301, "learning_rate": 4.832330785801227e-06, "loss": 0.1714775413274765, "step": 1157 }, { "epoch": 0.30062305295950154, "grad_norm": 5.448365688323975, "learning_rate": 4.8319722574760504e-06, "loss": 0.15798401832580566, "step": 1158 }, { "epoch": 0.3008826583592939, "grad_norm": 3.7219643592834473, "learning_rate": 4.8316133611058504e-06, "loss": 0.12812089920043945, "step": 1159 }, { "epoch": 0.3011422637590862, "grad_norm": 3.1815125942230225, "learning_rate": 4.831254096754072e-06, "loss": 0.16496552526950836, "step": 1160 }, { "epoch": 0.3014018691588785, "grad_norm": 2.7893259525299072, "learning_rate": 4.8308944644842225e-06, "loss": 0.15293899178504944, "step": 1161 }, { "epoch": 0.3016614745586708, "grad_norm": 1.339554786682129, "learning_rate": 4.8305344643598774e-06, "loss": 0.12890785932540894, "step": 1162 }, { "epoch": 0.3019210799584631, "grad_norm": 1.3083387613296509, "learning_rate": 4.830174096444678e-06, "loss": 0.154591903090477, "step": 1163 }, { "epoch": 0.30218068535825543, "grad_norm": 1.0837584733963013, "learning_rate": 4.8298133608023275e-06, "loss": 0.17410311102867126, "step": 1164 }, { "epoch": 0.3024402907580478, "grad_norm": 0.9100052118301392, "learning_rate": 4.829452257496596e-06, "loss": 0.12784646451473236, "step": 1165 }, { "epoch": 0.3026998961578401, "grad_norm": 1.0391088724136353, "learning_rate": 4.829090786591319e-06, "loss": 0.1270940601825714, "step": 1166 }, { "epoch": 0.3029595015576324, "grad_norm": 1.2332433462142944, "learning_rate": 4.828728948150395e-06, "loss": 0.12282796204090118, "step": 1167 }, { "epoch": 0.3032191069574247, "grad_norm": 1.1383298635482788, "learning_rate": 4.82836674223779e-06, "loss": 0.1306815892457962, "step": 1168 }, { "epoch": 0.303478712357217, "grad_norm": 0.7544129490852356, "learning_rate": 4.828004168917534e-06, "loss": 0.12048166990280151, "step": 1169 }, { "epoch": 0.3037383177570093, "grad_norm": 2.6285526752471924, "learning_rate": 4.8276412282537205e-06, "loss": 0.16235759854316711, "step": 1170 }, { "epoch": 0.3039979231568017, "grad_norm": 0.7261289954185486, "learning_rate": 4.8272779203105095e-06, "loss": 0.11611874401569366, "step": 1171 }, { "epoch": 0.304257528556594, "grad_norm": 0.901193380355835, "learning_rate": 4.826914245152126e-06, "loss": 0.12380799651145935, "step": 1172 }, { "epoch": 0.3045171339563863, "grad_norm": 1.6962225437164307, "learning_rate": 4.826550202842858e-06, "loss": 0.14867094159126282, "step": 1173 }, { "epoch": 0.3047767393561786, "grad_norm": 2.0318386554718018, "learning_rate": 4.826185793447061e-06, "loss": 0.15640956163406372, "step": 1174 }, { "epoch": 0.3050363447559709, "grad_norm": 1.5251089334487915, "learning_rate": 4.825821017029154e-06, "loss": 0.14483104646205902, "step": 1175 }, { "epoch": 0.3052959501557632, "grad_norm": 2.683154582977295, "learning_rate": 4.8254558736536216e-06, "loss": 0.13713480532169342, "step": 1176 }, { "epoch": 0.3055555555555556, "grad_norm": 1.4887639284133911, "learning_rate": 4.8250903633850125e-06, "loss": 0.1377815306186676, "step": 1177 }, { "epoch": 0.3058151609553479, "grad_norm": 1.6425056457519531, "learning_rate": 4.82472448628794e-06, "loss": 0.11737975478172302, "step": 1178 }, { "epoch": 0.3060747663551402, "grad_norm": 2.050257444381714, "learning_rate": 4.824358242427083e-06, "loss": 0.14865143597126007, "step": 1179 }, { "epoch": 0.3063343717549325, "grad_norm": 1.361522912979126, "learning_rate": 4.823991631867186e-06, "loss": 0.11425916850566864, "step": 1180 }, { "epoch": 0.3065939771547248, "grad_norm": 1.154189944267273, "learning_rate": 4.8236246546730555e-06, "loss": 0.16006995737552643, "step": 1181 }, { "epoch": 0.3068535825545171, "grad_norm": 2.538665533065796, "learning_rate": 4.823257310909567e-06, "loss": 0.13709375262260437, "step": 1182 }, { "epoch": 0.3071131879543095, "grad_norm": 2.0456340312957764, "learning_rate": 4.822889600641657e-06, "loss": 0.13022378087043762, "step": 1183 }, { "epoch": 0.3073727933541018, "grad_norm": 3.6981098651885986, "learning_rate": 4.822521523934328e-06, "loss": 0.16563481092453003, "step": 1184 }, { "epoch": 0.3076323987538941, "grad_norm": 1.1387580633163452, "learning_rate": 4.822153080852648e-06, "loss": 0.1135668158531189, "step": 1185 }, { "epoch": 0.3078920041536864, "grad_norm": 3.295551300048828, "learning_rate": 4.82178427146175e-06, "loss": 0.1657046675682068, "step": 1186 }, { "epoch": 0.3081516095534787, "grad_norm": 2.708751678466797, "learning_rate": 4.82141509582683e-06, "loss": 0.12813200056552887, "step": 1187 }, { "epoch": 0.308411214953271, "grad_norm": 2.1711230278015137, "learning_rate": 4.821045554013151e-06, "loss": 0.1447024792432785, "step": 1188 }, { "epoch": 0.30867082035306337, "grad_norm": 1.5606493949890137, "learning_rate": 4.820675646086037e-06, "loss": 0.1262846291065216, "step": 1189 }, { "epoch": 0.3089304257528557, "grad_norm": 3.2946224212646484, "learning_rate": 4.8203053721108815e-06, "loss": 0.1270962953567505, "step": 1190 }, { "epoch": 0.309190031152648, "grad_norm": 2.46060848236084, "learning_rate": 4.819934732153141e-06, "loss": 0.11133982241153717, "step": 1191 }, { "epoch": 0.3094496365524403, "grad_norm": 0.8282498121261597, "learning_rate": 4.819563726278334e-06, "loss": 0.11713571101427078, "step": 1192 }, { "epoch": 0.3097092419522326, "grad_norm": 1.380273461341858, "learning_rate": 4.819192354552047e-06, "loss": 0.11864530295133591, "step": 1193 }, { "epoch": 0.3099688473520249, "grad_norm": 3.7562060356140137, "learning_rate": 4.81882061703993e-06, "loss": 0.16588005423545837, "step": 1194 }, { "epoch": 0.31022845275181726, "grad_norm": 2.4314422607421875, "learning_rate": 4.818448513807698e-06, "loss": 0.1609063446521759, "step": 1195 }, { "epoch": 0.31048805815160957, "grad_norm": 0.9627255201339722, "learning_rate": 4.818076044921129e-06, "loss": 0.11850199103355408, "step": 1196 }, { "epoch": 0.3107476635514019, "grad_norm": 1.5962642431259155, "learning_rate": 4.817703210446069e-06, "loss": 0.14392386376857758, "step": 1197 }, { "epoch": 0.3110072689511942, "grad_norm": 1.9863688945770264, "learning_rate": 4.817330010448426e-06, "loss": 0.10814866423606873, "step": 1198 }, { "epoch": 0.3112668743509865, "grad_norm": 3.2910282611846924, "learning_rate": 4.816956444994172e-06, "loss": 0.16646629571914673, "step": 1199 }, { "epoch": 0.3115264797507788, "grad_norm": 1.752800703048706, "learning_rate": 4.816582514149344e-06, "loss": 0.12393022328615189, "step": 1200 }, { "epoch": 0.31178608515057116, "grad_norm": 0.8882256150245667, "learning_rate": 4.816208217980049e-06, "loss": 0.12165125459432602, "step": 1201 }, { "epoch": 0.31204569055036346, "grad_norm": 2.5050480365753174, "learning_rate": 4.815833556552448e-06, "loss": 0.127735435962677, "step": 1202 }, { "epoch": 0.31230529595015577, "grad_norm": 1.6320512294769287, "learning_rate": 4.815458529932778e-06, "loss": 0.11947804689407349, "step": 1203 }, { "epoch": 0.3125649013499481, "grad_norm": 3.557981014251709, "learning_rate": 4.815083138187331e-06, "loss": 0.1189470887184143, "step": 1204 }, { "epoch": 0.3128245067497404, "grad_norm": 1.261776089668274, "learning_rate": 4.8147073813824695e-06, "loss": 0.13601377606391907, "step": 1205 }, { "epoch": 0.3130841121495327, "grad_norm": 1.3534977436065674, "learning_rate": 4.814331259584619e-06, "loss": 0.13731437921524048, "step": 1206 }, { "epoch": 0.31334371754932505, "grad_norm": 2.419095516204834, "learning_rate": 4.8139547728602685e-06, "loss": 0.16012579202651978, "step": 1207 }, { "epoch": 0.31360332294911736, "grad_norm": 1.2338627576828003, "learning_rate": 4.8135779212759725e-06, "loss": 0.13848257064819336, "step": 1208 }, { "epoch": 0.31386292834890966, "grad_norm": 0.7788304686546326, "learning_rate": 4.81320070489835e-06, "loss": 0.11260983347892761, "step": 1209 }, { "epoch": 0.31412253374870197, "grad_norm": 2.286978006362915, "learning_rate": 4.812823123794084e-06, "loss": 0.1477309763431549, "step": 1210 }, { "epoch": 0.3143821391484943, "grad_norm": 1.6199878454208374, "learning_rate": 4.8124451780299215e-06, "loss": 0.12214980274438858, "step": 1211 }, { "epoch": 0.3146417445482866, "grad_norm": 2.025137424468994, "learning_rate": 4.812066867672676e-06, "loss": 0.15028049051761627, "step": 1212 }, { "epoch": 0.31490134994807895, "grad_norm": 2.8511483669281006, "learning_rate": 4.8116881927892225e-06, "loss": 0.15604159235954285, "step": 1213 }, { "epoch": 0.31516095534787125, "grad_norm": 2.3156819343566895, "learning_rate": 4.811309153446503e-06, "loss": 0.1698848307132721, "step": 1214 }, { "epoch": 0.31542056074766356, "grad_norm": 1.027448058128357, "learning_rate": 4.810929749711523e-06, "loss": 0.1443614959716797, "step": 1215 }, { "epoch": 0.31568016614745587, "grad_norm": 2.7933313846588135, "learning_rate": 4.810549981651353e-06, "loss": 0.16587494313716888, "step": 1216 }, { "epoch": 0.31593977154724817, "grad_norm": 2.9944450855255127, "learning_rate": 4.810169849333126e-06, "loss": 0.14645197987556458, "step": 1217 }, { "epoch": 0.3161993769470405, "grad_norm": 1.0448315143585205, "learning_rate": 4.8097893528240415e-06, "loss": 0.13049563765525818, "step": 1218 }, { "epoch": 0.31645898234683284, "grad_norm": 2.2791478633880615, "learning_rate": 4.809408492191362e-06, "loss": 0.16910390555858612, "step": 1219 }, { "epoch": 0.31671858774662515, "grad_norm": 0.7823804020881653, "learning_rate": 4.809027267502415e-06, "loss": 0.12132275104522705, "step": 1220 }, { "epoch": 0.31697819314641745, "grad_norm": 0.8538275957107544, "learning_rate": 4.808645678824593e-06, "loss": 0.1107749491930008, "step": 1221 }, { "epoch": 0.31723779854620976, "grad_norm": 2.8357832431793213, "learning_rate": 4.8082637262253515e-06, "loss": 0.16825082898139954, "step": 1222 }, { "epoch": 0.31749740394600207, "grad_norm": 3.5897269248962402, "learning_rate": 4.807881409772211e-06, "loss": 0.12211591750383377, "step": 1223 }, { "epoch": 0.3177570093457944, "grad_norm": 1.4337667226791382, "learning_rate": 4.807498729532758e-06, "loss": 0.16595113277435303, "step": 1224 }, { "epoch": 0.31801661474558673, "grad_norm": 2.192711114883423, "learning_rate": 4.807115685574638e-06, "loss": 0.15316280722618103, "step": 1225 }, { "epoch": 0.31827622014537904, "grad_norm": 1.4771168231964111, "learning_rate": 4.806732277965569e-06, "loss": 0.11703802645206451, "step": 1226 }, { "epoch": 0.31853582554517135, "grad_norm": 1.55703866481781, "learning_rate": 4.806348506773325e-06, "loss": 0.17370861768722534, "step": 1227 }, { "epoch": 0.31879543094496365, "grad_norm": 3.2410802841186523, "learning_rate": 4.805964372065748e-06, "loss": 0.14462541043758392, "step": 1228 }, { "epoch": 0.31905503634475596, "grad_norm": 1.615551233291626, "learning_rate": 4.805579873910746e-06, "loss": 0.14140111207962036, "step": 1229 }, { "epoch": 0.31931464174454827, "grad_norm": 2.040297031402588, "learning_rate": 4.805195012376289e-06, "loss": 0.1576792299747467, "step": 1230 }, { "epoch": 0.31957424714434063, "grad_norm": 1.7814483642578125, "learning_rate": 4.804809787530411e-06, "loss": 0.16035588085651398, "step": 1231 }, { "epoch": 0.31983385254413293, "grad_norm": 4.280364513397217, "learning_rate": 4.804424199441213e-06, "loss": 0.1916554868221283, "step": 1232 }, { "epoch": 0.32009345794392524, "grad_norm": 1.5616950988769531, "learning_rate": 4.804038248176855e-06, "loss": 0.12091472744941711, "step": 1233 }, { "epoch": 0.32035306334371755, "grad_norm": 0.8234257698059082, "learning_rate": 4.803651933805567e-06, "loss": 0.13106811046600342, "step": 1234 }, { "epoch": 0.32061266874350985, "grad_norm": 0.6501259207725525, "learning_rate": 4.803265256395639e-06, "loss": 0.13009518384933472, "step": 1235 }, { "epoch": 0.32087227414330216, "grad_norm": 2.119117259979248, "learning_rate": 4.802878216015428e-06, "loss": 0.1261034458875656, "step": 1236 }, { "epoch": 0.3211318795430945, "grad_norm": 3.6058013439178467, "learning_rate": 4.802490812733352e-06, "loss": 0.14252270758152008, "step": 1237 }, { "epoch": 0.32139148494288683, "grad_norm": 2.467608690261841, "learning_rate": 4.8021030466178965e-06, "loss": 0.155976340174675, "step": 1238 }, { "epoch": 0.32165109034267914, "grad_norm": 3.2493202686309814, "learning_rate": 4.80171491773761e-06, "loss": 0.1371782273054123, "step": 1239 }, { "epoch": 0.32191069574247144, "grad_norm": 4.578697204589844, "learning_rate": 4.801326426161103e-06, "loss": 0.1555701494216919, "step": 1240 }, { "epoch": 0.32217030114226375, "grad_norm": 3.294595718383789, "learning_rate": 4.800937571957053e-06, "loss": 0.127311110496521, "step": 1241 }, { "epoch": 0.32242990654205606, "grad_norm": 1.7859108448028564, "learning_rate": 4.800548355194202e-06, "loss": 0.1380906105041504, "step": 1242 }, { "epoch": 0.3226895119418484, "grad_norm": 3.3397231101989746, "learning_rate": 4.800158775941351e-06, "loss": 0.12855514883995056, "step": 1243 }, { "epoch": 0.3229491173416407, "grad_norm": 3.8905255794525146, "learning_rate": 4.799768834267372e-06, "loss": 0.14307370781898499, "step": 1244 }, { "epoch": 0.32320872274143303, "grad_norm": 2.393688440322876, "learning_rate": 4.799378530241197e-06, "loss": 0.12869326770305634, "step": 1245 }, { "epoch": 0.32346832814122534, "grad_norm": 1.72889244556427, "learning_rate": 4.7989878639318225e-06, "loss": 0.14145344495773315, "step": 1246 }, { "epoch": 0.32372793354101764, "grad_norm": 2.4065866470336914, "learning_rate": 4.798596835408309e-06, "loss": 0.1338874101638794, "step": 1247 }, { "epoch": 0.32398753894080995, "grad_norm": 2.1728196144104004, "learning_rate": 4.79820544473978e-06, "loss": 0.12385382503271103, "step": 1248 }, { "epoch": 0.3242471443406023, "grad_norm": 2.4295332431793213, "learning_rate": 4.797813691995428e-06, "loss": 0.1392567902803421, "step": 1249 }, { "epoch": 0.3245067497403946, "grad_norm": 1.835083246231079, "learning_rate": 4.7974215772445035e-06, "loss": 0.1707586646080017, "step": 1250 }, { "epoch": 0.3247663551401869, "grad_norm": 0.7262897491455078, "learning_rate": 4.797029100556324e-06, "loss": 0.13398241996765137, "step": 1251 }, { "epoch": 0.32502596053997923, "grad_norm": 1.0519407987594604, "learning_rate": 4.796636262000269e-06, "loss": 0.1263769418001175, "step": 1252 }, { "epoch": 0.32528556593977154, "grad_norm": 1.593798279762268, "learning_rate": 4.796243061645786e-06, "loss": 0.12935389578342438, "step": 1253 }, { "epoch": 0.32554517133956384, "grad_norm": 2.902144193649292, "learning_rate": 4.7958494995623824e-06, "loss": 0.16345329582691193, "step": 1254 }, { "epoch": 0.3258047767393562, "grad_norm": 0.5780671834945679, "learning_rate": 4.79545557581963e-06, "loss": 0.12394073605537415, "step": 1255 }, { "epoch": 0.3260643821391485, "grad_norm": 1.7297046184539795, "learning_rate": 4.795061290487168e-06, "loss": 0.1419844925403595, "step": 1256 }, { "epoch": 0.3263239875389408, "grad_norm": 5.287092208862305, "learning_rate": 4.794666643634695e-06, "loss": 0.14462020993232727, "step": 1257 }, { "epoch": 0.3265835929387331, "grad_norm": 1.9768671989440918, "learning_rate": 4.794271635331975e-06, "loss": 0.12273897975683212, "step": 1258 }, { "epoch": 0.32684319833852543, "grad_norm": 1.2797590494155884, "learning_rate": 4.793876265648839e-06, "loss": 0.13773639500141144, "step": 1259 }, { "epoch": 0.32710280373831774, "grad_norm": 3.1101276874542236, "learning_rate": 4.793480534655177e-06, "loss": 0.16100609302520752, "step": 1260 }, { "epoch": 0.3273624091381101, "grad_norm": 1.3066203594207764, "learning_rate": 4.793084442420946e-06, "loss": 0.12313488870859146, "step": 1261 }, { "epoch": 0.3276220145379024, "grad_norm": 2.597257614135742, "learning_rate": 4.792687989016166e-06, "loss": 0.15711688995361328, "step": 1262 }, { "epoch": 0.3278816199376947, "grad_norm": 1.1482195854187012, "learning_rate": 4.792291174510921e-06, "loss": 0.1329621970653534, "step": 1263 }, { "epoch": 0.328141225337487, "grad_norm": 0.7208447456359863, "learning_rate": 4.791893998975359e-06, "loss": 0.12204167246818542, "step": 1264 }, { "epoch": 0.3284008307372793, "grad_norm": 2.573315382003784, "learning_rate": 4.791496462479691e-06, "loss": 0.1488124430179596, "step": 1265 }, { "epoch": 0.32866043613707163, "grad_norm": 2.0069620609283447, "learning_rate": 4.791098565094192e-06, "loss": 0.17241007089614868, "step": 1266 }, { "epoch": 0.328920041536864, "grad_norm": 2.499105930328369, "learning_rate": 4.790700306889202e-06, "loss": 0.16793833673000336, "step": 1267 }, { "epoch": 0.3291796469366563, "grad_norm": 3.4656240940093994, "learning_rate": 4.790301687935124e-06, "loss": 0.14980822801589966, "step": 1268 }, { "epoch": 0.3294392523364486, "grad_norm": 1.5764633417129517, "learning_rate": 4.789902708302423e-06, "loss": 0.12308566272258759, "step": 1269 }, { "epoch": 0.3296988577362409, "grad_norm": 2.0909392833709717, "learning_rate": 4.789503368061632e-06, "loss": 0.16330337524414062, "step": 1270 }, { "epoch": 0.3299584631360332, "grad_norm": 2.2861530780792236, "learning_rate": 4.789103667283344e-06, "loss": 0.13089537620544434, "step": 1271 }, { "epoch": 0.3302180685358255, "grad_norm": 2.3099777698516846, "learning_rate": 4.788703606038216e-06, "loss": 0.15255342423915863, "step": 1272 }, { "epoch": 0.3304776739356179, "grad_norm": 1.4425644874572754, "learning_rate": 4.788303184396971e-06, "loss": 0.14259129762649536, "step": 1273 }, { "epoch": 0.3307372793354102, "grad_norm": 1.7258577346801758, "learning_rate": 4.787902402430395e-06, "loss": 0.1344154179096222, "step": 1274 }, { "epoch": 0.3309968847352025, "grad_norm": 1.650156021118164, "learning_rate": 4.787501260209334e-06, "loss": 0.13293211162090302, "step": 1275 }, { "epoch": 0.3312564901349948, "grad_norm": 1.4520039558410645, "learning_rate": 4.7870997578047035e-06, "loss": 0.1292339265346527, "step": 1276 }, { "epoch": 0.3315160955347871, "grad_norm": 3.0375101566314697, "learning_rate": 4.7866978952874795e-06, "loss": 0.1500752568244934, "step": 1277 }, { "epoch": 0.3317757009345794, "grad_norm": 1.2126034498214722, "learning_rate": 4.786295672728702e-06, "loss": 0.1268511265516281, "step": 1278 }, { "epoch": 0.3320353063343718, "grad_norm": 2.5875957012176514, "learning_rate": 4.785893090199475e-06, "loss": 0.14120572805404663, "step": 1279 }, { "epoch": 0.3322949117341641, "grad_norm": 1.9161930084228516, "learning_rate": 4.785490147770964e-06, "loss": 0.14769606292247772, "step": 1280 }, { "epoch": 0.3325545171339564, "grad_norm": 0.8727218508720398, "learning_rate": 4.785086845514403e-06, "loss": 0.1502564251422882, "step": 1281 }, { "epoch": 0.3328141225337487, "grad_norm": 2.1722683906555176, "learning_rate": 4.784683183501084e-06, "loss": 0.167103573679924, "step": 1282 }, { "epoch": 0.333073727933541, "grad_norm": 2.5589241981506348, "learning_rate": 4.7842791618023666e-06, "loss": 0.15857678651809692, "step": 1283 }, { "epoch": 0.3333333333333333, "grad_norm": 1.8198060989379883, "learning_rate": 4.783874780489673e-06, "loss": 0.140297994017601, "step": 1284 }, { "epoch": 0.3335929387331256, "grad_norm": 1.423288106918335, "learning_rate": 4.783470039634486e-06, "loss": 0.14889055490493774, "step": 1285 }, { "epoch": 0.333852544132918, "grad_norm": 1.5581185817718506, "learning_rate": 4.783064939308357e-06, "loss": 0.1789550930261612, "step": 1286 }, { "epoch": 0.3341121495327103, "grad_norm": 0.8999871015548706, "learning_rate": 4.782659479582898e-06, "loss": 0.12421291321516037, "step": 1287 }, { "epoch": 0.3343717549325026, "grad_norm": 1.0669244527816772, "learning_rate": 4.782253660529785e-06, "loss": 0.11859692633152008, "step": 1288 }, { "epoch": 0.3346313603322949, "grad_norm": 2.6939969062805176, "learning_rate": 4.781847482220757e-06, "loss": 0.1837959587574005, "step": 1289 }, { "epoch": 0.3348909657320872, "grad_norm": 1.4871585369110107, "learning_rate": 4.781440944727617e-06, "loss": 0.14437925815582275, "step": 1290 }, { "epoch": 0.3351505711318795, "grad_norm": 0.6521194577217102, "learning_rate": 4.781034048122232e-06, "loss": 0.11695507168769836, "step": 1291 }, { "epoch": 0.3354101765316719, "grad_norm": 1.4824033975601196, "learning_rate": 4.780626792476532e-06, "loss": 0.1649455428123474, "step": 1292 }, { "epoch": 0.3356697819314642, "grad_norm": 2.5639336109161377, "learning_rate": 4.78021917786251e-06, "loss": 0.12667036056518555, "step": 1293 }, { "epoch": 0.3359293873312565, "grad_norm": 0.9500283002853394, "learning_rate": 4.779811204352223e-06, "loss": 0.1342286467552185, "step": 1294 }, { "epoch": 0.3361889927310488, "grad_norm": 1.125359058380127, "learning_rate": 4.779402872017792e-06, "loss": 0.11428298056125641, "step": 1295 }, { "epoch": 0.3364485981308411, "grad_norm": 1.0341105461120605, "learning_rate": 4.778994180931401e-06, "loss": 0.13259917497634888, "step": 1296 }, { "epoch": 0.3367082035306334, "grad_norm": 1.5116490125656128, "learning_rate": 4.778585131165295e-06, "loss": 0.14181368052959442, "step": 1297 }, { "epoch": 0.33696780893042577, "grad_norm": 2.4272727966308594, "learning_rate": 4.778175722791788e-06, "loss": 0.09018256515264511, "step": 1298 }, { "epoch": 0.3372274143302181, "grad_norm": 2.39936900138855, "learning_rate": 4.777765955883251e-06, "loss": 0.1346638798713684, "step": 1299 }, { "epoch": 0.3374870197300104, "grad_norm": 3.0090746879577637, "learning_rate": 4.777355830512122e-06, "loss": 0.19671767950057983, "step": 1300 }, { "epoch": 0.3377466251298027, "grad_norm": 2.6772360801696777, "learning_rate": 4.776945346750905e-06, "loss": 0.1416684240102768, "step": 1301 }, { "epoch": 0.338006230529595, "grad_norm": 1.708342432975769, "learning_rate": 4.776534504672159e-06, "loss": 0.13049528002738953, "step": 1302 }, { "epoch": 0.3382658359293873, "grad_norm": 2.9070448875427246, "learning_rate": 4.7761233043485165e-06, "loss": 0.13189242780208588, "step": 1303 }, { "epoch": 0.33852544132917967, "grad_norm": 1.6024528741836548, "learning_rate": 4.7757117458526645e-06, "loss": 0.1418723464012146, "step": 1304 }, { "epoch": 0.338785046728972, "grad_norm": 1.9828472137451172, "learning_rate": 4.775299829257358e-06, "loss": 0.16095124185085297, "step": 1305 }, { "epoch": 0.3390446521287643, "grad_norm": 3.5170235633850098, "learning_rate": 4.7748875546354155e-06, "loss": 0.16891957819461823, "step": 1306 }, { "epoch": 0.3393042575285566, "grad_norm": 1.2898000478744507, "learning_rate": 4.774474922059718e-06, "loss": 0.12418253719806671, "step": 1307 }, { "epoch": 0.3395638629283489, "grad_norm": 1.5417203903198242, "learning_rate": 4.774061931603207e-06, "loss": 0.11584381759166718, "step": 1308 }, { "epoch": 0.3398234683281412, "grad_norm": 0.926440954208374, "learning_rate": 4.773648583338892e-06, "loss": 0.1255704164505005, "step": 1309 }, { "epoch": 0.34008307372793356, "grad_norm": 1.80246901512146, "learning_rate": 4.773234877339843e-06, "loss": 0.1318824291229248, "step": 1310 }, { "epoch": 0.34034267912772587, "grad_norm": 1.3899426460266113, "learning_rate": 4.772820813679194e-06, "loss": 0.1297493577003479, "step": 1311 }, { "epoch": 0.3406022845275182, "grad_norm": 2.7666404247283936, "learning_rate": 4.772406392430141e-06, "loss": 0.15328797698020935, "step": 1312 }, { "epoch": 0.3408618899273105, "grad_norm": 0.9853875041007996, "learning_rate": 4.771991613665945e-06, "loss": 0.14753557741641998, "step": 1313 }, { "epoch": 0.3411214953271028, "grad_norm": 1.1346651315689087, "learning_rate": 4.771576477459929e-06, "loss": 0.12399102747440338, "step": 1314 }, { "epoch": 0.3413811007268951, "grad_norm": 1.095422625541687, "learning_rate": 4.77116098388548e-06, "loss": 0.13485689461231232, "step": 1315 }, { "epoch": 0.34164070612668745, "grad_norm": 3.294250249862671, "learning_rate": 4.770745133016048e-06, "loss": 0.15660813450813293, "step": 1316 }, { "epoch": 0.34190031152647976, "grad_norm": 0.9630849957466125, "learning_rate": 4.770328924925145e-06, "loss": 0.10371631383895874, "step": 1317 }, { "epoch": 0.34215991692627207, "grad_norm": 1.803703784942627, "learning_rate": 4.769912359686346e-06, "loss": 0.1425597071647644, "step": 1318 }, { "epoch": 0.3424195223260644, "grad_norm": 0.8067253232002258, "learning_rate": 4.7694954373732935e-06, "loss": 0.12254443019628525, "step": 1319 }, { "epoch": 0.3426791277258567, "grad_norm": 0.9203118681907654, "learning_rate": 4.769078158059687e-06, "loss": 0.1313229501247406, "step": 1320 }, { "epoch": 0.342938733125649, "grad_norm": 1.2898863554000854, "learning_rate": 4.7686605218192935e-06, "loss": 0.16108828783035278, "step": 1321 }, { "epoch": 0.34319833852544135, "grad_norm": 1.5232168436050415, "learning_rate": 4.768242528725941e-06, "loss": 0.15034128725528717, "step": 1322 }, { "epoch": 0.34345794392523366, "grad_norm": 0.9859563708305359, "learning_rate": 4.767824178853521e-06, "loss": 0.13163189589977264, "step": 1323 }, { "epoch": 0.34371754932502596, "grad_norm": 1.7541477680206299, "learning_rate": 4.767405472275988e-06, "loss": 0.13238310813903809, "step": 1324 }, { "epoch": 0.34397715472481827, "grad_norm": 0.8576701283454895, "learning_rate": 4.76698640906736e-06, "loss": 0.12129935622215271, "step": 1325 }, { "epoch": 0.3442367601246106, "grad_norm": 1.2783854007720947, "learning_rate": 4.766566989301718e-06, "loss": 0.16442321240901947, "step": 1326 }, { "epoch": 0.3444963655244029, "grad_norm": 1.5155298709869385, "learning_rate": 4.7661472130532056e-06, "loss": 0.14474134147167206, "step": 1327 }, { "epoch": 0.34475597092419524, "grad_norm": 1.6600216627120972, "learning_rate": 4.7657270803960285e-06, "loss": 0.13445571064949036, "step": 1328 }, { "epoch": 0.34501557632398755, "grad_norm": 2.5421981811523438, "learning_rate": 4.765306591404459e-06, "loss": 0.15589158236980438, "step": 1329 }, { "epoch": 0.34527518172377986, "grad_norm": 0.8950066566467285, "learning_rate": 4.764885746152828e-06, "loss": 0.11974549293518066, "step": 1330 }, { "epoch": 0.34553478712357216, "grad_norm": 2.372218608856201, "learning_rate": 4.764464544715532e-06, "loss": 0.1347641944885254, "step": 1331 }, { "epoch": 0.34579439252336447, "grad_norm": 1.2899540662765503, "learning_rate": 4.764042987167029e-06, "loss": 0.14560332894325256, "step": 1332 }, { "epoch": 0.3460539979231568, "grad_norm": 1.1336381435394287, "learning_rate": 4.763621073581842e-06, "loss": 0.13245759904384613, "step": 1333 }, { "epoch": 0.34631360332294914, "grad_norm": 1.8693974018096924, "learning_rate": 4.7631988040345554e-06, "loss": 0.12749557197093964, "step": 1334 }, { "epoch": 0.34657320872274144, "grad_norm": 1.743302345275879, "learning_rate": 4.762776178599815e-06, "loss": 0.11958493292331696, "step": 1335 }, { "epoch": 0.34683281412253375, "grad_norm": 1.5014848709106445, "learning_rate": 4.762353197352333e-06, "loss": 0.14992734789848328, "step": 1336 }, { "epoch": 0.34709241952232606, "grad_norm": 0.8668684363365173, "learning_rate": 4.761929860366883e-06, "loss": 0.13081175088882446, "step": 1337 }, { "epoch": 0.34735202492211836, "grad_norm": 0.9219000339508057, "learning_rate": 4.761506167718302e-06, "loss": 0.11549510806798935, "step": 1338 }, { "epoch": 0.34761163032191067, "grad_norm": 1.4452623128890991, "learning_rate": 4.761082119481486e-06, "loss": 0.12769025564193726, "step": 1339 }, { "epoch": 0.34787123572170303, "grad_norm": 2.1471855640411377, "learning_rate": 4.7606577157314e-06, "loss": 0.13981936872005463, "step": 1340 }, { "epoch": 0.34813084112149534, "grad_norm": 1.4036061763763428, "learning_rate": 4.760232956543067e-06, "loss": 0.12775994837284088, "step": 1341 }, { "epoch": 0.34839044652128764, "grad_norm": 1.7334316968917847, "learning_rate": 4.759807841991576e-06, "loss": 0.1386134773492813, "step": 1342 }, { "epoch": 0.34865005192107995, "grad_norm": 1.588472604751587, "learning_rate": 4.759382372152077e-06, "loss": 0.14931440353393555, "step": 1343 }, { "epoch": 0.34890965732087226, "grad_norm": 1.821330189704895, "learning_rate": 4.7589565470997836e-06, "loss": 0.1328338086605072, "step": 1344 }, { "epoch": 0.34916926272066456, "grad_norm": 2.29375958442688, "learning_rate": 4.758530366909971e-06, "loss": 0.12774908542633057, "step": 1345 }, { "epoch": 0.3494288681204569, "grad_norm": 1.3897892236709595, "learning_rate": 4.758103831657979e-06, "loss": 0.1445748507976532, "step": 1346 }, { "epoch": 0.34968847352024923, "grad_norm": 2.4409546852111816, "learning_rate": 4.75767694141921e-06, "loss": 0.15677881240844727, "step": 1347 }, { "epoch": 0.34994807892004154, "grad_norm": 2.8395965099334717, "learning_rate": 4.757249696269126e-06, "loss": 0.152114599943161, "step": 1348 }, { "epoch": 0.35020768431983385, "grad_norm": 3.006497859954834, "learning_rate": 4.756822096283257e-06, "loss": 0.16385534405708313, "step": 1349 }, { "epoch": 0.35046728971962615, "grad_norm": 3.8133203983306885, "learning_rate": 4.7563941415371905e-06, "loss": 0.1404397040605545, "step": 1350 }, { "epoch": 0.35072689511941846, "grad_norm": 2.4440627098083496, "learning_rate": 4.755965832106581e-06, "loss": 0.16792690753936768, "step": 1351 }, { "epoch": 0.3509865005192108, "grad_norm": 1.8934270143508911, "learning_rate": 4.755537168067142e-06, "loss": 0.15626996755599976, "step": 1352 }, { "epoch": 0.3512461059190031, "grad_norm": 1.500353217124939, "learning_rate": 4.755108149494653e-06, "loss": 0.1306118667125702, "step": 1353 }, { "epoch": 0.35150571131879543, "grad_norm": 1.0677390098571777, "learning_rate": 4.754678776464954e-06, "loss": 0.1309526264667511, "step": 1354 }, { "epoch": 0.35176531671858774, "grad_norm": 1.974844217300415, "learning_rate": 4.754249049053948e-06, "loss": 0.15721184015274048, "step": 1355 }, { "epoch": 0.35202492211838005, "grad_norm": 2.850886583328247, "learning_rate": 4.753818967337602e-06, "loss": 0.14708837866783142, "step": 1356 }, { "epoch": 0.35228452751817235, "grad_norm": 2.1670591831207275, "learning_rate": 4.753388531391943e-06, "loss": 0.16091424226760864, "step": 1357 }, { "epoch": 0.3525441329179647, "grad_norm": 1.3657532930374146, "learning_rate": 4.752957741293064e-06, "loss": 0.13871629536151886, "step": 1358 }, { "epoch": 0.352803738317757, "grad_norm": 1.5206961631774902, "learning_rate": 4.752526597117117e-06, "loss": 0.13283176720142365, "step": 1359 }, { "epoch": 0.3530633437175493, "grad_norm": 2.839966297149658, "learning_rate": 4.7520950989403205e-06, "loss": 0.1477258801460266, "step": 1360 }, { "epoch": 0.35332294911734163, "grad_norm": 0.7281484007835388, "learning_rate": 4.751663246838953e-06, "loss": 0.14452379941940308, "step": 1361 }, { "epoch": 0.35358255451713394, "grad_norm": 0.6857381463050842, "learning_rate": 4.751231040889354e-06, "loss": 0.12289783358573914, "step": 1362 }, { "epoch": 0.35384215991692625, "grad_norm": 3.408419370651245, "learning_rate": 4.7507984811679296e-06, "loss": 0.15522095561027527, "step": 1363 }, { "epoch": 0.3541017653167186, "grad_norm": 2.5224545001983643, "learning_rate": 4.750365567751146e-06, "loss": 0.16979631781578064, "step": 1364 }, { "epoch": 0.3543613707165109, "grad_norm": 0.7207149267196655, "learning_rate": 4.749932300715532e-06, "loss": 0.11652833223342896, "step": 1365 }, { "epoch": 0.3546209761163032, "grad_norm": 1.0920740365982056, "learning_rate": 4.74949868013768e-06, "loss": 0.12907835841178894, "step": 1366 }, { "epoch": 0.35488058151609553, "grad_norm": 2.3590409755706787, "learning_rate": 4.749064706094244e-06, "loss": 0.13918092846870422, "step": 1367 }, { "epoch": 0.35514018691588783, "grad_norm": 2.2834415435791016, "learning_rate": 4.74863037866194e-06, "loss": 0.11684690415859222, "step": 1368 }, { "epoch": 0.35539979231568014, "grad_norm": 3.634422779083252, "learning_rate": 4.748195697917547e-06, "loss": 0.1622171550989151, "step": 1369 }, { "epoch": 0.3556593977154725, "grad_norm": 1.151496171951294, "learning_rate": 4.747760663937908e-06, "loss": 0.15057110786437988, "step": 1370 }, { "epoch": 0.3559190031152648, "grad_norm": 4.996406078338623, "learning_rate": 4.747325276799927e-06, "loss": 0.12955227494239807, "step": 1371 }, { "epoch": 0.3561786085150571, "grad_norm": 5.763355255126953, "learning_rate": 4.746889536580569e-06, "loss": 0.17308726906776428, "step": 1372 }, { "epoch": 0.3564382139148494, "grad_norm": 1.8401639461517334, "learning_rate": 4.7464534433568634e-06, "loss": 0.13222019374370575, "step": 1373 }, { "epoch": 0.35669781931464173, "grad_norm": 2.3036446571350098, "learning_rate": 4.7460169972059015e-06, "loss": 0.12125842273235321, "step": 1374 }, { "epoch": 0.35695742471443404, "grad_norm": 4.078943252563477, "learning_rate": 4.7455801982048374e-06, "loss": 0.13230416178703308, "step": 1375 }, { "epoch": 0.3572170301142264, "grad_norm": 3.8070573806762695, "learning_rate": 4.7451430464308875e-06, "loss": 0.12720265984535217, "step": 1376 }, { "epoch": 0.3574766355140187, "grad_norm": 2.5415573120117188, "learning_rate": 4.74470554196133e-06, "loss": 0.158138245344162, "step": 1377 }, { "epoch": 0.357736240913811, "grad_norm": 1.3037865161895752, "learning_rate": 4.744267684873505e-06, "loss": 0.13463668525218964, "step": 1378 }, { "epoch": 0.3579958463136033, "grad_norm": 3.40617036819458, "learning_rate": 4.743829475244816e-06, "loss": 0.12234634160995483, "step": 1379 }, { "epoch": 0.3582554517133956, "grad_norm": 3.3238439559936523, "learning_rate": 4.743390913152728e-06, "loss": 0.12534287571907043, "step": 1380 }, { "epoch": 0.35851505711318793, "grad_norm": 2.9031271934509277, "learning_rate": 4.74295199867477e-06, "loss": 0.14251083135604858, "step": 1381 }, { "epoch": 0.3587746625129803, "grad_norm": 2.6653871536254883, "learning_rate": 4.74251273188853e-06, "loss": 0.14871007204055786, "step": 1382 }, { "epoch": 0.3590342679127726, "grad_norm": 2.6732168197631836, "learning_rate": 4.742073112871663e-06, "loss": 0.12976732850074768, "step": 1383 }, { "epoch": 0.3592938733125649, "grad_norm": 2.233290433883667, "learning_rate": 4.741633141701881e-06, "loss": 0.15777629613876343, "step": 1384 }, { "epoch": 0.3595534787123572, "grad_norm": 2.268446445465088, "learning_rate": 4.741192818456961e-06, "loss": 0.12168967723846436, "step": 1385 }, { "epoch": 0.3598130841121495, "grad_norm": 1.375584602355957, "learning_rate": 4.740752143214744e-06, "loss": 0.1462942659854889, "step": 1386 }, { "epoch": 0.3600726895119418, "grad_norm": 1.5971466302871704, "learning_rate": 4.74031111605313e-06, "loss": 0.156875342130661, "step": 1387 }, { "epoch": 0.3603322949117342, "grad_norm": 1.4102096557617188, "learning_rate": 4.739869737050084e-06, "loss": 0.12747366726398468, "step": 1388 }, { "epoch": 0.3605919003115265, "grad_norm": 1.061420202255249, "learning_rate": 4.73942800628363e-06, "loss": 0.13630619645118713, "step": 1389 }, { "epoch": 0.3608515057113188, "grad_norm": 1.4038760662078857, "learning_rate": 4.7389859238318555e-06, "loss": 0.1330459713935852, "step": 1390 }, { "epoch": 0.3611111111111111, "grad_norm": 2.601490020751953, "learning_rate": 4.738543489772912e-06, "loss": 0.1581585705280304, "step": 1391 }, { "epoch": 0.3613707165109034, "grad_norm": 2.318917989730835, "learning_rate": 4.738100704185011e-06, "loss": 0.14516854286193848, "step": 1392 }, { "epoch": 0.3616303219106957, "grad_norm": 1.4558967351913452, "learning_rate": 4.737657567146427e-06, "loss": 0.13519462943077087, "step": 1393 }, { "epoch": 0.3618899273104881, "grad_norm": 1.845494031906128, "learning_rate": 4.737214078735496e-06, "loss": 0.1430090218782425, "step": 1394 }, { "epoch": 0.3621495327102804, "grad_norm": 4.645568370819092, "learning_rate": 4.736770239030618e-06, "loss": 0.14028999209403992, "step": 1395 }, { "epoch": 0.3624091381100727, "grad_norm": 1.8139145374298096, "learning_rate": 4.736326048110252e-06, "loss": 0.1516065001487732, "step": 1396 }, { "epoch": 0.362668743509865, "grad_norm": 1.4425171613693237, "learning_rate": 4.735881506052922e-06, "loss": 0.15620160102844238, "step": 1397 }, { "epoch": 0.3629283489096573, "grad_norm": 1.708578109741211, "learning_rate": 4.735436612937211e-06, "loss": 0.11878371983766556, "step": 1398 }, { "epoch": 0.3631879543094496, "grad_norm": 1.3652048110961914, "learning_rate": 4.734991368841769e-06, "loss": 0.12637263536453247, "step": 1399 }, { "epoch": 0.363447559709242, "grad_norm": 2.7847471237182617, "learning_rate": 4.7345457738453025e-06, "loss": 0.13188377022743225, "step": 1400 }, { "epoch": 0.3637071651090343, "grad_norm": 1.4255253076553345, "learning_rate": 4.734099828026584e-06, "loss": 0.1474292278289795, "step": 1401 }, { "epoch": 0.3639667705088266, "grad_norm": 1.8270832300186157, "learning_rate": 4.733653531464445e-06, "loss": 0.13149330019950867, "step": 1402 }, { "epoch": 0.3642263759086189, "grad_norm": 1.3387913703918457, "learning_rate": 4.7332068842377815e-06, "loss": 0.1298515796661377, "step": 1403 }, { "epoch": 0.3644859813084112, "grad_norm": 1.5661206245422363, "learning_rate": 4.73275988642555e-06, "loss": 0.141943097114563, "step": 1404 }, { "epoch": 0.3647455867082035, "grad_norm": 2.8560144901275635, "learning_rate": 4.73231253810677e-06, "loss": 0.12116041779518127, "step": 1405 }, { "epoch": 0.36500519210799587, "grad_norm": 2.2923874855041504, "learning_rate": 4.731864839360523e-06, "loss": 0.14285975694656372, "step": 1406 }, { "epoch": 0.3652647975077882, "grad_norm": 1.9606233835220337, "learning_rate": 4.73141679026595e-06, "loss": 0.1252773106098175, "step": 1407 }, { "epoch": 0.3655244029075805, "grad_norm": 2.666598320007324, "learning_rate": 4.730968390902257e-06, "loss": 0.15289156138896942, "step": 1408 }, { "epoch": 0.3657840083073728, "grad_norm": 1.3361725807189941, "learning_rate": 4.730519641348711e-06, "loss": 0.16213396191596985, "step": 1409 }, { "epoch": 0.3660436137071651, "grad_norm": 1.1253831386566162, "learning_rate": 4.73007054168464e-06, "loss": 0.12252065539360046, "step": 1410 }, { "epoch": 0.3663032191069574, "grad_norm": 1.4750527143478394, "learning_rate": 4.729621091989434e-06, "loss": 0.11701301485300064, "step": 1411 }, { "epoch": 0.36656282450674976, "grad_norm": 1.1709141731262207, "learning_rate": 4.729171292342549e-06, "loss": 0.12834641337394714, "step": 1412 }, { "epoch": 0.36682242990654207, "grad_norm": 1.053848385810852, "learning_rate": 4.728721142823495e-06, "loss": 0.12184201180934906, "step": 1413 }, { "epoch": 0.3670820353063344, "grad_norm": 1.555461049079895, "learning_rate": 4.728270643511849e-06, "loss": 0.1362358033657074, "step": 1414 }, { "epoch": 0.3673416407061267, "grad_norm": 1.6544016599655151, "learning_rate": 4.72781979448725e-06, "loss": 0.1220775693655014, "step": 1415 }, { "epoch": 0.367601246105919, "grad_norm": 1.16851806640625, "learning_rate": 4.727368595829397e-06, "loss": 0.1259147971868515, "step": 1416 }, { "epoch": 0.3678608515057113, "grad_norm": 1.3974179029464722, "learning_rate": 4.726917047618053e-06, "loss": 0.1335514932870865, "step": 1417 }, { "epoch": 0.36812045690550366, "grad_norm": 2.458430767059326, "learning_rate": 4.72646514993304e-06, "loss": 0.11718528717756271, "step": 1418 }, { "epoch": 0.36838006230529596, "grad_norm": 1.667514443397522, "learning_rate": 4.726012902854243e-06, "loss": 0.14440305531024933, "step": 1419 }, { "epoch": 0.36863966770508827, "grad_norm": 1.8830738067626953, "learning_rate": 4.72556030646161e-06, "loss": 0.14590775966644287, "step": 1420 }, { "epoch": 0.3688992731048806, "grad_norm": 1.2633479833602905, "learning_rate": 4.7251073608351485e-06, "loss": 0.12919539213180542, "step": 1421 }, { "epoch": 0.3691588785046729, "grad_norm": 0.9750471115112305, "learning_rate": 4.724654066054931e-06, "loss": 0.11745976656675339, "step": 1422 }, { "epoch": 0.3694184839044652, "grad_norm": 1.1073474884033203, "learning_rate": 4.724200422201087e-06, "loss": 0.14667800068855286, "step": 1423 }, { "epoch": 0.36967808930425755, "grad_norm": 1.5758402347564697, "learning_rate": 4.7237464293538115e-06, "loss": 0.12686170637607574, "step": 1424 }, { "epoch": 0.36993769470404986, "grad_norm": 1.2935242652893066, "learning_rate": 4.72329208759336e-06, "loss": 0.15352660417556763, "step": 1425 }, { "epoch": 0.37019730010384216, "grad_norm": 1.186810851097107, "learning_rate": 4.722837397000052e-06, "loss": 0.11128885298967361, "step": 1426 }, { "epoch": 0.37045690550363447, "grad_norm": 1.1488932371139526, "learning_rate": 4.722382357654262e-06, "loss": 0.11820584535598755, "step": 1427 }, { "epoch": 0.3707165109034268, "grad_norm": 2.4021921157836914, "learning_rate": 4.721926969636433e-06, "loss": 0.16409865021705627, "step": 1428 }, { "epoch": 0.3709761163032191, "grad_norm": 2.4265120029449463, "learning_rate": 4.721471233027067e-06, "loss": 0.14705246686935425, "step": 1429 }, { "epoch": 0.37123572170301145, "grad_norm": 8.106634140014648, "learning_rate": 4.721015147906727e-06, "loss": 0.13117097318172455, "step": 1430 }, { "epoch": 0.37149532710280375, "grad_norm": 1.304664969444275, "learning_rate": 4.720558714356041e-06, "loss": 0.13853931427001953, "step": 1431 }, { "epoch": 0.37175493250259606, "grad_norm": 3.0270156860351562, "learning_rate": 4.720101932455693e-06, "loss": 0.1312447488307953, "step": 1432 }, { "epoch": 0.37201453790238836, "grad_norm": 2.639616012573242, "learning_rate": 4.719644802286433e-06, "loss": 0.1771467924118042, "step": 1433 }, { "epoch": 0.37227414330218067, "grad_norm": 2.9524712562561035, "learning_rate": 4.71918732392907e-06, "loss": 0.1515834778547287, "step": 1434 }, { "epoch": 0.372533748701973, "grad_norm": 1.0109511613845825, "learning_rate": 4.718729497464477e-06, "loss": 0.12721553444862366, "step": 1435 }, { "epoch": 0.37279335410176534, "grad_norm": 2.4598655700683594, "learning_rate": 4.7182713229735874e-06, "loss": 0.13370202481746674, "step": 1436 }, { "epoch": 0.37305295950155765, "grad_norm": 2.64630126953125, "learning_rate": 4.717812800537394e-06, "loss": 0.16204464435577393, "step": 1437 }, { "epoch": 0.37331256490134995, "grad_norm": 1.024637222290039, "learning_rate": 4.717353930236956e-06, "loss": 0.13159281015396118, "step": 1438 }, { "epoch": 0.37357217030114226, "grad_norm": 1.5957506895065308, "learning_rate": 4.716894712153388e-06, "loss": 0.1821521520614624, "step": 1439 }, { "epoch": 0.37383177570093457, "grad_norm": 2.1126868724823, "learning_rate": 4.716435146367872e-06, "loss": 0.14300858974456787, "step": 1440 }, { "epoch": 0.37409138110072687, "grad_norm": 1.4588181972503662, "learning_rate": 4.715975232961647e-06, "loss": 0.14706221222877502, "step": 1441 }, { "epoch": 0.37435098650051923, "grad_norm": 2.3774564266204834, "learning_rate": 4.715514972016016e-06, "loss": 0.1390371322631836, "step": 1442 }, { "epoch": 0.37461059190031154, "grad_norm": 2.9452176094055176, "learning_rate": 4.715054363612343e-06, "loss": 0.15603214502334595, "step": 1443 }, { "epoch": 0.37487019730010385, "grad_norm": 1.6543198823928833, "learning_rate": 4.714593407832051e-06, "loss": 0.12770912051200867, "step": 1444 }, { "epoch": 0.37512980269989615, "grad_norm": 1.9352573156356812, "learning_rate": 4.714132104756628e-06, "loss": 0.13745909929275513, "step": 1445 }, { "epoch": 0.37538940809968846, "grad_norm": 2.036879777908325, "learning_rate": 4.7136704544676215e-06, "loss": 0.15202127397060394, "step": 1446 }, { "epoch": 0.37564901349948077, "grad_norm": 1.5030885934829712, "learning_rate": 4.713208457046641e-06, "loss": 0.12040311098098755, "step": 1447 }, { "epoch": 0.37590861889927313, "grad_norm": 2.5620110034942627, "learning_rate": 4.7127461125753575e-06, "loss": 0.16054658591747284, "step": 1448 }, { "epoch": 0.37616822429906543, "grad_norm": 1.2556710243225098, "learning_rate": 4.7122834211355e-06, "loss": 0.11344773322343826, "step": 1449 }, { "epoch": 0.37642782969885774, "grad_norm": 1.718722939491272, "learning_rate": 4.711820382808866e-06, "loss": 0.14846569299697876, "step": 1450 }, { "epoch": 0.37668743509865005, "grad_norm": 2.5296008586883545, "learning_rate": 4.711356997677308e-06, "loss": 0.1369132697582245, "step": 1451 }, { "epoch": 0.37694704049844235, "grad_norm": 1.1606630086898804, "learning_rate": 4.710893265822741e-06, "loss": 0.09881296753883362, "step": 1452 }, { "epoch": 0.37720664589823466, "grad_norm": 2.5636425018310547, "learning_rate": 4.710429187327144e-06, "loss": 0.13747920095920563, "step": 1453 }, { "epoch": 0.377466251298027, "grad_norm": 2.8748297691345215, "learning_rate": 4.709964762272554e-06, "loss": 0.12680113315582275, "step": 1454 }, { "epoch": 0.37772585669781933, "grad_norm": 0.8536306023597717, "learning_rate": 4.709499990741071e-06, "loss": 0.12343338131904602, "step": 1455 }, { "epoch": 0.37798546209761164, "grad_norm": 1.1384601593017578, "learning_rate": 4.709034872814856e-06, "loss": 0.10837668925523758, "step": 1456 }, { "epoch": 0.37824506749740394, "grad_norm": 2.9269297122955322, "learning_rate": 4.708569408576133e-06, "loss": 0.16306468844413757, "step": 1457 }, { "epoch": 0.37850467289719625, "grad_norm": 2.378326177597046, "learning_rate": 4.708103598107182e-06, "loss": 0.11638054251670837, "step": 1458 }, { "epoch": 0.37876427829698855, "grad_norm": 1.8692078590393066, "learning_rate": 4.707637441490351e-06, "loss": 0.12314299494028091, "step": 1459 }, { "epoch": 0.3790238836967809, "grad_norm": 5.274399757385254, "learning_rate": 4.707170938808044e-06, "loss": 0.15907688438892365, "step": 1460 }, { "epoch": 0.3792834890965732, "grad_norm": 1.3595058917999268, "learning_rate": 4.706704090142728e-06, "loss": 0.11099734157323837, "step": 1461 }, { "epoch": 0.37954309449636553, "grad_norm": 1.6196904182434082, "learning_rate": 4.706236895576931e-06, "loss": 0.12691813707351685, "step": 1462 }, { "epoch": 0.37980269989615784, "grad_norm": 2.0623791217803955, "learning_rate": 4.705769355193244e-06, "loss": 0.12270665168762207, "step": 1463 }, { "epoch": 0.38006230529595014, "grad_norm": 1.4348036050796509, "learning_rate": 4.705301469074314e-06, "loss": 0.1403878778219223, "step": 1464 }, { "epoch": 0.38032191069574245, "grad_norm": 1.3474550247192383, "learning_rate": 4.704833237302857e-06, "loss": 0.12354833632707596, "step": 1465 }, { "epoch": 0.3805815160955348, "grad_norm": 1.5929323434829712, "learning_rate": 4.7043646599616425e-06, "loss": 0.12518435716629028, "step": 1466 }, { "epoch": 0.3808411214953271, "grad_norm": 3.0720536708831787, "learning_rate": 4.703895737133505e-06, "loss": 0.17017394304275513, "step": 1467 }, { "epoch": 0.3811007268951194, "grad_norm": 2.0949363708496094, "learning_rate": 4.703426468901339e-06, "loss": 0.16092772781848907, "step": 1468 }, { "epoch": 0.38136033229491173, "grad_norm": 1.3004974126815796, "learning_rate": 4.702956855348102e-06, "loss": 0.1328549087047577, "step": 1469 }, { "epoch": 0.38161993769470404, "grad_norm": 0.9433514475822449, "learning_rate": 4.702486896556809e-06, "loss": 0.1361847221851349, "step": 1470 }, { "epoch": 0.38187954309449634, "grad_norm": 3.0358052253723145, "learning_rate": 4.702016592610539e-06, "loss": 0.15296846628189087, "step": 1471 }, { "epoch": 0.3821391484942887, "grad_norm": 1.7040942907333374, "learning_rate": 4.70154594359243e-06, "loss": 0.1577446162700653, "step": 1472 }, { "epoch": 0.382398753894081, "grad_norm": 1.2612566947937012, "learning_rate": 4.701074949585684e-06, "loss": 0.15045207738876343, "step": 1473 }, { "epoch": 0.3826583592938733, "grad_norm": 1.8295769691467285, "learning_rate": 4.7006036106735594e-06, "loss": 0.12513406574726105, "step": 1474 }, { "epoch": 0.3829179646936656, "grad_norm": 0.9226046800613403, "learning_rate": 4.700131926939379e-06, "loss": 0.13019371032714844, "step": 1475 }, { "epoch": 0.38317757009345793, "grad_norm": 1.3432929515838623, "learning_rate": 4.699659898466527e-06, "loss": 0.15303421020507812, "step": 1476 }, { "epoch": 0.38343717549325024, "grad_norm": 2.245724678039551, "learning_rate": 4.699187525338447e-06, "loss": 0.1443779170513153, "step": 1477 }, { "epoch": 0.3836967808930426, "grad_norm": 1.5594427585601807, "learning_rate": 4.698714807638641e-06, "loss": 0.15638116002082825, "step": 1478 }, { "epoch": 0.3839563862928349, "grad_norm": 1.10682213306427, "learning_rate": 4.698241745450678e-06, "loss": 0.1507757604122162, "step": 1479 }, { "epoch": 0.3842159916926272, "grad_norm": 1.8751260042190552, "learning_rate": 4.697768338858183e-06, "loss": 0.12118469923734665, "step": 1480 }, { "epoch": 0.3844755970924195, "grad_norm": 1.258566975593567, "learning_rate": 4.697294587944844e-06, "loss": 0.15610408782958984, "step": 1481 }, { "epoch": 0.3847352024922118, "grad_norm": 2.578094482421875, "learning_rate": 4.696820492794408e-06, "loss": 0.12743058800697327, "step": 1482 }, { "epoch": 0.38499480789200413, "grad_norm": 0.9899160861968994, "learning_rate": 4.6963460534906865e-06, "loss": 0.13191713392734528, "step": 1483 }, { "epoch": 0.3852544132917965, "grad_norm": 2.9616482257843018, "learning_rate": 4.6958712701175465e-06, "loss": 0.11973781883716583, "step": 1484 }, { "epoch": 0.3855140186915888, "grad_norm": 1.7027384042739868, "learning_rate": 4.695396142758922e-06, "loss": 0.12935782968997955, "step": 1485 }, { "epoch": 0.3857736240913811, "grad_norm": 1.7454665899276733, "learning_rate": 4.694920671498802e-06, "loss": 0.13675430417060852, "step": 1486 }, { "epoch": 0.3860332294911734, "grad_norm": 2.1090164184570312, "learning_rate": 4.694444856421242e-06, "loss": 0.13646087050437927, "step": 1487 }, { "epoch": 0.3862928348909657, "grad_norm": 2.505250930786133, "learning_rate": 4.693968697610351e-06, "loss": 0.15111282467842102, "step": 1488 }, { "epoch": 0.386552440290758, "grad_norm": 1.1283352375030518, "learning_rate": 4.693492195150306e-06, "loss": 0.14481067657470703, "step": 1489 }, { "epoch": 0.3868120456905504, "grad_norm": 2.1669654846191406, "learning_rate": 4.693015349125341e-06, "loss": 0.13179005682468414, "step": 1490 }, { "epoch": 0.3870716510903427, "grad_norm": 1.4215424060821533, "learning_rate": 4.692538159619751e-06, "loss": 0.1480793058872223, "step": 1491 }, { "epoch": 0.387331256490135, "grad_norm": 1.0852744579315186, "learning_rate": 4.692060626717894e-06, "loss": 0.11800035834312439, "step": 1492 }, { "epoch": 0.3875908618899273, "grad_norm": 1.6385070085525513, "learning_rate": 4.691582750504184e-06, "loss": 0.1254182755947113, "step": 1493 }, { "epoch": 0.3878504672897196, "grad_norm": 4.759466171264648, "learning_rate": 4.691104531063101e-06, "loss": 0.15241765975952148, "step": 1494 }, { "epoch": 0.3881100726895119, "grad_norm": 2.794991970062256, "learning_rate": 4.6906259684791816e-06, "loss": 0.1262856274843216, "step": 1495 }, { "epoch": 0.3883696780893043, "grad_norm": 3.169562578201294, "learning_rate": 4.690147062837026e-06, "loss": 0.1277974247932434, "step": 1496 }, { "epoch": 0.3886292834890966, "grad_norm": 1.61216402053833, "learning_rate": 4.689667814221293e-06, "loss": 0.12365536391735077, "step": 1497 }, { "epoch": 0.3888888888888889, "grad_norm": 1.108405590057373, "learning_rate": 4.689188222716703e-06, "loss": 0.12907859683036804, "step": 1498 }, { "epoch": 0.3891484942886812, "grad_norm": 2.5656189918518066, "learning_rate": 4.688708288408036e-06, "loss": 0.1580512523651123, "step": 1499 }, { "epoch": 0.3894080996884735, "grad_norm": 1.9580198526382446, "learning_rate": 4.688228011380135e-06, "loss": 0.13590966165065765, "step": 1500 }, { "epoch": 0.3896677050882658, "grad_norm": 2.406306743621826, "learning_rate": 4.687747391717901e-06, "loss": 0.13741710782051086, "step": 1501 }, { "epoch": 0.3899273104880582, "grad_norm": 2.0721609592437744, "learning_rate": 4.687266429506296e-06, "loss": 0.13152098655700684, "step": 1502 }, { "epoch": 0.3901869158878505, "grad_norm": 2.924036979675293, "learning_rate": 4.686785124830345e-06, "loss": 0.12176994979381561, "step": 1503 }, { "epoch": 0.3904465212876428, "grad_norm": 3.1319615840911865, "learning_rate": 4.686303477775129e-06, "loss": 0.1412050426006317, "step": 1504 }, { "epoch": 0.3907061266874351, "grad_norm": 1.7956922054290771, "learning_rate": 4.685821488425795e-06, "loss": 0.13295024633407593, "step": 1505 }, { "epoch": 0.3909657320872274, "grad_norm": 1.3604886531829834, "learning_rate": 4.6853391568675456e-06, "loss": 0.1189534068107605, "step": 1506 }, { "epoch": 0.3912253374870197, "grad_norm": 2.1305689811706543, "learning_rate": 4.684856483185647e-06, "loss": 0.12080289423465729, "step": 1507 }, { "epoch": 0.39148494288681207, "grad_norm": 2.2864792346954346, "learning_rate": 4.684373467465424e-06, "loss": 0.140235036611557, "step": 1508 }, { "epoch": 0.3917445482866044, "grad_norm": 1.948392629623413, "learning_rate": 4.683890109792264e-06, "loss": 0.15960611402988434, "step": 1509 }, { "epoch": 0.3920041536863967, "grad_norm": 2.289713144302368, "learning_rate": 4.683406410251612e-06, "loss": 0.14784133434295654, "step": 1510 }, { "epoch": 0.392263759086189, "grad_norm": 1.546944499015808, "learning_rate": 4.682922368928977e-06, "loss": 0.11731395870447159, "step": 1511 }, { "epoch": 0.3925233644859813, "grad_norm": 1.0828312635421753, "learning_rate": 4.682437985909924e-06, "loss": 0.13009855151176453, "step": 1512 }, { "epoch": 0.3927829698857736, "grad_norm": 1.5195566415786743, "learning_rate": 4.6819532612800824e-06, "loss": 0.1401260793209076, "step": 1513 }, { "epoch": 0.39304257528556596, "grad_norm": 3.1567606925964355, "learning_rate": 4.68146819512514e-06, "loss": 0.15417146682739258, "step": 1514 }, { "epoch": 0.39330218068535827, "grad_norm": 1.151792049407959, "learning_rate": 4.680982787530845e-06, "loss": 0.14613844454288483, "step": 1515 }, { "epoch": 0.3935617860851506, "grad_norm": 3.4229800701141357, "learning_rate": 4.680497038583008e-06, "loss": 0.1563015580177307, "step": 1516 }, { "epoch": 0.3938213914849429, "grad_norm": 1.9622750282287598, "learning_rate": 4.680010948367496e-06, "loss": 0.12447398155927658, "step": 1517 }, { "epoch": 0.3940809968847352, "grad_norm": 0.9794098138809204, "learning_rate": 4.67952451697024e-06, "loss": 0.12636446952819824, "step": 1518 }, { "epoch": 0.3943406022845275, "grad_norm": 2.3353536128997803, "learning_rate": 4.67903774447723e-06, "loss": 0.14529728889465332, "step": 1519 }, { "epoch": 0.39460020768431986, "grad_norm": 1.9981898069381714, "learning_rate": 4.678550630974515e-06, "loss": 0.1719006896018982, "step": 1520 }, { "epoch": 0.39485981308411217, "grad_norm": 0.853019118309021, "learning_rate": 4.678063176548209e-06, "loss": 0.11857080459594727, "step": 1521 }, { "epoch": 0.39511941848390447, "grad_norm": 1.5774879455566406, "learning_rate": 4.677575381284478e-06, "loss": 0.12030060589313507, "step": 1522 }, { "epoch": 0.3953790238836968, "grad_norm": 2.0239689350128174, "learning_rate": 4.6770872452695546e-06, "loss": 0.12405487149953842, "step": 1523 }, { "epoch": 0.3956386292834891, "grad_norm": 2.6780645847320557, "learning_rate": 4.6765987685897315e-06, "loss": 0.14475315809249878, "step": 1524 }, { "epoch": 0.3958982346832814, "grad_norm": 1.830981969833374, "learning_rate": 4.6761099513313595e-06, "loss": 0.14403954148292542, "step": 1525 }, { "epoch": 0.39615784008307375, "grad_norm": 2.0698959827423096, "learning_rate": 4.67562079358085e-06, "loss": 0.14536789059638977, "step": 1526 }, { "epoch": 0.39641744548286606, "grad_norm": 1.518005609512329, "learning_rate": 4.675131295424675e-06, "loss": 0.1266302615404129, "step": 1527 }, { "epoch": 0.39667705088265837, "grad_norm": 1.9104896783828735, "learning_rate": 4.674641456949368e-06, "loss": 0.13809308409690857, "step": 1528 }, { "epoch": 0.3969366562824507, "grad_norm": 1.962982177734375, "learning_rate": 4.674151278241518e-06, "loss": 0.13858498632907867, "step": 1529 }, { "epoch": 0.397196261682243, "grad_norm": 2.062655210494995, "learning_rate": 4.673660759387779e-06, "loss": 0.13385652005672455, "step": 1530 }, { "epoch": 0.3974558670820353, "grad_norm": 3.5989556312561035, "learning_rate": 4.673169900474865e-06, "loss": 0.12971894443035126, "step": 1531 }, { "epoch": 0.39771547248182765, "grad_norm": 0.7782934904098511, "learning_rate": 4.672678701589547e-06, "loss": 0.1292046159505844, "step": 1532 }, { "epoch": 0.39797507788161995, "grad_norm": 1.4850975275039673, "learning_rate": 4.672187162818658e-06, "loss": 0.11615432798862457, "step": 1533 }, { "epoch": 0.39823468328141226, "grad_norm": 2.3307857513427734, "learning_rate": 4.671695284249091e-06, "loss": 0.10375477373600006, "step": 1534 }, { "epoch": 0.39849428868120457, "grad_norm": 1.1348896026611328, "learning_rate": 4.671203065967799e-06, "loss": 0.10910165309906006, "step": 1535 }, { "epoch": 0.3987538940809969, "grad_norm": 2.8202106952667236, "learning_rate": 4.670710508061794e-06, "loss": 0.14035025238990784, "step": 1536 }, { "epoch": 0.3990134994807892, "grad_norm": 2.4536025524139404, "learning_rate": 4.67021761061815e-06, "loss": 0.12764236330986023, "step": 1537 }, { "epoch": 0.39927310488058154, "grad_norm": 2.4897468090057373, "learning_rate": 4.669724373724e-06, "loss": 0.1537627875804901, "step": 1538 }, { "epoch": 0.39953271028037385, "grad_norm": 2.0580813884735107, "learning_rate": 4.6692307974665355e-06, "loss": 0.15384186804294586, "step": 1539 }, { "epoch": 0.39979231568016615, "grad_norm": 1.942086935043335, "learning_rate": 4.668736881933012e-06, "loss": 0.12609711289405823, "step": 1540 }, { "epoch": 0.40005192107995846, "grad_norm": 2.139195680618286, "learning_rate": 4.66824262721074e-06, "loss": 0.13197234272956848, "step": 1541 }, { "epoch": 0.40031152647975077, "grad_norm": 1.6586185693740845, "learning_rate": 4.6677480333870935e-06, "loss": 0.12051133811473846, "step": 1542 }, { "epoch": 0.4005711318795431, "grad_norm": 0.7163363695144653, "learning_rate": 4.667253100549506e-06, "loss": 0.1336260735988617, "step": 1543 }, { "epoch": 0.40083073727933544, "grad_norm": 1.0022073984146118, "learning_rate": 4.666757828785469e-06, "loss": 0.13228586316108704, "step": 1544 }, { "epoch": 0.40109034267912774, "grad_norm": 0.9735524654388428, "learning_rate": 4.666262218182535e-06, "loss": 0.13548460602760315, "step": 1545 }, { "epoch": 0.40134994807892005, "grad_norm": 0.7149692177772522, "learning_rate": 4.665766268828319e-06, "loss": 0.12565815448760986, "step": 1546 }, { "epoch": 0.40160955347871236, "grad_norm": 1.2304208278656006, "learning_rate": 4.66526998081049e-06, "loss": 0.1264282912015915, "step": 1547 }, { "epoch": 0.40186915887850466, "grad_norm": 1.527355670928955, "learning_rate": 4.664773354216785e-06, "loss": 0.1241949275135994, "step": 1548 }, { "epoch": 0.40212876427829697, "grad_norm": 2.410564422607422, "learning_rate": 4.664276389134991e-06, "loss": 0.12741857767105103, "step": 1549 }, { "epoch": 0.40238836967808933, "grad_norm": 2.446528434753418, "learning_rate": 4.663779085652964e-06, "loss": 0.14470872282981873, "step": 1550 }, { "epoch": 0.40264797507788164, "grad_norm": 3.188675880432129, "learning_rate": 4.663281443858614e-06, "loss": 0.154939204454422, "step": 1551 }, { "epoch": 0.40290758047767394, "grad_norm": 3.7186479568481445, "learning_rate": 4.662783463839913e-06, "loss": 0.12230600416660309, "step": 1552 }, { "epoch": 0.40316718587746625, "grad_norm": 3.122513771057129, "learning_rate": 4.662285145684892e-06, "loss": 0.14145737886428833, "step": 1553 }, { "epoch": 0.40342679127725856, "grad_norm": 2.1632306575775146, "learning_rate": 4.661786489481644e-06, "loss": 0.17060966789722443, "step": 1554 }, { "epoch": 0.40368639667705086, "grad_norm": 6.105969429016113, "learning_rate": 4.661287495318317e-06, "loss": 0.1627768576145172, "step": 1555 }, { "epoch": 0.4039460020768432, "grad_norm": 4.993813514709473, "learning_rate": 4.660788163283125e-06, "loss": 0.1904834806919098, "step": 1556 }, { "epoch": 0.40420560747663553, "grad_norm": 3.003261089324951, "learning_rate": 4.6602884934643364e-06, "loss": 0.16173169016838074, "step": 1557 }, { "epoch": 0.40446521287642784, "grad_norm": 1.701500415802002, "learning_rate": 4.659788485950282e-06, "loss": 0.1492013782262802, "step": 1558 }, { "epoch": 0.40472481827622014, "grad_norm": 3.1943297386169434, "learning_rate": 4.659288140829351e-06, "loss": 0.1285531371831894, "step": 1559 }, { "epoch": 0.40498442367601245, "grad_norm": 1.9258382320404053, "learning_rate": 4.658787458189995e-06, "loss": 0.1442459225654602, "step": 1560 }, { "epoch": 0.40524402907580476, "grad_norm": 2.9891936779022217, "learning_rate": 4.658286438120721e-06, "loss": 0.16032439470291138, "step": 1561 }, { "epoch": 0.4055036344755971, "grad_norm": 2.4365696907043457, "learning_rate": 4.657785080710099e-06, "loss": 0.15746556222438812, "step": 1562 }, { "epoch": 0.4057632398753894, "grad_norm": 3.6938774585723877, "learning_rate": 4.657283386046757e-06, "loss": 0.12498936802148819, "step": 1563 }, { "epoch": 0.40602284527518173, "grad_norm": 1.4014562368392944, "learning_rate": 4.656781354219383e-06, "loss": 0.1585405468940735, "step": 1564 }, { "epoch": 0.40628245067497404, "grad_norm": 1.8489885330200195, "learning_rate": 4.656278985316725e-06, "loss": 0.12555839121341705, "step": 1565 }, { "epoch": 0.40654205607476634, "grad_norm": 1.279691457748413, "learning_rate": 4.655776279427591e-06, "loss": 0.12308369576931, "step": 1566 }, { "epoch": 0.40680166147455865, "grad_norm": 2.176668167114258, "learning_rate": 4.655273236640847e-06, "loss": 0.13111291825771332, "step": 1567 }, { "epoch": 0.407061266874351, "grad_norm": 2.1180260181427, "learning_rate": 4.6547698570454206e-06, "loss": 0.14599308371543884, "step": 1568 }, { "epoch": 0.4073208722741433, "grad_norm": 3.0751709938049316, "learning_rate": 4.654266140730297e-06, "loss": 0.15869663655757904, "step": 1569 }, { "epoch": 0.4075804776739356, "grad_norm": 1.0687483549118042, "learning_rate": 4.653762087784521e-06, "loss": 0.11780278384685516, "step": 1570 }, { "epoch": 0.40784008307372793, "grad_norm": 0.8945303559303284, "learning_rate": 4.653257698297198e-06, "loss": 0.12014713883399963, "step": 1571 }, { "epoch": 0.40809968847352024, "grad_norm": 1.5780185461044312, "learning_rate": 4.652752972357492e-06, "loss": 0.15834520757198334, "step": 1572 }, { "epoch": 0.40835929387331255, "grad_norm": 1.2632560729980469, "learning_rate": 4.652247910054629e-06, "loss": 0.13692152500152588, "step": 1573 }, { "epoch": 0.4086188992731049, "grad_norm": 2.551466464996338, "learning_rate": 4.6517425114778905e-06, "loss": 0.14322346448898315, "step": 1574 }, { "epoch": 0.4088785046728972, "grad_norm": 2.748467206954956, "learning_rate": 4.65123677671662e-06, "loss": 0.11911319941282272, "step": 1575 }, { "epoch": 0.4091381100726895, "grad_norm": 2.7181196212768555, "learning_rate": 4.6507307058602205e-06, "loss": 0.13264569640159607, "step": 1576 }, { "epoch": 0.4093977154724818, "grad_norm": 1.323907494544983, "learning_rate": 4.650224298998154e-06, "loss": 0.12190558761358261, "step": 1577 }, { "epoch": 0.40965732087227413, "grad_norm": 2.426690101623535, "learning_rate": 4.649717556219939e-06, "loss": 0.14690738916397095, "step": 1578 }, { "epoch": 0.40991692627206644, "grad_norm": 1.5253641605377197, "learning_rate": 4.649210477615159e-06, "loss": 0.12556345760822296, "step": 1579 }, { "epoch": 0.4101765316718588, "grad_norm": 2.1710076332092285, "learning_rate": 4.648703063273452e-06, "loss": 0.12746891379356384, "step": 1580 }, { "epoch": 0.4104361370716511, "grad_norm": 1.706214427947998, "learning_rate": 4.6481953132845184e-06, "loss": 0.15513285994529724, "step": 1581 }, { "epoch": 0.4106957424714434, "grad_norm": 1.6883137226104736, "learning_rate": 4.647687227738116e-06, "loss": 0.16075453162193298, "step": 1582 }, { "epoch": 0.4109553478712357, "grad_norm": 0.8340747356414795, "learning_rate": 4.647178806724064e-06, "loss": 0.12284763157367706, "step": 1583 }, { "epoch": 0.411214953271028, "grad_norm": 1.0409972667694092, "learning_rate": 4.646670050332238e-06, "loss": 0.12142769247293472, "step": 1584 }, { "epoch": 0.41147455867082033, "grad_norm": 4.175857067108154, "learning_rate": 4.646160958652575e-06, "loss": 0.14108595252037048, "step": 1585 }, { "epoch": 0.4117341640706127, "grad_norm": 1.7173020839691162, "learning_rate": 4.645651531775071e-06, "loss": 0.11925645917654037, "step": 1586 }, { "epoch": 0.411993769470405, "grad_norm": 2.9328417778015137, "learning_rate": 4.64514176978978e-06, "loss": 0.1510319709777832, "step": 1587 }, { "epoch": 0.4122533748701973, "grad_norm": 1.4539213180541992, "learning_rate": 4.6446316727868175e-06, "loss": 0.16489043831825256, "step": 1588 }, { "epoch": 0.4125129802699896, "grad_norm": 2.2537364959716797, "learning_rate": 4.644121240856357e-06, "loss": 0.13310301303863525, "step": 1589 }, { "epoch": 0.4127725856697819, "grad_norm": 3.4683187007904053, "learning_rate": 4.6436104740886305e-06, "loss": 0.1251639574766159, "step": 1590 }, { "epoch": 0.41303219106957423, "grad_norm": 2.4857566356658936, "learning_rate": 4.64309937257393e-06, "loss": 0.1590220183134079, "step": 1591 }, { "epoch": 0.4132917964693666, "grad_norm": 3.1810030937194824, "learning_rate": 4.642587936402607e-06, "loss": 0.1370125412940979, "step": 1592 }, { "epoch": 0.4135514018691589, "grad_norm": 1.9003909826278687, "learning_rate": 4.642076165665071e-06, "loss": 0.1340569108724594, "step": 1593 }, { "epoch": 0.4138110072689512, "grad_norm": 1.1769055128097534, "learning_rate": 4.641564060451792e-06, "loss": 0.13740918040275574, "step": 1594 }, { "epoch": 0.4140706126687435, "grad_norm": 1.1699144840240479, "learning_rate": 4.641051620853298e-06, "loss": 0.12913398444652557, "step": 1595 }, { "epoch": 0.4143302180685358, "grad_norm": 3.070997953414917, "learning_rate": 4.640538846960177e-06, "loss": 0.12431541085243225, "step": 1596 }, { "epoch": 0.4145898234683281, "grad_norm": 5.500311374664307, "learning_rate": 4.640025738863076e-06, "loss": 0.15546521544456482, "step": 1597 }, { "epoch": 0.4148494288681205, "grad_norm": 2.295590877532959, "learning_rate": 4.639512296652699e-06, "loss": 0.18519994616508484, "step": 1598 }, { "epoch": 0.4151090342679128, "grad_norm": 1.4254968166351318, "learning_rate": 4.638998520419812e-06, "loss": 0.10819556564092636, "step": 1599 }, { "epoch": 0.4153686396677051, "grad_norm": 0.9076033234596252, "learning_rate": 4.63848441025524e-06, "loss": 0.11775277554988861, "step": 1600 }, { "epoch": 0.4156282450674974, "grad_norm": 1.9913018941879272, "learning_rate": 4.6379699662498646e-06, "loss": 0.14760658144950867, "step": 1601 }, { "epoch": 0.4158878504672897, "grad_norm": 4.028369903564453, "learning_rate": 4.637455188494627e-06, "loss": 0.1618044376373291, "step": 1602 }, { "epoch": 0.416147455867082, "grad_norm": 0.6942084431648254, "learning_rate": 4.636940077080531e-06, "loss": 0.12638935446739197, "step": 1603 }, { "epoch": 0.4164070612668744, "grad_norm": 1.142223596572876, "learning_rate": 4.6364246320986325e-06, "loss": 0.13086344301700592, "step": 1604 }, { "epoch": 0.4166666666666667, "grad_norm": 2.3548154830932617, "learning_rate": 4.6359088536400534e-06, "loss": 0.13092859089374542, "step": 1605 }, { "epoch": 0.416926272066459, "grad_norm": 1.8774040937423706, "learning_rate": 4.635392741795971e-06, "loss": 0.13865861296653748, "step": 1606 }, { "epoch": 0.4171858774662513, "grad_norm": 1.5782153606414795, "learning_rate": 4.63487629665762e-06, "loss": 0.14593400061130524, "step": 1607 }, { "epoch": 0.4174454828660436, "grad_norm": 3.4370625019073486, "learning_rate": 4.634359518316299e-06, "loss": 0.18144822120666504, "step": 1608 }, { "epoch": 0.4177050882658359, "grad_norm": 3.220466375350952, "learning_rate": 4.6338424068633615e-06, "loss": 0.11008962988853455, "step": 1609 }, { "epoch": 0.4179646936656282, "grad_norm": 5.316440582275391, "learning_rate": 4.6333249623902195e-06, "loss": 0.12892493605613708, "step": 1610 }, { "epoch": 0.4182242990654206, "grad_norm": 3.8291661739349365, "learning_rate": 4.632807184988348e-06, "loss": 0.1095842719078064, "step": 1611 }, { "epoch": 0.4184839044652129, "grad_norm": 3.0517687797546387, "learning_rate": 4.6322890747492765e-06, "loss": 0.14555545151233673, "step": 1612 }, { "epoch": 0.4187435098650052, "grad_norm": 2.1414358615875244, "learning_rate": 4.631770631764595e-06, "loss": 0.12436568737030029, "step": 1613 }, { "epoch": 0.4190031152647975, "grad_norm": 3.509148359298706, "learning_rate": 4.631251856125952e-06, "loss": 0.16264155507087708, "step": 1614 }, { "epoch": 0.4192627206645898, "grad_norm": 3.054633378982544, "learning_rate": 4.630732747925057e-06, "loss": 0.14836585521697998, "step": 1615 }, { "epoch": 0.4195223260643821, "grad_norm": 1.9101730585098267, "learning_rate": 4.630213307253674e-06, "loss": 0.1317736804485321, "step": 1616 }, { "epoch": 0.4197819314641745, "grad_norm": 3.0398061275482178, "learning_rate": 4.6296935342036305e-06, "loss": 0.15678466856479645, "step": 1617 }, { "epoch": 0.4200415368639668, "grad_norm": 2.4573254585266113, "learning_rate": 4.629173428866808e-06, "loss": 0.1197967603802681, "step": 1618 }, { "epoch": 0.4203011422637591, "grad_norm": 2.625749349594116, "learning_rate": 4.628652991335153e-06, "loss": 0.1522383838891983, "step": 1619 }, { "epoch": 0.4205607476635514, "grad_norm": 1.160304307937622, "learning_rate": 4.628132221700663e-06, "loss": 0.1212373748421669, "step": 1620 }, { "epoch": 0.4208203530633437, "grad_norm": 0.8768002390861511, "learning_rate": 4.627611120055399e-06, "loss": 0.11352908611297607, "step": 1621 }, { "epoch": 0.421079958463136, "grad_norm": 3.3436760902404785, "learning_rate": 4.627089686491482e-06, "loss": 0.11984486877918243, "step": 1622 }, { "epoch": 0.42133956386292837, "grad_norm": 2.1725971698760986, "learning_rate": 4.626567921101087e-06, "loss": 0.1428084522485733, "step": 1623 }, { "epoch": 0.4215991692627207, "grad_norm": 0.9787641763687134, "learning_rate": 4.626045823976453e-06, "loss": 0.11755163222551346, "step": 1624 }, { "epoch": 0.421858774662513, "grad_norm": 0.9065307378768921, "learning_rate": 4.625523395209871e-06, "loss": 0.11264517903327942, "step": 1625 }, { "epoch": 0.4221183800623053, "grad_norm": 1.4115575551986694, "learning_rate": 4.625000634893698e-06, "loss": 0.12975120544433594, "step": 1626 }, { "epoch": 0.4223779854620976, "grad_norm": 3.6656079292297363, "learning_rate": 4.624477543120344e-06, "loss": 0.1442541629076004, "step": 1627 }, { "epoch": 0.4226375908618899, "grad_norm": 1.7553836107254028, "learning_rate": 4.62395411998228e-06, "loss": 0.13189351558685303, "step": 1628 }, { "epoch": 0.42289719626168226, "grad_norm": 1.5720317363739014, "learning_rate": 4.623430365572036e-06, "loss": 0.11098207533359528, "step": 1629 }, { "epoch": 0.42315680166147457, "grad_norm": 1.804314374923706, "learning_rate": 4.622906279982199e-06, "loss": 0.16460460424423218, "step": 1630 }, { "epoch": 0.4234164070612669, "grad_norm": 2.839451789855957, "learning_rate": 4.6223818633054155e-06, "loss": 0.1368458867073059, "step": 1631 }, { "epoch": 0.4236760124610592, "grad_norm": 1.7597509622573853, "learning_rate": 4.62185711563439e-06, "loss": 0.1285489797592163, "step": 1632 }, { "epoch": 0.4239356178608515, "grad_norm": 0.6872251033782959, "learning_rate": 4.621332037061887e-06, "loss": 0.12017548084259033, "step": 1633 }, { "epoch": 0.4241952232606438, "grad_norm": 1.3989812135696411, "learning_rate": 4.620806627680728e-06, "loss": 0.13385790586471558, "step": 1634 }, { "epoch": 0.42445482866043616, "grad_norm": 1.54410719871521, "learning_rate": 4.6202808875837915e-06, "loss": 0.15508083999156952, "step": 1635 }, { "epoch": 0.42471443406022846, "grad_norm": 1.8374974727630615, "learning_rate": 4.619754816864019e-06, "loss": 0.15832236409187317, "step": 1636 }, { "epoch": 0.42497403946002077, "grad_norm": 1.046613097190857, "learning_rate": 4.619228415614406e-06, "loss": 0.1466858685016632, "step": 1637 }, { "epoch": 0.4252336448598131, "grad_norm": 1.1566680669784546, "learning_rate": 4.61870168392801e-06, "loss": 0.1300225406885147, "step": 1638 }, { "epoch": 0.4254932502596054, "grad_norm": 2.0743067264556885, "learning_rate": 4.6181746218979424e-06, "loss": 0.11934428662061691, "step": 1639 }, { "epoch": 0.4257528556593977, "grad_norm": 2.099224805831909, "learning_rate": 4.6176472296173775e-06, "loss": 0.13234373927116394, "step": 1640 }, { "epoch": 0.42601246105919005, "grad_norm": 2.3820149898529053, "learning_rate": 4.617119507179546e-06, "loss": 0.1132281944155693, "step": 1641 }, { "epoch": 0.42627206645898236, "grad_norm": 0.9514886140823364, "learning_rate": 4.616591454677737e-06, "loss": 0.1163700595498085, "step": 1642 }, { "epoch": 0.42653167185877466, "grad_norm": 0.8687743544578552, "learning_rate": 4.616063072205298e-06, "loss": 0.12327706813812256, "step": 1643 }, { "epoch": 0.42679127725856697, "grad_norm": 2.1050233840942383, "learning_rate": 4.615534359855636e-06, "loss": 0.15641546249389648, "step": 1644 }, { "epoch": 0.4270508826583593, "grad_norm": 1.9259246587753296, "learning_rate": 4.615005317722213e-06, "loss": 0.12234854698181152, "step": 1645 }, { "epoch": 0.4273104880581516, "grad_norm": 1.8222986459732056, "learning_rate": 4.614475945898554e-06, "loss": 0.11539669334888458, "step": 1646 }, { "epoch": 0.42757009345794394, "grad_norm": 2.304982900619507, "learning_rate": 4.613946244478238e-06, "loss": 0.1187230572104454, "step": 1647 }, { "epoch": 0.42782969885773625, "grad_norm": 1.1863491535186768, "learning_rate": 4.613416213554904e-06, "loss": 0.10720635950565338, "step": 1648 }, { "epoch": 0.42808930425752856, "grad_norm": 0.7927123308181763, "learning_rate": 4.61288585322225e-06, "loss": 0.1048865020275116, "step": 1649 }, { "epoch": 0.42834890965732086, "grad_norm": 1.2274484634399414, "learning_rate": 4.6123551635740336e-06, "loss": 0.11776265501976013, "step": 1650 }, { "epoch": 0.42860851505711317, "grad_norm": 1.0594381093978882, "learning_rate": 4.611824144704064e-06, "loss": 0.1314769834280014, "step": 1651 }, { "epoch": 0.4288681204569055, "grad_norm": 2.5062241554260254, "learning_rate": 4.611292796706218e-06, "loss": 0.1257939636707306, "step": 1652 }, { "epoch": 0.42912772585669784, "grad_norm": 1.109006404876709, "learning_rate": 4.610761119674422e-06, "loss": 0.1197441816329956, "step": 1653 }, { "epoch": 0.42938733125649015, "grad_norm": 1.3449221849441528, "learning_rate": 4.610229113702666e-06, "loss": 0.12735679745674133, "step": 1654 }, { "epoch": 0.42964693665628245, "grad_norm": 1.9588850736618042, "learning_rate": 4.609696778884997e-06, "loss": 0.11957129091024399, "step": 1655 }, { "epoch": 0.42990654205607476, "grad_norm": 2.0965771675109863, "learning_rate": 4.6091641153155175e-06, "loss": 0.10609216243028641, "step": 1656 }, { "epoch": 0.43016614745586707, "grad_norm": 2.6801836490631104, "learning_rate": 4.608631123088393e-06, "loss": 0.13337092101573944, "step": 1657 }, { "epoch": 0.43042575285565937, "grad_norm": 2.121567487716675, "learning_rate": 4.6080978022978425e-06, "loss": 0.12014596909284592, "step": 1658 }, { "epoch": 0.43068535825545173, "grad_norm": 1.9428355693817139, "learning_rate": 4.607564153038146e-06, "loss": 0.1337609887123108, "step": 1659 }, { "epoch": 0.43094496365524404, "grad_norm": 2.1719117164611816, "learning_rate": 4.607030175403638e-06, "loss": 0.13542307913303375, "step": 1660 }, { "epoch": 0.43120456905503635, "grad_norm": 3.6154074668884277, "learning_rate": 4.606495869488716e-06, "loss": 0.1553022414445877, "step": 1661 }, { "epoch": 0.43146417445482865, "grad_norm": 2.063239574432373, "learning_rate": 4.605961235387834e-06, "loss": 0.14332973957061768, "step": 1662 }, { "epoch": 0.43172377985462096, "grad_norm": 3.610060214996338, "learning_rate": 4.605426273195499e-06, "loss": 0.14653152227401733, "step": 1663 }, { "epoch": 0.43198338525441327, "grad_norm": 3.07220458984375, "learning_rate": 4.604890983006285e-06, "loss": 0.1260397583246231, "step": 1664 }, { "epoch": 0.4322429906542056, "grad_norm": 3.3220956325531006, "learning_rate": 4.604355364914816e-06, "loss": 0.15730449557304382, "step": 1665 }, { "epoch": 0.43250259605399793, "grad_norm": 3.1383495330810547, "learning_rate": 4.603819419015776e-06, "loss": 0.1567794382572174, "step": 1666 }, { "epoch": 0.43276220145379024, "grad_norm": 3.2479867935180664, "learning_rate": 4.603283145403911e-06, "loss": 0.12320179492235184, "step": 1667 }, { "epoch": 0.43302180685358255, "grad_norm": 2.0668134689331055, "learning_rate": 4.602746544174021e-06, "loss": 0.12184067815542221, "step": 1668 }, { "epoch": 0.43328141225337485, "grad_norm": 2.364611864089966, "learning_rate": 4.602209615420963e-06, "loss": 0.1381237655878067, "step": 1669 }, { "epoch": 0.43354101765316716, "grad_norm": 1.3164323568344116, "learning_rate": 4.601672359239654e-06, "loss": 0.11787334084510803, "step": 1670 }, { "epoch": 0.4338006230529595, "grad_norm": 2.3559296131134033, "learning_rate": 4.6011347757250716e-06, "loss": 0.1451016068458557, "step": 1671 }, { "epoch": 0.43406022845275183, "grad_norm": 6.120560169219971, "learning_rate": 4.600596864972246e-06, "loss": 0.16190466284751892, "step": 1672 }, { "epoch": 0.43431983385254413, "grad_norm": 1.673815131187439, "learning_rate": 4.600058627076266e-06, "loss": 0.11458516120910645, "step": 1673 }, { "epoch": 0.43457943925233644, "grad_norm": 2.241781234741211, "learning_rate": 4.599520062132283e-06, "loss": 0.13378594815731049, "step": 1674 }, { "epoch": 0.43483904465212875, "grad_norm": 4.5161213874816895, "learning_rate": 4.598981170235501e-06, "loss": 0.1392088085412979, "step": 1675 }, { "epoch": 0.43509865005192105, "grad_norm": 2.136446237564087, "learning_rate": 4.598441951481184e-06, "loss": 0.11588843166828156, "step": 1676 }, { "epoch": 0.4353582554517134, "grad_norm": 2.568730115890503, "learning_rate": 4.597902405964654e-06, "loss": 0.12304569780826569, "step": 1677 }, { "epoch": 0.4356178608515057, "grad_norm": 2.272228240966797, "learning_rate": 4.59736253378129e-06, "loss": 0.12517066299915314, "step": 1678 }, { "epoch": 0.43587746625129803, "grad_norm": 2.515228271484375, "learning_rate": 4.596822335026529e-06, "loss": 0.12261714041233063, "step": 1679 }, { "epoch": 0.43613707165109034, "grad_norm": 3.7082772254943848, "learning_rate": 4.596281809795866e-06, "loss": 0.13825663924217224, "step": 1680 }, { "epoch": 0.43639667705088264, "grad_norm": 1.3873342275619507, "learning_rate": 4.5957409581848535e-06, "loss": 0.12682285904884338, "step": 1681 }, { "epoch": 0.43665628245067495, "grad_norm": 3.6899187564849854, "learning_rate": 4.595199780289101e-06, "loss": 0.16325832903385162, "step": 1682 }, { "epoch": 0.4369158878504673, "grad_norm": 3.0788979530334473, "learning_rate": 4.5946582762042765e-06, "loss": 0.1405901163816452, "step": 1683 }, { "epoch": 0.4371754932502596, "grad_norm": 2.351841926574707, "learning_rate": 4.594116446026106e-06, "loss": 0.13199692964553833, "step": 1684 }, { "epoch": 0.4374350986500519, "grad_norm": 2.8104891777038574, "learning_rate": 4.5935742898503725e-06, "loss": 0.13057103753089905, "step": 1685 }, { "epoch": 0.43769470404984423, "grad_norm": 1.9955378770828247, "learning_rate": 4.5930318077729165e-06, "loss": 0.12152154743671417, "step": 1686 }, { "epoch": 0.43795430944963654, "grad_norm": 1.1867367029190063, "learning_rate": 4.592488999889638e-06, "loss": 0.12240763008594513, "step": 1687 }, { "epoch": 0.43821391484942884, "grad_norm": 0.986609935760498, "learning_rate": 4.5919458662964905e-06, "loss": 0.11008930206298828, "step": 1688 }, { "epoch": 0.4384735202492212, "grad_norm": 1.6075668334960938, "learning_rate": 4.591402407089489e-06, "loss": 0.11339017003774643, "step": 1689 }, { "epoch": 0.4387331256490135, "grad_norm": 3.4001896381378174, "learning_rate": 4.590858622364705e-06, "loss": 0.11967264115810394, "step": 1690 }, { "epoch": 0.4389927310488058, "grad_norm": 1.348446249961853, "learning_rate": 4.590314512218266e-06, "loss": 0.1428850293159485, "step": 1691 }, { "epoch": 0.4392523364485981, "grad_norm": 0.888413667678833, "learning_rate": 4.58977007674636e-06, "loss": 0.1234244853258133, "step": 1692 }, { "epoch": 0.43951194184839043, "grad_norm": 1.9612904787063599, "learning_rate": 4.589225316045228e-06, "loss": 0.12498094886541367, "step": 1693 }, { "epoch": 0.43977154724818274, "grad_norm": 3.0118179321289062, "learning_rate": 4.588680230211174e-06, "loss": 0.1258326768875122, "step": 1694 }, { "epoch": 0.4400311526479751, "grad_norm": 1.5786560773849487, "learning_rate": 4.588134819340554e-06, "loss": 0.1242460384964943, "step": 1695 }, { "epoch": 0.4402907580477674, "grad_norm": 3.1718626022338867, "learning_rate": 4.587589083529788e-06, "loss": 0.1679912507534027, "step": 1696 }, { "epoch": 0.4405503634475597, "grad_norm": 1.8912794589996338, "learning_rate": 4.587043022875345e-06, "loss": 0.14026769995689392, "step": 1697 }, { "epoch": 0.440809968847352, "grad_norm": 4.3002800941467285, "learning_rate": 4.586496637473759e-06, "loss": 0.12959596514701843, "step": 1698 }, { "epoch": 0.4410695742471443, "grad_norm": 2.6899445056915283, "learning_rate": 4.585949927421618e-06, "loss": 0.13263605535030365, "step": 1699 }, { "epoch": 0.44132917964693663, "grad_norm": 2.810262441635132, "learning_rate": 4.5854028928155674e-06, "loss": 0.11321863532066345, "step": 1700 }, { "epoch": 0.441588785046729, "grad_norm": 1.574061632156372, "learning_rate": 4.584855533752311e-06, "loss": 0.12064475566148758, "step": 1701 }, { "epoch": 0.4418483904465213, "grad_norm": 1.8178093433380127, "learning_rate": 4.584307850328608e-06, "loss": 0.11880029737949371, "step": 1702 }, { "epoch": 0.4421079958463136, "grad_norm": 4.1219401359558105, "learning_rate": 4.5837598426412775e-06, "loss": 0.1182326227426529, "step": 1703 }, { "epoch": 0.4423676012461059, "grad_norm": 3.0597519874572754, "learning_rate": 4.583211510787194e-06, "loss": 0.12028147280216217, "step": 1704 }, { "epoch": 0.4426272066458982, "grad_norm": 0.829649031162262, "learning_rate": 4.582662854863291e-06, "loss": 0.1139221340417862, "step": 1705 }, { "epoch": 0.4428868120456905, "grad_norm": 1.6498602628707886, "learning_rate": 4.582113874966558e-06, "loss": 0.17425194382667542, "step": 1706 }, { "epoch": 0.4431464174454829, "grad_norm": 1.8177025318145752, "learning_rate": 4.581564571194041e-06, "loss": 0.12978611886501312, "step": 1707 }, { "epoch": 0.4434060228452752, "grad_norm": 4.5506815910339355, "learning_rate": 4.581014943642845e-06, "loss": 0.17067840695381165, "step": 1708 }, { "epoch": 0.4436656282450675, "grad_norm": 3.6685233116149902, "learning_rate": 4.580464992410132e-06, "loss": 0.14024467766284943, "step": 1709 }, { "epoch": 0.4439252336448598, "grad_norm": 5.68958854675293, "learning_rate": 4.57991471759312e-06, "loss": 0.16300855576992035, "step": 1710 }, { "epoch": 0.4441848390446521, "grad_norm": 3.0761635303497314, "learning_rate": 4.579364119289086e-06, "loss": 0.1559440642595291, "step": 1711 }, { "epoch": 0.4444444444444444, "grad_norm": 1.9395025968551636, "learning_rate": 4.578813197595363e-06, "loss": 0.15400739014148712, "step": 1712 }, { "epoch": 0.4447040498442368, "grad_norm": 1.454890489578247, "learning_rate": 4.578261952609339e-06, "loss": 0.1333225965499878, "step": 1713 }, { "epoch": 0.4449636552440291, "grad_norm": 2.777334213256836, "learning_rate": 4.577710384428466e-06, "loss": 0.1303519308567047, "step": 1714 }, { "epoch": 0.4452232606438214, "grad_norm": 7.23226261138916, "learning_rate": 4.577158493150244e-06, "loss": 0.16276849806308746, "step": 1715 }, { "epoch": 0.4454828660436137, "grad_norm": 6.832577228546143, "learning_rate": 4.576606278872238e-06, "loss": 0.14870019257068634, "step": 1716 }, { "epoch": 0.445742471443406, "grad_norm": 2.555166006088257, "learning_rate": 4.576053741692065e-06, "loss": 0.1420269012451172, "step": 1717 }, { "epoch": 0.4460020768431983, "grad_norm": 3.1285386085510254, "learning_rate": 4.575500881707402e-06, "loss": 0.1485336422920227, "step": 1718 }, { "epoch": 0.4462616822429907, "grad_norm": 3.8227880001068115, "learning_rate": 4.574947699015981e-06, "loss": 0.15184953808784485, "step": 1719 }, { "epoch": 0.446521287642783, "grad_norm": 4.688770294189453, "learning_rate": 4.574394193715593e-06, "loss": 0.16239303350448608, "step": 1720 }, { "epoch": 0.4467808930425753, "grad_norm": 4.501445293426514, "learning_rate": 4.573840365904084e-06, "loss": 0.131692036986351, "step": 1721 }, { "epoch": 0.4470404984423676, "grad_norm": 2.905322790145874, "learning_rate": 4.57328621567936e-06, "loss": 0.15249904990196228, "step": 1722 }, { "epoch": 0.4473001038421599, "grad_norm": 3.149751901626587, "learning_rate": 4.57273174313938e-06, "loss": 0.11797140538692474, "step": 1723 }, { "epoch": 0.4475597092419522, "grad_norm": 5.056156158447266, "learning_rate": 4.5721769483821625e-06, "loss": 0.188409686088562, "step": 1724 }, { "epoch": 0.44781931464174457, "grad_norm": 2.6835217475891113, "learning_rate": 4.571621831505783e-06, "loss": 0.13415029644966125, "step": 1725 }, { "epoch": 0.4480789200415369, "grad_norm": 3.890195846557617, "learning_rate": 4.571066392608374e-06, "loss": 0.16520902514457703, "step": 1726 }, { "epoch": 0.4483385254413292, "grad_norm": 1.5330986976623535, "learning_rate": 4.5705106317881225e-06, "loss": 0.15148845314979553, "step": 1727 }, { "epoch": 0.4485981308411215, "grad_norm": 2.3484201431274414, "learning_rate": 4.569954549143276e-06, "loss": 0.168706476688385, "step": 1728 }, { "epoch": 0.4488577362409138, "grad_norm": 1.3614164590835571, "learning_rate": 4.569398144772137e-06, "loss": 0.11801806092262268, "step": 1729 }, { "epoch": 0.4491173416407061, "grad_norm": 1.9768099784851074, "learning_rate": 4.568841418773064e-06, "loss": 0.11919497698545456, "step": 1730 }, { "epoch": 0.44937694704049846, "grad_norm": 2.0827701091766357, "learning_rate": 4.568284371244475e-06, "loss": 0.14085116982460022, "step": 1731 }, { "epoch": 0.44963655244029077, "grad_norm": 3.1338021755218506, "learning_rate": 4.567727002284842e-06, "loss": 0.14805138111114502, "step": 1732 }, { "epoch": 0.4498961578400831, "grad_norm": 1.3862848281860352, "learning_rate": 4.567169311992696e-06, "loss": 0.1300428807735443, "step": 1733 }, { "epoch": 0.4501557632398754, "grad_norm": 2.06801438331604, "learning_rate": 4.566611300466622e-06, "loss": 0.14482693374156952, "step": 1734 }, { "epoch": 0.4504153686396677, "grad_norm": 2.0088157653808594, "learning_rate": 4.566052967805266e-06, "loss": 0.12047277390956879, "step": 1735 }, { "epoch": 0.45067497403946, "grad_norm": 2.8782851696014404, "learning_rate": 4.565494314107327e-06, "loss": 0.11372576653957367, "step": 1736 }, { "epoch": 0.45093457943925236, "grad_norm": 1.9475449323654175, "learning_rate": 4.564935339471563e-06, "loss": 0.12626546621322632, "step": 1737 }, { "epoch": 0.45119418483904467, "grad_norm": 1.481911063194275, "learning_rate": 4.564376043996787e-06, "loss": 0.12896721065044403, "step": 1738 }, { "epoch": 0.45145379023883697, "grad_norm": 1.3144551515579224, "learning_rate": 4.5638164277818705e-06, "loss": 0.12805485725402832, "step": 1739 }, { "epoch": 0.4517133956386293, "grad_norm": 2.3615334033966064, "learning_rate": 4.56325649092574e-06, "loss": 0.14507758617401123, "step": 1740 }, { "epoch": 0.4519730010384216, "grad_norm": 2.302398681640625, "learning_rate": 4.56269623352738e-06, "loss": 0.13046476244926453, "step": 1741 }, { "epoch": 0.4522326064382139, "grad_norm": 1.793242335319519, "learning_rate": 4.562135655685831e-06, "loss": 0.14193874597549438, "step": 1742 }, { "epoch": 0.45249221183800625, "grad_norm": 0.8508290648460388, "learning_rate": 4.561574757500192e-06, "loss": 0.12954729795455933, "step": 1743 }, { "epoch": 0.45275181723779856, "grad_norm": 1.0479978322982788, "learning_rate": 4.5610135390696135e-06, "loss": 0.12880469858646393, "step": 1744 }, { "epoch": 0.45301142263759087, "grad_norm": 1.172229290008545, "learning_rate": 4.560452000493309e-06, "loss": 0.1082909107208252, "step": 1745 }, { "epoch": 0.4532710280373832, "grad_norm": 2.1447665691375732, "learning_rate": 4.559890141870545e-06, "loss": 0.14197266101837158, "step": 1746 }, { "epoch": 0.4535306334371755, "grad_norm": 0.8000888824462891, "learning_rate": 4.559327963300643e-06, "loss": 0.11919485032558441, "step": 1747 }, { "epoch": 0.4537902388369678, "grad_norm": 1.9591023921966553, "learning_rate": 4.558765464882987e-06, "loss": 0.1862388253211975, "step": 1748 }, { "epoch": 0.45404984423676015, "grad_norm": 1.2067800760269165, "learning_rate": 4.558202646717011e-06, "loss": 0.11650595813989639, "step": 1749 }, { "epoch": 0.45430944963655245, "grad_norm": 1.3411568403244019, "learning_rate": 4.557639508902211e-06, "loss": 0.12040483951568604, "step": 1750 }, { "epoch": 0.45456905503634476, "grad_norm": 1.254858374595642, "learning_rate": 4.557076051538134e-06, "loss": 0.11809828132390976, "step": 1751 }, { "epoch": 0.45482866043613707, "grad_norm": 1.631534218788147, "learning_rate": 4.556512274724387e-06, "loss": 0.14228996634483337, "step": 1752 }, { "epoch": 0.4550882658359294, "grad_norm": 2.170114517211914, "learning_rate": 4.555948178560635e-06, "loss": 0.1053905189037323, "step": 1753 }, { "epoch": 0.4553478712357217, "grad_norm": 1.174186110496521, "learning_rate": 4.555383763146596e-06, "loss": 0.1256886124610901, "step": 1754 }, { "epoch": 0.45560747663551404, "grad_norm": 2.588801145553589, "learning_rate": 4.554819028582044e-06, "loss": 0.11400370299816132, "step": 1755 }, { "epoch": 0.45586708203530635, "grad_norm": 1.3588180541992188, "learning_rate": 4.5542539749668134e-06, "loss": 0.12691529095172882, "step": 1756 }, { "epoch": 0.45612668743509865, "grad_norm": 1.8580654859542847, "learning_rate": 4.553688602400792e-06, "loss": 0.13331571221351624, "step": 1757 }, { "epoch": 0.45638629283489096, "grad_norm": 2.976173162460327, "learning_rate": 4.5531229109839244e-06, "loss": 0.15506546199321747, "step": 1758 }, { "epoch": 0.45664589823468327, "grad_norm": 1.211253046989441, "learning_rate": 4.552556900816212e-06, "loss": 0.1188487559556961, "step": 1759 }, { "epoch": 0.4569055036344756, "grad_norm": 1.8078867197036743, "learning_rate": 4.551990571997713e-06, "loss": 0.11933251470327377, "step": 1760 }, { "epoch": 0.45716510903426794, "grad_norm": 2.30973219871521, "learning_rate": 4.551423924628541e-06, "loss": 0.11851023882627487, "step": 1761 }, { "epoch": 0.45742471443406024, "grad_norm": 1.152445912361145, "learning_rate": 4.550856958808867e-06, "loss": 0.13867133855819702, "step": 1762 }, { "epoch": 0.45768431983385255, "grad_norm": 1.3883640766143799, "learning_rate": 4.550289674638915e-06, "loss": 0.15151527523994446, "step": 1763 }, { "epoch": 0.45794392523364486, "grad_norm": 1.447243094444275, "learning_rate": 4.549722072218971e-06, "loss": 0.13416612148284912, "step": 1764 }, { "epoch": 0.45820353063343716, "grad_norm": 2.8802692890167236, "learning_rate": 4.5491541516493725e-06, "loss": 0.12281207740306854, "step": 1765 }, { "epoch": 0.45846313603322947, "grad_norm": 1.3612003326416016, "learning_rate": 4.548585913030515e-06, "loss": 0.12391230463981628, "step": 1766 }, { "epoch": 0.45872274143302183, "grad_norm": 1.740522861480713, "learning_rate": 4.548017356462851e-06, "loss": 0.13488753139972687, "step": 1767 }, { "epoch": 0.45898234683281414, "grad_norm": 2.0142173767089844, "learning_rate": 4.547448482046887e-06, "loss": 0.1718573272228241, "step": 1768 }, { "epoch": 0.45924195223260644, "grad_norm": 1.7480159997940063, "learning_rate": 4.546879289883187e-06, "loss": 0.15273526310920715, "step": 1769 }, { "epoch": 0.45950155763239875, "grad_norm": 1.9966280460357666, "learning_rate": 4.5463097800723725e-06, "loss": 0.15012520551681519, "step": 1770 }, { "epoch": 0.45976116303219106, "grad_norm": 2.1895687580108643, "learning_rate": 4.545739952715118e-06, "loss": 0.12425090372562408, "step": 1771 }, { "epoch": 0.46002076843198336, "grad_norm": 0.9623545408248901, "learning_rate": 4.5451698079121575e-06, "loss": 0.13137242197990417, "step": 1772 }, { "epoch": 0.4602803738317757, "grad_norm": 2.537194013595581, "learning_rate": 4.544599345764279e-06, "loss": 0.13006509840488434, "step": 1773 }, { "epoch": 0.46053997923156803, "grad_norm": 2.5368075370788574, "learning_rate": 4.544028566372326e-06, "loss": 0.13507121801376343, "step": 1774 }, { "epoch": 0.46079958463136034, "grad_norm": 2.3170907497406006, "learning_rate": 4.543457469837202e-06, "loss": 0.13702505826950073, "step": 1775 }, { "epoch": 0.46105919003115264, "grad_norm": 2.306401252746582, "learning_rate": 4.542886056259861e-06, "loss": 0.11753755807876587, "step": 1776 }, { "epoch": 0.46131879543094495, "grad_norm": 1.9869096279144287, "learning_rate": 4.542314325741316e-06, "loss": 0.1272146999835968, "step": 1777 }, { "epoch": 0.46157840083073726, "grad_norm": 1.079573631286621, "learning_rate": 4.541742278382638e-06, "loss": 0.13747720420360565, "step": 1778 }, { "epoch": 0.4618380062305296, "grad_norm": 1.233639121055603, "learning_rate": 4.541169914284951e-06, "loss": 0.12878963351249695, "step": 1779 }, { "epoch": 0.4620976116303219, "grad_norm": 2.1224489212036133, "learning_rate": 4.5405972335494345e-06, "loss": 0.12854453921318054, "step": 1780 }, { "epoch": 0.46235721703011423, "grad_norm": 1.443693995475769, "learning_rate": 4.540024236277327e-06, "loss": 0.13573242723941803, "step": 1781 }, { "epoch": 0.46261682242990654, "grad_norm": 1.4073822498321533, "learning_rate": 4.539450922569921e-06, "loss": 0.15301945805549622, "step": 1782 }, { "epoch": 0.46287642782969884, "grad_norm": 1.7889728546142578, "learning_rate": 4.538877292528564e-06, "loss": 0.1384231001138687, "step": 1783 }, { "epoch": 0.46313603322949115, "grad_norm": 1.5508023500442505, "learning_rate": 4.538303346254662e-06, "loss": 0.1151675283908844, "step": 1784 }, { "epoch": 0.4633956386292835, "grad_norm": 1.5596325397491455, "learning_rate": 4.537729083849675e-06, "loss": 0.13685886561870575, "step": 1785 }, { "epoch": 0.4636552440290758, "grad_norm": 1.3101582527160645, "learning_rate": 4.53715450541512e-06, "loss": 0.12971529364585876, "step": 1786 }, { "epoch": 0.4639148494288681, "grad_norm": 1.046054482460022, "learning_rate": 4.5365796110525674e-06, "loss": 0.12674330174922943, "step": 1787 }, { "epoch": 0.46417445482866043, "grad_norm": 3.2546427249908447, "learning_rate": 4.5360044008636485e-06, "loss": 0.14296074211597443, "step": 1788 }, { "epoch": 0.46443406022845274, "grad_norm": 1.5914945602416992, "learning_rate": 4.535428874950045e-06, "loss": 0.1495247185230255, "step": 1789 }, { "epoch": 0.46469366562824505, "grad_norm": 1.2551467418670654, "learning_rate": 4.5348530334134975e-06, "loss": 0.13047176599502563, "step": 1790 }, { "epoch": 0.4649532710280374, "grad_norm": 1.7899101972579956, "learning_rate": 4.534276876355801e-06, "loss": 0.11948814243078232, "step": 1791 }, { "epoch": 0.4652128764278297, "grad_norm": 2.058189630508423, "learning_rate": 4.533700403878806e-06, "loss": 0.1594281643629074, "step": 1792 }, { "epoch": 0.465472481827622, "grad_norm": 1.8860793113708496, "learning_rate": 4.533123616084423e-06, "loss": 0.12206889688968658, "step": 1793 }, { "epoch": 0.4657320872274143, "grad_norm": 1.8544751405715942, "learning_rate": 4.5325465130746125e-06, "loss": 0.1470472514629364, "step": 1794 }, { "epoch": 0.46599169262720663, "grad_norm": 1.646803379058838, "learning_rate": 4.531969094951393e-06, "loss": 0.14958181977272034, "step": 1795 }, { "epoch": 0.46625129802699894, "grad_norm": 1.0237253904342651, "learning_rate": 4.5313913618168394e-06, "loss": 0.11670225858688354, "step": 1796 }, { "epoch": 0.4665109034267913, "grad_norm": 1.0502443313598633, "learning_rate": 4.530813313773081e-06, "loss": 0.12230505794286728, "step": 1797 }, { "epoch": 0.4667705088265836, "grad_norm": 1.7376047372817993, "learning_rate": 4.530234950922304e-06, "loss": 0.11482851207256317, "step": 1798 }, { "epoch": 0.4670301142263759, "grad_norm": 4.041429042816162, "learning_rate": 4.52965627336675e-06, "loss": 0.14313815534114838, "step": 1799 }, { "epoch": 0.4672897196261682, "grad_norm": 1.1964236497879028, "learning_rate": 4.529077281208716e-06, "loss": 0.11753951758146286, "step": 1800 }, { "epoch": 0.4675493250259605, "grad_norm": 1.1067167520523071, "learning_rate": 4.528497974550553e-06, "loss": 0.11983759701251984, "step": 1801 }, { "epoch": 0.46780893042575283, "grad_norm": 0.9783208966255188, "learning_rate": 4.527918353494672e-06, "loss": 0.11387097835540771, "step": 1802 }, { "epoch": 0.4680685358255452, "grad_norm": 1.1790072917938232, "learning_rate": 4.527338418143533e-06, "loss": 0.13721494376659393, "step": 1803 }, { "epoch": 0.4683281412253375, "grad_norm": 1.8877414464950562, "learning_rate": 4.526758168599658e-06, "loss": 0.11231281608343124, "step": 1804 }, { "epoch": 0.4685877466251298, "grad_norm": 2.5374553203582764, "learning_rate": 4.526177604965622e-06, "loss": 0.13095182180404663, "step": 1805 }, { "epoch": 0.4688473520249221, "grad_norm": 0.7665579319000244, "learning_rate": 4.525596727344054e-06, "loss": 0.12212171405553818, "step": 1806 }, { "epoch": 0.4691069574247144, "grad_norm": 2.1715543270111084, "learning_rate": 4.52501553583764e-06, "loss": 0.13110694289207458, "step": 1807 }, { "epoch": 0.46936656282450673, "grad_norm": 1.500449776649475, "learning_rate": 4.524434030549122e-06, "loss": 0.12129367887973785, "step": 1808 }, { "epoch": 0.4696261682242991, "grad_norm": 3.4054698944091797, "learning_rate": 4.523852211581297e-06, "loss": 0.14549849927425385, "step": 1809 }, { "epoch": 0.4698857736240914, "grad_norm": 1.3730676174163818, "learning_rate": 4.523270079037016e-06, "loss": 0.10873763263225555, "step": 1810 }, { "epoch": 0.4701453790238837, "grad_norm": 1.6430450677871704, "learning_rate": 4.522687633019189e-06, "loss": 0.11329399049282074, "step": 1811 }, { "epoch": 0.470404984423676, "grad_norm": 2.4003140926361084, "learning_rate": 4.522104873630776e-06, "loss": 0.130730539560318, "step": 1812 }, { "epoch": 0.4706645898234683, "grad_norm": 1.6646502017974854, "learning_rate": 4.521521800974798e-06, "loss": 0.15189805626869202, "step": 1813 }, { "epoch": 0.4709241952232606, "grad_norm": 2.3542447090148926, "learning_rate": 4.520938415154328e-06, "loss": 0.1534491777420044, "step": 1814 }, { "epoch": 0.471183800623053, "grad_norm": 0.9958761930465698, "learning_rate": 4.520354716272495e-06, "loss": 0.11925467848777771, "step": 1815 }, { "epoch": 0.4714434060228453, "grad_norm": 1.5300912857055664, "learning_rate": 4.519770704432484e-06, "loss": 0.13537171483039856, "step": 1816 }, { "epoch": 0.4717030114226376, "grad_norm": 3.569852590560913, "learning_rate": 4.519186379737535e-06, "loss": 0.1374935507774353, "step": 1817 }, { "epoch": 0.4719626168224299, "grad_norm": 1.4127411842346191, "learning_rate": 4.518601742290943e-06, "loss": 0.1418108195066452, "step": 1818 }, { "epoch": 0.4722222222222222, "grad_norm": 1.8163865804672241, "learning_rate": 4.518016792196059e-06, "loss": 0.15466485917568207, "step": 1819 }, { "epoch": 0.4724818276220145, "grad_norm": 1.6772536039352417, "learning_rate": 4.5174315295562885e-06, "loss": 0.15159955620765686, "step": 1820 }, { "epoch": 0.4727414330218069, "grad_norm": 1.5064277648925781, "learning_rate": 4.516845954475091e-06, "loss": 0.12283230572938919, "step": 1821 }, { "epoch": 0.4730010384215992, "grad_norm": 2.475276231765747, "learning_rate": 4.516260067055985e-06, "loss": 0.11998173594474792, "step": 1822 }, { "epoch": 0.4732606438213915, "grad_norm": 1.59848153591156, "learning_rate": 4.515673867402541e-06, "loss": 0.12648937106132507, "step": 1823 }, { "epoch": 0.4735202492211838, "grad_norm": 1.7140157222747803, "learning_rate": 4.515087355618387e-06, "loss": 0.15328319370746613, "step": 1824 }, { "epoch": 0.4737798546209761, "grad_norm": 0.9714024066925049, "learning_rate": 4.514500531807201e-06, "loss": 0.11758483946323395, "step": 1825 }, { "epoch": 0.4740394600207684, "grad_norm": 3.3638346195220947, "learning_rate": 4.513913396072725e-06, "loss": 0.1391327977180481, "step": 1826 }, { "epoch": 0.4742990654205608, "grad_norm": 2.3567140102386475, "learning_rate": 4.513325948518748e-06, "loss": 0.11844972521066666, "step": 1827 }, { "epoch": 0.4745586708203531, "grad_norm": 2.842729091644287, "learning_rate": 4.512738189249119e-06, "loss": 0.1370527446269989, "step": 1828 }, { "epoch": 0.4748182762201454, "grad_norm": 2.0010266304016113, "learning_rate": 4.512150118367739e-06, "loss": 0.1414237916469574, "step": 1829 }, { "epoch": 0.4750778816199377, "grad_norm": 1.8380976915359497, "learning_rate": 4.511561735978564e-06, "loss": 0.1344795823097229, "step": 1830 }, { "epoch": 0.47533748701973, "grad_norm": 1.5375643968582153, "learning_rate": 4.510973042185611e-06, "loss": 0.11952155083417892, "step": 1831 }, { "epoch": 0.4755970924195223, "grad_norm": 1.8081268072128296, "learning_rate": 4.5103840370929446e-06, "loss": 0.13495565950870514, "step": 1832 }, { "epoch": 0.47585669781931467, "grad_norm": 1.2899852991104126, "learning_rate": 4.509794720804688e-06, "loss": 0.127648264169693, "step": 1833 }, { "epoch": 0.476116303219107, "grad_norm": 0.7269832491874695, "learning_rate": 4.509205093425018e-06, "loss": 0.10804609954357147, "step": 1834 }, { "epoch": 0.4763759086188993, "grad_norm": 1.3218973875045776, "learning_rate": 4.508615155058168e-06, "loss": 0.13143400847911835, "step": 1835 }, { "epoch": 0.4766355140186916, "grad_norm": 1.683500051498413, "learning_rate": 4.508024905808426e-06, "loss": 0.13148194551467896, "step": 1836 }, { "epoch": 0.4768951194184839, "grad_norm": 0.7550497055053711, "learning_rate": 4.507434345780133e-06, "loss": 0.13219115138053894, "step": 1837 }, { "epoch": 0.4771547248182762, "grad_norm": 2.3475863933563232, "learning_rate": 4.506843475077689e-06, "loss": 0.14207537472248077, "step": 1838 }, { "epoch": 0.47741433021806856, "grad_norm": 1.5322939157485962, "learning_rate": 4.5062522938055444e-06, "loss": 0.15556687116622925, "step": 1839 }, { "epoch": 0.47767393561786087, "grad_norm": 1.4404213428497314, "learning_rate": 4.505660802068207e-06, "loss": 0.13098855316638947, "step": 1840 }, { "epoch": 0.4779335410176532, "grad_norm": 2.0558457374572754, "learning_rate": 4.50506899997024e-06, "loss": 0.11380185186862946, "step": 1841 }, { "epoch": 0.4781931464174455, "grad_norm": 2.2306289672851562, "learning_rate": 4.504476887616258e-06, "loss": 0.1383584439754486, "step": 1842 }, { "epoch": 0.4784527518172378, "grad_norm": 2.8078360557556152, "learning_rate": 4.503884465110935e-06, "loss": 0.156161367893219, "step": 1843 }, { "epoch": 0.4787123572170301, "grad_norm": 1.7781113386154175, "learning_rate": 4.503291732558998e-06, "loss": 0.13726924359798431, "step": 1844 }, { "epoch": 0.47897196261682246, "grad_norm": 4.1204376220703125, "learning_rate": 4.502698690065227e-06, "loss": 0.12751100957393646, "step": 1845 }, { "epoch": 0.47923156801661476, "grad_norm": 1.475135087966919, "learning_rate": 4.5021053377344596e-06, "loss": 0.12904295325279236, "step": 1846 }, { "epoch": 0.47949117341640707, "grad_norm": 1.8506187200546265, "learning_rate": 4.501511675671586e-06, "loss": 0.12135139107704163, "step": 1847 }, { "epoch": 0.4797507788161994, "grad_norm": 1.971703290939331, "learning_rate": 4.500917703981553e-06, "loss": 0.15856723487377167, "step": 1848 }, { "epoch": 0.4800103842159917, "grad_norm": 1.0850712060928345, "learning_rate": 4.5003234227693596e-06, "loss": 0.12597936391830444, "step": 1849 }, { "epoch": 0.480269989615784, "grad_norm": 1.5804654359817505, "learning_rate": 4.499728832140063e-06, "loss": 0.13922619819641113, "step": 1850 }, { "epoch": 0.48052959501557635, "grad_norm": 2.205052375793457, "learning_rate": 4.499133932198772e-06, "loss": 0.14374974370002747, "step": 1851 }, { "epoch": 0.48078920041536866, "grad_norm": 1.9782001972198486, "learning_rate": 4.498538723050651e-06, "loss": 0.1472829431295395, "step": 1852 }, { "epoch": 0.48104880581516096, "grad_norm": 3.488112211227417, "learning_rate": 4.4979432048009215e-06, "loss": 0.12657994031906128, "step": 1853 }, { "epoch": 0.48130841121495327, "grad_norm": 2.0685436725616455, "learning_rate": 4.497347377554855e-06, "loss": 0.12784500420093536, "step": 1854 }, { "epoch": 0.4815680166147456, "grad_norm": 2.506946086883545, "learning_rate": 4.4967512414177816e-06, "loss": 0.11215748637914658, "step": 1855 }, { "epoch": 0.4818276220145379, "grad_norm": 1.2836123704910278, "learning_rate": 4.496154796495083e-06, "loss": 0.12329847365617752, "step": 1856 }, { "epoch": 0.48208722741433024, "grad_norm": 1.5120419263839722, "learning_rate": 4.495558042892199e-06, "loss": 0.13524174690246582, "step": 1857 }, { "epoch": 0.48234683281412255, "grad_norm": 1.1256235837936401, "learning_rate": 4.49496098071462e-06, "loss": 0.12381589412689209, "step": 1858 }, { "epoch": 0.48260643821391486, "grad_norm": 1.6043531894683838, "learning_rate": 4.494363610067895e-06, "loss": 0.1388191431760788, "step": 1859 }, { "epoch": 0.48286604361370716, "grad_norm": 2.823514938354492, "learning_rate": 4.493765931057624e-06, "loss": 0.10931394249200821, "step": 1860 }, { "epoch": 0.48312564901349947, "grad_norm": 3.4384384155273438, "learning_rate": 4.493167943789463e-06, "loss": 0.1385069489479065, "step": 1861 }, { "epoch": 0.4833852544132918, "grad_norm": 1.128042459487915, "learning_rate": 4.492569648369124e-06, "loss": 0.11931697279214859, "step": 1862 }, { "epoch": 0.48364485981308414, "grad_norm": 2.9475293159484863, "learning_rate": 4.4919710449023695e-06, "loss": 0.12546682357788086, "step": 1863 }, { "epoch": 0.48390446521287644, "grad_norm": 3.437965154647827, "learning_rate": 4.491372133495021e-06, "loss": 0.15010903775691986, "step": 1864 }, { "epoch": 0.48416407061266875, "grad_norm": 3.342881441116333, "learning_rate": 4.49077291425295e-06, "loss": 0.15099437534809113, "step": 1865 }, { "epoch": 0.48442367601246106, "grad_norm": 1.9678329229354858, "learning_rate": 4.490173387282087e-06, "loss": 0.1250012218952179, "step": 1866 }, { "epoch": 0.48468328141225336, "grad_norm": 4.0903143882751465, "learning_rate": 4.489573552688414e-06, "loss": 0.11142571270465851, "step": 1867 }, { "epoch": 0.48494288681204567, "grad_norm": 3.503371477127075, "learning_rate": 4.488973410577968e-06, "loss": 0.13305675983428955, "step": 1868 }, { "epoch": 0.48520249221183803, "grad_norm": 2.6230080127716064, "learning_rate": 4.48837296105684e-06, "loss": 0.12051653861999512, "step": 1869 }, { "epoch": 0.48546209761163034, "grad_norm": 2.0339109897613525, "learning_rate": 4.4877722042311765e-06, "loss": 0.1416255682706833, "step": 1870 }, { "epoch": 0.48572170301142265, "grad_norm": 3.8004074096679688, "learning_rate": 4.487171140207176e-06, "loss": 0.13906581699848175, "step": 1871 }, { "epoch": 0.48598130841121495, "grad_norm": 2.6885178089141846, "learning_rate": 4.4865697690910945e-06, "loss": 0.13802775740623474, "step": 1872 }, { "epoch": 0.48624091381100726, "grad_norm": 2.971226215362549, "learning_rate": 4.48596809098924e-06, "loss": 0.14561234414577484, "step": 1873 }, { "epoch": 0.48650051921079956, "grad_norm": 1.1411761045455933, "learning_rate": 4.485366106007975e-06, "loss": 0.1329157054424286, "step": 1874 }, { "epoch": 0.4867601246105919, "grad_norm": 2.3378584384918213, "learning_rate": 4.484763814253717e-06, "loss": 0.1470731496810913, "step": 1875 }, { "epoch": 0.48701973001038423, "grad_norm": 2.170192003250122, "learning_rate": 4.484161215832938e-06, "loss": 0.11777830123901367, "step": 1876 }, { "epoch": 0.48727933541017654, "grad_norm": 4.5358710289001465, "learning_rate": 4.483558310852163e-06, "loss": 0.14980989694595337, "step": 1877 }, { "epoch": 0.48753894080996885, "grad_norm": 2.6674485206604004, "learning_rate": 4.482955099417971e-06, "loss": 0.10626136511564255, "step": 1878 }, { "epoch": 0.48779854620976115, "grad_norm": 1.6428887844085693, "learning_rate": 4.482351581636998e-06, "loss": 0.09366072714328766, "step": 1879 }, { "epoch": 0.48805815160955346, "grad_norm": 2.842782735824585, "learning_rate": 4.48174775761593e-06, "loss": 0.1247568130493164, "step": 1880 }, { "epoch": 0.4883177570093458, "grad_norm": 2.1200478076934814, "learning_rate": 4.48114362746151e-06, "loss": 0.13611744344234467, "step": 1881 }, { "epoch": 0.4885773624091381, "grad_norm": 3.306230306625366, "learning_rate": 4.4805391912805366e-06, "loss": 0.13161036372184753, "step": 1882 }, { "epoch": 0.48883696780893043, "grad_norm": 1.3173069953918457, "learning_rate": 4.4799344491798564e-06, "loss": 0.1263260394334793, "step": 1883 }, { "epoch": 0.48909657320872274, "grad_norm": 3.8388707637786865, "learning_rate": 4.4793294012663765e-06, "loss": 0.1685476303100586, "step": 1884 }, { "epoch": 0.48935617860851505, "grad_norm": 1.6008497476577759, "learning_rate": 4.478724047647054e-06, "loss": 0.1367577314376831, "step": 1885 }, { "epoch": 0.48961578400830735, "grad_norm": 3.310904026031494, "learning_rate": 4.478118388428904e-06, "loss": 0.12493300437927246, "step": 1886 }, { "epoch": 0.4898753894080997, "grad_norm": 0.9744511246681213, "learning_rate": 4.47751242371899e-06, "loss": 0.12402524054050446, "step": 1887 }, { "epoch": 0.490134994807892, "grad_norm": 1.0096354484558105, "learning_rate": 4.476906153624435e-06, "loss": 0.104520283639431, "step": 1888 }, { "epoch": 0.49039460020768433, "grad_norm": 1.9600595235824585, "learning_rate": 4.476299578252414e-06, "loss": 0.1355391889810562, "step": 1889 }, { "epoch": 0.49065420560747663, "grad_norm": 3.007842779159546, "learning_rate": 4.475692697710153e-06, "loss": 0.1331196278333664, "step": 1890 }, { "epoch": 0.49091381100726894, "grad_norm": 2.8215105533599854, "learning_rate": 4.475085512104937e-06, "loss": 0.14413517713546753, "step": 1891 }, { "epoch": 0.49117341640706125, "grad_norm": 3.8246161937713623, "learning_rate": 4.474478021544102e-06, "loss": 0.14548395574092865, "step": 1892 }, { "epoch": 0.4914330218068536, "grad_norm": 1.6433415412902832, "learning_rate": 4.473870226135036e-06, "loss": 0.14212098717689514, "step": 1893 }, { "epoch": 0.4916926272066459, "grad_norm": 1.3417394161224365, "learning_rate": 4.473262125985188e-06, "loss": 0.13046881556510925, "step": 1894 }, { "epoch": 0.4919522326064382, "grad_norm": 1.3143343925476074, "learning_rate": 4.472653721202052e-06, "loss": 0.12240210175514221, "step": 1895 }, { "epoch": 0.49221183800623053, "grad_norm": 2.157235860824585, "learning_rate": 4.472045011893182e-06, "loss": 0.16443225741386414, "step": 1896 }, { "epoch": 0.49247144340602284, "grad_norm": 2.9117181301116943, "learning_rate": 4.471435998166185e-06, "loss": 0.1261684000492096, "step": 1897 }, { "epoch": 0.49273104880581514, "grad_norm": 2.1276462078094482, "learning_rate": 4.470826680128717e-06, "loss": 0.12685278058052063, "step": 1898 }, { "epoch": 0.4929906542056075, "grad_norm": 2.3913934230804443, "learning_rate": 4.470217057888495e-06, "loss": 0.13423249125480652, "step": 1899 }, { "epoch": 0.4932502596053998, "grad_norm": 1.01007080078125, "learning_rate": 4.469607131553284e-06, "loss": 0.1294236183166504, "step": 1900 }, { "epoch": 0.4935098650051921, "grad_norm": 1.9595874547958374, "learning_rate": 4.468996901230907e-06, "loss": 0.10921456664800644, "step": 1901 }, { "epoch": 0.4937694704049844, "grad_norm": 3.9520766735076904, "learning_rate": 4.468386367029236e-06, "loss": 0.15937373042106628, "step": 1902 }, { "epoch": 0.49402907580477673, "grad_norm": 3.159067153930664, "learning_rate": 4.467775529056203e-06, "loss": 0.12233296036720276, "step": 1903 }, { "epoch": 0.49428868120456904, "grad_norm": 2.5450944900512695, "learning_rate": 4.467164387419787e-06, "loss": 0.09897300601005554, "step": 1904 }, { "epoch": 0.4945482866043614, "grad_norm": 1.2151128053665161, "learning_rate": 4.466552942228026e-06, "loss": 0.12075388431549072, "step": 1905 }, { "epoch": 0.4948078920041537, "grad_norm": 3.4943625926971436, "learning_rate": 4.465941193589008e-06, "loss": 0.10943889617919922, "step": 1906 }, { "epoch": 0.495067497403946, "grad_norm": 3.3338849544525146, "learning_rate": 4.465329141610877e-06, "loss": 0.13217684626579285, "step": 1907 }, { "epoch": 0.4953271028037383, "grad_norm": 4.419769763946533, "learning_rate": 4.46471678640183e-06, "loss": 0.1302943378686905, "step": 1908 }, { "epoch": 0.4955867082035306, "grad_norm": 3.6854662895202637, "learning_rate": 4.4641041280701164e-06, "loss": 0.12965121865272522, "step": 1909 }, { "epoch": 0.49584631360332293, "grad_norm": 2.3923821449279785, "learning_rate": 4.463491166724041e-06, "loss": 0.13455581665039062, "step": 1910 }, { "epoch": 0.4961059190031153, "grad_norm": 3.961033582687378, "learning_rate": 4.462877902471961e-06, "loss": 0.11737670004367828, "step": 1911 }, { "epoch": 0.4963655244029076, "grad_norm": 5.480629920959473, "learning_rate": 4.462264335422288e-06, "loss": 0.1257660686969757, "step": 1912 }, { "epoch": 0.4966251298026999, "grad_norm": 3.800637722015381, "learning_rate": 4.4616504656834855e-06, "loss": 0.11350401490926743, "step": 1913 }, { "epoch": 0.4968847352024922, "grad_norm": 1.2804008722305298, "learning_rate": 4.461036293364073e-06, "loss": 0.11705578863620758, "step": 1914 }, { "epoch": 0.4971443406022845, "grad_norm": 2.7816336154937744, "learning_rate": 4.46042181857262e-06, "loss": 0.10314309597015381, "step": 1915 }, { "epoch": 0.4974039460020768, "grad_norm": 4.254234790802002, "learning_rate": 4.459807041417755e-06, "loss": 0.13170133531093597, "step": 1916 }, { "epoch": 0.4976635514018692, "grad_norm": 3.7995729446411133, "learning_rate": 4.4591919620081526e-06, "loss": 0.13215228915214539, "step": 1917 }, { "epoch": 0.4979231568016615, "grad_norm": 2.554246664047241, "learning_rate": 4.458576580452547e-06, "loss": 0.1189437210559845, "step": 1918 }, { "epoch": 0.4981827622014538, "grad_norm": 2.17252516746521, "learning_rate": 4.457960896859724e-06, "loss": 0.11617685854434967, "step": 1919 }, { "epoch": 0.4984423676012461, "grad_norm": 2.067354440689087, "learning_rate": 4.4573449113385214e-06, "loss": 0.13466501235961914, "step": 1920 }, { "epoch": 0.4987019730010384, "grad_norm": 4.314873218536377, "learning_rate": 4.456728623997831e-06, "loss": 0.1468401849269867, "step": 1921 }, { "epoch": 0.4989615784008307, "grad_norm": 1.5115562677383423, "learning_rate": 4.456112034946599e-06, "loss": 0.12697643041610718, "step": 1922 }, { "epoch": 0.4992211838006231, "grad_norm": 3.218855857849121, "learning_rate": 4.455495144293824e-06, "loss": 0.1326196789741516, "step": 1923 }, { "epoch": 0.4994807892004154, "grad_norm": 2.6999876499176025, "learning_rate": 4.454877952148558e-06, "loss": 0.15216974914073944, "step": 1924 }, { "epoch": 0.4997403946002077, "grad_norm": 1.5764477252960205, "learning_rate": 4.454260458619907e-06, "loss": 0.13664758205413818, "step": 1925 }, { "epoch": 0.5, "grad_norm": 2.37035870552063, "learning_rate": 4.45364266381703e-06, "loss": 0.11732691526412964, "step": 1926 }, { "epoch": 0.5002596053997923, "grad_norm": 2.1298909187316895, "learning_rate": 4.453024567849136e-06, "loss": 0.12035813927650452, "step": 1927 }, { "epoch": 0.5005192107995846, "grad_norm": 1.83112370967865, "learning_rate": 4.452406170825494e-06, "loss": 0.13722676038742065, "step": 1928 }, { "epoch": 0.5007788161993769, "grad_norm": 3.1022207736968994, "learning_rate": 4.451787472855421e-06, "loss": 0.11727265268564224, "step": 1929 }, { "epoch": 0.5010384215991692, "grad_norm": 2.221402645111084, "learning_rate": 4.451168474048288e-06, "loss": 0.14655019342899323, "step": 1930 }, { "epoch": 0.5012980269989615, "grad_norm": 2.011758804321289, "learning_rate": 4.450549174513521e-06, "loss": 0.15005192160606384, "step": 1931 }, { "epoch": 0.5015576323987538, "grad_norm": 3.264101982116699, "learning_rate": 4.449929574360597e-06, "loss": 0.17065021395683289, "step": 1932 }, { "epoch": 0.5018172377985463, "grad_norm": 3.5309336185455322, "learning_rate": 4.4493096736990476e-06, "loss": 0.11612365394830704, "step": 1933 }, { "epoch": 0.5020768431983386, "grad_norm": 2.778428077697754, "learning_rate": 4.448689472638457e-06, "loss": 0.12449654191732407, "step": 1934 }, { "epoch": 0.5023364485981309, "grad_norm": 1.557695984840393, "learning_rate": 4.448068971288462e-06, "loss": 0.1302122175693512, "step": 1935 }, { "epoch": 0.5025960539979232, "grad_norm": 1.7975997924804688, "learning_rate": 4.447448169758754e-06, "loss": 0.10834506899118423, "step": 1936 }, { "epoch": 0.5028556593977155, "grad_norm": 1.7614353895187378, "learning_rate": 4.446827068159076e-06, "loss": 0.11664547026157379, "step": 1937 }, { "epoch": 0.5031152647975078, "grad_norm": 1.7713512182235718, "learning_rate": 4.446205666599224e-06, "loss": 0.13940495252609253, "step": 1938 }, { "epoch": 0.5033748701973001, "grad_norm": 1.5844104290008545, "learning_rate": 4.445583965189049e-06, "loss": 0.13762161135673523, "step": 1939 }, { "epoch": 0.5036344755970924, "grad_norm": 1.0340474843978882, "learning_rate": 4.444961964038452e-06, "loss": 0.11518019437789917, "step": 1940 }, { "epoch": 0.5038940809968847, "grad_norm": 2.6947176456451416, "learning_rate": 4.444339663257389e-06, "loss": 0.13216650485992432, "step": 1941 }, { "epoch": 0.504153686396677, "grad_norm": 2.029714822769165, "learning_rate": 4.443717062955869e-06, "loss": 0.15063157677650452, "step": 1942 }, { "epoch": 0.5044132917964693, "grad_norm": 2.320695400238037, "learning_rate": 4.443094163243954e-06, "loss": 0.1156151071190834, "step": 1943 }, { "epoch": 0.5046728971962616, "grad_norm": 2.6626124382019043, "learning_rate": 4.442470964231758e-06, "loss": 0.12984055280685425, "step": 1944 }, { "epoch": 0.504932502596054, "grad_norm": 1.2014793157577515, "learning_rate": 4.441847466029447e-06, "loss": 0.1153816431760788, "step": 1945 }, { "epoch": 0.5051921079958464, "grad_norm": 2.8761181831359863, "learning_rate": 4.441223668747241e-06, "loss": 0.11928251385688782, "step": 1946 }, { "epoch": 0.5054517133956387, "grad_norm": 0.7820672392845154, "learning_rate": 4.440599572495416e-06, "loss": 0.11976119875907898, "step": 1947 }, { "epoch": 0.505711318795431, "grad_norm": 0.6482967138290405, "learning_rate": 4.439975177384296e-06, "loss": 0.11287827789783478, "step": 1948 }, { "epoch": 0.5059709241952233, "grad_norm": 1.2984399795532227, "learning_rate": 4.439350483524259e-06, "loss": 0.11193607747554779, "step": 1949 }, { "epoch": 0.5062305295950156, "grad_norm": 1.2527164220809937, "learning_rate": 4.4387254910257386e-06, "loss": 0.10791832208633423, "step": 1950 }, { "epoch": 0.5064901349948079, "grad_norm": 2.2145750522613525, "learning_rate": 4.438100199999217e-06, "loss": 0.099252849817276, "step": 1951 }, { "epoch": 0.5067497403946002, "grad_norm": 2.540639638900757, "learning_rate": 4.437474610555234e-06, "loss": 0.13323643803596497, "step": 1952 }, { "epoch": 0.5070093457943925, "grad_norm": 1.8335678577423096, "learning_rate": 4.436848722804376e-06, "loss": 0.125696063041687, "step": 1953 }, { "epoch": 0.5072689511941848, "grad_norm": 3.217804193496704, "learning_rate": 4.436222536857289e-06, "loss": 0.12012387812137604, "step": 1954 }, { "epoch": 0.5075285565939771, "grad_norm": 2.706469774246216, "learning_rate": 4.435596052824666e-06, "loss": 0.10806752741336823, "step": 1955 }, { "epoch": 0.5077881619937694, "grad_norm": 2.020537853240967, "learning_rate": 4.434969270817257e-06, "loss": 0.12082043290138245, "step": 1956 }, { "epoch": 0.5080477673935618, "grad_norm": 2.207646369934082, "learning_rate": 4.4343421909458605e-06, "loss": 0.13762953877449036, "step": 1957 }, { "epoch": 0.5083073727933541, "grad_norm": 3.0833613872528076, "learning_rate": 4.433714813321331e-06, "loss": 0.1563483476638794, "step": 1958 }, { "epoch": 0.5085669781931464, "grad_norm": 2.3175477981567383, "learning_rate": 4.433087138054575e-06, "loss": 0.10970904678106308, "step": 1959 }, { "epoch": 0.5088265835929388, "grad_norm": 2.043138265609741, "learning_rate": 4.4324591652565505e-06, "loss": 0.13325972855091095, "step": 1960 }, { "epoch": 0.5090861889927311, "grad_norm": 2.416938543319702, "learning_rate": 4.4318308950382685e-06, "loss": 0.13365259766578674, "step": 1961 }, { "epoch": 0.5093457943925234, "grad_norm": 1.0642400979995728, "learning_rate": 4.431202327510793e-06, "loss": 0.12018389999866486, "step": 1962 }, { "epoch": 0.5096053997923157, "grad_norm": 2.4185874462127686, "learning_rate": 4.43057346278524e-06, "loss": 0.11299005895853043, "step": 1963 }, { "epoch": 0.509865005192108, "grad_norm": 1.9935023784637451, "learning_rate": 4.42994430097278e-06, "loss": 0.1316833198070526, "step": 1964 }, { "epoch": 0.5101246105919003, "grad_norm": 2.6950786113739014, "learning_rate": 4.429314842184632e-06, "loss": 0.0999017208814621, "step": 1965 }, { "epoch": 0.5103842159916926, "grad_norm": 1.831247091293335, "learning_rate": 4.428685086532072e-06, "loss": 0.14226099848747253, "step": 1966 }, { "epoch": 0.5106438213914849, "grad_norm": 2.01126766204834, "learning_rate": 4.428055034126426e-06, "loss": 0.12295082211494446, "step": 1967 }, { "epoch": 0.5109034267912772, "grad_norm": 1.8341865539550781, "learning_rate": 4.427424685079071e-06, "loss": 0.11978497356176376, "step": 1968 }, { "epoch": 0.5111630321910696, "grad_norm": 1.769567608833313, "learning_rate": 4.42679403950144e-06, "loss": 0.1030375212430954, "step": 1969 }, { "epoch": 0.5114226375908619, "grad_norm": 1.9966286420822144, "learning_rate": 4.426163097505017e-06, "loss": 0.12421675026416779, "step": 1970 }, { "epoch": 0.5116822429906542, "grad_norm": 1.3665857315063477, "learning_rate": 4.425531859201337e-06, "loss": 0.12965890765190125, "step": 1971 }, { "epoch": 0.5119418483904465, "grad_norm": 0.9989604949951172, "learning_rate": 4.424900324701989e-06, "loss": 0.11011995375156403, "step": 1972 }, { "epoch": 0.5122014537902388, "grad_norm": 1.3482486009597778, "learning_rate": 4.424268494118613e-06, "loss": 0.12359024584293365, "step": 1973 }, { "epoch": 0.5124610591900312, "grad_norm": 1.6183147430419922, "learning_rate": 4.423636367562903e-06, "loss": 0.12576669454574585, "step": 1974 }, { "epoch": 0.5127206645898235, "grad_norm": 1.273245930671692, "learning_rate": 4.423003945146605e-06, "loss": 0.11192028224468231, "step": 1975 }, { "epoch": 0.5129802699896158, "grad_norm": 2.3131792545318604, "learning_rate": 4.422371226981517e-06, "loss": 0.1151144802570343, "step": 1976 }, { "epoch": 0.5132398753894081, "grad_norm": 2.384157657623291, "learning_rate": 4.421738213179487e-06, "loss": 0.13977397978305817, "step": 1977 }, { "epoch": 0.5134994807892004, "grad_norm": 1.7363415956497192, "learning_rate": 4.421104903852419e-06, "loss": 0.1381976306438446, "step": 1978 }, { "epoch": 0.5137590861889927, "grad_norm": 1.6680487394332886, "learning_rate": 4.420471299112267e-06, "loss": 0.1220884919166565, "step": 1979 }, { "epoch": 0.514018691588785, "grad_norm": 1.1457489728927612, "learning_rate": 4.419837399071038e-06, "loss": 0.10799053311347961, "step": 1980 }, { "epoch": 0.5142782969885774, "grad_norm": 1.1439363956451416, "learning_rate": 4.419203203840792e-06, "loss": 0.11146117746829987, "step": 1981 }, { "epoch": 0.5145379023883697, "grad_norm": 1.2045220136642456, "learning_rate": 4.418568713533638e-06, "loss": 0.11459086835384369, "step": 1982 }, { "epoch": 0.514797507788162, "grad_norm": 1.6444525718688965, "learning_rate": 4.4179339282617416e-06, "loss": 0.1375177949666977, "step": 1983 }, { "epoch": 0.5150571131879543, "grad_norm": 2.0190718173980713, "learning_rate": 4.417298848137317e-06, "loss": 0.09372123330831528, "step": 1984 }, { "epoch": 0.5153167185877466, "grad_norm": 1.9976204633712769, "learning_rate": 4.416663473272632e-06, "loss": 0.13410833477973938, "step": 1985 }, { "epoch": 0.5155763239875389, "grad_norm": 4.15676212310791, "learning_rate": 4.416027803780006e-06, "loss": 0.14498470723628998, "step": 1986 }, { "epoch": 0.5158359293873312, "grad_norm": 1.5574636459350586, "learning_rate": 4.415391839771812e-06, "loss": 0.13168556988239288, "step": 1987 }, { "epoch": 0.5160955347871236, "grad_norm": 1.0476816892623901, "learning_rate": 4.414755581360473e-06, "loss": 0.11701834201812744, "step": 1988 }, { "epoch": 0.5163551401869159, "grad_norm": 1.5351554155349731, "learning_rate": 4.414119028658465e-06, "loss": 0.12253548204898834, "step": 1989 }, { "epoch": 0.5166147455867082, "grad_norm": 1.722923994064331, "learning_rate": 4.4134821817783155e-06, "loss": 0.11354363709688187, "step": 1990 }, { "epoch": 0.5168743509865005, "grad_norm": 2.5070762634277344, "learning_rate": 4.4128450408326065e-06, "loss": 0.13325995206832886, "step": 1991 }, { "epoch": 0.5171339563862928, "grad_norm": 3.2576723098754883, "learning_rate": 4.412207605933967e-06, "loss": 0.12087902426719666, "step": 1992 }, { "epoch": 0.5173935617860852, "grad_norm": 2.246246814727783, "learning_rate": 4.411569877195082e-06, "loss": 0.12047795951366425, "step": 1993 }, { "epoch": 0.5176531671858775, "grad_norm": 2.559702157974243, "learning_rate": 4.410931854728689e-06, "loss": 0.12377485632896423, "step": 1994 }, { "epoch": 0.5179127725856698, "grad_norm": 4.421638488769531, "learning_rate": 4.410293538647573e-06, "loss": 0.13450735807418823, "step": 1995 }, { "epoch": 0.5181723779854621, "grad_norm": 3.061326742172241, "learning_rate": 4.409654929064575e-06, "loss": 0.11723657697439194, "step": 1996 }, { "epoch": 0.5184319833852544, "grad_norm": 1.2576550245285034, "learning_rate": 4.4090160260925874e-06, "loss": 0.1132296770811081, "step": 1997 }, { "epoch": 0.5186915887850467, "grad_norm": 3.549909830093384, "learning_rate": 4.408376829844553e-06, "loss": 0.12965938448905945, "step": 1998 }, { "epoch": 0.518951194184839, "grad_norm": 5.019473075866699, "learning_rate": 4.407737340433466e-06, "loss": 0.1509074568748474, "step": 1999 }, { "epoch": 0.5192107995846313, "grad_norm": 1.8335580825805664, "learning_rate": 4.407097557972375e-06, "loss": 0.1299402415752411, "step": 2000 }, { "epoch": 0.5194704049844237, "grad_norm": 3.791388750076294, "learning_rate": 4.406457482574379e-06, "loss": 0.12844833731651306, "step": 2001 }, { "epoch": 0.519730010384216, "grad_norm": 2.4530041217803955, "learning_rate": 4.405817114352627e-06, "loss": 0.15154282748699188, "step": 2002 }, { "epoch": 0.5199896157840083, "grad_norm": 3.2025341987609863, "learning_rate": 4.405176453420322e-06, "loss": 0.12950174510478973, "step": 2003 }, { "epoch": 0.5202492211838006, "grad_norm": 1.8781867027282715, "learning_rate": 4.4045354998907196e-06, "loss": 0.13546739518642426, "step": 2004 }, { "epoch": 0.520508826583593, "grad_norm": 1.3082679510116577, "learning_rate": 4.4038942538771235e-06, "loss": 0.10126572847366333, "step": 2005 }, { "epoch": 0.5207684319833853, "grad_norm": 1.9086027145385742, "learning_rate": 4.403252715492893e-06, "loss": 0.11705589294433594, "step": 2006 }, { "epoch": 0.5210280373831776, "grad_norm": 0.9397906064987183, "learning_rate": 4.402610884851437e-06, "loss": 0.10652218014001846, "step": 2007 }, { "epoch": 0.5212876427829699, "grad_norm": 1.1561118364334106, "learning_rate": 4.401968762066216e-06, "loss": 0.11239989101886749, "step": 2008 }, { "epoch": 0.5215472481827622, "grad_norm": 2.0997495651245117, "learning_rate": 4.4013263472507434e-06, "loss": 0.10893050581216812, "step": 2009 }, { "epoch": 0.5218068535825545, "grad_norm": 0.9591972827911377, "learning_rate": 4.400683640518583e-06, "loss": 0.11834415793418884, "step": 2010 }, { "epoch": 0.5220664589823468, "grad_norm": 2.5448615550994873, "learning_rate": 4.40004064198335e-06, "loss": 0.1117473617196083, "step": 2011 }, { "epoch": 0.5223260643821391, "grad_norm": 1.2061433792114258, "learning_rate": 4.399397351758714e-06, "loss": 0.14312857389450073, "step": 2012 }, { "epoch": 0.5225856697819314, "grad_norm": 2.619658946990967, "learning_rate": 4.39875376995839e-06, "loss": 0.13084600865840912, "step": 2013 }, { "epoch": 0.5228452751817237, "grad_norm": 1.7502739429473877, "learning_rate": 4.3981098966961524e-06, "loss": 0.11939692497253418, "step": 2014 }, { "epoch": 0.523104880581516, "grad_norm": 2.354868173599243, "learning_rate": 4.397465732085822e-06, "loss": 0.1443760097026825, "step": 2015 }, { "epoch": 0.5233644859813084, "grad_norm": 2.8455073833465576, "learning_rate": 4.396821276241272e-06, "loss": 0.12544244527816772, "step": 2016 }, { "epoch": 0.5236240913811008, "grad_norm": 3.053485631942749, "learning_rate": 4.396176529276426e-06, "loss": 0.1363816261291504, "step": 2017 }, { "epoch": 0.5238836967808931, "grad_norm": 2.912365674972534, "learning_rate": 4.3955314913052636e-06, "loss": 0.12455637007951736, "step": 2018 }, { "epoch": 0.5241433021806854, "grad_norm": 3.2694642543792725, "learning_rate": 4.394886162441811e-06, "loss": 0.09791822731494904, "step": 2019 }, { "epoch": 0.5244029075804777, "grad_norm": 2.3188624382019043, "learning_rate": 4.394240542800148e-06, "loss": 0.12641456723213196, "step": 2020 }, { "epoch": 0.52466251298027, "grad_norm": 1.1268826723098755, "learning_rate": 4.393594632494404e-06, "loss": 0.1120649054646492, "step": 2021 }, { "epoch": 0.5249221183800623, "grad_norm": 2.940020799636841, "learning_rate": 4.3929484316387636e-06, "loss": 0.12668535113334656, "step": 2022 }, { "epoch": 0.5251817237798546, "grad_norm": 2.5016376972198486, "learning_rate": 4.392301940347458e-06, "loss": 0.11386948823928833, "step": 2023 }, { "epoch": 0.5254413291796469, "grad_norm": 3.8838601112365723, "learning_rate": 4.391655158734774e-06, "loss": 0.1545712947845459, "step": 2024 }, { "epoch": 0.5257009345794392, "grad_norm": 3.3186585903167725, "learning_rate": 4.391008086915046e-06, "loss": 0.15694017708301544, "step": 2025 }, { "epoch": 0.5259605399792315, "grad_norm": 2.222515821456909, "learning_rate": 4.390360725002663e-06, "loss": 0.11846349388360977, "step": 2026 }, { "epoch": 0.5262201453790238, "grad_norm": 2.3581862449645996, "learning_rate": 4.389713073112063e-06, "loss": 0.11909611523151398, "step": 2027 }, { "epoch": 0.5264797507788161, "grad_norm": 3.6687605381011963, "learning_rate": 4.389065131357736e-06, "loss": 0.13871710002422333, "step": 2028 }, { "epoch": 0.5267393561786086, "grad_norm": 1.549666166305542, "learning_rate": 4.388416899854224e-06, "loss": 0.12446251511573792, "step": 2029 }, { "epoch": 0.5269989615784009, "grad_norm": 1.782500982284546, "learning_rate": 4.387768378716117e-06, "loss": 0.12821589410305023, "step": 2030 }, { "epoch": 0.5272585669781932, "grad_norm": 1.7808338403701782, "learning_rate": 4.387119568058063e-06, "loss": 0.12541377544403076, "step": 2031 }, { "epoch": 0.5275181723779855, "grad_norm": 11.79625129699707, "learning_rate": 4.386470467994753e-06, "loss": 0.11434774100780487, "step": 2032 }, { "epoch": 0.5277777777777778, "grad_norm": 2.4708564281463623, "learning_rate": 4.385821078640935e-06, "loss": 0.13311977684497833, "step": 2033 }, { "epoch": 0.5280373831775701, "grad_norm": 1.533882737159729, "learning_rate": 4.385171400111405e-06, "loss": 0.11239848285913467, "step": 2034 }, { "epoch": 0.5282969885773624, "grad_norm": 0.9933056235313416, "learning_rate": 4.3845214325210124e-06, "loss": 0.12220822274684906, "step": 2035 }, { "epoch": 0.5285565939771547, "grad_norm": 2.326352119445801, "learning_rate": 4.383871175984656e-06, "loss": 0.11856043338775635, "step": 2036 }, { "epoch": 0.528816199376947, "grad_norm": 2.262390613555908, "learning_rate": 4.383220630617287e-06, "loss": 0.12899243831634521, "step": 2037 }, { "epoch": 0.5290758047767393, "grad_norm": 2.6319687366485596, "learning_rate": 4.3825697965339055e-06, "loss": 0.12810233235359192, "step": 2038 }, { "epoch": 0.5293354101765316, "grad_norm": 2.5091395378112793, "learning_rate": 4.3819186738495645e-06, "loss": 0.13436132669448853, "step": 2039 }, { "epoch": 0.5295950155763239, "grad_norm": 1.1971381902694702, "learning_rate": 4.381267262679369e-06, "loss": 0.12344398349523544, "step": 2040 }, { "epoch": 0.5298546209761164, "grad_norm": 2.007255792617798, "learning_rate": 4.380615563138472e-06, "loss": 0.12793320417404175, "step": 2041 }, { "epoch": 0.5301142263759087, "grad_norm": 4.860598564147949, "learning_rate": 4.3799635753420794e-06, "loss": 0.1209569051861763, "step": 2042 }, { "epoch": 0.530373831775701, "grad_norm": 2.2177631855010986, "learning_rate": 4.379311299405449e-06, "loss": 0.11245104670524597, "step": 2043 }, { "epoch": 0.5306334371754933, "grad_norm": 1.2386932373046875, "learning_rate": 4.378658735443886e-06, "loss": 0.11700718849897385, "step": 2044 }, { "epoch": 0.5308930425752856, "grad_norm": 2.2273294925689697, "learning_rate": 4.37800588357275e-06, "loss": 0.11553338915109634, "step": 2045 }, { "epoch": 0.5311526479750779, "grad_norm": 2.781486749649048, "learning_rate": 4.37735274390745e-06, "loss": 0.1393412947654724, "step": 2046 }, { "epoch": 0.5314122533748702, "grad_norm": 3.784506320953369, "learning_rate": 4.376699316563447e-06, "loss": 0.11426542699337006, "step": 2047 }, { "epoch": 0.5316718587746625, "grad_norm": 2.373750925064087, "learning_rate": 4.3760456016562505e-06, "loss": 0.11350499838590622, "step": 2048 }, { "epoch": 0.5319314641744548, "grad_norm": 1.0834479331970215, "learning_rate": 4.375391599301424e-06, "loss": 0.10924850404262543, "step": 2049 }, { "epoch": 0.5321910695742471, "grad_norm": 2.3175480365753174, "learning_rate": 4.374737309614579e-06, "loss": 0.12909837067127228, "step": 2050 }, { "epoch": 0.5324506749740394, "grad_norm": 3.702164888381958, "learning_rate": 4.374082732711379e-06, "loss": 0.13260118663311005, "step": 2051 }, { "epoch": 0.5327102803738317, "grad_norm": 1.9755738973617554, "learning_rate": 4.373427868707539e-06, "loss": 0.10646931827068329, "step": 2052 }, { "epoch": 0.5329698857736241, "grad_norm": 4.146246910095215, "learning_rate": 4.3727727177188226e-06, "loss": 0.11755113303661346, "step": 2053 }, { "epoch": 0.5332294911734164, "grad_norm": 2.7128517627716064, "learning_rate": 4.372117279861047e-06, "loss": 0.09673839807510376, "step": 2054 }, { "epoch": 0.5334890965732088, "grad_norm": 2.6609034538269043, "learning_rate": 4.371461555250077e-06, "loss": 0.11379005014896393, "step": 2055 }, { "epoch": 0.5337487019730011, "grad_norm": 3.9680263996124268, "learning_rate": 4.370805544001831e-06, "loss": 0.123557448387146, "step": 2056 }, { "epoch": 0.5340083073727934, "grad_norm": 3.985316753387451, "learning_rate": 4.370149246232277e-06, "loss": 0.12413358688354492, "step": 2057 }, { "epoch": 0.5342679127725857, "grad_norm": 2.3803749084472656, "learning_rate": 4.3694926620574315e-06, "loss": 0.13004858791828156, "step": 2058 }, { "epoch": 0.534527518172378, "grad_norm": 4.03987979888916, "learning_rate": 4.3688357915933656e-06, "loss": 0.13832950592041016, "step": 2059 }, { "epoch": 0.5347871235721703, "grad_norm": 2.172118902206421, "learning_rate": 4.368178634956199e-06, "loss": 0.1403358280658722, "step": 2060 }, { "epoch": 0.5350467289719626, "grad_norm": 2.373370409011841, "learning_rate": 4.3675211922621e-06, "loss": 0.11190677434206009, "step": 2061 }, { "epoch": 0.5353063343717549, "grad_norm": 1.948409080505371, "learning_rate": 4.36686346362729e-06, "loss": 0.10380979627370834, "step": 2062 }, { "epoch": 0.5355659397715472, "grad_norm": 2.621903657913208, "learning_rate": 4.366205449168042e-06, "loss": 0.10987437516450882, "step": 2063 }, { "epoch": 0.5358255451713395, "grad_norm": 1.0829923152923584, "learning_rate": 4.3655471490006766e-06, "loss": 0.1034424751996994, "step": 2064 }, { "epoch": 0.5360851505711319, "grad_norm": 0.6836061477661133, "learning_rate": 4.364888563241566e-06, "loss": 0.11419211328029633, "step": 2065 }, { "epoch": 0.5363447559709242, "grad_norm": 0.7354099154472351, "learning_rate": 4.364229692007134e-06, "loss": 0.13045546412467957, "step": 2066 }, { "epoch": 0.5366043613707165, "grad_norm": 2.0977935791015625, "learning_rate": 4.363570535413852e-06, "loss": 0.11799634993076324, "step": 2067 }, { "epoch": 0.5368639667705088, "grad_norm": 2.90828537940979, "learning_rate": 4.362911093578246e-06, "loss": 0.12028857320547104, "step": 2068 }, { "epoch": 0.5371235721703012, "grad_norm": 2.873861312866211, "learning_rate": 4.362251366616891e-06, "loss": 0.15318918228149414, "step": 2069 }, { "epoch": 0.5373831775700935, "grad_norm": 2.7353644371032715, "learning_rate": 4.361591354646409e-06, "loss": 0.1299131214618683, "step": 2070 }, { "epoch": 0.5376427829698858, "grad_norm": 2.8716912269592285, "learning_rate": 4.360931057783475e-06, "loss": 0.1261885166168213, "step": 2071 }, { "epoch": 0.5379023883696781, "grad_norm": 4.315973281860352, "learning_rate": 4.360270476144817e-06, "loss": 0.11583228409290314, "step": 2072 }, { "epoch": 0.5381619937694704, "grad_norm": 1.2511696815490723, "learning_rate": 4.359609609847208e-06, "loss": 0.11592555046081543, "step": 2073 }, { "epoch": 0.5384215991692627, "grad_norm": 1.4542690515518188, "learning_rate": 4.358948459007475e-06, "loss": 0.1122203916311264, "step": 2074 }, { "epoch": 0.538681204569055, "grad_norm": 3.798694610595703, "learning_rate": 4.358287023742495e-06, "loss": 0.09695657342672348, "step": 2075 }, { "epoch": 0.5389408099688473, "grad_norm": 1.8683383464813232, "learning_rate": 4.357625304169194e-06, "loss": 0.12539097666740417, "step": 2076 }, { "epoch": 0.5392004153686397, "grad_norm": 2.26434326171875, "learning_rate": 4.356963300404549e-06, "loss": 0.10829802602529526, "step": 2077 }, { "epoch": 0.539460020768432, "grad_norm": 1.873244047164917, "learning_rate": 4.356301012565588e-06, "loss": 0.12590232491493225, "step": 2078 }, { "epoch": 0.5397196261682243, "grad_norm": 2.5433783531188965, "learning_rate": 4.355638440769385e-06, "loss": 0.14357513189315796, "step": 2079 }, { "epoch": 0.5399792315680166, "grad_norm": 1.8011234998703003, "learning_rate": 4.354975585133072e-06, "loss": 0.1191687136888504, "step": 2080 }, { "epoch": 0.5402388369678089, "grad_norm": 1.7253737449645996, "learning_rate": 4.354312445773825e-06, "loss": 0.13926811516284943, "step": 2081 }, { "epoch": 0.5404984423676013, "grad_norm": 2.621243715286255, "learning_rate": 4.353649022808869e-06, "loss": 0.14776107668876648, "step": 2082 }, { "epoch": 0.5407580477673936, "grad_norm": 3.398630380630493, "learning_rate": 4.352985316355487e-06, "loss": 0.1317748725414276, "step": 2083 }, { "epoch": 0.5410176531671859, "grad_norm": 2.847623825073242, "learning_rate": 4.3523213265310036e-06, "loss": 0.11185447871685028, "step": 2084 }, { "epoch": 0.5412772585669782, "grad_norm": 0.8078081011772156, "learning_rate": 4.351657053452798e-06, "loss": 0.13034655153751373, "step": 2085 }, { "epoch": 0.5415368639667705, "grad_norm": 3.3543200492858887, "learning_rate": 4.350992497238298e-06, "loss": 0.12680351734161377, "step": 2086 }, { "epoch": 0.5417964693665628, "grad_norm": 2.1460328102111816, "learning_rate": 4.350327658004983e-06, "loss": 0.1273733228445053, "step": 2087 }, { "epoch": 0.5420560747663551, "grad_norm": 2.6774280071258545, "learning_rate": 4.349662535870381e-06, "loss": 0.12423699349164963, "step": 2088 }, { "epoch": 0.5423156801661475, "grad_norm": 3.92419695854187, "learning_rate": 4.348997130952069e-06, "loss": 0.14475244283676147, "step": 2089 }, { "epoch": 0.5425752855659398, "grad_norm": 2.8560268878936768, "learning_rate": 4.348331443367677e-06, "loss": 0.1331464648246765, "step": 2090 }, { "epoch": 0.5428348909657321, "grad_norm": 1.0946851968765259, "learning_rate": 4.3476654732348826e-06, "loss": 0.10649845004081726, "step": 2091 }, { "epoch": 0.5430944963655244, "grad_norm": 2.8409996032714844, "learning_rate": 4.346999220671414e-06, "loss": 0.13894377648830414, "step": 2092 }, { "epoch": 0.5433541017653167, "grad_norm": 3.1691644191741943, "learning_rate": 4.34633268579505e-06, "loss": 0.11493811756372452, "step": 2093 }, { "epoch": 0.543613707165109, "grad_norm": 1.9689100980758667, "learning_rate": 4.345665868723618e-06, "loss": 0.12609517574310303, "step": 2094 }, { "epoch": 0.5438733125649013, "grad_norm": 2.858330011367798, "learning_rate": 4.3449987695749964e-06, "loss": 0.10196562111377716, "step": 2095 }, { "epoch": 0.5441329179646937, "grad_norm": 2.145916223526001, "learning_rate": 4.344331388467113e-06, "loss": 0.1323206126689911, "step": 2096 }, { "epoch": 0.544392523364486, "grad_norm": 1.0318974256515503, "learning_rate": 4.343663725517946e-06, "loss": 0.11680027842521667, "step": 2097 }, { "epoch": 0.5446521287642783, "grad_norm": 4.0464043617248535, "learning_rate": 4.342995780845522e-06, "loss": 0.15165141224861145, "step": 2098 }, { "epoch": 0.5449117341640706, "grad_norm": 4.3744916915893555, "learning_rate": 4.342327554567918e-06, "loss": 0.1447189450263977, "step": 2099 }, { "epoch": 0.5451713395638629, "grad_norm": 1.8389066457748413, "learning_rate": 4.341659046803263e-06, "loss": 0.12200476229190826, "step": 2100 }, { "epoch": 0.5454309449636553, "grad_norm": 1.3541665077209473, "learning_rate": 4.340990257669732e-06, "loss": 0.12621702253818512, "step": 2101 }, { "epoch": 0.5456905503634476, "grad_norm": 2.197845697402954, "learning_rate": 4.340321187285553e-06, "loss": 0.12566611170768738, "step": 2102 }, { "epoch": 0.5459501557632399, "grad_norm": 4.110098838806152, "learning_rate": 4.339651835769e-06, "loss": 0.13013538718223572, "step": 2103 }, { "epoch": 0.5462097611630322, "grad_norm": 1.5482085943222046, "learning_rate": 4.338982203238403e-06, "loss": 0.11798806488513947, "step": 2104 }, { "epoch": 0.5464693665628245, "grad_norm": 1.7798962593078613, "learning_rate": 4.338312289812134e-06, "loss": 0.1260923594236374, "step": 2105 }, { "epoch": 0.5467289719626168, "grad_norm": 3.214982271194458, "learning_rate": 4.33764209560862e-06, "loss": 0.14067772030830383, "step": 2106 }, { "epoch": 0.5469885773624091, "grad_norm": 1.4959263801574707, "learning_rate": 4.336971620746334e-06, "loss": 0.11355575174093246, "step": 2107 }, { "epoch": 0.5472481827622014, "grad_norm": 1.5359292030334473, "learning_rate": 4.336300865343805e-06, "loss": 0.134963259100914, "step": 2108 }, { "epoch": 0.5475077881619937, "grad_norm": 2.2568209171295166, "learning_rate": 4.335629829519602e-06, "loss": 0.12655769288539886, "step": 2109 }, { "epoch": 0.547767393561786, "grad_norm": 1.71578049659729, "learning_rate": 4.334958513392353e-06, "loss": 0.1225573867559433, "step": 2110 }, { "epoch": 0.5480269989615784, "grad_norm": 1.0033475160598755, "learning_rate": 4.33428691708073e-06, "loss": 0.12131151556968689, "step": 2111 }, { "epoch": 0.5482866043613707, "grad_norm": 1.5657466650009155, "learning_rate": 4.333615040703454e-06, "loss": 0.10470591485500336, "step": 2112 }, { "epoch": 0.5485462097611631, "grad_norm": 3.2996907234191895, "learning_rate": 4.332942884379299e-06, "loss": 0.11997067183256149, "step": 2113 }, { "epoch": 0.5488058151609554, "grad_norm": 1.9409831762313843, "learning_rate": 4.332270448227087e-06, "loss": 0.10943956673145294, "step": 2114 }, { "epoch": 0.5490654205607477, "grad_norm": 0.8689941167831421, "learning_rate": 4.331597732365689e-06, "loss": 0.11278876662254333, "step": 2115 }, { "epoch": 0.54932502596054, "grad_norm": 2.2704861164093018, "learning_rate": 4.330924736914026e-06, "loss": 0.12895268201828003, "step": 2116 }, { "epoch": 0.5495846313603323, "grad_norm": 1.395727515220642, "learning_rate": 4.3302514619910665e-06, "loss": 0.11531613767147064, "step": 2117 }, { "epoch": 0.5498442367601246, "grad_norm": 3.4282681941986084, "learning_rate": 4.329577907715833e-06, "loss": 0.13339054584503174, "step": 2118 }, { "epoch": 0.5501038421599169, "grad_norm": 3.6261024475097656, "learning_rate": 4.328904074207392e-06, "loss": 0.12402568757534027, "step": 2119 }, { "epoch": 0.5503634475597092, "grad_norm": 6.362729072570801, "learning_rate": 4.328229961584864e-06, "loss": 0.15267261862754822, "step": 2120 }, { "epoch": 0.5506230529595015, "grad_norm": 1.6152969598770142, "learning_rate": 4.327555569967414e-06, "loss": 0.12226212024688721, "step": 2121 }, { "epoch": 0.5508826583592938, "grad_norm": 2.1877899169921875, "learning_rate": 4.326880899474261e-06, "loss": 0.1302894800901413, "step": 2122 }, { "epoch": 0.5511422637590861, "grad_norm": 3.2196614742279053, "learning_rate": 4.32620595022467e-06, "loss": 0.12382028251886368, "step": 2123 }, { "epoch": 0.5514018691588785, "grad_norm": 5.112706661224365, "learning_rate": 4.325530722337959e-06, "loss": 0.1473696231842041, "step": 2124 }, { "epoch": 0.5516614745586709, "grad_norm": 2.91556453704834, "learning_rate": 4.32485521593349e-06, "loss": 0.12168692797422409, "step": 2125 }, { "epoch": 0.5519210799584632, "grad_norm": 1.0398328304290771, "learning_rate": 4.324179431130679e-06, "loss": 0.11521132290363312, "step": 2126 }, { "epoch": 0.5521806853582555, "grad_norm": 2.255669355392456, "learning_rate": 4.323503368048987e-06, "loss": 0.11359703540802002, "step": 2127 }, { "epoch": 0.5524402907580478, "grad_norm": 1.6864395141601562, "learning_rate": 4.32282702680793e-06, "loss": 0.08895222842693329, "step": 2128 }, { "epoch": 0.5526998961578401, "grad_norm": 4.052178382873535, "learning_rate": 4.322150407527067e-06, "loss": 0.12491011619567871, "step": 2129 }, { "epoch": 0.5529595015576324, "grad_norm": 2.5124387741088867, "learning_rate": 4.321473510326009e-06, "loss": 0.11957038193941116, "step": 2130 }, { "epoch": 0.5532191069574247, "grad_norm": 2.677018404006958, "learning_rate": 4.320796335324416e-06, "loss": 0.14086134731769562, "step": 2131 }, { "epoch": 0.553478712357217, "grad_norm": 1.2880157232284546, "learning_rate": 4.3201188826419985e-06, "loss": 0.12631268799304962, "step": 2132 }, { "epoch": 0.5537383177570093, "grad_norm": 2.135772466659546, "learning_rate": 4.319441152398513e-06, "loss": 0.11709444224834442, "step": 2133 }, { "epoch": 0.5539979231568016, "grad_norm": 1.9017390012741089, "learning_rate": 4.318763144713767e-06, "loss": 0.13992831110954285, "step": 2134 }, { "epoch": 0.5542575285565939, "grad_norm": 3.372443914413452, "learning_rate": 4.318084859707617e-06, "loss": 0.12325534224510193, "step": 2135 }, { "epoch": 0.5545171339563862, "grad_norm": 1.0538227558135986, "learning_rate": 4.317406297499968e-06, "loss": 0.11285422742366791, "step": 2136 }, { "epoch": 0.5547767393561787, "grad_norm": 1.8764805793762207, "learning_rate": 4.316727458210774e-06, "loss": 0.10791990160942078, "step": 2137 }, { "epoch": 0.555036344755971, "grad_norm": 1.5492355823516846, "learning_rate": 4.31604834196004e-06, "loss": 0.11716180294752121, "step": 2138 }, { "epoch": 0.5552959501557633, "grad_norm": 2.724609136581421, "learning_rate": 4.315368948867816e-06, "loss": 0.1100931316614151, "step": 2139 }, { "epoch": 0.5555555555555556, "grad_norm": 2.084449529647827, "learning_rate": 4.3146892790542035e-06, "loss": 0.1150374561548233, "step": 2140 }, { "epoch": 0.5558151609553479, "grad_norm": 1.5541870594024658, "learning_rate": 4.314009332639355e-06, "loss": 0.12259181588888168, "step": 2141 }, { "epoch": 0.5560747663551402, "grad_norm": 2.0706801414489746, "learning_rate": 4.313329109743466e-06, "loss": 0.12528789043426514, "step": 2142 }, { "epoch": 0.5563343717549325, "grad_norm": 2.2475931644439697, "learning_rate": 4.312648610486786e-06, "loss": 0.11917687952518463, "step": 2143 }, { "epoch": 0.5565939771547248, "grad_norm": 2.5143167972564697, "learning_rate": 4.311967834989612e-06, "loss": 0.12797856330871582, "step": 2144 }, { "epoch": 0.5568535825545171, "grad_norm": 1.9656398296356201, "learning_rate": 4.311286783372289e-06, "loss": 0.1289362609386444, "step": 2145 }, { "epoch": 0.5571131879543094, "grad_norm": 1.4458121061325073, "learning_rate": 4.310605455755211e-06, "loss": 0.13147804141044617, "step": 2146 }, { "epoch": 0.5573727933541017, "grad_norm": 1.155295729637146, "learning_rate": 4.3099238522588225e-06, "loss": 0.11799715459346771, "step": 2147 }, { "epoch": 0.557632398753894, "grad_norm": 1.8225914239883423, "learning_rate": 4.309241973003614e-06, "loss": 0.10698229819536209, "step": 2148 }, { "epoch": 0.5578920041536864, "grad_norm": 2.5340347290039062, "learning_rate": 4.3085598181101256e-06, "loss": 0.1327984482049942, "step": 2149 }, { "epoch": 0.5581516095534788, "grad_norm": 2.230046510696411, "learning_rate": 4.307877387698949e-06, "loss": 0.11672541499137878, "step": 2150 }, { "epoch": 0.5584112149532711, "grad_norm": 3.175961494445801, "learning_rate": 4.30719468189072e-06, "loss": 0.12959827482700348, "step": 2151 }, { "epoch": 0.5586708203530634, "grad_norm": 1.321337342262268, "learning_rate": 4.306511700806127e-06, "loss": 0.1115468442440033, "step": 2152 }, { "epoch": 0.5589304257528557, "grad_norm": 2.706083059310913, "learning_rate": 4.305828444565904e-06, "loss": 0.10385219007730484, "step": 2153 }, { "epoch": 0.559190031152648, "grad_norm": 1.6292436122894287, "learning_rate": 4.305144913290837e-06, "loss": 0.11553502082824707, "step": 2154 }, { "epoch": 0.5594496365524403, "grad_norm": 2.9299211502075195, "learning_rate": 4.304461107101756e-06, "loss": 0.15711957216262817, "step": 2155 }, { "epoch": 0.5597092419522326, "grad_norm": 2.3729848861694336, "learning_rate": 4.303777026119545e-06, "loss": 0.1348770558834076, "step": 2156 }, { "epoch": 0.5599688473520249, "grad_norm": 4.112486839294434, "learning_rate": 4.3030926704651314e-06, "loss": 0.1272743046283722, "step": 2157 }, { "epoch": 0.5602284527518172, "grad_norm": 1.5399588346481323, "learning_rate": 4.302408040259495e-06, "loss": 0.12724781036376953, "step": 2158 }, { "epoch": 0.5604880581516095, "grad_norm": 2.268338918685913, "learning_rate": 4.301723135623662e-06, "loss": 0.15374168753623962, "step": 2159 }, { "epoch": 0.5607476635514018, "grad_norm": 2.0762157440185547, "learning_rate": 4.30103795667871e-06, "loss": 0.07743412256240845, "step": 2160 }, { "epoch": 0.5610072689511942, "grad_norm": 2.737025499343872, "learning_rate": 4.300352503545761e-06, "loss": 0.12104471027851105, "step": 2161 }, { "epoch": 0.5612668743509865, "grad_norm": 5.113266944885254, "learning_rate": 4.299666776345987e-06, "loss": 0.1360730230808258, "step": 2162 }, { "epoch": 0.5615264797507789, "grad_norm": 1.0738364458084106, "learning_rate": 4.298980775200609e-06, "loss": 0.10465465486049652, "step": 2163 }, { "epoch": 0.5617860851505712, "grad_norm": 1.0099011659622192, "learning_rate": 4.298294500230898e-06, "loss": 0.11854273080825806, "step": 2164 }, { "epoch": 0.5620456905503635, "grad_norm": 2.515598773956299, "learning_rate": 4.29760795155817e-06, "loss": 0.13250425457954407, "step": 2165 }, { "epoch": 0.5623052959501558, "grad_norm": 3.5537638664245605, "learning_rate": 4.296921129303792e-06, "loss": 0.12485432624816895, "step": 2166 }, { "epoch": 0.5625649013499481, "grad_norm": 2.3967459201812744, "learning_rate": 4.296234033589178e-06, "loss": 0.1207912415266037, "step": 2167 }, { "epoch": 0.5628245067497404, "grad_norm": 2.2460734844207764, "learning_rate": 4.295546664535791e-06, "loss": 0.12181563675403595, "step": 2168 }, { "epoch": 0.5630841121495327, "grad_norm": 1.5142055749893188, "learning_rate": 4.2948590222651424e-06, "loss": 0.11780738830566406, "step": 2169 }, { "epoch": 0.563343717549325, "grad_norm": 2.7858991622924805, "learning_rate": 4.294171106898791e-06, "loss": 0.12791508436203003, "step": 2170 }, { "epoch": 0.5636033229491173, "grad_norm": 2.167301893234253, "learning_rate": 4.293482918558344e-06, "loss": 0.11547235399484634, "step": 2171 }, { "epoch": 0.5638629283489096, "grad_norm": 1.688668966293335, "learning_rate": 4.2927944573654595e-06, "loss": 0.1303727626800537, "step": 2172 }, { "epoch": 0.564122533748702, "grad_norm": 2.407670736312866, "learning_rate": 4.292105723441839e-06, "loss": 0.12447008490562439, "step": 2173 }, { "epoch": 0.5643821391484943, "grad_norm": 3.352027416229248, "learning_rate": 4.291416716909237e-06, "loss": 0.12352519482374191, "step": 2174 }, { "epoch": 0.5646417445482866, "grad_norm": 1.4168672561645508, "learning_rate": 4.290727437889453e-06, "loss": 0.11939871311187744, "step": 2175 }, { "epoch": 0.564901349948079, "grad_norm": 2.085803747177124, "learning_rate": 4.290037886504335e-06, "loss": 0.10829664021730423, "step": 2176 }, { "epoch": 0.5651609553478713, "grad_norm": 3.5152347087860107, "learning_rate": 4.289348062875782e-06, "loss": 0.1166018545627594, "step": 2177 }, { "epoch": 0.5654205607476636, "grad_norm": 1.957512617111206, "learning_rate": 4.288657967125738e-06, "loss": 0.1428908407688141, "step": 2178 }, { "epoch": 0.5656801661474559, "grad_norm": 2.336308240890503, "learning_rate": 4.2879675993761954e-06, "loss": 0.14787103235721588, "step": 2179 }, { "epoch": 0.5659397715472482, "grad_norm": 2.146068811416626, "learning_rate": 4.287276959749196e-06, "loss": 0.1154857724905014, "step": 2180 }, { "epoch": 0.5661993769470405, "grad_norm": 1.6414262056350708, "learning_rate": 4.286586048366829e-06, "loss": 0.10152725875377655, "step": 2181 }, { "epoch": 0.5664589823468328, "grad_norm": 3.8629143238067627, "learning_rate": 4.285894865351233e-06, "loss": 0.10513446480035782, "step": 2182 }, { "epoch": 0.5667185877466251, "grad_norm": 2.241304636001587, "learning_rate": 4.2852034108245895e-06, "loss": 0.11155714839696884, "step": 2183 }, { "epoch": 0.5669781931464174, "grad_norm": 1.1080377101898193, "learning_rate": 4.284511684909136e-06, "loss": 0.1128615289926529, "step": 2184 }, { "epoch": 0.5672377985462098, "grad_norm": 2.9894721508026123, "learning_rate": 4.2838196877271536e-06, "loss": 0.11858758330345154, "step": 2185 }, { "epoch": 0.5674974039460021, "grad_norm": 2.616866111755371, "learning_rate": 4.283127419400969e-06, "loss": 0.11943817138671875, "step": 2186 }, { "epoch": 0.5677570093457944, "grad_norm": 1.1518170833587646, "learning_rate": 4.28243488005296e-06, "loss": 0.11338556557893753, "step": 2187 }, { "epoch": 0.5680166147455867, "grad_norm": 2.7441232204437256, "learning_rate": 4.281742069805553e-06, "loss": 0.07293090224266052, "step": 2188 }, { "epoch": 0.568276220145379, "grad_norm": 2.467839241027832, "learning_rate": 4.281048988781221e-06, "loss": 0.1279602348804474, "step": 2189 }, { "epoch": 0.5685358255451713, "grad_norm": 2.4034011363983154, "learning_rate": 4.280355637102483e-06, "loss": 0.06455725431442261, "step": 2190 }, { "epoch": 0.5687954309449637, "grad_norm": 2.494563579559326, "learning_rate": 4.279662014891909e-06, "loss": 0.11460642516613007, "step": 2191 }, { "epoch": 0.569055036344756, "grad_norm": 2.550480842590332, "learning_rate": 4.2789681222721155e-06, "loss": 0.13016042113304138, "step": 2192 }, { "epoch": 0.5693146417445483, "grad_norm": 1.5156950950622559, "learning_rate": 4.278273959365768e-06, "loss": 0.11467255651950836, "step": 2193 }, { "epoch": 0.5695742471443406, "grad_norm": 2.6244592666625977, "learning_rate": 4.277579526295575e-06, "loss": 0.0985720157623291, "step": 2194 }, { "epoch": 0.5698338525441329, "grad_norm": 2.005815267562866, "learning_rate": 4.2768848231843e-06, "loss": 0.1333007663488388, "step": 2195 }, { "epoch": 0.5700934579439252, "grad_norm": 2.9505677223205566, "learning_rate": 4.27618985015475e-06, "loss": 0.13755570352077484, "step": 2196 }, { "epoch": 0.5703530633437176, "grad_norm": 3.228346109390259, "learning_rate": 4.275494607329778e-06, "loss": 0.1277502030134201, "step": 2197 }, { "epoch": 0.5706126687435099, "grad_norm": 2.679226875305176, "learning_rate": 4.27479909483229e-06, "loss": 0.10689784586429596, "step": 2198 }, { "epoch": 0.5708722741433022, "grad_norm": 3.0219857692718506, "learning_rate": 4.274103312785235e-06, "loss": 0.12469355762004852, "step": 2199 }, { "epoch": 0.5711318795430945, "grad_norm": 4.326538562774658, "learning_rate": 4.27340726131161e-06, "loss": 0.13191834092140198, "step": 2200 }, { "epoch": 0.5713914849428868, "grad_norm": 4.161297798156738, "learning_rate": 4.272710940534465e-06, "loss": 0.13748610019683838, "step": 2201 }, { "epoch": 0.5716510903426791, "grad_norm": 2.5720977783203125, "learning_rate": 4.27201435057689e-06, "loss": 0.11870089173316956, "step": 2202 }, { "epoch": 0.5719106957424714, "grad_norm": 1.1194697618484497, "learning_rate": 4.271317491562028e-06, "loss": 0.10776330530643463, "step": 2203 }, { "epoch": 0.5721703011422637, "grad_norm": 1.3567180633544922, "learning_rate": 4.270620363613066e-06, "loss": 0.11261685192584991, "step": 2204 }, { "epoch": 0.572429906542056, "grad_norm": 3.003782033920288, "learning_rate": 4.269922966853242e-06, "loss": 0.09438121318817139, "step": 2205 }, { "epoch": 0.5726895119418484, "grad_norm": 1.4651082754135132, "learning_rate": 4.269225301405838e-06, "loss": 0.1193796694278717, "step": 2206 }, { "epoch": 0.5729491173416407, "grad_norm": 3.1041653156280518, "learning_rate": 4.268527367394188e-06, "loss": 0.13547328114509583, "step": 2207 }, { "epoch": 0.573208722741433, "grad_norm": 3.280543565750122, "learning_rate": 4.267829164941669e-06, "loss": 0.11290401220321655, "step": 2208 }, { "epoch": 0.5734683281412254, "grad_norm": 1.4876707792282104, "learning_rate": 4.267130694171706e-06, "loss": 0.11578759551048279, "step": 2209 }, { "epoch": 0.5737279335410177, "grad_norm": 2.3287441730499268, "learning_rate": 4.266431955207774e-06, "loss": 0.13622649013996124, "step": 2210 }, { "epoch": 0.57398753894081, "grad_norm": 1.8955572843551636, "learning_rate": 4.265732948173395e-06, "loss": 0.11412569880485535, "step": 2211 }, { "epoch": 0.5742471443406023, "grad_norm": 2.202094316482544, "learning_rate": 4.265033673192136e-06, "loss": 0.12384625524282455, "step": 2212 }, { "epoch": 0.5745067497403946, "grad_norm": 2.6410293579101562, "learning_rate": 4.264334130387614e-06, "loss": 0.13039235770702362, "step": 2213 }, { "epoch": 0.5747663551401869, "grad_norm": 3.7167515754699707, "learning_rate": 4.26363431988349e-06, "loss": 0.15357813239097595, "step": 2214 }, { "epoch": 0.5750259605399792, "grad_norm": 2.082618474960327, "learning_rate": 4.262934241803477e-06, "loss": 0.11512379348278046, "step": 2215 }, { "epoch": 0.5752855659397715, "grad_norm": 1.3506933450698853, "learning_rate": 4.2622338962713315e-06, "loss": 0.11679190397262573, "step": 2216 }, { "epoch": 0.5755451713395638, "grad_norm": 2.117025375366211, "learning_rate": 4.261533283410858e-06, "loss": 0.12267552316188812, "step": 2217 }, { "epoch": 0.5758047767393562, "grad_norm": 2.7335612773895264, "learning_rate": 4.26083240334591e-06, "loss": 0.12503227591514587, "step": 2218 }, { "epoch": 0.5760643821391485, "grad_norm": 2.26762318611145, "learning_rate": 4.260131256200387e-06, "loss": 0.12634141743183136, "step": 2219 }, { "epoch": 0.5763239875389408, "grad_norm": 2.2423882484436035, "learning_rate": 4.259429842098235e-06, "loss": 0.11465147137641907, "step": 2220 }, { "epoch": 0.5765835929387332, "grad_norm": 1.5286779403686523, "learning_rate": 4.258728161163448e-06, "loss": 0.11594849824905396, "step": 2221 }, { "epoch": 0.5768431983385255, "grad_norm": 1.379412055015564, "learning_rate": 4.258026213520067e-06, "loss": 0.11798246204853058, "step": 2222 }, { "epoch": 0.5771028037383178, "grad_norm": 1.4699456691741943, "learning_rate": 4.257323999292182e-06, "loss": 0.13136297464370728, "step": 2223 }, { "epoch": 0.5773624091381101, "grad_norm": 2.6539101600646973, "learning_rate": 4.256621518603926e-06, "loss": 0.11158650368452072, "step": 2224 }, { "epoch": 0.5776220145379024, "grad_norm": 2.8797409534454346, "learning_rate": 4.255918771579482e-06, "loss": 0.13457217812538147, "step": 2225 }, { "epoch": 0.5778816199376947, "grad_norm": 1.9330500364303589, "learning_rate": 4.255215758343082e-06, "loss": 0.11543191969394684, "step": 2226 }, { "epoch": 0.578141225337487, "grad_norm": 1.7452068328857422, "learning_rate": 4.2545124790189995e-06, "loss": 0.11609496921300888, "step": 2227 }, { "epoch": 0.5784008307372793, "grad_norm": 1.1912760734558105, "learning_rate": 4.2538089337315594e-06, "loss": 0.1303730607032776, "step": 2228 }, { "epoch": 0.5786604361370716, "grad_norm": 1.1996606588363647, "learning_rate": 4.253105122605134e-06, "loss": 0.11128851771354675, "step": 2229 }, { "epoch": 0.5789200415368639, "grad_norm": 2.622762441635132, "learning_rate": 4.252401045764138e-06, "loss": 0.12067253142595291, "step": 2230 }, { "epoch": 0.5791796469366562, "grad_norm": 2.767427444458008, "learning_rate": 4.251696703333037e-06, "loss": 0.12315056473016739, "step": 2231 }, { "epoch": 0.5794392523364486, "grad_norm": 0.9873501658439636, "learning_rate": 4.250992095436345e-06, "loss": 0.11385875940322876, "step": 2232 }, { "epoch": 0.579698857736241, "grad_norm": 1.3155627250671387, "learning_rate": 4.250287222198618e-06, "loss": 0.1052393764257431, "step": 2233 }, { "epoch": 0.5799584631360333, "grad_norm": 1.8522413969039917, "learning_rate": 4.249582083744461e-06, "loss": 0.12051886320114136, "step": 2234 }, { "epoch": 0.5802180685358256, "grad_norm": 5.5124688148498535, "learning_rate": 4.248876680198529e-06, "loss": 0.13308385014533997, "step": 2235 }, { "epoch": 0.5804776739356179, "grad_norm": 1.187656283378601, "learning_rate": 4.248171011685519e-06, "loss": 0.11458157002925873, "step": 2236 }, { "epoch": 0.5807372793354102, "grad_norm": 1.9387832880020142, "learning_rate": 4.247465078330177e-06, "loss": 0.11115805059671402, "step": 2237 }, { "epoch": 0.5809968847352025, "grad_norm": 2.696012496948242, "learning_rate": 4.246758880257299e-06, "loss": 0.1253272294998169, "step": 2238 }, { "epoch": 0.5812564901349948, "grad_norm": 2.0274813175201416, "learning_rate": 4.24605241759172e-06, "loss": 0.1227392852306366, "step": 2239 }, { "epoch": 0.5815160955347871, "grad_norm": 1.760373592376709, "learning_rate": 4.245345690458329e-06, "loss": 0.10347358882427216, "step": 2240 }, { "epoch": 0.5817757009345794, "grad_norm": 3.1988794803619385, "learning_rate": 4.244638698982059e-06, "loss": 0.1035258024930954, "step": 2241 }, { "epoch": 0.5820353063343717, "grad_norm": 3.419285774230957, "learning_rate": 4.243931443287889e-06, "loss": 0.11938076466321945, "step": 2242 }, { "epoch": 0.582294911734164, "grad_norm": 1.3173737525939941, "learning_rate": 4.243223923500846e-06, "loss": 0.11547505855560303, "step": 2243 }, { "epoch": 0.5825545171339563, "grad_norm": 2.367753744125366, "learning_rate": 4.2425161397460055e-06, "loss": 0.11567345261573792, "step": 2244 }, { "epoch": 0.5828141225337488, "grad_norm": 4.231690883636475, "learning_rate": 4.241808092148485e-06, "loss": 0.13188795745372772, "step": 2245 }, { "epoch": 0.5830737279335411, "grad_norm": 3.1237950325012207, "learning_rate": 4.241099780833451e-06, "loss": 0.10416639596223831, "step": 2246 }, { "epoch": 0.5833333333333334, "grad_norm": 0.7652198076248169, "learning_rate": 4.240391205926118e-06, "loss": 0.1146698147058487, "step": 2247 }, { "epoch": 0.5835929387331257, "grad_norm": 1.541367769241333, "learning_rate": 4.239682367551745e-06, "loss": 0.12359844148159027, "step": 2248 }, { "epoch": 0.583852544132918, "grad_norm": 1.9841262102127075, "learning_rate": 4.238973265835639e-06, "loss": 0.11376328021287918, "step": 2249 }, { "epoch": 0.5841121495327103, "grad_norm": 2.304330825805664, "learning_rate": 4.238263900903152e-06, "loss": 0.11995550245046616, "step": 2250 }, { "epoch": 0.5843717549325026, "grad_norm": 2.3177759647369385, "learning_rate": 4.2375542728796835e-06, "loss": 0.1211945116519928, "step": 2251 }, { "epoch": 0.5846313603322949, "grad_norm": 1.522262454032898, "learning_rate": 4.236844381890681e-06, "loss": 0.10683278739452362, "step": 2252 }, { "epoch": 0.5848909657320872, "grad_norm": 1.266281247138977, "learning_rate": 4.236134228061636e-06, "loss": 0.12862619757652283, "step": 2253 }, { "epoch": 0.5851505711318795, "grad_norm": 1.256664752960205, "learning_rate": 4.235423811518087e-06, "loss": 0.10564941167831421, "step": 2254 }, { "epoch": 0.5854101765316718, "grad_norm": 3.774021863937378, "learning_rate": 4.2347131323856195e-06, "loss": 0.13191930949687958, "step": 2255 }, { "epoch": 0.5856697819314641, "grad_norm": 2.445443630218506, "learning_rate": 4.234002190789867e-06, "loss": 0.11766310036182404, "step": 2256 }, { "epoch": 0.5859293873312564, "grad_norm": 4.111094951629639, "learning_rate": 4.233290986856504e-06, "loss": 0.11054445058107376, "step": 2257 }, { "epoch": 0.5861889927310489, "grad_norm": 3.1919047832489014, "learning_rate": 4.23257952071126e-06, "loss": 0.11786589026451111, "step": 2258 }, { "epoch": 0.5864485981308412, "grad_norm": 4.147726058959961, "learning_rate": 4.231867792479902e-06, "loss": 0.10611233115196228, "step": 2259 }, { "epoch": 0.5867082035306335, "grad_norm": 1.5961579084396362, "learning_rate": 4.2311558022882485e-06, "loss": 0.11879999190568924, "step": 2260 }, { "epoch": 0.5869678089304258, "grad_norm": 3.1489787101745605, "learning_rate": 4.2304435502621635e-06, "loss": 0.11620903015136719, "step": 2261 }, { "epoch": 0.5872274143302181, "grad_norm": 3.2613282203674316, "learning_rate": 4.229731036527556e-06, "loss": 0.1208723708987236, "step": 2262 }, { "epoch": 0.5874870197300104, "grad_norm": 3.05991792678833, "learning_rate": 4.229018261210383e-06, "loss": 0.12586244940757751, "step": 2263 }, { "epoch": 0.5877466251298027, "grad_norm": 2.5867373943328857, "learning_rate": 4.2283052244366465e-06, "loss": 0.12392392754554749, "step": 2264 }, { "epoch": 0.588006230529595, "grad_norm": 2.725491523742676, "learning_rate": 4.227591926332394e-06, "loss": 0.12416565418243408, "step": 2265 }, { "epoch": 0.5882658359293873, "grad_norm": 2.498992443084717, "learning_rate": 4.226878367023724e-06, "loss": 0.11551336944103241, "step": 2266 }, { "epoch": 0.5885254413291796, "grad_norm": 3.5118420124053955, "learning_rate": 4.226164546636772e-06, "loss": 0.12637107074260712, "step": 2267 }, { "epoch": 0.5887850467289719, "grad_norm": 3.0358448028564453, "learning_rate": 4.2254504652977285e-06, "loss": 0.12614791095256805, "step": 2268 }, { "epoch": 0.5890446521287642, "grad_norm": 2.502393960952759, "learning_rate": 4.224736123132826e-06, "loss": 0.08575240522623062, "step": 2269 }, { "epoch": 0.5893042575285566, "grad_norm": 3.2325637340545654, "learning_rate": 4.224021520268345e-06, "loss": 0.1347004771232605, "step": 2270 }, { "epoch": 0.589563862928349, "grad_norm": 1.5359398126602173, "learning_rate": 4.2233066568306084e-06, "loss": 0.12043347954750061, "step": 2271 }, { "epoch": 0.5898234683281413, "grad_norm": 3.218644142150879, "learning_rate": 4.22259153294599e-06, "loss": 0.12925823032855988, "step": 2272 }, { "epoch": 0.5900830737279336, "grad_norm": 2.0368006229400635, "learning_rate": 4.221876148740906e-06, "loss": 0.11261364817619324, "step": 2273 }, { "epoch": 0.5903426791277259, "grad_norm": 2.246157646179199, "learning_rate": 4.221160504341821e-06, "loss": 0.1257249414920807, "step": 2274 }, { "epoch": 0.5906022845275182, "grad_norm": 2.300896644592285, "learning_rate": 4.220444599875243e-06, "loss": 0.1088237538933754, "step": 2275 }, { "epoch": 0.5908618899273105, "grad_norm": 2.1020469665527344, "learning_rate": 4.219728435467729e-06, "loss": 0.1111314445734024, "step": 2276 }, { "epoch": 0.5911214953271028, "grad_norm": 1.732020616531372, "learning_rate": 4.21901201124588e-06, "loss": 0.12968391180038452, "step": 2277 }, { "epoch": 0.5913811007268951, "grad_norm": 3.468928098678589, "learning_rate": 4.2182953273363445e-06, "loss": 0.10750147700309753, "step": 2278 }, { "epoch": 0.5916407061266874, "grad_norm": 4.058762073516846, "learning_rate": 4.217578383865813e-06, "loss": 0.07783559709787369, "step": 2279 }, { "epoch": 0.5919003115264797, "grad_norm": 3.1651980876922607, "learning_rate": 4.216861180961026e-06, "loss": 0.12033742666244507, "step": 2280 }, { "epoch": 0.592159916926272, "grad_norm": 2.2093985080718994, "learning_rate": 4.21614371874877e-06, "loss": 0.10666847229003906, "step": 2281 }, { "epoch": 0.5924195223260644, "grad_norm": 1.8752493858337402, "learning_rate": 4.215425997355875e-06, "loss": 0.10457861423492432, "step": 2282 }, { "epoch": 0.5926791277258567, "grad_norm": 3.6775801181793213, "learning_rate": 4.214708016909216e-06, "loss": 0.0737149566411972, "step": 2283 }, { "epoch": 0.592938733125649, "grad_norm": 3.1083714962005615, "learning_rate": 4.213989777535719e-06, "loss": 0.09765734523534775, "step": 2284 }, { "epoch": 0.5931983385254413, "grad_norm": 2.6758155822753906, "learning_rate": 4.2132712793623485e-06, "loss": 0.1115216314792633, "step": 2285 }, { "epoch": 0.5934579439252337, "grad_norm": 1.7930736541748047, "learning_rate": 4.2125525225161205e-06, "loss": 0.11011287569999695, "step": 2286 }, { "epoch": 0.593717549325026, "grad_norm": 3.0121712684631348, "learning_rate": 4.211833507124095e-06, "loss": 0.09135228395462036, "step": 2287 }, { "epoch": 0.5939771547248183, "grad_norm": 4.031459808349609, "learning_rate": 4.211114233313377e-06, "loss": 0.1297735869884491, "step": 2288 }, { "epoch": 0.5942367601246106, "grad_norm": 2.2665531635284424, "learning_rate": 4.2103947012111176e-06, "loss": 0.11208131909370422, "step": 2289 }, { "epoch": 0.5944963655244029, "grad_norm": 3.5743818283081055, "learning_rate": 4.209674910944513e-06, "loss": 0.12391429394483566, "step": 2290 }, { "epoch": 0.5947559709241952, "grad_norm": 3.6724562644958496, "learning_rate": 4.208954862640807e-06, "loss": 0.11039601266384125, "step": 2291 }, { "epoch": 0.5950155763239875, "grad_norm": 2.8539206981658936, "learning_rate": 4.208234556427288e-06, "loss": 0.1135542020201683, "step": 2292 }, { "epoch": 0.5952751817237798, "grad_norm": 2.411240339279175, "learning_rate": 4.207513992431286e-06, "loss": 0.13764724135398865, "step": 2293 }, { "epoch": 0.5955347871235722, "grad_norm": 1.2670934200286865, "learning_rate": 4.206793170780185e-06, "loss": 0.09853613376617432, "step": 2294 }, { "epoch": 0.5957943925233645, "grad_norm": 4.400768280029297, "learning_rate": 4.206072091601406e-06, "loss": 0.15529164671897888, "step": 2295 }, { "epoch": 0.5960539979231568, "grad_norm": 4.04338264465332, "learning_rate": 4.2053507550224224e-06, "loss": 0.12844771146774292, "step": 2296 }, { "epoch": 0.5963136033229491, "grad_norm": 2.184166431427002, "learning_rate": 4.204629161170747e-06, "loss": 0.10399468243122101, "step": 2297 }, { "epoch": 0.5965732087227414, "grad_norm": 2.913213014602661, "learning_rate": 4.2039073101739434e-06, "loss": 0.1023319810628891, "step": 2298 }, { "epoch": 0.5968328141225338, "grad_norm": 1.2399955987930298, "learning_rate": 4.203185202159617e-06, "loss": 0.10579229891300201, "step": 2299 }, { "epoch": 0.5970924195223261, "grad_norm": 1.683995008468628, "learning_rate": 4.202462837255421e-06, "loss": 0.10815688222646713, "step": 2300 }, { "epoch": 0.5973520249221184, "grad_norm": 1.6725739240646362, "learning_rate": 4.201740215589051e-06, "loss": 0.10627228766679764, "step": 2301 }, { "epoch": 0.5976116303219107, "grad_norm": 6.613651752471924, "learning_rate": 4.201017337288253e-06, "loss": 0.11981799453496933, "step": 2302 }, { "epoch": 0.597871235721703, "grad_norm": 2.1106624603271484, "learning_rate": 4.2002942024808125e-06, "loss": 0.1229412853717804, "step": 2303 }, { "epoch": 0.5981308411214953, "grad_norm": 1.0883939266204834, "learning_rate": 4.199570811294564e-06, "loss": 0.09731995314359665, "step": 2304 }, { "epoch": 0.5983904465212876, "grad_norm": 0.9696819186210632, "learning_rate": 4.198847163857388e-06, "loss": 0.1288156807422638, "step": 2305 }, { "epoch": 0.59865005192108, "grad_norm": 3.7976412773132324, "learning_rate": 4.1981232602972075e-06, "loss": 0.10268138349056244, "step": 2306 }, { "epoch": 0.5989096573208723, "grad_norm": 1.7860389947891235, "learning_rate": 4.197399100741991e-06, "loss": 0.11203936487436295, "step": 2307 }, { "epoch": 0.5991692627206646, "grad_norm": 3.127276659011841, "learning_rate": 4.196674685319755e-06, "loss": 0.14103715121746063, "step": 2308 }, { "epoch": 0.5994288681204569, "grad_norm": 1.2885689735412598, "learning_rate": 4.19595001415856e-06, "loss": 0.11049780249595642, "step": 2309 }, { "epoch": 0.5996884735202492, "grad_norm": 1.5126214027404785, "learning_rate": 4.19522508738651e-06, "loss": 0.12195058166980743, "step": 2310 }, { "epoch": 0.5999480789200415, "grad_norm": 1.7269829511642456, "learning_rate": 4.194499905131755e-06, "loss": 0.11263781785964966, "step": 2311 }, { "epoch": 0.6002076843198338, "grad_norm": 1.9149593114852905, "learning_rate": 4.1937744675224925e-06, "loss": 0.11154990643262863, "step": 2312 }, { "epoch": 0.6004672897196262, "grad_norm": 1.6166435480117798, "learning_rate": 4.193048774686961e-06, "loss": 0.107577383518219, "step": 2313 }, { "epoch": 0.6007268951194185, "grad_norm": 3.5099854469299316, "learning_rate": 4.192322826753449e-06, "loss": 0.08871482312679291, "step": 2314 }, { "epoch": 0.6009865005192108, "grad_norm": 1.6384830474853516, "learning_rate": 4.1915966238502855e-06, "loss": 0.12016880512237549, "step": 2315 }, { "epoch": 0.6012461059190031, "grad_norm": 2.5616087913513184, "learning_rate": 4.1908701661058465e-06, "loss": 0.08494527637958527, "step": 2316 }, { "epoch": 0.6015057113187954, "grad_norm": 3.2135417461395264, "learning_rate": 4.190143453648554e-06, "loss": 0.11099688708782196, "step": 2317 }, { "epoch": 0.6017653167185878, "grad_norm": 3.466843843460083, "learning_rate": 4.1894164866068745e-06, "loss": 0.09915198385715485, "step": 2318 }, { "epoch": 0.6020249221183801, "grad_norm": 1.6789419651031494, "learning_rate": 4.188689265109318e-06, "loss": 0.11048753559589386, "step": 2319 }, { "epoch": 0.6022845275181724, "grad_norm": 2.4356284141540527, "learning_rate": 4.187961789284441e-06, "loss": 0.13741931319236755, "step": 2320 }, { "epoch": 0.6025441329179647, "grad_norm": 2.730924129486084, "learning_rate": 4.187234059260845e-06, "loss": 0.10589483380317688, "step": 2321 }, { "epoch": 0.602803738317757, "grad_norm": 3.9635610580444336, "learning_rate": 4.186506075167174e-06, "loss": 0.15282946825027466, "step": 2322 }, { "epoch": 0.6030633437175493, "grad_norm": 1.4319742918014526, "learning_rate": 4.1857778371321225e-06, "loss": 0.12609155476093292, "step": 2323 }, { "epoch": 0.6033229491173416, "grad_norm": 3.459131956100464, "learning_rate": 4.185049345284423e-06, "loss": 0.14902636408805847, "step": 2324 }, { "epoch": 0.6035825545171339, "grad_norm": 0.8590538501739502, "learning_rate": 4.184320599752858e-06, "loss": 0.11107891798019409, "step": 2325 }, { "epoch": 0.6038421599169262, "grad_norm": 2.01106595993042, "learning_rate": 4.183591600666253e-06, "loss": 0.1062285304069519, "step": 2326 }, { "epoch": 0.6041017653167186, "grad_norm": 2.426548719406128, "learning_rate": 4.182862348153478e-06, "loss": 0.09381373971700668, "step": 2327 }, { "epoch": 0.6043613707165109, "grad_norm": 3.0485892295837402, "learning_rate": 4.182132842343448e-06, "loss": 0.14089378714561462, "step": 2328 }, { "epoch": 0.6046209761163032, "grad_norm": 2.5609214305877686, "learning_rate": 4.181403083365122e-06, "loss": 0.07632999122142792, "step": 2329 }, { "epoch": 0.6048805815160956, "grad_norm": 1.6398981809616089, "learning_rate": 4.180673071347506e-06, "loss": 0.11096066236495972, "step": 2330 }, { "epoch": 0.6051401869158879, "grad_norm": 2.810267925262451, "learning_rate": 4.17994280641965e-06, "loss": 0.1342170387506485, "step": 2331 }, { "epoch": 0.6053997923156802, "grad_norm": 1.115018367767334, "learning_rate": 4.179212288710646e-06, "loss": 0.11738033592700958, "step": 2332 }, { "epoch": 0.6056593977154725, "grad_norm": 3.5168821811676025, "learning_rate": 4.178481518349634e-06, "loss": 0.12901544570922852, "step": 2333 }, { "epoch": 0.6059190031152648, "grad_norm": 1.2770525217056274, "learning_rate": 4.1777504954657975e-06, "loss": 0.1368236243724823, "step": 2334 }, { "epoch": 0.6061786085150571, "grad_norm": 2.6845991611480713, "learning_rate": 4.1770192201883656e-06, "loss": 0.1326020210981369, "step": 2335 }, { "epoch": 0.6064382139148494, "grad_norm": 3.6163129806518555, "learning_rate": 4.17628769264661e-06, "loss": 0.13994266092777252, "step": 2336 }, { "epoch": 0.6066978193146417, "grad_norm": 3.393641471862793, "learning_rate": 4.175555912969846e-06, "loss": 0.13585346937179565, "step": 2337 }, { "epoch": 0.606957424714434, "grad_norm": 0.9932310581207275, "learning_rate": 4.17482388128744e-06, "loss": 0.10121428966522217, "step": 2338 }, { "epoch": 0.6072170301142263, "grad_norm": 2.6496219635009766, "learning_rate": 4.174091597728793e-06, "loss": 0.13067948818206787, "step": 2339 }, { "epoch": 0.6074766355140186, "grad_norm": 2.6408801078796387, "learning_rate": 4.173359062423361e-06, "loss": 0.11043630540370941, "step": 2340 }, { "epoch": 0.607736240913811, "grad_norm": 1.121127724647522, "learning_rate": 4.172626275500637e-06, "loss": 0.11497874557971954, "step": 2341 }, { "epoch": 0.6079958463136034, "grad_norm": 1.4335216283798218, "learning_rate": 4.171893237090161e-06, "loss": 0.12105932831764221, "step": 2342 }, { "epoch": 0.6082554517133957, "grad_norm": 2.682232141494751, "learning_rate": 4.171159947321519e-06, "loss": 0.10390397906303406, "step": 2343 }, { "epoch": 0.608515057113188, "grad_norm": 1.7509633302688599, "learning_rate": 4.170426406324337e-06, "loss": 0.11217943578958511, "step": 2344 }, { "epoch": 0.6087746625129803, "grad_norm": 1.2895848751068115, "learning_rate": 4.16969261422829e-06, "loss": 0.1028963029384613, "step": 2345 }, { "epoch": 0.6090342679127726, "grad_norm": 0.9471580982208252, "learning_rate": 4.168958571163097e-06, "loss": 0.12043648213148117, "step": 2346 }, { "epoch": 0.6092938733125649, "grad_norm": 2.980677843093872, "learning_rate": 4.168224277258517e-06, "loss": 0.11261536180973053, "step": 2347 }, { "epoch": 0.6095534787123572, "grad_norm": 2.94033145904541, "learning_rate": 4.167489732644358e-06, "loss": 0.15828722715377808, "step": 2348 }, { "epoch": 0.6098130841121495, "grad_norm": 2.844053268432617, "learning_rate": 4.166754937450471e-06, "loss": 0.1021883636713028, "step": 2349 }, { "epoch": 0.6100726895119418, "grad_norm": 2.64320707321167, "learning_rate": 4.1660198918067504e-06, "loss": 0.10242857784032822, "step": 2350 }, { "epoch": 0.6103322949117341, "grad_norm": 3.253553867340088, "learning_rate": 4.165284595843135e-06, "loss": 0.11243066936731339, "step": 2351 }, { "epoch": 0.6105919003115264, "grad_norm": 4.171022415161133, "learning_rate": 4.1645490496896085e-06, "loss": 0.11146079748868942, "step": 2352 }, { "epoch": 0.6108515057113187, "grad_norm": 2.8368079662323, "learning_rate": 4.1638132534762e-06, "loss": 0.08520844578742981, "step": 2353 }, { "epoch": 0.6111111111111112, "grad_norm": 2.1902964115142822, "learning_rate": 4.163077207332979e-06, "loss": 0.12316787242889404, "step": 2354 }, { "epoch": 0.6113707165109035, "grad_norm": 1.8582535982131958, "learning_rate": 4.1623409113900636e-06, "loss": 0.1217663511633873, "step": 2355 }, { "epoch": 0.6116303219106958, "grad_norm": 1.1217025518417358, "learning_rate": 4.161604365777613e-06, "loss": 0.1130741685628891, "step": 2356 }, { "epoch": 0.6118899273104881, "grad_norm": 4.9271321296691895, "learning_rate": 4.160867570625833e-06, "loss": 0.10631755739450455, "step": 2357 }, { "epoch": 0.6121495327102804, "grad_norm": 2.753629207611084, "learning_rate": 4.16013052606497e-06, "loss": 0.11592347174882889, "step": 2358 }, { "epoch": 0.6124091381100727, "grad_norm": 1.5429620742797852, "learning_rate": 4.159393232225318e-06, "loss": 0.11265977472066879, "step": 2359 }, { "epoch": 0.612668743509865, "grad_norm": 2.7995641231536865, "learning_rate": 4.1586556892372135e-06, "loss": 0.13083896040916443, "step": 2360 }, { "epoch": 0.6129283489096573, "grad_norm": 1.1026923656463623, "learning_rate": 4.157917897231037e-06, "loss": 0.10630190372467041, "step": 2361 }, { "epoch": 0.6131879543094496, "grad_norm": 2.874312162399292, "learning_rate": 4.157179856337214e-06, "loss": 0.1285717785358429, "step": 2362 }, { "epoch": 0.6134475597092419, "grad_norm": 2.2944414615631104, "learning_rate": 4.156441566686212e-06, "loss": 0.10695111751556396, "step": 2363 }, { "epoch": 0.6137071651090342, "grad_norm": 2.964691162109375, "learning_rate": 4.155703028408544e-06, "loss": 0.1322634518146515, "step": 2364 }, { "epoch": 0.6139667705088265, "grad_norm": 2.470837116241455, "learning_rate": 4.1549642416347676e-06, "loss": 0.12746615707874298, "step": 2365 }, { "epoch": 0.614226375908619, "grad_norm": 2.0078203678131104, "learning_rate": 4.154225206495482e-06, "loss": 0.11011289060115814, "step": 2366 }, { "epoch": 0.6144859813084113, "grad_norm": 2.4442567825317383, "learning_rate": 4.153485923121332e-06, "loss": 0.11510468274354935, "step": 2367 }, { "epoch": 0.6147455867082036, "grad_norm": 2.837341547012329, "learning_rate": 4.152746391643008e-06, "loss": 0.11706633865833282, "step": 2368 }, { "epoch": 0.6150051921079959, "grad_norm": 3.4113314151763916, "learning_rate": 4.152006612191239e-06, "loss": 0.09511412680149078, "step": 2369 }, { "epoch": 0.6152647975077882, "grad_norm": 2.4313228130340576, "learning_rate": 4.151266584896803e-06, "loss": 0.11326707899570465, "step": 2370 }, { "epoch": 0.6155244029075805, "grad_norm": 2.0722970962524414, "learning_rate": 4.1505263098905196e-06, "loss": 0.11837894469499588, "step": 2371 }, { "epoch": 0.6157840083073728, "grad_norm": 2.215348720550537, "learning_rate": 4.149785787303253e-06, "loss": 0.12628358602523804, "step": 2372 }, { "epoch": 0.6160436137071651, "grad_norm": 2.082366466522217, "learning_rate": 4.14904501726591e-06, "loss": 0.12109658122062683, "step": 2373 }, { "epoch": 0.6163032191069574, "grad_norm": 2.6438920497894287, "learning_rate": 4.148303999909441e-06, "loss": 0.11130949854850769, "step": 2374 }, { "epoch": 0.6165628245067497, "grad_norm": 2.6695291996002197, "learning_rate": 4.147562735364842e-06, "loss": 0.10980008542537689, "step": 2375 }, { "epoch": 0.616822429906542, "grad_norm": 1.277039885520935, "learning_rate": 4.146821223763151e-06, "loss": 0.11415329575538635, "step": 2376 }, { "epoch": 0.6170820353063343, "grad_norm": 1.5030258893966675, "learning_rate": 4.146079465235452e-06, "loss": 0.09610656648874283, "step": 2377 }, { "epoch": 0.6173416407061267, "grad_norm": 1.205726981163025, "learning_rate": 4.145337459912868e-06, "loss": 0.12259785830974579, "step": 2378 }, { "epoch": 0.617601246105919, "grad_norm": 1.1831632852554321, "learning_rate": 4.144595207926571e-06, "loss": 0.10067649185657501, "step": 2379 }, { "epoch": 0.6178608515057114, "grad_norm": 1.7338058948516846, "learning_rate": 4.143852709407773e-06, "loss": 0.09020690619945526, "step": 2380 }, { "epoch": 0.6181204569055037, "grad_norm": 1.6217554807662964, "learning_rate": 4.143109964487731e-06, "loss": 0.11287280172109604, "step": 2381 }, { "epoch": 0.618380062305296, "grad_norm": 1.8206324577331543, "learning_rate": 4.1423669732977455e-06, "loss": 0.09938935935497284, "step": 2382 }, { "epoch": 0.6186396677050883, "grad_norm": 1.912185788154602, "learning_rate": 4.1416237359691595e-06, "loss": 0.10576717555522919, "step": 2383 }, { "epoch": 0.6188992731048806, "grad_norm": 3.187006950378418, "learning_rate": 4.140880252633362e-06, "loss": 0.1221565306186676, "step": 2384 }, { "epoch": 0.6191588785046729, "grad_norm": 3.7202212810516357, "learning_rate": 4.140136523421782e-06, "loss": 0.1314665675163269, "step": 2385 }, { "epoch": 0.6194184839044652, "grad_norm": 1.2421488761901855, "learning_rate": 4.139392548465895e-06, "loss": 0.11505751311779022, "step": 2386 }, { "epoch": 0.6196780893042575, "grad_norm": 4.554296016693115, "learning_rate": 4.138648327897216e-06, "loss": 0.10136662423610687, "step": 2387 }, { "epoch": 0.6199376947040498, "grad_norm": 1.5339739322662354, "learning_rate": 4.137903861847311e-06, "loss": 0.106984943151474, "step": 2388 }, { "epoch": 0.6201973001038421, "grad_norm": 0.9875327944755554, "learning_rate": 4.137159150447782e-06, "loss": 0.11349909007549286, "step": 2389 }, { "epoch": 0.6204569055036345, "grad_norm": 3.8345134258270264, "learning_rate": 4.136414193830276e-06, "loss": 0.1114337369799614, "step": 2390 }, { "epoch": 0.6207165109034268, "grad_norm": 4.291127681732178, "learning_rate": 4.135668992126485e-06, "loss": 0.12058256566524506, "step": 2391 }, { "epoch": 0.6209761163032191, "grad_norm": 2.8308770656585693, "learning_rate": 4.134923545468144e-06, "loss": 0.1295843869447708, "step": 2392 }, { "epoch": 0.6212357217030114, "grad_norm": 0.8603518605232239, "learning_rate": 4.134177853987032e-06, "loss": 0.11609930545091629, "step": 2393 }, { "epoch": 0.6214953271028038, "grad_norm": 2.4153549671173096, "learning_rate": 4.133431917814967e-06, "loss": 0.10646555572748184, "step": 2394 }, { "epoch": 0.6217549325025961, "grad_norm": 1.6794886589050293, "learning_rate": 4.132685737083817e-06, "loss": 0.11292620003223419, "step": 2395 }, { "epoch": 0.6220145379023884, "grad_norm": 2.7246270179748535, "learning_rate": 4.131939311925486e-06, "loss": 0.08066504448652267, "step": 2396 }, { "epoch": 0.6222741433021807, "grad_norm": 2.5227835178375244, "learning_rate": 4.131192642471929e-06, "loss": 0.10102330148220062, "step": 2397 }, { "epoch": 0.622533748701973, "grad_norm": 1.0160808563232422, "learning_rate": 4.130445728855137e-06, "loss": 0.11593824625015259, "step": 2398 }, { "epoch": 0.6227933541017653, "grad_norm": 0.7993999123573303, "learning_rate": 4.129698571207148e-06, "loss": 0.12138842791318893, "step": 2399 }, { "epoch": 0.6230529595015576, "grad_norm": 2.639807939529419, "learning_rate": 4.128951169660043e-06, "loss": 0.09931397438049316, "step": 2400 }, { "epoch": 0.6233125649013499, "grad_norm": 2.710111618041992, "learning_rate": 4.1282035243459446e-06, "loss": 0.10452260076999664, "step": 2401 }, { "epoch": 0.6235721703011423, "grad_norm": 2.2584381103515625, "learning_rate": 4.1274556353970196e-06, "loss": 0.11975239217281342, "step": 2402 }, { "epoch": 0.6238317757009346, "grad_norm": 1.9669506549835205, "learning_rate": 4.126707502945478e-06, "loss": 0.11609973013401031, "step": 2403 }, { "epoch": 0.6240913811007269, "grad_norm": 2.423336982727051, "learning_rate": 4.125959127123572e-06, "loss": 0.10506601631641388, "step": 2404 }, { "epoch": 0.6243509865005192, "grad_norm": 1.0578587055206299, "learning_rate": 4.125210508063599e-06, "loss": 0.10567115992307663, "step": 2405 }, { "epoch": 0.6246105919003115, "grad_norm": 1.284490704536438, "learning_rate": 4.1244616458978945e-06, "loss": 0.10633797943592072, "step": 2406 }, { "epoch": 0.6248701973001038, "grad_norm": 1.3499926328659058, "learning_rate": 4.1237125407588434e-06, "loss": 0.10948944091796875, "step": 2407 }, { "epoch": 0.6251298026998962, "grad_norm": 2.823208808898926, "learning_rate": 4.1229631927788684e-06, "loss": 0.11397570371627808, "step": 2408 }, { "epoch": 0.6253894080996885, "grad_norm": 3.132354974746704, "learning_rate": 4.122213602090438e-06, "loss": 0.1101984828710556, "step": 2409 }, { "epoch": 0.6256490134994808, "grad_norm": 1.8308669328689575, "learning_rate": 4.121463768826061e-06, "loss": 0.11270247399806976, "step": 2410 }, { "epoch": 0.6259086188992731, "grad_norm": 3.437232494354248, "learning_rate": 4.120713693118293e-06, "loss": 0.12771230936050415, "step": 2411 }, { "epoch": 0.6261682242990654, "grad_norm": 3.506573438644409, "learning_rate": 4.119963375099729e-06, "loss": 0.13248470425605774, "step": 2412 }, { "epoch": 0.6264278296988577, "grad_norm": 2.5936295986175537, "learning_rate": 4.119212814903009e-06, "loss": 0.12110534310340881, "step": 2413 }, { "epoch": 0.6266874350986501, "grad_norm": 1.6743969917297363, "learning_rate": 4.118462012660813e-06, "loss": 0.10701224207878113, "step": 2414 }, { "epoch": 0.6269470404984424, "grad_norm": 2.3279869556427, "learning_rate": 4.117710968505867e-06, "loss": 0.10701055824756622, "step": 2415 }, { "epoch": 0.6272066458982347, "grad_norm": 1.6287683248519897, "learning_rate": 4.116959682570938e-06, "loss": 0.11649294942617416, "step": 2416 }, { "epoch": 0.627466251298027, "grad_norm": 2.472733497619629, "learning_rate": 4.1162081549888375e-06, "loss": 0.12135770916938782, "step": 2417 }, { "epoch": 0.6277258566978193, "grad_norm": 2.732025146484375, "learning_rate": 4.1154563858924164e-06, "loss": 0.1081213653087616, "step": 2418 }, { "epoch": 0.6279854620976116, "grad_norm": 4.4107747077941895, "learning_rate": 4.114704375414572e-06, "loss": 0.1304505467414856, "step": 2419 }, { "epoch": 0.6282450674974039, "grad_norm": 3.122053384780884, "learning_rate": 4.1139521236882405e-06, "loss": 0.10321103781461716, "step": 2420 }, { "epoch": 0.6285046728971962, "grad_norm": 2.827517032623291, "learning_rate": 4.113199630846405e-06, "loss": 0.11645980179309845, "step": 2421 }, { "epoch": 0.6287642782969886, "grad_norm": 2.5741124153137207, "learning_rate": 4.112446897022086e-06, "loss": 0.11344864964485168, "step": 2422 }, { "epoch": 0.6290238836967809, "grad_norm": 2.195220470428467, "learning_rate": 4.111693922348353e-06, "loss": 0.09869270026683807, "step": 2423 }, { "epoch": 0.6292834890965732, "grad_norm": 2.7201461791992188, "learning_rate": 4.110940706958313e-06, "loss": 0.10414762049913406, "step": 2424 }, { "epoch": 0.6295430944963655, "grad_norm": 0.9239470362663269, "learning_rate": 4.110187250985117e-06, "loss": 0.11266662180423737, "step": 2425 }, { "epoch": 0.6298026998961579, "grad_norm": 3.820622444152832, "learning_rate": 4.10943355456196e-06, "loss": 0.12390120327472687, "step": 2426 }, { "epoch": 0.6300623052959502, "grad_norm": 1.2812671661376953, "learning_rate": 4.108679617822077e-06, "loss": 0.11669141054153442, "step": 2427 }, { "epoch": 0.6303219106957425, "grad_norm": 1.0921603441238403, "learning_rate": 4.107925440898747e-06, "loss": 0.11939734220504761, "step": 2428 }, { "epoch": 0.6305815160955348, "grad_norm": 2.5453226566314697, "learning_rate": 4.107171023925292e-06, "loss": 0.12583093345165253, "step": 2429 }, { "epoch": 0.6308411214953271, "grad_norm": 1.4600743055343628, "learning_rate": 4.106416367035075e-06, "loss": 0.1242239773273468, "step": 2430 }, { "epoch": 0.6311007268951194, "grad_norm": 3.243936538696289, "learning_rate": 4.1056614703615015e-06, "loss": 0.09874925017356873, "step": 2431 }, { "epoch": 0.6313603322949117, "grad_norm": 3.1706011295318604, "learning_rate": 4.1049063340380205e-06, "loss": 0.07160268723964691, "step": 2432 }, { "epoch": 0.631619937694704, "grad_norm": 1.2224576473236084, "learning_rate": 4.104150958198123e-06, "loss": 0.1135963499546051, "step": 2433 }, { "epoch": 0.6318795430944963, "grad_norm": 4.158326625823975, "learning_rate": 4.103395342975343e-06, "loss": 0.11739329993724823, "step": 2434 }, { "epoch": 0.6321391484942886, "grad_norm": 1.1308174133300781, "learning_rate": 4.102639488503254e-06, "loss": 0.10712175071239471, "step": 2435 }, { "epoch": 0.632398753894081, "grad_norm": 2.8410937786102295, "learning_rate": 4.101883394915475e-06, "loss": 0.0582936629652977, "step": 2436 }, { "epoch": 0.6326583592938733, "grad_norm": 3.3234920501708984, "learning_rate": 4.101127062345666e-06, "loss": 0.10853930562734604, "step": 2437 }, { "epoch": 0.6329179646936657, "grad_norm": 3.366262912750244, "learning_rate": 4.100370490927529e-06, "loss": 0.12845349311828613, "step": 2438 }, { "epoch": 0.633177570093458, "grad_norm": 2.8844964504241943, "learning_rate": 4.0996136807948085e-06, "loss": 0.11855687946081161, "step": 2439 }, { "epoch": 0.6334371754932503, "grad_norm": 1.9488321542739868, "learning_rate": 4.0988566320812915e-06, "loss": 0.09904433786869049, "step": 2440 }, { "epoch": 0.6336967808930426, "grad_norm": 1.2793165445327759, "learning_rate": 4.098099344920806e-06, "loss": 0.09893874824047089, "step": 2441 }, { "epoch": 0.6339563862928349, "grad_norm": 2.7927470207214355, "learning_rate": 4.097341819447224e-06, "loss": 0.1333141028881073, "step": 2442 }, { "epoch": 0.6342159916926272, "grad_norm": 2.63271164894104, "learning_rate": 4.096584055794459e-06, "loss": 0.09080912172794342, "step": 2443 }, { "epoch": 0.6344755970924195, "grad_norm": 3.1120123863220215, "learning_rate": 4.095826054096465e-06, "loss": 0.11244494467973709, "step": 2444 }, { "epoch": 0.6347352024922118, "grad_norm": 3.9722208976745605, "learning_rate": 4.09506781448724e-06, "loss": 0.12820738554000854, "step": 2445 }, { "epoch": 0.6349948078920041, "grad_norm": 3.189586877822876, "learning_rate": 4.094309337100823e-06, "loss": 0.11254953593015671, "step": 2446 }, { "epoch": 0.6352544132917964, "grad_norm": 2.1161677837371826, "learning_rate": 4.093550622071296e-06, "loss": 0.10774984955787659, "step": 2447 }, { "epoch": 0.6355140186915887, "grad_norm": 4.007640361785889, "learning_rate": 4.092791669532784e-06, "loss": 0.10942401736974716, "step": 2448 }, { "epoch": 0.635773624091381, "grad_norm": 0.9504082798957825, "learning_rate": 4.0920324796194485e-06, "loss": 0.11885164678096771, "step": 2449 }, { "epoch": 0.6360332294911735, "grad_norm": 2.6990346908569336, "learning_rate": 4.091273052465501e-06, "loss": 0.12527696788311005, "step": 2450 }, { "epoch": 0.6362928348909658, "grad_norm": 2.0620739459991455, "learning_rate": 4.090513388205187e-06, "loss": 0.10673211514949799, "step": 2451 }, { "epoch": 0.6365524402907581, "grad_norm": 3.048065185546875, "learning_rate": 4.0897534869728e-06, "loss": 0.09092481434345245, "step": 2452 }, { "epoch": 0.6368120456905504, "grad_norm": 3.0410969257354736, "learning_rate": 4.088993348902673e-06, "loss": 0.1088966578245163, "step": 2453 }, { "epoch": 0.6370716510903427, "grad_norm": 3.68351149559021, "learning_rate": 4.0882329741291815e-06, "loss": 0.1585153043270111, "step": 2454 }, { "epoch": 0.637331256490135, "grad_norm": 1.1922258138656616, "learning_rate": 4.087472362786741e-06, "loss": 0.11275622248649597, "step": 2455 }, { "epoch": 0.6375908618899273, "grad_norm": 0.9668871164321899, "learning_rate": 4.086711515009811e-06, "loss": 0.11813725531101227, "step": 2456 }, { "epoch": 0.6378504672897196, "grad_norm": 1.7933844327926636, "learning_rate": 4.085950430932892e-06, "loss": 0.12449765205383301, "step": 2457 }, { "epoch": 0.6381100726895119, "grad_norm": 1.1045253276824951, "learning_rate": 4.085189110690526e-06, "loss": 0.10964415967464447, "step": 2458 }, { "epoch": 0.6383696780893042, "grad_norm": 2.5226919651031494, "learning_rate": 4.084427554417296e-06, "loss": 0.09667417407035828, "step": 2459 }, { "epoch": 0.6386292834890965, "grad_norm": 1.6698132753372192, "learning_rate": 4.08366576224783e-06, "loss": 0.09741085022687912, "step": 2460 }, { "epoch": 0.6388888888888888, "grad_norm": 1.9673842191696167, "learning_rate": 4.082903734316793e-06, "loss": 0.1330081969499588, "step": 2461 }, { "epoch": 0.6391484942886813, "grad_norm": 2.457829475402832, "learning_rate": 4.082141470758896e-06, "loss": 0.1100817322731018, "step": 2462 }, { "epoch": 0.6394080996884736, "grad_norm": 2.935563564300537, "learning_rate": 4.081378971708889e-06, "loss": 0.09471064805984497, "step": 2463 }, { "epoch": 0.6396677050882659, "grad_norm": 4.142972469329834, "learning_rate": 4.080616237301563e-06, "loss": 0.1275705099105835, "step": 2464 }, { "epoch": 0.6399273104880582, "grad_norm": 1.1354490518569946, "learning_rate": 4.079853267671755e-06, "loss": 0.10056313872337341, "step": 2465 }, { "epoch": 0.6401869158878505, "grad_norm": 3.316654920578003, "learning_rate": 4.079090062954339e-06, "loss": 0.07178083062171936, "step": 2466 }, { "epoch": 0.6404465212876428, "grad_norm": 1.497634768486023, "learning_rate": 4.078326623284231e-06, "loss": 0.11849270761013031, "step": 2467 }, { "epoch": 0.6407061266874351, "grad_norm": 4.547255516052246, "learning_rate": 4.077562948796391e-06, "loss": 0.12534306943416595, "step": 2468 }, { "epoch": 0.6409657320872274, "grad_norm": 2.0811660289764404, "learning_rate": 4.076799039625819e-06, "loss": 0.07711420953273773, "step": 2469 }, { "epoch": 0.6412253374870197, "grad_norm": 2.832413673400879, "learning_rate": 4.076034895907556e-06, "loss": 0.12445701658725739, "step": 2470 }, { "epoch": 0.641484942886812, "grad_norm": 3.2347054481506348, "learning_rate": 4.075270517776686e-06, "loss": 0.11404621601104736, "step": 2471 }, { "epoch": 0.6417445482866043, "grad_norm": 2.580995559692383, "learning_rate": 4.0745059053683325e-06, "loss": 0.07827005535364151, "step": 2472 }, { "epoch": 0.6420041536863966, "grad_norm": 1.2138116359710693, "learning_rate": 4.073741058817663e-06, "loss": 0.11065740138292313, "step": 2473 }, { "epoch": 0.642263759086189, "grad_norm": 1.2020912170410156, "learning_rate": 4.072975978259883e-06, "loss": 0.10474913567304611, "step": 2474 }, { "epoch": 0.6425233644859814, "grad_norm": 3.335925340652466, "learning_rate": 4.072210663830243e-06, "loss": 0.08536167442798615, "step": 2475 }, { "epoch": 0.6427829698857737, "grad_norm": 2.6854498386383057, "learning_rate": 4.0714451156640314e-06, "loss": 0.13123324513435364, "step": 2476 }, { "epoch": 0.643042575285566, "grad_norm": 2.883512496948242, "learning_rate": 4.070679333896582e-06, "loss": 0.11587280035018921, "step": 2477 }, { "epoch": 0.6433021806853583, "grad_norm": 2.592310905456543, "learning_rate": 4.069913318663264e-06, "loss": 0.12261156737804413, "step": 2478 }, { "epoch": 0.6435617860851506, "grad_norm": 1.8025580644607544, "learning_rate": 4.069147070099495e-06, "loss": 0.11490131914615631, "step": 2479 }, { "epoch": 0.6438213914849429, "grad_norm": 0.7918742299079895, "learning_rate": 4.068380588340728e-06, "loss": 0.10731633007526398, "step": 2480 }, { "epoch": 0.6440809968847352, "grad_norm": 1.266193151473999, "learning_rate": 4.067613873522459e-06, "loss": 0.11669045686721802, "step": 2481 }, { "epoch": 0.6443406022845275, "grad_norm": 4.566499710083008, "learning_rate": 4.066846925780226e-06, "loss": 0.10181733965873718, "step": 2482 }, { "epoch": 0.6446002076843198, "grad_norm": 1.4927541017532349, "learning_rate": 4.06607974524961e-06, "loss": 0.11173133552074432, "step": 2483 }, { "epoch": 0.6448598130841121, "grad_norm": 1.5140697956085205, "learning_rate": 4.065312332066229e-06, "loss": 0.1220349371433258, "step": 2484 }, { "epoch": 0.6451194184839044, "grad_norm": 0.8532938957214355, "learning_rate": 4.064544686365744e-06, "loss": 0.10836829245090485, "step": 2485 }, { "epoch": 0.6453790238836968, "grad_norm": 3.4011402130126953, "learning_rate": 4.063776808283857e-06, "loss": 0.10736466944217682, "step": 2486 }, { "epoch": 0.6456386292834891, "grad_norm": 1.9608283042907715, "learning_rate": 4.0630086979563125e-06, "loss": 0.10534514486789703, "step": 2487 }, { "epoch": 0.6458982346832814, "grad_norm": 1.2769083976745605, "learning_rate": 4.062240355518895e-06, "loss": 0.10836469382047653, "step": 2488 }, { "epoch": 0.6461578400830738, "grad_norm": 0.929313063621521, "learning_rate": 4.061471781107427e-06, "loss": 0.11626248061656952, "step": 2489 }, { "epoch": 0.6464174454828661, "grad_norm": 2.2555954456329346, "learning_rate": 4.060702974857778e-06, "loss": 0.14371046423912048, "step": 2490 }, { "epoch": 0.6466770508826584, "grad_norm": 1.467341661453247, "learning_rate": 4.059933936905854e-06, "loss": 0.11852259188890457, "step": 2491 }, { "epoch": 0.6469366562824507, "grad_norm": 1.8376678228378296, "learning_rate": 4.059164667387603e-06, "loss": 0.11664407700300217, "step": 2492 }, { "epoch": 0.647196261682243, "grad_norm": 1.7390687465667725, "learning_rate": 4.058395166439015e-06, "loss": 0.10338589549064636, "step": 2493 }, { "epoch": 0.6474558670820353, "grad_norm": 1.7700591087341309, "learning_rate": 4.05762543419612e-06, "loss": 0.13265152275562286, "step": 2494 }, { "epoch": 0.6477154724818276, "grad_norm": 0.9675712585449219, "learning_rate": 4.056855470794988e-06, "loss": 0.12417837977409363, "step": 2495 }, { "epoch": 0.6479750778816199, "grad_norm": 0.7276498675346375, "learning_rate": 4.0560852763717335e-06, "loss": 0.10509141534566879, "step": 2496 }, { "epoch": 0.6482346832814122, "grad_norm": 0.83101487159729, "learning_rate": 4.0553148510625065e-06, "loss": 0.10136529803276062, "step": 2497 }, { "epoch": 0.6484942886812046, "grad_norm": 1.9112061262130737, "learning_rate": 4.054544195003502e-06, "loss": 0.10036835074424744, "step": 2498 }, { "epoch": 0.6487538940809969, "grad_norm": 1.388360857963562, "learning_rate": 4.053773308330953e-06, "loss": 0.1285250186920166, "step": 2499 }, { "epoch": 0.6490134994807892, "grad_norm": 1.1283316612243652, "learning_rate": 4.053002191181137e-06, "loss": 0.11163295060396194, "step": 2500 }, { "epoch": 0.6492731048805815, "grad_norm": 2.8093628883361816, "learning_rate": 4.052230843690367e-06, "loss": 0.08926809579133987, "step": 2501 }, { "epoch": 0.6495327102803738, "grad_norm": 2.5296170711517334, "learning_rate": 4.051459265995001e-06, "loss": 0.13116809725761414, "step": 2502 }, { "epoch": 0.6497923156801662, "grad_norm": 3.0098860263824463, "learning_rate": 4.050687458231437e-06, "loss": 0.08985130488872528, "step": 2503 }, { "epoch": 0.6500519210799585, "grad_norm": 1.2341980934143066, "learning_rate": 4.049915420536112e-06, "loss": 0.10115595906972885, "step": 2504 }, { "epoch": 0.6503115264797508, "grad_norm": 0.7517531514167786, "learning_rate": 4.049143153045505e-06, "loss": 0.10815861821174622, "step": 2505 }, { "epoch": 0.6505711318795431, "grad_norm": 1.1453818082809448, "learning_rate": 4.048370655896135e-06, "loss": 0.10577622056007385, "step": 2506 }, { "epoch": 0.6508307372793354, "grad_norm": 1.2276463508605957, "learning_rate": 4.047597929224561e-06, "loss": 0.11658546328544617, "step": 2507 }, { "epoch": 0.6510903426791277, "grad_norm": 2.5805366039276123, "learning_rate": 4.0468249731673845e-06, "loss": 0.09896531701087952, "step": 2508 }, { "epoch": 0.65134994807892, "grad_norm": 3.2110235691070557, "learning_rate": 4.046051787861247e-06, "loss": 0.12326984107494354, "step": 2509 }, { "epoch": 0.6516095534787124, "grad_norm": 1.1662342548370361, "learning_rate": 4.045278373442828e-06, "loss": 0.11595095694065094, "step": 2510 }, { "epoch": 0.6518691588785047, "grad_norm": 1.3729496002197266, "learning_rate": 4.04450473004885e-06, "loss": 0.10874217748641968, "step": 2511 }, { "epoch": 0.652128764278297, "grad_norm": 2.7599849700927734, "learning_rate": 4.043730857816077e-06, "loss": 0.12014112621545792, "step": 2512 }, { "epoch": 0.6523883696780893, "grad_norm": 0.9200965166091919, "learning_rate": 4.042956756881311e-06, "loss": 0.11230093240737915, "step": 2513 }, { "epoch": 0.6526479750778816, "grad_norm": 2.786323070526123, "learning_rate": 4.042182427381395e-06, "loss": 0.11034088581800461, "step": 2514 }, { "epoch": 0.6529075804776739, "grad_norm": 2.025259256362915, "learning_rate": 4.041407869453212e-06, "loss": 0.10200373828411102, "step": 2515 }, { "epoch": 0.6531671858774662, "grad_norm": 2.7179176807403564, "learning_rate": 4.0406330832336885e-06, "loss": 0.11223803460597992, "step": 2516 }, { "epoch": 0.6534267912772586, "grad_norm": 1.86257004737854, "learning_rate": 4.039858068859787e-06, "loss": 0.12235403805971146, "step": 2517 }, { "epoch": 0.6536863966770509, "grad_norm": 1.0650081634521484, "learning_rate": 4.039082826468511e-06, "loss": 0.11108312010765076, "step": 2518 }, { "epoch": 0.6539460020768432, "grad_norm": 1.0086982250213623, "learning_rate": 4.03830735619691e-06, "loss": 0.11308903992176056, "step": 2519 }, { "epoch": 0.6542056074766355, "grad_norm": 1.3967623710632324, "learning_rate": 4.037531658182065e-06, "loss": 0.11262474954128265, "step": 2520 }, { "epoch": 0.6544652128764278, "grad_norm": 1.5756263732910156, "learning_rate": 4.036755732561103e-06, "loss": 0.09819203615188599, "step": 2521 }, { "epoch": 0.6547248182762202, "grad_norm": 1.8330105543136597, "learning_rate": 4.035979579471192e-06, "loss": 0.12269338220357895, "step": 2522 }, { "epoch": 0.6549844236760125, "grad_norm": 1.2806183099746704, "learning_rate": 4.035203199049536e-06, "loss": 0.1059032529592514, "step": 2523 }, { "epoch": 0.6552440290758048, "grad_norm": 1.3368381261825562, "learning_rate": 4.0344265914333805e-06, "loss": 0.09786635637283325, "step": 2524 }, { "epoch": 0.6555036344755971, "grad_norm": 4.234493732452393, "learning_rate": 4.033649756760014e-06, "loss": 0.09888598322868347, "step": 2525 }, { "epoch": 0.6557632398753894, "grad_norm": 2.6742663383483887, "learning_rate": 4.032872695166763e-06, "loss": 0.1265220195055008, "step": 2526 }, { "epoch": 0.6560228452751817, "grad_norm": 1.7169300317764282, "learning_rate": 4.032095406790992e-06, "loss": 0.1068001389503479, "step": 2527 }, { "epoch": 0.656282450674974, "grad_norm": 3.0329105854034424, "learning_rate": 4.031317891770109e-06, "loss": 0.12530192732810974, "step": 2528 }, { "epoch": 0.6565420560747663, "grad_norm": 2.918457508087158, "learning_rate": 4.030540150241563e-06, "loss": 0.1287350058555603, "step": 2529 }, { "epoch": 0.6568016614745587, "grad_norm": 1.2256656885147095, "learning_rate": 4.029762182342836e-06, "loss": 0.11068689823150635, "step": 2530 }, { "epoch": 0.657061266874351, "grad_norm": 0.991975724697113, "learning_rate": 4.02898398821146e-06, "loss": 0.11475177854299545, "step": 2531 }, { "epoch": 0.6573208722741433, "grad_norm": 4.957997798919678, "learning_rate": 4.028205567984999e-06, "loss": 0.11444362998008728, "step": 2532 }, { "epoch": 0.6575804776739356, "grad_norm": 1.7151265144348145, "learning_rate": 4.0274269218010606e-06, "loss": 0.11199599504470825, "step": 2533 }, { "epoch": 0.657840083073728, "grad_norm": 1.0345284938812256, "learning_rate": 4.026648049797292e-06, "loss": 0.10761061310768127, "step": 2534 }, { "epoch": 0.6580996884735203, "grad_norm": 3.037505865097046, "learning_rate": 4.025868952111379e-06, "loss": 0.13051797449588776, "step": 2535 }, { "epoch": 0.6583592938733126, "grad_norm": 1.9683552980422974, "learning_rate": 4.025089628881049e-06, "loss": 0.11573933064937592, "step": 2536 }, { "epoch": 0.6586188992731049, "grad_norm": 1.8074074983596802, "learning_rate": 4.024310080244069e-06, "loss": 0.11245442926883698, "step": 2537 }, { "epoch": 0.6588785046728972, "grad_norm": 2.882195234298706, "learning_rate": 4.023530306338244e-06, "loss": 0.12413782626390457, "step": 2538 }, { "epoch": 0.6591381100726895, "grad_norm": 1.99857497215271, "learning_rate": 4.02275030730142e-06, "loss": 0.12345445901155472, "step": 2539 }, { "epoch": 0.6593977154724818, "grad_norm": 0.6250925660133362, "learning_rate": 4.021970083271486e-06, "loss": 0.11296898126602173, "step": 2540 }, { "epoch": 0.6596573208722741, "grad_norm": 1.4767870903015137, "learning_rate": 4.021189634386363e-06, "loss": 0.10344643890857697, "step": 2541 }, { "epoch": 0.6599169262720664, "grad_norm": 2.45908522605896, "learning_rate": 4.0204089607840204e-06, "loss": 0.1173844113945961, "step": 2542 }, { "epoch": 0.6601765316718587, "grad_norm": 1.1681652069091797, "learning_rate": 4.019628062602462e-06, "loss": 0.1191173568367958, "step": 2543 }, { "epoch": 0.660436137071651, "grad_norm": 1.2240779399871826, "learning_rate": 4.018846939979732e-06, "loss": 0.111060231924057, "step": 2544 }, { "epoch": 0.6606957424714434, "grad_norm": 1.6455307006835938, "learning_rate": 4.018065593053917e-06, "loss": 0.09539200365543365, "step": 2545 }, { "epoch": 0.6609553478712358, "grad_norm": 1.8793141841888428, "learning_rate": 4.01728402196314e-06, "loss": 0.11226354539394379, "step": 2546 }, { "epoch": 0.6612149532710281, "grad_norm": 1.8115079402923584, "learning_rate": 4.016502226845564e-06, "loss": 0.09958600252866745, "step": 2547 }, { "epoch": 0.6614745586708204, "grad_norm": 1.1703940629959106, "learning_rate": 4.015720207839394e-06, "loss": 0.10281555354595184, "step": 2548 }, { "epoch": 0.6617341640706127, "grad_norm": 1.3949555158615112, "learning_rate": 4.014937965082871e-06, "loss": 0.10268077254295349, "step": 2549 }, { "epoch": 0.661993769470405, "grad_norm": 1.3265498876571655, "learning_rate": 4.014155498714281e-06, "loss": 0.10624847561120987, "step": 2550 }, { "epoch": 0.6622533748701973, "grad_norm": 2.087836742401123, "learning_rate": 4.0133728088719425e-06, "loss": 0.10553441941738129, "step": 2551 }, { "epoch": 0.6625129802699896, "grad_norm": 2.2492527961730957, "learning_rate": 4.0125898956942195e-06, "loss": 0.10230719298124313, "step": 2552 }, { "epoch": 0.6627725856697819, "grad_norm": 1.026728630065918, "learning_rate": 4.011806759319513e-06, "loss": 0.1138225644826889, "step": 2553 }, { "epoch": 0.6630321910695742, "grad_norm": 2.637371301651001, "learning_rate": 4.011023399886262e-06, "loss": 0.09435145556926727, "step": 2554 }, { "epoch": 0.6632917964693665, "grad_norm": 2.4982500076293945, "learning_rate": 4.010239817532947e-06, "loss": 0.11059005558490753, "step": 2555 }, { "epoch": 0.6635514018691588, "grad_norm": 2.5434961318969727, "learning_rate": 4.009456012398089e-06, "loss": 0.112345851957798, "step": 2556 }, { "epoch": 0.6638110072689511, "grad_norm": 2.3753793239593506, "learning_rate": 4.008671984620246e-06, "loss": 0.12323826551437378, "step": 2557 }, { "epoch": 0.6640706126687436, "grad_norm": 3.0086162090301514, "learning_rate": 4.007887734338016e-06, "loss": 0.12781786918640137, "step": 2558 }, { "epoch": 0.6643302180685359, "grad_norm": 3.067392349243164, "learning_rate": 4.007103261690035e-06, "loss": 0.07726483792066574, "step": 2559 }, { "epoch": 0.6645898234683282, "grad_norm": 2.358229637145996, "learning_rate": 4.006318566814982e-06, "loss": 0.11816590279340744, "step": 2560 }, { "epoch": 0.6648494288681205, "grad_norm": 1.6239477396011353, "learning_rate": 4.005533649851573e-06, "loss": 0.09235645830631256, "step": 2561 }, { "epoch": 0.6651090342679128, "grad_norm": 4.638926982879639, "learning_rate": 4.004748510938562e-06, "loss": 0.14532706141471863, "step": 2562 }, { "epoch": 0.6653686396677051, "grad_norm": 2.762049436569214, "learning_rate": 4.003963150214743e-06, "loss": 0.11242619156837463, "step": 2563 }, { "epoch": 0.6656282450674974, "grad_norm": 2.0056636333465576, "learning_rate": 4.003177567818952e-06, "loss": 0.10004330426454544, "step": 2564 }, { "epoch": 0.6658878504672897, "grad_norm": 4.002059459686279, "learning_rate": 4.002391763890062e-06, "loss": 0.11926723271608353, "step": 2565 }, { "epoch": 0.666147455867082, "grad_norm": 2.745884418487549, "learning_rate": 4.001605738566983e-06, "loss": 0.08517098426818848, "step": 2566 }, { "epoch": 0.6664070612668743, "grad_norm": 2.977607250213623, "learning_rate": 4.000819491988667e-06, "loss": 0.08607587218284607, "step": 2567 }, { "epoch": 0.6666666666666666, "grad_norm": 1.1533851623535156, "learning_rate": 4.000033024294105e-06, "loss": 0.10600558668375015, "step": 2568 }, { "epoch": 0.6669262720664589, "grad_norm": 1.354213833808899, "learning_rate": 3.999246335622327e-06, "loss": 0.11016593873500824, "step": 2569 }, { "epoch": 0.6671858774662512, "grad_norm": 1.8288910388946533, "learning_rate": 3.9984594261124e-06, "loss": 0.10960103571414948, "step": 2570 }, { "epoch": 0.6674454828660437, "grad_norm": 1.3312463760375977, "learning_rate": 3.997672295903433e-06, "loss": 0.11303269118070602, "step": 2571 }, { "epoch": 0.667705088265836, "grad_norm": 4.22428035736084, "learning_rate": 3.996884945134572e-06, "loss": 0.09502914547920227, "step": 2572 }, { "epoch": 0.6679646936656283, "grad_norm": 2.927180051803589, "learning_rate": 3.996097373945003e-06, "loss": 0.12315721809864044, "step": 2573 }, { "epoch": 0.6682242990654206, "grad_norm": 4.209303379058838, "learning_rate": 3.995309582473949e-06, "loss": 0.09988897293806076, "step": 2574 }, { "epoch": 0.6684839044652129, "grad_norm": 2.9944446086883545, "learning_rate": 3.994521570860676e-06, "loss": 0.11794088035821915, "step": 2575 }, { "epoch": 0.6687435098650052, "grad_norm": 3.0344343185424805, "learning_rate": 3.993733339244484e-06, "loss": 0.11870270222425461, "step": 2576 }, { "epoch": 0.6690031152647975, "grad_norm": 4.210941791534424, "learning_rate": 3.992944887764715e-06, "loss": 0.11267003417015076, "step": 2577 }, { "epoch": 0.6692627206645898, "grad_norm": 2.5528934001922607, "learning_rate": 3.9921562165607494e-06, "loss": 0.12148834764957428, "step": 2578 }, { "epoch": 0.6695223260643821, "grad_norm": 2.1470789909362793, "learning_rate": 3.991367325772007e-06, "loss": 0.10851157456636429, "step": 2579 }, { "epoch": 0.6697819314641744, "grad_norm": 2.4864754676818848, "learning_rate": 3.990578215537945e-06, "loss": 0.09021884202957153, "step": 2580 }, { "epoch": 0.6700415368639667, "grad_norm": 3.359562635421753, "learning_rate": 3.98978888599806e-06, "loss": 0.12899695336818695, "step": 2581 }, { "epoch": 0.670301142263759, "grad_norm": 4.949881553649902, "learning_rate": 3.988999337291887e-06, "loss": 0.12461593747138977, "step": 2582 }, { "epoch": 0.6705607476635514, "grad_norm": 3.4609642028808594, "learning_rate": 3.988209569559001e-06, "loss": 0.06274864077568054, "step": 2583 }, { "epoch": 0.6708203530633438, "grad_norm": 3.1548757553100586, "learning_rate": 3.987419582939014e-06, "loss": 0.10540279746055603, "step": 2584 }, { "epoch": 0.6710799584631361, "grad_norm": 1.4228497743606567, "learning_rate": 3.986629377571579e-06, "loss": 0.13045714795589447, "step": 2585 }, { "epoch": 0.6713395638629284, "grad_norm": 2.5285139083862305, "learning_rate": 3.985838953596383e-06, "loss": 0.10622957348823547, "step": 2586 }, { "epoch": 0.6715991692627207, "grad_norm": 3.1006524562835693, "learning_rate": 3.985048311153158e-06, "loss": 0.14703187346458435, "step": 2587 }, { "epoch": 0.671858774662513, "grad_norm": 3.701615333557129, "learning_rate": 3.98425745038167e-06, "loss": 0.12741820514202118, "step": 2588 }, { "epoch": 0.6721183800623053, "grad_norm": 2.2593791484832764, "learning_rate": 3.983466371421727e-06, "loss": 0.10214335471391678, "step": 2589 }, { "epoch": 0.6723779854620976, "grad_norm": 1.034011721611023, "learning_rate": 3.982675074413171e-06, "loss": 0.10179382562637329, "step": 2590 }, { "epoch": 0.6726375908618899, "grad_norm": 2.7182843685150146, "learning_rate": 3.981883559495886e-06, "loss": 0.11878903210163116, "step": 2591 }, { "epoch": 0.6728971962616822, "grad_norm": 2.338479995727539, "learning_rate": 3.9810918268097945e-06, "loss": 0.10570254176855087, "step": 2592 }, { "epoch": 0.6731568016614745, "grad_norm": 5.854763031005859, "learning_rate": 3.980299876494857e-06, "loss": 0.1181628555059433, "step": 2593 }, { "epoch": 0.6734164070612668, "grad_norm": 2.3691914081573486, "learning_rate": 3.979507708691071e-06, "loss": 0.1156572699546814, "step": 2594 }, { "epoch": 0.6736760124610592, "grad_norm": 2.008756637573242, "learning_rate": 3.978715323538474e-06, "loss": 0.10311847925186157, "step": 2595 }, { "epoch": 0.6739356178608515, "grad_norm": 2.9519498348236084, "learning_rate": 3.977922721177141e-06, "loss": 0.0968085378408432, "step": 2596 }, { "epoch": 0.6741952232606438, "grad_norm": 2.304828643798828, "learning_rate": 3.977129901747187e-06, "loss": 0.10946156084537506, "step": 2597 }, { "epoch": 0.6744548286604362, "grad_norm": 2.31506085395813, "learning_rate": 3.976336865388765e-06, "loss": 0.12743857502937317, "step": 2598 }, { "epoch": 0.6747144340602285, "grad_norm": 2.790595769882202, "learning_rate": 3.975543612242063e-06, "loss": 0.12376056611537933, "step": 2599 }, { "epoch": 0.6749740394600208, "grad_norm": 2.505930185317993, "learning_rate": 3.974750142447312e-06, "loss": 0.11113718152046204, "step": 2600 }, { "epoch": 0.6752336448598131, "grad_norm": 1.304606556892395, "learning_rate": 3.97395645614478e-06, "loss": 0.09448065608739853, "step": 2601 }, { "epoch": 0.6754932502596054, "grad_norm": 2.909994602203369, "learning_rate": 3.97316255347477e-06, "loss": 0.11180821061134338, "step": 2602 }, { "epoch": 0.6757528556593977, "grad_norm": 2.7465009689331055, "learning_rate": 3.972368434577628e-06, "loss": 0.11770598590373993, "step": 2603 }, { "epoch": 0.67601246105919, "grad_norm": 2.719707727432251, "learning_rate": 3.971574099593735e-06, "loss": 0.10794394463300705, "step": 2604 }, { "epoch": 0.6762720664589823, "grad_norm": 1.6223835945129395, "learning_rate": 3.970779548663512e-06, "loss": 0.11147408932447433, "step": 2605 }, { "epoch": 0.6765316718587746, "grad_norm": 3.7048659324645996, "learning_rate": 3.969984781927417e-06, "loss": 0.11474983394145966, "step": 2606 }, { "epoch": 0.676791277258567, "grad_norm": 2.3060379028320312, "learning_rate": 3.969189799525946e-06, "loss": 0.103848397731781, "step": 2607 }, { "epoch": 0.6770508826583593, "grad_norm": 3.44224214553833, "learning_rate": 3.9683946015996335e-06, "loss": 0.12788698077201843, "step": 2608 }, { "epoch": 0.6773104880581516, "grad_norm": 2.095566749572754, "learning_rate": 3.967599188289054e-06, "loss": 0.09231531620025635, "step": 2609 }, { "epoch": 0.677570093457944, "grad_norm": 2.4491465091705322, "learning_rate": 3.966803559734818e-06, "loss": 0.11003296077251434, "step": 2610 }, { "epoch": 0.6778296988577363, "grad_norm": 2.2371490001678467, "learning_rate": 3.966007716077572e-06, "loss": 0.12511613965034485, "step": 2611 }, { "epoch": 0.6780893042575286, "grad_norm": 3.078883171081543, "learning_rate": 3.965211657458006e-06, "loss": 0.10931220650672913, "step": 2612 }, { "epoch": 0.6783489096573209, "grad_norm": 2.6481246948242188, "learning_rate": 3.964415384016843e-06, "loss": 0.11487013101577759, "step": 2613 }, { "epoch": 0.6786085150571132, "grad_norm": 2.209070920944214, "learning_rate": 3.963618895894847e-06, "loss": 0.10220393538475037, "step": 2614 }, { "epoch": 0.6788681204569055, "grad_norm": 2.285799980163574, "learning_rate": 3.962822193232818e-06, "loss": 0.1084493026137352, "step": 2615 }, { "epoch": 0.6791277258566978, "grad_norm": 3.7932376861572266, "learning_rate": 3.9620252761715955e-06, "loss": 0.12696228921413422, "step": 2616 }, { "epoch": 0.6793873312564901, "grad_norm": 2.2044758796691895, "learning_rate": 3.961228144852056e-06, "loss": 0.11890202760696411, "step": 2617 }, { "epoch": 0.6796469366562824, "grad_norm": 1.133423924446106, "learning_rate": 3.960430799415114e-06, "loss": 0.11563566327095032, "step": 2618 }, { "epoch": 0.6799065420560748, "grad_norm": 3.11417555809021, "learning_rate": 3.959633240001721e-06, "loss": 0.08050739765167236, "step": 2619 }, { "epoch": 0.6801661474558671, "grad_norm": 1.4126085042953491, "learning_rate": 3.958835466752868e-06, "loss": 0.08901552855968475, "step": 2620 }, { "epoch": 0.6804257528556594, "grad_norm": 3.1345694065093994, "learning_rate": 3.958037479809583e-06, "loss": 0.10962417721748352, "step": 2621 }, { "epoch": 0.6806853582554517, "grad_norm": 4.3358659744262695, "learning_rate": 3.957239279312934e-06, "loss": 0.10267441719770432, "step": 2622 }, { "epoch": 0.680944963655244, "grad_norm": 1.7750675678253174, "learning_rate": 3.9564408654040206e-06, "loss": 0.10625563561916351, "step": 2623 }, { "epoch": 0.6812045690550363, "grad_norm": 1.7265973091125488, "learning_rate": 3.955642238223987e-06, "loss": 0.1128811165690422, "step": 2624 }, { "epoch": 0.6814641744548287, "grad_norm": 3.0946335792541504, "learning_rate": 3.954843397914011e-06, "loss": 0.08655519783496857, "step": 2625 }, { "epoch": 0.681723779854621, "grad_norm": 2.9963433742523193, "learning_rate": 3.954044344615309e-06, "loss": 0.08480294048786163, "step": 2626 }, { "epoch": 0.6819833852544133, "grad_norm": 2.615424871444702, "learning_rate": 3.953245078469137e-06, "loss": 0.11144412308931351, "step": 2627 }, { "epoch": 0.6822429906542056, "grad_norm": 1.6566394567489624, "learning_rate": 3.952445599616784e-06, "loss": 0.09436122328042984, "step": 2628 }, { "epoch": 0.6825025960539979, "grad_norm": 1.1613216400146484, "learning_rate": 3.951645908199583e-06, "loss": 0.10488475114107132, "step": 2629 }, { "epoch": 0.6827622014537902, "grad_norm": 1.157365322113037, "learning_rate": 3.950846004358899e-06, "loss": 0.1101980060338974, "step": 2630 }, { "epoch": 0.6830218068535826, "grad_norm": 2.7055184841156006, "learning_rate": 3.950045888236137e-06, "loss": 0.1188296303153038, "step": 2631 }, { "epoch": 0.6832814122533749, "grad_norm": 1.46186363697052, "learning_rate": 3.949245559972739e-06, "loss": 0.1206350177526474, "step": 2632 }, { "epoch": 0.6835410176531672, "grad_norm": 1.1229736804962158, "learning_rate": 3.9484450197101856e-06, "loss": 0.1072206050157547, "step": 2633 }, { "epoch": 0.6838006230529595, "grad_norm": 1.9902207851409912, "learning_rate": 3.947644267589993e-06, "loss": 0.11984089016914368, "step": 2634 }, { "epoch": 0.6840602284527518, "grad_norm": 2.1464998722076416, "learning_rate": 3.946843303753716e-06, "loss": 0.11075077205896378, "step": 2635 }, { "epoch": 0.6843198338525441, "grad_norm": 2.8856983184814453, "learning_rate": 3.946042128342948e-06, "loss": 0.11970583349466324, "step": 2636 }, { "epoch": 0.6845794392523364, "grad_norm": 1.3912707567214966, "learning_rate": 3.945240741499317e-06, "loss": 0.1112305074930191, "step": 2637 }, { "epoch": 0.6848390446521287, "grad_norm": 2.3106203079223633, "learning_rate": 3.94443914336449e-06, "loss": 0.08088315278291702, "step": 2638 }, { "epoch": 0.685098650051921, "grad_norm": 2.972991943359375, "learning_rate": 3.9436373340801724e-06, "loss": 0.11391018331050873, "step": 2639 }, { "epoch": 0.6853582554517134, "grad_norm": 3.033417224884033, "learning_rate": 3.9428353137881034e-06, "loss": 0.12053833901882172, "step": 2640 }, { "epoch": 0.6856178608515057, "grad_norm": 2.4989113807678223, "learning_rate": 3.942033082630064e-06, "loss": 0.11001831293106079, "step": 2641 }, { "epoch": 0.685877466251298, "grad_norm": 2.552408456802368, "learning_rate": 3.94123064074787e-06, "loss": 0.11274685710668564, "step": 2642 }, { "epoch": 0.6861370716510904, "grad_norm": 3.6497671604156494, "learning_rate": 3.940427988283374e-06, "loss": 0.11366653442382812, "step": 2643 }, { "epoch": 0.6863966770508827, "grad_norm": 3.6640613079071045, "learning_rate": 3.939625125378467e-06, "loss": 0.09401433914899826, "step": 2644 }, { "epoch": 0.686656282450675, "grad_norm": 3.6316752433776855, "learning_rate": 3.938822052175076e-06, "loss": 0.09965358674526215, "step": 2645 }, { "epoch": 0.6869158878504673, "grad_norm": 2.193735122680664, "learning_rate": 3.9380187688151664e-06, "loss": 0.09573409706354141, "step": 2646 }, { "epoch": 0.6871754932502596, "grad_norm": 3.1822283267974854, "learning_rate": 3.9372152754407415e-06, "loss": 0.09998277574777603, "step": 2647 }, { "epoch": 0.6874350986500519, "grad_norm": 2.782541036605835, "learning_rate": 3.936411572193839e-06, "loss": 0.1070258617401123, "step": 2648 }, { "epoch": 0.6876947040498442, "grad_norm": 1.8982607126235962, "learning_rate": 3.9356076592165355e-06, "loss": 0.12338162958621979, "step": 2649 }, { "epoch": 0.6879543094496365, "grad_norm": 2.51055908203125, "learning_rate": 3.934803536650945e-06, "loss": 0.1047324538230896, "step": 2650 }, { "epoch": 0.6882139148494288, "grad_norm": 4.923798084259033, "learning_rate": 3.933999204639218e-06, "loss": 0.1260550618171692, "step": 2651 }, { "epoch": 0.6884735202492211, "grad_norm": 2.6418964862823486, "learning_rate": 3.933194663323541e-06, "loss": 0.09646866470575333, "step": 2652 }, { "epoch": 0.6887331256490135, "grad_norm": 2.7791519165039062, "learning_rate": 3.93238991284614e-06, "loss": 0.09347735345363617, "step": 2653 }, { "epoch": 0.6889927310488058, "grad_norm": 3.1269118785858154, "learning_rate": 3.931584953349274e-06, "loss": 0.08589434623718262, "step": 2654 }, { "epoch": 0.6892523364485982, "grad_norm": 1.3476835489273071, "learning_rate": 3.930779784975244e-06, "loss": 0.09844345599412918, "step": 2655 }, { "epoch": 0.6895119418483905, "grad_norm": 3.6410608291625977, "learning_rate": 3.929974407866384e-06, "loss": 0.10305215418338776, "step": 2656 }, { "epoch": 0.6897715472481828, "grad_norm": 4.261623859405518, "learning_rate": 3.929168822165067e-06, "loss": 0.10346640646457672, "step": 2657 }, { "epoch": 0.6900311526479751, "grad_norm": 4.202638626098633, "learning_rate": 3.9283630280137e-06, "loss": 0.08278807997703552, "step": 2658 }, { "epoch": 0.6902907580477674, "grad_norm": 3.633680582046509, "learning_rate": 3.927557025554733e-06, "loss": 0.10137151926755905, "step": 2659 }, { "epoch": 0.6905503634475597, "grad_norm": 3.8395187854766846, "learning_rate": 3.926750814930645e-06, "loss": 0.08875657618045807, "step": 2660 }, { "epoch": 0.690809968847352, "grad_norm": 1.4966607093811035, "learning_rate": 3.925944396283957e-06, "loss": 0.12230460345745087, "step": 2661 }, { "epoch": 0.6910695742471443, "grad_norm": 3.6352603435516357, "learning_rate": 3.9251377697572254e-06, "loss": 0.09872160851955414, "step": 2662 }, { "epoch": 0.6913291796469366, "grad_norm": 1.8431122303009033, "learning_rate": 3.9243309354930426e-06, "loss": 0.11768735200166702, "step": 2663 }, { "epoch": 0.6915887850467289, "grad_norm": 2.22102427482605, "learning_rate": 3.9235238936340405e-06, "loss": 0.07794348895549774, "step": 2664 }, { "epoch": 0.6918483904465212, "grad_norm": 1.931336760520935, "learning_rate": 3.922716644322883e-06, "loss": 0.10087833553552628, "step": 2665 }, { "epoch": 0.6921079958463136, "grad_norm": 3.3812642097473145, "learning_rate": 3.921909187702275e-06, "loss": 0.10564454644918442, "step": 2666 }, { "epoch": 0.692367601246106, "grad_norm": 3.9121224880218506, "learning_rate": 3.921101523914956e-06, "loss": 0.10459357500076294, "step": 2667 }, { "epoch": 0.6926272066458983, "grad_norm": 1.8757303953170776, "learning_rate": 3.920293653103701e-06, "loss": 0.10538080334663391, "step": 2668 }, { "epoch": 0.6928868120456906, "grad_norm": 1.5413708686828613, "learning_rate": 3.9194855754113245e-06, "loss": 0.14285877346992493, "step": 2669 }, { "epoch": 0.6931464174454829, "grad_norm": 3.329824924468994, "learning_rate": 3.918677290980678e-06, "loss": 0.1197686493396759, "step": 2670 }, { "epoch": 0.6934060228452752, "grad_norm": 3.940807580947876, "learning_rate": 3.917868799954644e-06, "loss": 0.08431804180145264, "step": 2671 }, { "epoch": 0.6936656282450675, "grad_norm": 1.100966215133667, "learning_rate": 3.9170601024761465e-06, "loss": 0.10387438535690308, "step": 2672 }, { "epoch": 0.6939252336448598, "grad_norm": 2.423975706100464, "learning_rate": 3.916251198688145e-06, "loss": 0.10034414380788803, "step": 2673 }, { "epoch": 0.6941848390446521, "grad_norm": 2.4808356761932373, "learning_rate": 3.915442088733636e-06, "loss": 0.076689213514328, "step": 2674 }, { "epoch": 0.6944444444444444, "grad_norm": 4.30159330368042, "learning_rate": 3.9146327727556504e-06, "loss": 0.11398166418075562, "step": 2675 }, { "epoch": 0.6947040498442367, "grad_norm": 2.684572458267212, "learning_rate": 3.913823250897257e-06, "loss": 0.11322621256113052, "step": 2676 }, { "epoch": 0.694963655244029, "grad_norm": 2.6880013942718506, "learning_rate": 3.913013523301561e-06, "loss": 0.10151808708906174, "step": 2677 }, { "epoch": 0.6952232606438213, "grad_norm": 2.8971476554870605, "learning_rate": 3.912203590111703e-06, "loss": 0.12281651794910431, "step": 2678 }, { "epoch": 0.6954828660436138, "grad_norm": 2.5180845260620117, "learning_rate": 3.911393451470863e-06, "loss": 0.10950504243373871, "step": 2679 }, { "epoch": 0.6957424714434061, "grad_norm": 2.787876844406128, "learning_rate": 3.910583107522251e-06, "loss": 0.1154162809252739, "step": 2680 }, { "epoch": 0.6960020768431984, "grad_norm": 1.7111152410507202, "learning_rate": 3.90977255840912e-06, "loss": 0.12621039152145386, "step": 2681 }, { "epoch": 0.6962616822429907, "grad_norm": 3.4599127769470215, "learning_rate": 3.9089618042747564e-06, "loss": 0.11311116814613342, "step": 2682 }, { "epoch": 0.696521287642783, "grad_norm": 2.645983934402466, "learning_rate": 3.908150845262483e-06, "loss": 0.09499059617519379, "step": 2683 }, { "epoch": 0.6967808930425753, "grad_norm": 3.3331358432769775, "learning_rate": 3.907339681515657e-06, "loss": 0.0821327343583107, "step": 2684 }, { "epoch": 0.6970404984423676, "grad_norm": 3.318755865097046, "learning_rate": 3.906528313177676e-06, "loss": 0.11690552532672882, "step": 2685 }, { "epoch": 0.6973001038421599, "grad_norm": 3.657212972640991, "learning_rate": 3.90571674039197e-06, "loss": 0.07906030118465424, "step": 2686 }, { "epoch": 0.6975597092419522, "grad_norm": 2.074388265609741, "learning_rate": 3.904904963302007e-06, "loss": 0.09871930629014969, "step": 2687 }, { "epoch": 0.6978193146417445, "grad_norm": 4.502304553985596, "learning_rate": 3.904092982051289e-06, "loss": 0.07280980050563812, "step": 2688 }, { "epoch": 0.6980789200415368, "grad_norm": 3.323577404022217, "learning_rate": 3.903280796783358e-06, "loss": 0.09623292833566666, "step": 2689 }, { "epoch": 0.6983385254413291, "grad_norm": 2.3784289360046387, "learning_rate": 3.902468407641788e-06, "loss": 0.10575594753026962, "step": 2690 }, { "epoch": 0.6985981308411215, "grad_norm": 3.0778634548187256, "learning_rate": 3.901655814770193e-06, "loss": 0.1211918443441391, "step": 2691 }, { "epoch": 0.6988577362409139, "grad_norm": 3.012620687484741, "learning_rate": 3.900843018312218e-06, "loss": 0.07979834079742432, "step": 2692 }, { "epoch": 0.6991173416407062, "grad_norm": 2.4413416385650635, "learning_rate": 3.90003001841155e-06, "loss": 0.12438015639781952, "step": 2693 }, { "epoch": 0.6993769470404985, "grad_norm": 3.171713352203369, "learning_rate": 3.899216815211905e-06, "loss": 0.11312232911586761, "step": 2694 }, { "epoch": 0.6996365524402908, "grad_norm": 1.6234039068222046, "learning_rate": 3.898403408857042e-06, "loss": 0.10338546335697174, "step": 2695 }, { "epoch": 0.6998961578400831, "grad_norm": 1.6632604598999023, "learning_rate": 3.897589799490749e-06, "loss": 0.10813607275485992, "step": 2696 }, { "epoch": 0.7001557632398754, "grad_norm": 1.948502779006958, "learning_rate": 3.8967759872568584e-06, "loss": 0.09961187839508057, "step": 2697 }, { "epoch": 0.7004153686396677, "grad_norm": 4.1100053787231445, "learning_rate": 3.89596197229923e-06, "loss": 0.10853981971740723, "step": 2698 }, { "epoch": 0.70067497403946, "grad_norm": 1.9486130475997925, "learning_rate": 3.895147754761763e-06, "loss": 0.10268291085958481, "step": 2699 }, { "epoch": 0.7009345794392523, "grad_norm": 0.9964583516120911, "learning_rate": 3.894333334788395e-06, "loss": 0.11909031867980957, "step": 2700 }, { "epoch": 0.7011941848390446, "grad_norm": 2.950227737426758, "learning_rate": 3.893518712523094e-06, "loss": 0.10068878531455994, "step": 2701 }, { "epoch": 0.7014537902388369, "grad_norm": 3.4004087448120117, "learning_rate": 3.8927038881098675e-06, "loss": 0.09066332131624222, "step": 2702 }, { "epoch": 0.7017133956386293, "grad_norm": 3.0000574588775635, "learning_rate": 3.891888861692759e-06, "loss": 0.08240465819835663, "step": 2703 }, { "epoch": 0.7019730010384216, "grad_norm": 3.8210456371307373, "learning_rate": 3.891073633415845e-06, "loss": 0.10956662148237228, "step": 2704 }, { "epoch": 0.702232606438214, "grad_norm": 2.7000033855438232, "learning_rate": 3.89025820342324e-06, "loss": 0.1386263221502304, "step": 2705 }, { "epoch": 0.7024922118380063, "grad_norm": 2.5287363529205322, "learning_rate": 3.889442571859093e-06, "loss": 0.10486093908548355, "step": 2706 }, { "epoch": 0.7027518172377986, "grad_norm": 1.8585814237594604, "learning_rate": 3.8886267388675885e-06, "loss": 0.10852442681789398, "step": 2707 }, { "epoch": 0.7030114226375909, "grad_norm": 0.919890284538269, "learning_rate": 3.887810704592949e-06, "loss": 0.11756318807601929, "step": 2708 }, { "epoch": 0.7032710280373832, "grad_norm": 1.1244748830795288, "learning_rate": 3.886994469179429e-06, "loss": 0.11166876554489136, "step": 2709 }, { "epoch": 0.7035306334371755, "grad_norm": 3.378155469894409, "learning_rate": 3.886178032771319e-06, "loss": 0.07913902401924133, "step": 2710 }, { "epoch": 0.7037902388369678, "grad_norm": 1.7232939004898071, "learning_rate": 3.8853613955129495e-06, "loss": 0.09655199199914932, "step": 2711 }, { "epoch": 0.7040498442367601, "grad_norm": 1.429857611656189, "learning_rate": 3.88454455754868e-06, "loss": 0.10172915458679199, "step": 2712 }, { "epoch": 0.7043094496365524, "grad_norm": 1.439439296722412, "learning_rate": 3.883727519022912e-06, "loss": 0.10703031718730927, "step": 2713 }, { "epoch": 0.7045690550363447, "grad_norm": 3.1970527172088623, "learning_rate": 3.882910280080077e-06, "loss": 0.10315737128257751, "step": 2714 }, { "epoch": 0.7048286604361371, "grad_norm": 1.9712058305740356, "learning_rate": 3.882092840864645e-06, "loss": 0.09980090707540512, "step": 2715 }, { "epoch": 0.7050882658359294, "grad_norm": 1.8158446550369263, "learning_rate": 3.88127520152112e-06, "loss": 0.12589237093925476, "step": 2716 }, { "epoch": 0.7053478712357217, "grad_norm": 4.366458415985107, "learning_rate": 3.880457362194041e-06, "loss": 0.10150478780269623, "step": 2717 }, { "epoch": 0.705607476635514, "grad_norm": 1.4471814632415771, "learning_rate": 3.879639323027986e-06, "loss": 0.09866726398468018, "step": 2718 }, { "epoch": 0.7058670820353063, "grad_norm": 2.141592502593994, "learning_rate": 3.878821084167563e-06, "loss": 0.0872131958603859, "step": 2719 }, { "epoch": 0.7061266874350987, "grad_norm": 3.075852870941162, "learning_rate": 3.8780026457574195e-06, "loss": 0.11515463143587112, "step": 2720 }, { "epoch": 0.706386292834891, "grad_norm": 1.4178986549377441, "learning_rate": 3.877184007942236e-06, "loss": 0.10400737076997757, "step": 2721 }, { "epoch": 0.7066458982346833, "grad_norm": 1.3464869260787964, "learning_rate": 3.8763651708667285e-06, "loss": 0.10763978213071823, "step": 2722 }, { "epoch": 0.7069055036344756, "grad_norm": 1.2059191465377808, "learning_rate": 3.8755461346756505e-06, "loss": 0.11260108649730682, "step": 2723 }, { "epoch": 0.7071651090342679, "grad_norm": 2.7702994346618652, "learning_rate": 3.874726899513786e-06, "loss": 0.07819484174251556, "step": 2724 }, { "epoch": 0.7074247144340602, "grad_norm": 1.8766816854476929, "learning_rate": 3.873907465525959e-06, "loss": 0.10260186344385147, "step": 2725 }, { "epoch": 0.7076843198338525, "grad_norm": 1.949192762374878, "learning_rate": 3.873087832857026e-06, "loss": 0.10572902858257294, "step": 2726 }, { "epoch": 0.7079439252336449, "grad_norm": 3.0642378330230713, "learning_rate": 3.8722680016518796e-06, "loss": 0.10575363039970398, "step": 2727 }, { "epoch": 0.7082035306334372, "grad_norm": 1.6648547649383545, "learning_rate": 3.8714479720554475e-06, "loss": 0.09605471789836884, "step": 2728 }, { "epoch": 0.7084631360332295, "grad_norm": 2.31380558013916, "learning_rate": 3.870627744212691e-06, "loss": 0.08459538221359253, "step": 2729 }, { "epoch": 0.7087227414330218, "grad_norm": 0.9579712152481079, "learning_rate": 3.869807318268609e-06, "loss": 0.11096762120723724, "step": 2730 }, { "epoch": 0.7089823468328141, "grad_norm": 1.3447799682617188, "learning_rate": 3.868986694368232e-06, "loss": 0.10022825747728348, "step": 2731 }, { "epoch": 0.7092419522326064, "grad_norm": 3.9165563583374023, "learning_rate": 3.8681658726566305e-06, "loss": 0.11373269557952881, "step": 2732 }, { "epoch": 0.7095015576323987, "grad_norm": 1.46726655960083, "learning_rate": 3.8673448532789044e-06, "loss": 0.0899481400847435, "step": 2733 }, { "epoch": 0.7097611630321911, "grad_norm": 1.6357710361480713, "learning_rate": 3.8665236363801915e-06, "loss": 0.10023827850818634, "step": 2734 }, { "epoch": 0.7100207684319834, "grad_norm": 2.7382874488830566, "learning_rate": 3.865702222105666e-06, "loss": 0.05044989287853241, "step": 2735 }, { "epoch": 0.7102803738317757, "grad_norm": 2.916696310043335, "learning_rate": 3.8648806106005335e-06, "loss": 0.11247558891773224, "step": 2736 }, { "epoch": 0.710539979231568, "grad_norm": 2.23681378364563, "learning_rate": 3.864058802010036e-06, "loss": 0.10296593606472015, "step": 2737 }, { "epoch": 0.7107995846313603, "grad_norm": 4.8586745262146, "learning_rate": 3.86323679647945e-06, "loss": 0.1105566918849945, "step": 2738 }, { "epoch": 0.7110591900311527, "grad_norm": 1.5288853645324707, "learning_rate": 3.8624145941540896e-06, "loss": 0.09905754029750824, "step": 2739 }, { "epoch": 0.711318795430945, "grad_norm": 2.3794264793395996, "learning_rate": 3.861592195179299e-06, "loss": 0.11115409433841705, "step": 2740 }, { "epoch": 0.7115784008307373, "grad_norm": 0.9858195781707764, "learning_rate": 3.860769599700461e-06, "loss": 0.09717093408107758, "step": 2741 }, { "epoch": 0.7118380062305296, "grad_norm": 3.8672711849212646, "learning_rate": 3.859946807862991e-06, "loss": 0.12321125715970993, "step": 2742 }, { "epoch": 0.7120976116303219, "grad_norm": 3.9010117053985596, "learning_rate": 3.8591238198123395e-06, "loss": 0.09753057360649109, "step": 2743 }, { "epoch": 0.7123572170301142, "grad_norm": 2.550074577331543, "learning_rate": 3.8583006356939926e-06, "loss": 0.07805978506803513, "step": 2744 }, { "epoch": 0.7126168224299065, "grad_norm": 2.8748326301574707, "learning_rate": 3.85747725565347e-06, "loss": 0.10775183141231537, "step": 2745 }, { "epoch": 0.7128764278296988, "grad_norm": 0.819820761680603, "learning_rate": 3.856653679836326e-06, "loss": 0.11561444401741028, "step": 2746 }, { "epoch": 0.7131360332294912, "grad_norm": 1.5918657779693604, "learning_rate": 3.85582990838815e-06, "loss": 0.11567071080207825, "step": 2747 }, { "epoch": 0.7133956386292835, "grad_norm": 3.038869857788086, "learning_rate": 3.855005941454567e-06, "loss": 0.13688261806964874, "step": 2748 }, { "epoch": 0.7136552440290758, "grad_norm": 1.1514039039611816, "learning_rate": 3.854181779181235e-06, "loss": 0.10373882949352264, "step": 2749 }, { "epoch": 0.7139148494288681, "grad_norm": 3.159343719482422, "learning_rate": 3.853357421713847e-06, "loss": 0.07923203706741333, "step": 2750 }, { "epoch": 0.7141744548286605, "grad_norm": 1.8013052940368652, "learning_rate": 3.852532869198129e-06, "loss": 0.11630862951278687, "step": 2751 }, { "epoch": 0.7144340602284528, "grad_norm": 1.1375579833984375, "learning_rate": 3.851708121779846e-06, "loss": 0.10667221993207932, "step": 2752 }, { "epoch": 0.7146936656282451, "grad_norm": 1.053174376487732, "learning_rate": 3.8508831796047925e-06, "loss": 0.11194384843111038, "step": 2753 }, { "epoch": 0.7149532710280374, "grad_norm": 0.9845998287200928, "learning_rate": 3.850058042818799e-06, "loss": 0.10572679340839386, "step": 2754 }, { "epoch": 0.7152128764278297, "grad_norm": 0.8461138606071472, "learning_rate": 3.849232711567732e-06, "loss": 0.11064106225967407, "step": 2755 }, { "epoch": 0.715472481827622, "grad_norm": 2.7195701599121094, "learning_rate": 3.848407185997491e-06, "loss": 0.12751497328281403, "step": 2756 }, { "epoch": 0.7157320872274143, "grad_norm": 1.6675596237182617, "learning_rate": 3.847581466254009e-06, "loss": 0.11237034201622009, "step": 2757 }, { "epoch": 0.7159916926272066, "grad_norm": 2.978447675704956, "learning_rate": 3.846755552483256e-06, "loss": 0.07962803542613983, "step": 2758 }, { "epoch": 0.7162512980269989, "grad_norm": 1.1139154434204102, "learning_rate": 3.845929444831234e-06, "loss": 0.10716073215007782, "step": 2759 }, { "epoch": 0.7165109034267912, "grad_norm": 2.6711618900299072, "learning_rate": 3.845103143443978e-06, "loss": 0.10233110934495926, "step": 2760 }, { "epoch": 0.7167705088265836, "grad_norm": 3.253835916519165, "learning_rate": 3.844276648467561e-06, "loss": 0.08792249858379364, "step": 2761 }, { "epoch": 0.7170301142263759, "grad_norm": 1.5829057693481445, "learning_rate": 3.843449960048089e-06, "loss": 0.11748982220888138, "step": 2762 }, { "epoch": 0.7172897196261683, "grad_norm": 1.4742183685302734, "learning_rate": 3.8426230783317e-06, "loss": 0.10385671257972717, "step": 2763 }, { "epoch": 0.7175493250259606, "grad_norm": 2.2735235691070557, "learning_rate": 3.841796003464569e-06, "loss": 0.11900417506694794, "step": 2764 }, { "epoch": 0.7178089304257529, "grad_norm": 4.0259294509887695, "learning_rate": 3.8409687355929035e-06, "loss": 0.08735252916812897, "step": 2765 }, { "epoch": 0.7180685358255452, "grad_norm": 3.2503201961517334, "learning_rate": 3.840141274862945e-06, "loss": 0.1302987039089203, "step": 2766 }, { "epoch": 0.7183281412253375, "grad_norm": 1.6530214548110962, "learning_rate": 3.83931362142097e-06, "loss": 0.11049473285675049, "step": 2767 }, { "epoch": 0.7185877466251298, "grad_norm": 3.574805736541748, "learning_rate": 3.83848577541329e-06, "loss": 0.09640197455883026, "step": 2768 }, { "epoch": 0.7188473520249221, "grad_norm": 3.66477108001709, "learning_rate": 3.837657736986246e-06, "loss": 0.10290022939443588, "step": 2769 }, { "epoch": 0.7191069574247144, "grad_norm": 1.7770711183547974, "learning_rate": 3.8368295062862196e-06, "loss": 0.10762137174606323, "step": 2770 }, { "epoch": 0.7193665628245067, "grad_norm": 2.8178365230560303, "learning_rate": 3.836001083459621e-06, "loss": 0.10344080626964569, "step": 2771 }, { "epoch": 0.719626168224299, "grad_norm": 3.9995229244232178, "learning_rate": 3.835172468652897e-06, "loss": 0.08632077276706696, "step": 2772 }, { "epoch": 0.7198857736240913, "grad_norm": 1.5581188201904297, "learning_rate": 3.834343662012528e-06, "loss": 0.1130707859992981, "step": 2773 }, { "epoch": 0.7201453790238836, "grad_norm": 3.4635183811187744, "learning_rate": 3.833514663685027e-06, "loss": 0.12291866540908813, "step": 2774 }, { "epoch": 0.7204049844236761, "grad_norm": 3.7774782180786133, "learning_rate": 3.832685473816944e-06, "loss": 0.10940267890691757, "step": 2775 }, { "epoch": 0.7206645898234684, "grad_norm": 2.2497847080230713, "learning_rate": 3.831856092554858e-06, "loss": 0.13052590191364288, "step": 2776 }, { "epoch": 0.7209241952232607, "grad_norm": 1.5022754669189453, "learning_rate": 3.8310265200453865e-06, "loss": 0.09163884073495865, "step": 2777 }, { "epoch": 0.721183800623053, "grad_norm": 1.8117084503173828, "learning_rate": 3.830196756435179e-06, "loss": 0.1135864108800888, "step": 2778 }, { "epoch": 0.7214434060228453, "grad_norm": 4.212275981903076, "learning_rate": 3.829366801870917e-06, "loss": 0.08752350509166718, "step": 2779 }, { "epoch": 0.7217030114226376, "grad_norm": 1.941206693649292, "learning_rate": 3.828536656499321e-06, "loss": 0.09294375777244568, "step": 2780 }, { "epoch": 0.7219626168224299, "grad_norm": 1.6390998363494873, "learning_rate": 3.827706320467137e-06, "loss": 0.10574904084205627, "step": 2781 }, { "epoch": 0.7222222222222222, "grad_norm": 1.244584321975708, "learning_rate": 3.826875793921152e-06, "loss": 0.10260152816772461, "step": 2782 }, { "epoch": 0.7224818276220145, "grad_norm": 2.363142490386963, "learning_rate": 3.826045077008185e-06, "loss": 0.10980024933815002, "step": 2783 }, { "epoch": 0.7227414330218068, "grad_norm": 5.7231292724609375, "learning_rate": 3.825214169875085e-06, "loss": 0.10381408035755157, "step": 2784 }, { "epoch": 0.7230010384215991, "grad_norm": 3.5762939453125, "learning_rate": 3.82438307266874e-06, "loss": 0.10672629624605179, "step": 2785 }, { "epoch": 0.7232606438213914, "grad_norm": 1.5819578170776367, "learning_rate": 3.8235517855360685e-06, "loss": 0.11694730818271637, "step": 2786 }, { "epoch": 0.7235202492211839, "grad_norm": 1.311306357383728, "learning_rate": 3.822720308624021e-06, "loss": 0.10843084752559662, "step": 2787 }, { "epoch": 0.7237798546209762, "grad_norm": 2.507401943206787, "learning_rate": 3.821888642079586e-06, "loss": 0.08314020931720734, "step": 2788 }, { "epoch": 0.7240394600207685, "grad_norm": 1.5879955291748047, "learning_rate": 3.821056786049781e-06, "loss": 0.10366164147853851, "step": 2789 }, { "epoch": 0.7242990654205608, "grad_norm": 3.495288848876953, "learning_rate": 3.8202247406816614e-06, "loss": 0.08982621133327484, "step": 2790 }, { "epoch": 0.7245586708203531, "grad_norm": 3.4328579902648926, "learning_rate": 3.819392506122312e-06, "loss": 0.0677318200469017, "step": 2791 }, { "epoch": 0.7248182762201454, "grad_norm": 2.620083808898926, "learning_rate": 3.8185600825188526e-06, "loss": 0.1314588189125061, "step": 2792 }, { "epoch": 0.7250778816199377, "grad_norm": 1.8669675588607788, "learning_rate": 3.817727470018439e-06, "loss": 0.11831019073724747, "step": 2793 }, { "epoch": 0.72533748701973, "grad_norm": 1.1050056219100952, "learning_rate": 3.816894668768255e-06, "loss": 0.09774964302778244, "step": 2794 }, { "epoch": 0.7255970924195223, "grad_norm": 2.526639938354492, "learning_rate": 3.816061678915523e-06, "loss": 0.10820743441581726, "step": 2795 }, { "epoch": 0.7258566978193146, "grad_norm": 3.4413928985595703, "learning_rate": 3.815228500607495e-06, "loss": 0.07924079895019531, "step": 2796 }, { "epoch": 0.7261163032191069, "grad_norm": 3.588965654373169, "learning_rate": 3.8143951339914596e-06, "loss": 0.11432598531246185, "step": 2797 }, { "epoch": 0.7263759086188992, "grad_norm": 2.3759820461273193, "learning_rate": 3.813561579214736e-06, "loss": 0.09303559362888336, "step": 2798 }, { "epoch": 0.7266355140186916, "grad_norm": 3.4763779640197754, "learning_rate": 3.8127278364246766e-06, "loss": 0.0838470309972763, "step": 2799 }, { "epoch": 0.726895119418484, "grad_norm": 3.0228333473205566, "learning_rate": 3.811893905768669e-06, "loss": 0.08957874774932861, "step": 2800 }, { "epoch": 0.7271547248182763, "grad_norm": 1.9403070211410522, "learning_rate": 3.811059787394133e-06, "loss": 0.10913510620594025, "step": 2801 }, { "epoch": 0.7274143302180686, "grad_norm": 3.5054867267608643, "learning_rate": 3.8102254814485225e-06, "loss": 0.09098200500011444, "step": 2802 }, { "epoch": 0.7276739356178609, "grad_norm": 1.836917757987976, "learning_rate": 3.8093909880793223e-06, "loss": 0.09614309668540955, "step": 2803 }, { "epoch": 0.7279335410176532, "grad_norm": 1.6213709115982056, "learning_rate": 3.8085563074340513e-06, "loss": 0.11707530915737152, "step": 2804 }, { "epoch": 0.7281931464174455, "grad_norm": 4.91257381439209, "learning_rate": 3.8077214396602642e-06, "loss": 0.15143528580665588, "step": 2805 }, { "epoch": 0.7284527518172378, "grad_norm": 1.9703805446624756, "learning_rate": 3.806886384905545e-06, "loss": 0.11764497309923172, "step": 2806 }, { "epoch": 0.7287123572170301, "grad_norm": 4.028010845184326, "learning_rate": 3.806051143317512e-06, "loss": 0.11239975690841675, "step": 2807 }, { "epoch": 0.7289719626168224, "grad_norm": 3.1499178409576416, "learning_rate": 3.805215715043818e-06, "loss": 0.09811331331729889, "step": 2808 }, { "epoch": 0.7292315680166147, "grad_norm": 1.902392029762268, "learning_rate": 3.8043801002321457e-06, "loss": 0.10830742120742798, "step": 2809 }, { "epoch": 0.729491173416407, "grad_norm": 3.467135190963745, "learning_rate": 3.8035442990302145e-06, "loss": 0.09962897002696991, "step": 2810 }, { "epoch": 0.7297507788161994, "grad_norm": 3.3187553882598877, "learning_rate": 3.802708311585774e-06, "loss": 0.10398229956626892, "step": 2811 }, { "epoch": 0.7300103842159917, "grad_norm": 3.278446674346924, "learning_rate": 3.801872138046608e-06, "loss": 0.11261804401874542, "step": 2812 }, { "epoch": 0.730269989615784, "grad_norm": 2.291872024536133, "learning_rate": 3.801035778560532e-06, "loss": 0.110722616314888, "step": 2813 }, { "epoch": 0.7305295950155763, "grad_norm": 3.623713493347168, "learning_rate": 3.800199233275397e-06, "loss": 0.08239197731018066, "step": 2814 }, { "epoch": 0.7307892004153687, "grad_norm": 2.9818015098571777, "learning_rate": 3.7993625023390835e-06, "loss": 0.07083603739738464, "step": 2815 }, { "epoch": 0.731048805815161, "grad_norm": 2.9548513889312744, "learning_rate": 3.7985255858995067e-06, "loss": 0.08923764526844025, "step": 2816 }, { "epoch": 0.7313084112149533, "grad_norm": 2.8367397785186768, "learning_rate": 3.7976884841046135e-06, "loss": 0.1115785539150238, "step": 2817 }, { "epoch": 0.7315680166147456, "grad_norm": 2.5702013969421387, "learning_rate": 3.7968511971023863e-06, "loss": 0.08456835150718689, "step": 2818 }, { "epoch": 0.7318276220145379, "grad_norm": 1.9633586406707764, "learning_rate": 3.7960137250408374e-06, "loss": 0.11754372715950012, "step": 2819 }, { "epoch": 0.7320872274143302, "grad_norm": 2.5469400882720947, "learning_rate": 3.795176068068012e-06, "loss": 0.12590737640857697, "step": 2820 }, { "epoch": 0.7323468328141225, "grad_norm": 0.9902964234352112, "learning_rate": 3.794338226331989e-06, "loss": 0.10988862812519073, "step": 2821 }, { "epoch": 0.7326064382139148, "grad_norm": 3.031482696533203, "learning_rate": 3.7935001999808796e-06, "loss": 0.11914017796516418, "step": 2822 }, { "epoch": 0.7328660436137072, "grad_norm": 2.9271953105926514, "learning_rate": 3.792661989162828e-06, "loss": 0.10910244286060333, "step": 2823 }, { "epoch": 0.7331256490134995, "grad_norm": 2.9784023761749268, "learning_rate": 3.7918235940260108e-06, "loss": 0.11560700833797455, "step": 2824 }, { "epoch": 0.7333852544132918, "grad_norm": 1.8450846672058105, "learning_rate": 3.790985014718636e-06, "loss": 0.10975612699985504, "step": 2825 }, { "epoch": 0.7336448598130841, "grad_norm": 1.551566481590271, "learning_rate": 3.790146251388947e-06, "loss": 0.10800719261169434, "step": 2826 }, { "epoch": 0.7339044652128764, "grad_norm": 2.3923838138580322, "learning_rate": 3.789307304185216e-06, "loss": 0.11217284202575684, "step": 2827 }, { "epoch": 0.7341640706126688, "grad_norm": 2.9608359336853027, "learning_rate": 3.7884681732557517e-06, "loss": 0.08683094382286072, "step": 2828 }, { "epoch": 0.7344236760124611, "grad_norm": 4.897519111633301, "learning_rate": 3.7876288587488913e-06, "loss": 0.08374205231666565, "step": 2829 }, { "epoch": 0.7346832814122534, "grad_norm": 4.281473159790039, "learning_rate": 3.786789360813007e-06, "loss": 0.14125008881092072, "step": 2830 }, { "epoch": 0.7349428868120457, "grad_norm": 3.4559760093688965, "learning_rate": 3.785949679596504e-06, "loss": 0.09369005262851715, "step": 2831 }, { "epoch": 0.735202492211838, "grad_norm": 0.8403342962265015, "learning_rate": 3.785109815247817e-06, "loss": 0.10981343686580658, "step": 2832 }, { "epoch": 0.7354620976116303, "grad_norm": 2.970121383666992, "learning_rate": 3.784269767915416e-06, "loss": 0.08806820958852768, "step": 2833 }, { "epoch": 0.7357217030114226, "grad_norm": 2.124500036239624, "learning_rate": 3.7834295377478013e-06, "loss": 0.11244557797908783, "step": 2834 }, { "epoch": 0.735981308411215, "grad_norm": 2.304290533065796, "learning_rate": 3.7825891248935066e-06, "loss": 0.09708891808986664, "step": 2835 }, { "epoch": 0.7362409138110073, "grad_norm": 1.5852423906326294, "learning_rate": 3.7817485295010982e-06, "loss": 0.08985982835292816, "step": 2836 }, { "epoch": 0.7365005192107996, "grad_norm": 3.6862540245056152, "learning_rate": 3.7809077517191733e-06, "loss": 0.1213279590010643, "step": 2837 }, { "epoch": 0.7367601246105919, "grad_norm": 1.3172852993011475, "learning_rate": 3.780066791696363e-06, "loss": 0.10066276788711548, "step": 2838 }, { "epoch": 0.7370197300103842, "grad_norm": 1.4475303888320923, "learning_rate": 3.779225649581329e-06, "loss": 0.13273948431015015, "step": 2839 }, { "epoch": 0.7372793354101765, "grad_norm": 2.511626958847046, "learning_rate": 3.778384325522767e-06, "loss": 0.07135617733001709, "step": 2840 }, { "epoch": 0.7375389408099688, "grad_norm": 2.125908613204956, "learning_rate": 3.7775428196694023e-06, "loss": 0.1069670021533966, "step": 2841 }, { "epoch": 0.7377985462097612, "grad_norm": 3.4286601543426514, "learning_rate": 3.776701132169995e-06, "loss": 0.07413871586322784, "step": 2842 }, { "epoch": 0.7380581516095535, "grad_norm": 2.5345160961151123, "learning_rate": 3.7758592631733354e-06, "loss": 0.10193830728530884, "step": 2843 }, { "epoch": 0.7383177570093458, "grad_norm": 2.182591438293457, "learning_rate": 3.775017212828248e-06, "loss": 0.1116432398557663, "step": 2844 }, { "epoch": 0.7385773624091381, "grad_norm": 2.3436152935028076, "learning_rate": 3.774174981283587e-06, "loss": 0.13373295962810516, "step": 2845 }, { "epoch": 0.7388369678089304, "grad_norm": 1.5754666328430176, "learning_rate": 3.773332568688239e-06, "loss": 0.10292110592126846, "step": 2846 }, { "epoch": 0.7390965732087228, "grad_norm": 2.3767142295837402, "learning_rate": 3.7724899751911254e-06, "loss": 0.09725654870271683, "step": 2847 }, { "epoch": 0.7393561786085151, "grad_norm": 1.4181160926818848, "learning_rate": 3.7716472009411944e-06, "loss": 0.10518939793109894, "step": 2848 }, { "epoch": 0.7396157840083074, "grad_norm": 1.5031139850616455, "learning_rate": 3.7708042460874315e-06, "loss": 0.10414066910743713, "step": 2849 }, { "epoch": 0.7398753894080997, "grad_norm": 1.0079232454299927, "learning_rate": 3.769961110778852e-06, "loss": 0.11045560240745544, "step": 2850 }, { "epoch": 0.740134994807892, "grad_norm": 2.8490989208221436, "learning_rate": 3.7691177951645018e-06, "loss": 0.10307091474533081, "step": 2851 }, { "epoch": 0.7403946002076843, "grad_norm": 1.7189230918884277, "learning_rate": 3.7682742993934595e-06, "loss": 0.12079839408397675, "step": 2852 }, { "epoch": 0.7406542056074766, "grad_norm": 0.8911322951316833, "learning_rate": 3.7674306236148374e-06, "loss": 0.11292248964309692, "step": 2853 }, { "epoch": 0.7409138110072689, "grad_norm": 1.8973257541656494, "learning_rate": 3.7665867679777763e-06, "loss": 0.09595488756895065, "step": 2854 }, { "epoch": 0.7411734164070612, "grad_norm": 3.0064473152160645, "learning_rate": 3.7657427326314523e-06, "loss": 0.10769408196210861, "step": 2855 }, { "epoch": 0.7414330218068536, "grad_norm": 2.302384614944458, "learning_rate": 3.7648985177250697e-06, "loss": 0.11657702922821045, "step": 2856 }, { "epoch": 0.7416926272066459, "grad_norm": 2.9511260986328125, "learning_rate": 3.764054123407867e-06, "loss": 0.08341602981090546, "step": 2857 }, { "epoch": 0.7419522326064382, "grad_norm": 2.9326767921447754, "learning_rate": 3.763209549829115e-06, "loss": 0.09270106256008148, "step": 2858 }, { "epoch": 0.7422118380062306, "grad_norm": 1.9639595746994019, "learning_rate": 3.7623647971381132e-06, "loss": 0.1005304679274559, "step": 2859 }, { "epoch": 0.7424714434060229, "grad_norm": 3.058485507965088, "learning_rate": 3.7615198654841957e-06, "loss": 0.10017469525337219, "step": 2860 }, { "epoch": 0.7427310488058152, "grad_norm": 2.411797523498535, "learning_rate": 3.760674755016727e-06, "loss": 0.1092609316110611, "step": 2861 }, { "epoch": 0.7429906542056075, "grad_norm": 2.1844851970672607, "learning_rate": 3.7598294658851025e-06, "loss": 0.08838585019111633, "step": 2862 }, { "epoch": 0.7432502596053998, "grad_norm": 2.655369758605957, "learning_rate": 3.758983998238751e-06, "loss": 0.0856640487909317, "step": 2863 }, { "epoch": 0.7435098650051921, "grad_norm": 2.8278560638427734, "learning_rate": 3.758138352227131e-06, "loss": 0.08790716528892517, "step": 2864 }, { "epoch": 0.7437694704049844, "grad_norm": 2.2621114253997803, "learning_rate": 3.757292527999734e-06, "loss": 0.0967196524143219, "step": 2865 }, { "epoch": 0.7440290758047767, "grad_norm": 3.613792657852173, "learning_rate": 3.756446525706081e-06, "loss": 0.08625313639640808, "step": 2866 }, { "epoch": 0.744288681204569, "grad_norm": 3.3010003566741943, "learning_rate": 3.7556003454957273e-06, "loss": 0.08924783021211624, "step": 2867 }, { "epoch": 0.7445482866043613, "grad_norm": 1.8750574588775635, "learning_rate": 3.754753987518259e-06, "loss": 0.11334583163261414, "step": 2868 }, { "epoch": 0.7448078920041536, "grad_norm": 3.1985702514648438, "learning_rate": 3.75390745192329e-06, "loss": 0.12413933128118515, "step": 2869 }, { "epoch": 0.745067497403946, "grad_norm": 2.5565285682678223, "learning_rate": 3.75306073886047e-06, "loss": 0.08030608296394348, "step": 2870 }, { "epoch": 0.7453271028037384, "grad_norm": 2.5197598934173584, "learning_rate": 3.752213848479479e-06, "loss": 0.11105579137802124, "step": 2871 }, { "epoch": 0.7455867082035307, "grad_norm": 1.528658151626587, "learning_rate": 3.751366780930027e-06, "loss": 0.0837993174791336, "step": 2872 }, { "epoch": 0.745846313603323, "grad_norm": 4.164549350738525, "learning_rate": 3.7505195363618563e-06, "loss": 0.12426165491342545, "step": 2873 }, { "epoch": 0.7461059190031153, "grad_norm": 1.7572671175003052, "learning_rate": 3.7496721149247407e-06, "loss": 0.10208229720592499, "step": 2874 }, { "epoch": 0.7463655244029076, "grad_norm": 3.201951503753662, "learning_rate": 3.7488245167684846e-06, "loss": 0.08395642042160034, "step": 2875 }, { "epoch": 0.7466251298026999, "grad_norm": 4.053391456604004, "learning_rate": 3.7479767420429235e-06, "loss": 0.12886686623096466, "step": 2876 }, { "epoch": 0.7468847352024922, "grad_norm": 1.0484927892684937, "learning_rate": 3.7471287908979254e-06, "loss": 0.11710744351148605, "step": 2877 }, { "epoch": 0.7471443406022845, "grad_norm": 2.7039437294006348, "learning_rate": 3.746280663483389e-06, "loss": 0.08133582770824432, "step": 2878 }, { "epoch": 0.7474039460020768, "grad_norm": 2.0536093711853027, "learning_rate": 3.7454323599492414e-06, "loss": 0.10146204382181168, "step": 2879 }, { "epoch": 0.7476635514018691, "grad_norm": 4.16386604309082, "learning_rate": 3.744583880445446e-06, "loss": 0.10528405010700226, "step": 2880 }, { "epoch": 0.7479231568016614, "grad_norm": 2.4379069805145264, "learning_rate": 3.743735225121994e-06, "loss": 0.08415671437978745, "step": 2881 }, { "epoch": 0.7481827622014537, "grad_norm": 2.0268986225128174, "learning_rate": 3.7428863941289067e-06, "loss": 0.08348539471626282, "step": 2882 }, { "epoch": 0.7484423676012462, "grad_norm": 3.1258323192596436, "learning_rate": 3.742037387616239e-06, "loss": 0.059309810400009155, "step": 2883 }, { "epoch": 0.7487019730010385, "grad_norm": 3.8109750747680664, "learning_rate": 3.7411882057340766e-06, "loss": 0.10640650987625122, "step": 2884 }, { "epoch": 0.7489615784008308, "grad_norm": 2.617828130722046, "learning_rate": 3.7403388486325345e-06, "loss": 0.11125980317592621, "step": 2885 }, { "epoch": 0.7492211838006231, "grad_norm": 4.845392227172852, "learning_rate": 3.739489316461759e-06, "loss": 0.09379630535840988, "step": 2886 }, { "epoch": 0.7494807892004154, "grad_norm": 2.3189735412597656, "learning_rate": 3.7386396093719297e-06, "loss": 0.10794258117675781, "step": 2887 }, { "epoch": 0.7497403946002077, "grad_norm": 1.6032100915908813, "learning_rate": 3.7377897275132534e-06, "loss": 0.1107177734375, "step": 2888 }, { "epoch": 0.75, "grad_norm": 5.523347854614258, "learning_rate": 3.736939671035971e-06, "loss": 0.10984098166227341, "step": 2889 }, { "epoch": 0.7502596053997923, "grad_norm": 5.083263874053955, "learning_rate": 3.736089440090353e-06, "loss": 0.09571047127246857, "step": 2890 }, { "epoch": 0.7505192107995846, "grad_norm": 2.954146385192871, "learning_rate": 3.7352390348266997e-06, "loss": 0.13465404510498047, "step": 2891 }, { "epoch": 0.7507788161993769, "grad_norm": 3.0914971828460693, "learning_rate": 3.734388455395344e-06, "loss": 0.10798351466655731, "step": 2892 }, { "epoch": 0.7510384215991692, "grad_norm": 3.506319284439087, "learning_rate": 3.7335377019466497e-06, "loss": 0.08895230293273926, "step": 2893 }, { "epoch": 0.7512980269989615, "grad_norm": 3.661663055419922, "learning_rate": 3.73268677463101e-06, "loss": 0.08101990073919296, "step": 2894 }, { "epoch": 0.7515576323987538, "grad_norm": 2.5731115341186523, "learning_rate": 3.731835673598848e-06, "loss": 0.08985523879528046, "step": 2895 }, { "epoch": 0.7518172377985463, "grad_norm": 1.6908835172653198, "learning_rate": 3.7309843990006207e-06, "loss": 0.09930292516946793, "step": 2896 }, { "epoch": 0.7520768431983386, "grad_norm": 3.668304443359375, "learning_rate": 3.7301329509868124e-06, "loss": 0.07817874848842621, "step": 2897 }, { "epoch": 0.7523364485981309, "grad_norm": 1.3470027446746826, "learning_rate": 3.7292813297079416e-06, "loss": 0.1043965294957161, "step": 2898 }, { "epoch": 0.7525960539979232, "grad_norm": 2.0349998474121094, "learning_rate": 3.728429535314554e-06, "loss": 0.1019938737154007, "step": 2899 }, { "epoch": 0.7528556593977155, "grad_norm": 2.60859751701355, "learning_rate": 3.727577567957227e-06, "loss": 0.1082518994808197, "step": 2900 }, { "epoch": 0.7531152647975078, "grad_norm": 3.118514060974121, "learning_rate": 3.72672542778657e-06, "loss": 0.0998685210943222, "step": 2901 }, { "epoch": 0.7533748701973001, "grad_norm": 3.9874908924102783, "learning_rate": 3.7258731149532207e-06, "loss": 0.07648082822561264, "step": 2902 }, { "epoch": 0.7536344755970924, "grad_norm": 2.2727158069610596, "learning_rate": 3.72502062960785e-06, "loss": 0.13703709840774536, "step": 2903 }, { "epoch": 0.7538940809968847, "grad_norm": 1.979538083076477, "learning_rate": 3.724167971901157e-06, "loss": 0.11555030941963196, "step": 2904 }, { "epoch": 0.754153686396677, "grad_norm": 1.0119446516036987, "learning_rate": 3.7233151419838713e-06, "loss": 0.11023278534412384, "step": 2905 }, { "epoch": 0.7544132917964693, "grad_norm": 3.147200584411621, "learning_rate": 3.722462140006755e-06, "loss": 0.09306813776493073, "step": 2906 }, { "epoch": 0.7546728971962616, "grad_norm": 3.104562282562256, "learning_rate": 3.721608966120599e-06, "loss": 0.07630135864019394, "step": 2907 }, { "epoch": 0.754932502596054, "grad_norm": 1.159058928489685, "learning_rate": 3.7207556204762248e-06, "loss": 0.11913936585187912, "step": 2908 }, { "epoch": 0.7551921079958464, "grad_norm": 1.2143927812576294, "learning_rate": 3.7199021032244843e-06, "loss": 0.09827344864606857, "step": 2909 }, { "epoch": 0.7554517133956387, "grad_norm": 2.177509069442749, "learning_rate": 3.7190484145162585e-06, "loss": 0.08551530539989471, "step": 2910 }, { "epoch": 0.755711318795431, "grad_norm": 3.118557929992676, "learning_rate": 3.7181945545024622e-06, "loss": 0.07155989110469818, "step": 2911 }, { "epoch": 0.7559709241952233, "grad_norm": 3.1462392807006836, "learning_rate": 3.717340523334037e-06, "loss": 0.10707814991474152, "step": 2912 }, { "epoch": 0.7562305295950156, "grad_norm": 1.479500412940979, "learning_rate": 3.716486321161957e-06, "loss": 0.10164058208465576, "step": 2913 }, { "epoch": 0.7564901349948079, "grad_norm": 1.1262139081954956, "learning_rate": 3.7156319481372247e-06, "loss": 0.10606968402862549, "step": 2914 }, { "epoch": 0.7567497403946002, "grad_norm": 1.041907548904419, "learning_rate": 3.7147774044108743e-06, "loss": 0.11065249890089035, "step": 2915 }, { "epoch": 0.7570093457943925, "grad_norm": 2.7689831256866455, "learning_rate": 3.7139226901339685e-06, "loss": 0.12040722370147705, "step": 2916 }, { "epoch": 0.7572689511941848, "grad_norm": 1.7826355695724487, "learning_rate": 3.713067805457603e-06, "loss": 0.11711964011192322, "step": 2917 }, { "epoch": 0.7575285565939771, "grad_norm": 1.6533182859420776, "learning_rate": 3.7122127505329003e-06, "loss": 0.09132131934165955, "step": 2918 }, { "epoch": 0.7577881619937694, "grad_norm": 1.857042670249939, "learning_rate": 3.7113575255110155e-06, "loss": 0.11386314779520035, "step": 2919 }, { "epoch": 0.7580477673935618, "grad_norm": 0.9303894639015198, "learning_rate": 3.7105021305431317e-06, "loss": 0.11560770869255066, "step": 2920 }, { "epoch": 0.7583073727933541, "grad_norm": 0.9555432200431824, "learning_rate": 3.709646565780464e-06, "loss": 0.11797737330198288, "step": 2921 }, { "epoch": 0.7585669781931464, "grad_norm": 4.312989711761475, "learning_rate": 3.708790831374257e-06, "loss": 0.09881985187530518, "step": 2922 }, { "epoch": 0.7588265835929388, "grad_norm": 2.181360960006714, "learning_rate": 3.7079349274757838e-06, "loss": 0.10747683793306351, "step": 2923 }, { "epoch": 0.7590861889927311, "grad_norm": 2.5312998294830322, "learning_rate": 3.7070788542363495e-06, "loss": 0.13428246974945068, "step": 2924 }, { "epoch": 0.7593457943925234, "grad_norm": 2.6478383541107178, "learning_rate": 3.706222611807288e-06, "loss": 0.09591735154390335, "step": 2925 }, { "epoch": 0.7596053997923157, "grad_norm": 2.431934118270874, "learning_rate": 3.705366200339964e-06, "loss": 0.13408318161964417, "step": 2926 }, { "epoch": 0.759865005192108, "grad_norm": 2.421419620513916, "learning_rate": 3.7045096199857694e-06, "loss": 0.08024560660123825, "step": 2927 }, { "epoch": 0.7601246105919003, "grad_norm": 2.294677734375, "learning_rate": 3.7036528708961297e-06, "loss": 0.09621189534664154, "step": 2928 }, { "epoch": 0.7603842159916926, "grad_norm": 0.9781589508056641, "learning_rate": 3.702795953222498e-06, "loss": 0.11766180396080017, "step": 2929 }, { "epoch": 0.7606438213914849, "grad_norm": 1.3111447095870972, "learning_rate": 3.7019388671163588e-06, "loss": 0.11237214505672455, "step": 2930 }, { "epoch": 0.7609034267912772, "grad_norm": 2.185152769088745, "learning_rate": 3.7010816127292245e-06, "loss": 0.10756832361221313, "step": 2931 }, { "epoch": 0.7611630321910696, "grad_norm": 1.8220503330230713, "learning_rate": 3.7002241902126366e-06, "loss": 0.08906662464141846, "step": 2932 }, { "epoch": 0.7614226375908619, "grad_norm": 1.6537283658981323, "learning_rate": 3.6993665997181698e-06, "loss": 0.09596209228038788, "step": 2933 }, { "epoch": 0.7616822429906542, "grad_norm": 1.7654250860214233, "learning_rate": 3.698508841397426e-06, "loss": 0.11199717968702316, "step": 2934 }, { "epoch": 0.7619418483904465, "grad_norm": 2.2600696086883545, "learning_rate": 3.6976509154020375e-06, "loss": 0.09552156180143356, "step": 2935 }, { "epoch": 0.7622014537902388, "grad_norm": 0.8954067826271057, "learning_rate": 3.6967928218836646e-06, "loss": 0.11923463642597198, "step": 2936 }, { "epoch": 0.7624610591900312, "grad_norm": 1.5645650625228882, "learning_rate": 3.695934560993999e-06, "loss": 0.09252370893955231, "step": 2937 }, { "epoch": 0.7627206645898235, "grad_norm": 3.672638416290283, "learning_rate": 3.6950761328847634e-06, "loss": 0.0966067835688591, "step": 2938 }, { "epoch": 0.7629802699896158, "grad_norm": 2.74927020072937, "learning_rate": 3.694217537707706e-06, "loss": 0.0954061895608902, "step": 2939 }, { "epoch": 0.7632398753894081, "grad_norm": 2.9994986057281494, "learning_rate": 3.6933587756146085e-06, "loss": 0.10144755244255066, "step": 2940 }, { "epoch": 0.7634994807892004, "grad_norm": 1.9040025472640991, "learning_rate": 3.6924998467572782e-06, "loss": 0.09393909573554993, "step": 2941 }, { "epoch": 0.7637590861889927, "grad_norm": 1.0077964067459106, "learning_rate": 3.6916407512875567e-06, "loss": 0.12026695907115936, "step": 2942 }, { "epoch": 0.764018691588785, "grad_norm": 3.3938887119293213, "learning_rate": 3.6907814893573102e-06, "loss": 0.10436130315065384, "step": 2943 }, { "epoch": 0.7642782969885774, "grad_norm": 2.666609525680542, "learning_rate": 3.6899220611184384e-06, "loss": 0.09763148427009583, "step": 2944 }, { "epoch": 0.7645379023883697, "grad_norm": 2.8780829906463623, "learning_rate": 3.6890624667228664e-06, "loss": 0.07986794412136078, "step": 2945 }, { "epoch": 0.764797507788162, "grad_norm": 2.3785903453826904, "learning_rate": 3.6882027063225527e-06, "loss": 0.09206196665763855, "step": 2946 }, { "epoch": 0.7650571131879543, "grad_norm": 4.014995098114014, "learning_rate": 3.687342780069482e-06, "loss": 0.07714106142520905, "step": 2947 }, { "epoch": 0.7653167185877466, "grad_norm": 2.161240816116333, "learning_rate": 3.68648268811567e-06, "loss": 0.11449620127677917, "step": 2948 }, { "epoch": 0.7655763239875389, "grad_norm": 2.642658233642578, "learning_rate": 3.6856224306131622e-06, "loss": 0.06252627074718475, "step": 2949 }, { "epoch": 0.7658359293873312, "grad_norm": 2.664013385772705, "learning_rate": 3.684762007714031e-06, "loss": 0.0930061936378479, "step": 2950 }, { "epoch": 0.7660955347871236, "grad_norm": 2.265045166015625, "learning_rate": 3.68390141957038e-06, "loss": 0.09524445235729218, "step": 2951 }, { "epoch": 0.7663551401869159, "grad_norm": 0.9181973338127136, "learning_rate": 3.683040666334342e-06, "loss": 0.10551735758781433, "step": 2952 }, { "epoch": 0.7666147455867082, "grad_norm": 1.1506824493408203, "learning_rate": 3.682179748158078e-06, "loss": 0.10072849690914154, "step": 2953 }, { "epoch": 0.7668743509865005, "grad_norm": 3.240983486175537, "learning_rate": 3.6813186651937775e-06, "loss": 0.07650710642337799, "step": 2954 }, { "epoch": 0.7671339563862928, "grad_norm": 3.2343220710754395, "learning_rate": 3.6804574175936623e-06, "loss": 0.06779979914426804, "step": 2955 }, { "epoch": 0.7673935617860852, "grad_norm": 3.734875440597534, "learning_rate": 3.6795960055099804e-06, "loss": 0.08885161578655243, "step": 2956 }, { "epoch": 0.7676531671858775, "grad_norm": 1.625309944152832, "learning_rate": 3.6787344290950094e-06, "loss": 0.11074478924274445, "step": 2957 }, { "epoch": 0.7679127725856698, "grad_norm": 3.447133779525757, "learning_rate": 3.677872688501057e-06, "loss": 0.10433461517095566, "step": 2958 }, { "epoch": 0.7681723779854621, "grad_norm": 3.2092125415802, "learning_rate": 3.6770107838804577e-06, "loss": 0.11979880928993225, "step": 2959 }, { "epoch": 0.7684319833852544, "grad_norm": 1.351920247077942, "learning_rate": 3.676148715385578e-06, "loss": 0.10523565858602524, "step": 2960 }, { "epoch": 0.7686915887850467, "grad_norm": 3.0454211235046387, "learning_rate": 3.6752864831688124e-06, "loss": 0.09317687153816223, "step": 2961 }, { "epoch": 0.768951194184839, "grad_norm": 3.499523878097534, "learning_rate": 3.6744240873825816e-06, "loss": 0.0918852761387825, "step": 2962 }, { "epoch": 0.7692107995846313, "grad_norm": 1.8697429895401, "learning_rate": 3.673561528179339e-06, "loss": 0.07587509602308273, "step": 2963 }, { "epoch": 0.7694704049844237, "grad_norm": 2.023601531982422, "learning_rate": 3.6726988057115646e-06, "loss": 0.10624277591705322, "step": 2964 }, { "epoch": 0.769730010384216, "grad_norm": 1.372132658958435, "learning_rate": 3.6718359201317695e-06, "loss": 0.1227259635925293, "step": 2965 }, { "epoch": 0.7699896157840083, "grad_norm": 4.454433441162109, "learning_rate": 3.67097287159249e-06, "loss": 0.08249668776988983, "step": 2966 }, { "epoch": 0.7702492211838006, "grad_norm": 1.862219214439392, "learning_rate": 3.6701096602462944e-06, "loss": 0.09852619469165802, "step": 2967 }, { "epoch": 0.770508826583593, "grad_norm": 5.119838714599609, "learning_rate": 3.6692462862457784e-06, "loss": 0.14718753099441528, "step": 2968 }, { "epoch": 0.7707684319833853, "grad_norm": 2.3596575260162354, "learning_rate": 3.6683827497435678e-06, "loss": 0.1008148118853569, "step": 2969 }, { "epoch": 0.7710280373831776, "grad_norm": 1.2220491170883179, "learning_rate": 3.667519050892315e-06, "loss": 0.10756462812423706, "step": 2970 }, { "epoch": 0.7712876427829699, "grad_norm": 1.5294890403747559, "learning_rate": 3.666655189844701e-06, "loss": 0.10288725793361664, "step": 2971 }, { "epoch": 0.7715472481827622, "grad_norm": 3.358030080795288, "learning_rate": 3.6657911667534397e-06, "loss": 0.0854019820690155, "step": 2972 }, { "epoch": 0.7718068535825545, "grad_norm": 3.9474973678588867, "learning_rate": 3.6649269817712686e-06, "loss": 0.080097496509552, "step": 2973 }, { "epoch": 0.7720664589823468, "grad_norm": 4.344934940338135, "learning_rate": 3.6640626350509557e-06, "loss": 0.06586246937513351, "step": 2974 }, { "epoch": 0.7723260643821391, "grad_norm": 2.1154332160949707, "learning_rate": 3.663198126745299e-06, "loss": 0.09364597499370575, "step": 2975 }, { "epoch": 0.7725856697819314, "grad_norm": 4.039140701293945, "learning_rate": 3.6623334570071216e-06, "loss": 0.07470813393592834, "step": 2976 }, { "epoch": 0.7728452751817237, "grad_norm": 1.9570895433425903, "learning_rate": 3.6614686259892794e-06, "loss": 0.10254158079624176, "step": 2977 }, { "epoch": 0.773104880581516, "grad_norm": 2.9779789447784424, "learning_rate": 3.660603633844654e-06, "loss": 0.06473740190267563, "step": 2978 }, { "epoch": 0.7733644859813084, "grad_norm": 3.1434390544891357, "learning_rate": 3.6597384807261564e-06, "loss": 0.1137549877166748, "step": 2979 }, { "epoch": 0.7736240913811008, "grad_norm": 1.7000155448913574, "learning_rate": 3.658873166786726e-06, "loss": 0.10410165786743164, "step": 2980 }, { "epoch": 0.7738836967808931, "grad_norm": 3.4851646423339844, "learning_rate": 3.6580076921793295e-06, "loss": 0.07427465915679932, "step": 2981 }, { "epoch": 0.7741433021806854, "grad_norm": 2.849783420562744, "learning_rate": 3.6571420570569643e-06, "loss": 0.08545643091201782, "step": 2982 }, { "epoch": 0.7744029075804777, "grad_norm": 1.1431385278701782, "learning_rate": 3.6562762615726538e-06, "loss": 0.1119021400809288, "step": 2983 }, { "epoch": 0.77466251298027, "grad_norm": 2.7249937057495117, "learning_rate": 3.6554103058794504e-06, "loss": 0.08424293994903564, "step": 2984 }, { "epoch": 0.7749221183800623, "grad_norm": 2.853069305419922, "learning_rate": 3.6545441901304366e-06, "loss": 0.10661962628364563, "step": 2985 }, { "epoch": 0.7751817237798546, "grad_norm": 1.019778847694397, "learning_rate": 3.6536779144787223e-06, "loss": 0.10528077930212021, "step": 2986 }, { "epoch": 0.7754413291796469, "grad_norm": 1.3021138906478882, "learning_rate": 3.652811479077443e-06, "loss": 0.09734150767326355, "step": 2987 }, { "epoch": 0.7757009345794392, "grad_norm": 1.0076359510421753, "learning_rate": 3.6519448840797667e-06, "loss": 0.10821489989757538, "step": 2988 }, { "epoch": 0.7759605399792315, "grad_norm": 3.481985092163086, "learning_rate": 3.651078129638886e-06, "loss": 0.09131082147359848, "step": 2989 }, { "epoch": 0.7762201453790238, "grad_norm": 4.287659645080566, "learning_rate": 3.6502112159080243e-06, "loss": 0.11928585171699524, "step": 2990 }, { "epoch": 0.7764797507788161, "grad_norm": 1.8532692193984985, "learning_rate": 3.6493441430404315e-06, "loss": 0.12151935696601868, "step": 2991 }, { "epoch": 0.7767393561786086, "grad_norm": 4.8209943771362305, "learning_rate": 3.648476911189387e-06, "loss": 0.11428537219762802, "step": 2992 }, { "epoch": 0.7769989615784009, "grad_norm": 2.9815852642059326, "learning_rate": 3.6476095205081963e-06, "loss": 0.1008681058883667, "step": 2993 }, { "epoch": 0.7772585669781932, "grad_norm": 1.1590739488601685, "learning_rate": 3.646741971150195e-06, "loss": 0.1046576201915741, "step": 2994 }, { "epoch": 0.7775181723779855, "grad_norm": 2.699599504470825, "learning_rate": 3.6458742632687462e-06, "loss": 0.10159999877214432, "step": 2995 }, { "epoch": 0.7777777777777778, "grad_norm": 1.160440444946289, "learning_rate": 3.64500639701724e-06, "loss": 0.1302075833082199, "step": 2996 }, { "epoch": 0.7780373831775701, "grad_norm": 3.5405967235565186, "learning_rate": 3.6441383725490957e-06, "loss": 0.10022155195474625, "step": 2997 }, { "epoch": 0.7782969885773624, "grad_norm": 2.847740650177002, "learning_rate": 3.643270190017761e-06, "loss": 0.1346202939748764, "step": 2998 }, { "epoch": 0.7785565939771547, "grad_norm": 3.0170533657073975, "learning_rate": 3.642401849576709e-06, "loss": 0.1266183853149414, "step": 2999 }, { "epoch": 0.778816199376947, "grad_norm": 2.712376117706299, "learning_rate": 3.6415333513794442e-06, "loss": 0.08590130507946014, "step": 3000 }, { "epoch": 0.7790758047767393, "grad_norm": 1.6798666715621948, "learning_rate": 3.640664695579496e-06, "loss": 0.11118367314338684, "step": 3001 }, { "epoch": 0.7793354101765316, "grad_norm": 1.4713114500045776, "learning_rate": 3.639795882330423e-06, "loss": 0.11839605867862701, "step": 3002 }, { "epoch": 0.7795950155763239, "grad_norm": 1.2980194091796875, "learning_rate": 3.638926911785811e-06, "loss": 0.09955711662769318, "step": 3003 }, { "epoch": 0.7798546209761164, "grad_norm": 3.2383437156677246, "learning_rate": 3.6380577840992758e-06, "loss": 0.08119405061006546, "step": 3004 }, { "epoch": 0.7801142263759087, "grad_norm": 1.1694902181625366, "learning_rate": 3.637188499424457e-06, "loss": 0.09168047457933426, "step": 3005 }, { "epoch": 0.780373831775701, "grad_norm": 2.6135470867156982, "learning_rate": 3.636319057915027e-06, "loss": 0.11936288326978683, "step": 3006 }, { "epoch": 0.7806334371754933, "grad_norm": 1.4657838344573975, "learning_rate": 3.63544945972468e-06, "loss": 0.09832021594047546, "step": 3007 }, { "epoch": 0.7808930425752856, "grad_norm": 2.948629140853882, "learning_rate": 3.634579705007144e-06, "loss": 0.07606088370084763, "step": 3008 }, { "epoch": 0.7811526479750779, "grad_norm": 1.6514383554458618, "learning_rate": 3.6337097939161698e-06, "loss": 0.12860165536403656, "step": 3009 }, { "epoch": 0.7814122533748702, "grad_norm": 2.8327219486236572, "learning_rate": 3.632839726605538e-06, "loss": 0.06736224889755249, "step": 3010 }, { "epoch": 0.7816718587746625, "grad_norm": 4.004268646240234, "learning_rate": 3.6319695032290565e-06, "loss": 0.09080299735069275, "step": 3011 }, { "epoch": 0.7819314641744548, "grad_norm": 2.7940173149108887, "learning_rate": 3.6310991239405625e-06, "loss": 0.09317360818386078, "step": 3012 }, { "epoch": 0.7821910695742471, "grad_norm": 2.814335584640503, "learning_rate": 3.630228588893917e-06, "loss": 0.08734457194805145, "step": 3013 }, { "epoch": 0.7824506749740394, "grad_norm": 3.069042205810547, "learning_rate": 3.629357898243011e-06, "loss": 0.11576278507709503, "step": 3014 }, { "epoch": 0.7827102803738317, "grad_norm": 4.664558410644531, "learning_rate": 3.6284870521417646e-06, "loss": 0.09412916004657745, "step": 3015 }, { "epoch": 0.7829698857736241, "grad_norm": 2.195801019668579, "learning_rate": 3.627616050744121e-06, "loss": 0.10193650424480438, "step": 3016 }, { "epoch": 0.7832294911734164, "grad_norm": 2.8695068359375, "learning_rate": 3.6267448942040553e-06, "loss": 0.06979714334011078, "step": 3017 }, { "epoch": 0.7834890965732088, "grad_norm": 3.1100826263427734, "learning_rate": 3.6258735826755663e-06, "loss": 0.037633661180734634, "step": 3018 }, { "epoch": 0.7837487019730011, "grad_norm": 3.0254783630371094, "learning_rate": 3.625002116312683e-06, "loss": 0.06476493179798126, "step": 3019 }, { "epoch": 0.7840083073727934, "grad_norm": 2.262652635574341, "learning_rate": 3.6241304952694606e-06, "loss": 0.11893518269062042, "step": 3020 }, { "epoch": 0.7842679127725857, "grad_norm": 3.271146535873413, "learning_rate": 3.623258719699982e-06, "loss": 0.11545949429273605, "step": 3021 }, { "epoch": 0.784527518172378, "grad_norm": 1.414630651473999, "learning_rate": 3.6223867897583555e-06, "loss": 0.11362513154745102, "step": 3022 }, { "epoch": 0.7847871235721703, "grad_norm": 3.2385902404785156, "learning_rate": 3.6215147055987204e-06, "loss": 0.12119675427675247, "step": 3023 }, { "epoch": 0.7850467289719626, "grad_norm": 2.5657968521118164, "learning_rate": 3.62064246737524e-06, "loss": 0.05989743769168854, "step": 3024 }, { "epoch": 0.7853063343717549, "grad_norm": 1.0580463409423828, "learning_rate": 3.6197700752421066e-06, "loss": 0.11075031757354736, "step": 3025 }, { "epoch": 0.7855659397715472, "grad_norm": 2.8513808250427246, "learning_rate": 3.6188975293535393e-06, "loss": 0.08572736382484436, "step": 3026 }, { "epoch": 0.7858255451713395, "grad_norm": 2.1042258739471436, "learning_rate": 3.618024829863784e-06, "loss": 0.10996528714895248, "step": 3027 }, { "epoch": 0.7860851505711319, "grad_norm": 3.075373888015747, "learning_rate": 3.617151976927113e-06, "loss": 0.0859973281621933, "step": 3028 }, { "epoch": 0.7863447559709242, "grad_norm": 3.49137282371521, "learning_rate": 3.6162789706978287e-06, "loss": 0.08184543251991272, "step": 3029 }, { "epoch": 0.7866043613707165, "grad_norm": 4.158303260803223, "learning_rate": 3.6154058113302566e-06, "loss": 0.14897099137306213, "step": 3030 }, { "epoch": 0.7868639667705088, "grad_norm": 1.9229977130889893, "learning_rate": 3.614532498978753e-06, "loss": 0.0930103063583374, "step": 3031 }, { "epoch": 0.7871235721703012, "grad_norm": 2.9654488563537598, "learning_rate": 3.6136590337976985e-06, "loss": 0.09036377817392349, "step": 3032 }, { "epoch": 0.7873831775700935, "grad_norm": 2.3121843338012695, "learning_rate": 3.6127854159415014e-06, "loss": 0.10560373961925507, "step": 3033 }, { "epoch": 0.7876427829698858, "grad_norm": 2.3486135005950928, "learning_rate": 3.611911645564599e-06, "loss": 0.10903488099575043, "step": 3034 }, { "epoch": 0.7879023883696781, "grad_norm": 3.3372790813446045, "learning_rate": 3.611037722821452e-06, "loss": 0.060187991708517075, "step": 3035 }, { "epoch": 0.7881619937694704, "grad_norm": 2.622335910797119, "learning_rate": 3.610163647866552e-06, "loss": 0.1158914715051651, "step": 3036 }, { "epoch": 0.7884215991692627, "grad_norm": 3.1881327629089355, "learning_rate": 3.6092894208544128e-06, "loss": 0.09522099792957306, "step": 3037 }, { "epoch": 0.788681204569055, "grad_norm": 1.7806915044784546, "learning_rate": 3.608415041939579e-06, "loss": 0.10816433280706406, "step": 3038 }, { "epoch": 0.7889408099688473, "grad_norm": 1.718750238418579, "learning_rate": 3.6075405112766213e-06, "loss": 0.11435578763484955, "step": 3039 }, { "epoch": 0.7892004153686397, "grad_norm": 3.1589269638061523, "learning_rate": 3.606665829020137e-06, "loss": 0.09884078800678253, "step": 3040 }, { "epoch": 0.789460020768432, "grad_norm": 2.0312108993530273, "learning_rate": 3.605790995324749e-06, "loss": 0.10343371331691742, "step": 3041 }, { "epoch": 0.7897196261682243, "grad_norm": 4.464409828186035, "learning_rate": 3.604916010345107e-06, "loss": 0.07402541488409042, "step": 3042 }, { "epoch": 0.7899792315680166, "grad_norm": 2.122652769088745, "learning_rate": 3.60404087423589e-06, "loss": 0.10801702737808228, "step": 3043 }, { "epoch": 0.7902388369678089, "grad_norm": 4.04167366027832, "learning_rate": 3.6031655871518017e-06, "loss": 0.0869961827993393, "step": 3044 }, { "epoch": 0.7904984423676013, "grad_norm": 2.2865307331085205, "learning_rate": 3.602290149247572e-06, "loss": 0.10443350672721863, "step": 3045 }, { "epoch": 0.7907580477673936, "grad_norm": 1.5821603536605835, "learning_rate": 3.6014145606779586e-06, "loss": 0.09638147056102753, "step": 3046 }, { "epoch": 0.7910176531671859, "grad_norm": 3.9024524688720703, "learning_rate": 3.600538821597746e-06, "loss": 0.0636407881975174, "step": 3047 }, { "epoch": 0.7912772585669782, "grad_norm": 3.525395393371582, "learning_rate": 3.599662932161745e-06, "loss": 0.10352268815040588, "step": 3048 }, { "epoch": 0.7915368639667705, "grad_norm": 4.869919776916504, "learning_rate": 3.5987868925247916e-06, "loss": 0.08826947212219238, "step": 3049 }, { "epoch": 0.7917964693665628, "grad_norm": 1.4795678853988647, "learning_rate": 3.597910702841751e-06, "loss": 0.11712922900915146, "step": 3050 }, { "epoch": 0.7920560747663551, "grad_norm": 3.6457910537719727, "learning_rate": 3.5970343632675115e-06, "loss": 0.06191570684313774, "step": 3051 }, { "epoch": 0.7923156801661475, "grad_norm": 2.782831907272339, "learning_rate": 3.5961578739569926e-06, "loss": 0.10179929435253143, "step": 3052 }, { "epoch": 0.7925752855659398, "grad_norm": 3.8145041465759277, "learning_rate": 3.5952812350651346e-06, "loss": 0.09644141793251038, "step": 3053 }, { "epoch": 0.7928348909657321, "grad_norm": 1.8937793970108032, "learning_rate": 3.5944044467469095e-06, "loss": 0.10096970200538635, "step": 3054 }, { "epoch": 0.7930944963655244, "grad_norm": 1.6431527137756348, "learning_rate": 3.5935275091573124e-06, "loss": 0.11308233439922333, "step": 3055 }, { "epoch": 0.7933541017653167, "grad_norm": 1.2572379112243652, "learning_rate": 3.5926504224513655e-06, "loss": 0.11159790307283401, "step": 3056 }, { "epoch": 0.793613707165109, "grad_norm": 2.131256580352783, "learning_rate": 3.591773186784118e-06, "loss": 0.11236591637134552, "step": 3057 }, { "epoch": 0.7938733125649013, "grad_norm": 3.25308895111084, "learning_rate": 3.590895802310646e-06, "loss": 0.07088253647089005, "step": 3058 }, { "epoch": 0.7941329179646937, "grad_norm": 2.2298245429992676, "learning_rate": 3.590018269186049e-06, "loss": 0.09106875956058502, "step": 3059 }, { "epoch": 0.794392523364486, "grad_norm": 2.6630029678344727, "learning_rate": 3.589140587565456e-06, "loss": 0.059052206575870514, "step": 3060 }, { "epoch": 0.7946521287642783, "grad_norm": 1.1406359672546387, "learning_rate": 3.5882627576040212e-06, "loss": 0.098714180290699, "step": 3061 }, { "epoch": 0.7949117341640706, "grad_norm": 2.9692649841308594, "learning_rate": 3.587384779456925e-06, "loss": 0.09605531394481659, "step": 3062 }, { "epoch": 0.7951713395638629, "grad_norm": 0.9660643339157104, "learning_rate": 3.5865066532793737e-06, "loss": 0.09509019553661346, "step": 3063 }, { "epoch": 0.7954309449636553, "grad_norm": 2.9851129055023193, "learning_rate": 3.5856283792265984e-06, "loss": 0.11479080468416214, "step": 3064 }, { "epoch": 0.7956905503634476, "grad_norm": 2.4156758785247803, "learning_rate": 3.5847499574538598e-06, "loss": 0.11895948648452759, "step": 3065 }, { "epoch": 0.7959501557632399, "grad_norm": 2.9414985179901123, "learning_rate": 3.5838713881164413e-06, "loss": 0.09655547142028809, "step": 3066 }, { "epoch": 0.7962097611630322, "grad_norm": 0.9962708353996277, "learning_rate": 3.582992671369655e-06, "loss": 0.1156526729464531, "step": 3067 }, { "epoch": 0.7964693665628245, "grad_norm": 2.283120632171631, "learning_rate": 3.5821138073688373e-06, "loss": 0.07222352921962738, "step": 3068 }, { "epoch": 0.7967289719626168, "grad_norm": 1.2907345294952393, "learning_rate": 3.581234796269351e-06, "loss": 0.08731058239936829, "step": 3069 }, { "epoch": 0.7969885773624091, "grad_norm": 3.057199716567993, "learning_rate": 3.5803556382265854e-06, "loss": 0.06899356842041016, "step": 3070 }, { "epoch": 0.7972481827622014, "grad_norm": 3.6880640983581543, "learning_rate": 3.5794763333959564e-06, "loss": 0.10278193652629852, "step": 3071 }, { "epoch": 0.7975077881619937, "grad_norm": 1.6629056930541992, "learning_rate": 3.578596881932904e-06, "loss": 0.11765210330486298, "step": 3072 }, { "epoch": 0.797767393561786, "grad_norm": 3.611692190170288, "learning_rate": 3.5777172839928943e-06, "loss": 0.109833724796772, "step": 3073 }, { "epoch": 0.7980269989615784, "grad_norm": 2.3036646842956543, "learning_rate": 3.5768375397314215e-06, "loss": 0.06439931690692902, "step": 3074 }, { "epoch": 0.7982866043613707, "grad_norm": 2.6754374504089355, "learning_rate": 3.5759576493040045e-06, "loss": 0.09832631796598434, "step": 3075 }, { "epoch": 0.7985462097611631, "grad_norm": 1.2295087575912476, "learning_rate": 3.5750776128661865e-06, "loss": 0.10923463106155396, "step": 3076 }, { "epoch": 0.7988058151609554, "grad_norm": 2.80505633354187, "learning_rate": 3.5741974305735377e-06, "loss": 0.11372864991426468, "step": 3077 }, { "epoch": 0.7990654205607477, "grad_norm": 2.82680344581604, "learning_rate": 3.5733171025816547e-06, "loss": 0.08534911274909973, "step": 3078 }, { "epoch": 0.79932502596054, "grad_norm": 1.3361382484436035, "learning_rate": 3.5724366290461604e-06, "loss": 0.10124523937702179, "step": 3079 }, { "epoch": 0.7995846313603323, "grad_norm": 3.3639371395111084, "learning_rate": 3.5715560101227013e-06, "loss": 0.0751732662320137, "step": 3080 }, { "epoch": 0.7998442367601246, "grad_norm": 1.7478219270706177, "learning_rate": 3.5706752459669506e-06, "loss": 0.11703671514987946, "step": 3081 }, { "epoch": 0.8001038421599169, "grad_norm": 2.526831865310669, "learning_rate": 3.5697943367346066e-06, "loss": 0.126052588224411, "step": 3082 }, { "epoch": 0.8003634475597092, "grad_norm": 1.0346978902816772, "learning_rate": 3.5689132825813953e-06, "loss": 0.11510608345270157, "step": 3083 }, { "epoch": 0.8006230529595015, "grad_norm": 3.2859034538269043, "learning_rate": 3.5680320836630667e-06, "loss": 0.0874086394906044, "step": 3084 }, { "epoch": 0.8008826583592938, "grad_norm": 2.2764499187469482, "learning_rate": 3.567150740135396e-06, "loss": 0.08553677797317505, "step": 3085 }, { "epoch": 0.8011422637590861, "grad_norm": 2.2251222133636475, "learning_rate": 3.566269252154184e-06, "loss": 0.09262358397245407, "step": 3086 }, { "epoch": 0.8014018691588785, "grad_norm": 3.0390827655792236, "learning_rate": 3.565387619875259e-06, "loss": 0.09512477368116379, "step": 3087 }, { "epoch": 0.8016614745586709, "grad_norm": 3.609097480773926, "learning_rate": 3.564505843454472e-06, "loss": 0.13425898551940918, "step": 3088 }, { "epoch": 0.8019210799584632, "grad_norm": 3.1772327423095703, "learning_rate": 3.5636239230477025e-06, "loss": 0.04855071008205414, "step": 3089 }, { "epoch": 0.8021806853582555, "grad_norm": 3.867915630340576, "learning_rate": 3.562741858810852e-06, "loss": 0.07306812703609467, "step": 3090 }, { "epoch": 0.8024402907580478, "grad_norm": 3.9015629291534424, "learning_rate": 3.561859650899851e-06, "loss": 0.08039970695972443, "step": 3091 }, { "epoch": 0.8026998961578401, "grad_norm": 4.139793872833252, "learning_rate": 3.560977299470653e-06, "loss": 0.06385506689548492, "step": 3092 }, { "epoch": 0.8029595015576324, "grad_norm": 1.9434043169021606, "learning_rate": 3.560094804679237e-06, "loss": 0.10455536842346191, "step": 3093 }, { "epoch": 0.8032191069574247, "grad_norm": 1.4213489294052124, "learning_rate": 3.5592121666816083e-06, "loss": 0.10782720148563385, "step": 3094 }, { "epoch": 0.803478712357217, "grad_norm": 1.8510541915893555, "learning_rate": 3.5583293856337963e-06, "loss": 0.09306807816028595, "step": 3095 }, { "epoch": 0.8037383177570093, "grad_norm": 2.7850894927978516, "learning_rate": 3.557446461691858e-06, "loss": 0.05951489880681038, "step": 3096 }, { "epoch": 0.8039979231568016, "grad_norm": 3.0771384239196777, "learning_rate": 3.5565633950118727e-06, "loss": 0.08433304727077484, "step": 3097 }, { "epoch": 0.8042575285565939, "grad_norm": 2.8691720962524414, "learning_rate": 3.5556801857499474e-06, "loss": 0.08895410597324371, "step": 3098 }, { "epoch": 0.8045171339563862, "grad_norm": 1.8892730474472046, "learning_rate": 3.554796834062213e-06, "loss": 0.11717177927494049, "step": 3099 }, { "epoch": 0.8047767393561787, "grad_norm": 1.8470412492752075, "learning_rate": 3.553913340104826e-06, "loss": 0.09919220209121704, "step": 3100 }, { "epoch": 0.805036344755971, "grad_norm": 5.093517780303955, "learning_rate": 3.5530297040339667e-06, "loss": 0.09606429934501648, "step": 3101 }, { "epoch": 0.8052959501557633, "grad_norm": 2.978886365890503, "learning_rate": 3.5521459260058435e-06, "loss": 0.06314022094011307, "step": 3102 }, { "epoch": 0.8055555555555556, "grad_norm": 2.095074415206909, "learning_rate": 3.5512620061766866e-06, "loss": 0.1082669049501419, "step": 3103 }, { "epoch": 0.8058151609553479, "grad_norm": 2.7637531757354736, "learning_rate": 3.550377944702755e-06, "loss": 0.1008409783244133, "step": 3104 }, { "epoch": 0.8060747663551402, "grad_norm": 1.9396451711654663, "learning_rate": 3.549493741740327e-06, "loss": 0.09942790865898132, "step": 3105 }, { "epoch": 0.8063343717549325, "grad_norm": 2.2054569721221924, "learning_rate": 3.5486093974457133e-06, "loss": 0.09403643012046814, "step": 3106 }, { "epoch": 0.8065939771547248, "grad_norm": 2.6678109169006348, "learning_rate": 3.5477249119752426e-06, "loss": 0.10706629604101181, "step": 3107 }, { "epoch": 0.8068535825545171, "grad_norm": 3.3354604244232178, "learning_rate": 3.546840285485273e-06, "loss": 0.10194860398769379, "step": 3108 }, { "epoch": 0.8071131879543094, "grad_norm": 2.553461790084839, "learning_rate": 3.5459555181321856e-06, "loss": 0.09108313918113708, "step": 3109 }, { "epoch": 0.8073727933541017, "grad_norm": 1.4970227479934692, "learning_rate": 3.545070610072389e-06, "loss": 0.09861800074577332, "step": 3110 }, { "epoch": 0.807632398753894, "grad_norm": 1.5211634635925293, "learning_rate": 3.5441855614623126e-06, "loss": 0.12168699502944946, "step": 3111 }, { "epoch": 0.8078920041536864, "grad_norm": 3.834301233291626, "learning_rate": 3.5433003724584143e-06, "loss": 0.10977654159069061, "step": 3112 }, { "epoch": 0.8081516095534788, "grad_norm": 3.8870046138763428, "learning_rate": 3.5424150432171727e-06, "loss": 0.10651694238185883, "step": 3113 }, { "epoch": 0.8084112149532711, "grad_norm": 3.8995018005371094, "learning_rate": 3.5415295738950965e-06, "loss": 0.08231543749570847, "step": 3114 }, { "epoch": 0.8086708203530634, "grad_norm": 2.828267812728882, "learning_rate": 3.5406439646487144e-06, "loss": 0.08933176100254059, "step": 3115 }, { "epoch": 0.8089304257528557, "grad_norm": 4.051664352416992, "learning_rate": 3.539758215634584e-06, "loss": 0.07670502364635468, "step": 3116 }, { "epoch": 0.809190031152648, "grad_norm": 1.6902697086334229, "learning_rate": 3.5388723270092824e-06, "loss": 0.10730834305286407, "step": 3117 }, { "epoch": 0.8094496365524403, "grad_norm": 3.3740780353546143, "learning_rate": 3.5379862989294174e-06, "loss": 0.07989045977592468, "step": 3118 }, { "epoch": 0.8097092419522326, "grad_norm": 2.570972204208374, "learning_rate": 3.537100131551617e-06, "loss": 0.10712545365095139, "step": 3119 }, { "epoch": 0.8099688473520249, "grad_norm": 2.9407219886779785, "learning_rate": 3.536213825032535e-06, "loss": 0.08957277238368988, "step": 3120 }, { "epoch": 0.8102284527518172, "grad_norm": 2.7370660305023193, "learning_rate": 3.535327379528851e-06, "loss": 0.09242887049913406, "step": 3121 }, { "epoch": 0.8104880581516095, "grad_norm": 2.021672487258911, "learning_rate": 3.5344407951972687e-06, "loss": 0.08682551980018616, "step": 3122 }, { "epoch": 0.8107476635514018, "grad_norm": 2.6364946365356445, "learning_rate": 3.5335540721945145e-06, "loss": 0.12303638458251953, "step": 3123 }, { "epoch": 0.8110072689511942, "grad_norm": 3.041273355484009, "learning_rate": 3.532667210677341e-06, "loss": 0.10228997468948364, "step": 3124 }, { "epoch": 0.8112668743509865, "grad_norm": 3.8143885135650635, "learning_rate": 3.531780210802526e-06, "loss": 0.08445747941732407, "step": 3125 }, { "epoch": 0.8115264797507789, "grad_norm": 2.077611207962036, "learning_rate": 3.5308930727268693e-06, "loss": 0.10852006077766418, "step": 3126 }, { "epoch": 0.8117860851505712, "grad_norm": 3.695866584777832, "learning_rate": 3.530005796607198e-06, "loss": 0.09678800404071808, "step": 3127 }, { "epoch": 0.8120456905503635, "grad_norm": 4.584177494049072, "learning_rate": 3.529118382600361e-06, "loss": 0.11717663705348969, "step": 3128 }, { "epoch": 0.8123052959501558, "grad_norm": 1.2189722061157227, "learning_rate": 3.5282308308632348e-06, "loss": 0.1080881655216217, "step": 3129 }, { "epoch": 0.8125649013499481, "grad_norm": 3.3322770595550537, "learning_rate": 3.527343141552715e-06, "loss": 0.0779213011264801, "step": 3130 }, { "epoch": 0.8128245067497404, "grad_norm": 1.5680060386657715, "learning_rate": 3.526455314825728e-06, "loss": 0.1098969429731369, "step": 3131 }, { "epoch": 0.8130841121495327, "grad_norm": 1.3196884393692017, "learning_rate": 3.525567350839219e-06, "loss": 0.0939682349562645, "step": 3132 }, { "epoch": 0.813343717549325, "grad_norm": 3.1863255500793457, "learning_rate": 3.5246792497501615e-06, "loss": 0.11531957983970642, "step": 3133 }, { "epoch": 0.8136033229491173, "grad_norm": 2.3954906463623047, "learning_rate": 3.5237910117155494e-06, "loss": 0.07746727764606476, "step": 3134 }, { "epoch": 0.8138629283489096, "grad_norm": 2.7822422981262207, "learning_rate": 3.5229026368924043e-06, "loss": 0.08588462322950363, "step": 3135 }, { "epoch": 0.814122533748702, "grad_norm": 2.1958253383636475, "learning_rate": 3.5220141254377697e-06, "loss": 0.027894796803593636, "step": 3136 }, { "epoch": 0.8143821391484943, "grad_norm": 1.5405951738357544, "learning_rate": 3.521125477508715e-06, "loss": 0.11247137933969498, "step": 3137 }, { "epoch": 0.8146417445482866, "grad_norm": 1.6926440000534058, "learning_rate": 3.5202366932623327e-06, "loss": 0.10435294359922409, "step": 3138 }, { "epoch": 0.814901349948079, "grad_norm": 0.9669217467308044, "learning_rate": 3.519347772855738e-06, "loss": 0.10613667964935303, "step": 3139 }, { "epoch": 0.8151609553478713, "grad_norm": 3.2839972972869873, "learning_rate": 3.518458716446074e-06, "loss": 0.08819913864135742, "step": 3140 }, { "epoch": 0.8154205607476636, "grad_norm": 1.6951308250427246, "learning_rate": 3.517569524190503e-06, "loss": 0.09690261632204056, "step": 3141 }, { "epoch": 0.8156801661474559, "grad_norm": 2.856832981109619, "learning_rate": 3.516680196246216e-06, "loss": 0.11062329262495041, "step": 3142 }, { "epoch": 0.8159397715472482, "grad_norm": 1.9256727695465088, "learning_rate": 3.5157907327704244e-06, "loss": 0.10804706811904907, "step": 3143 }, { "epoch": 0.8161993769470405, "grad_norm": 4.441931247711182, "learning_rate": 3.514901133920366e-06, "loss": 0.09777020663022995, "step": 3144 }, { "epoch": 0.8164589823468328, "grad_norm": 1.071760892868042, "learning_rate": 3.5140113998533015e-06, "loss": 0.11236352473497391, "step": 3145 }, { "epoch": 0.8167185877466251, "grad_norm": 1.407532811164856, "learning_rate": 3.513121530726515e-06, "loss": 0.1044665277004242, "step": 3146 }, { "epoch": 0.8169781931464174, "grad_norm": 1.4046117067337036, "learning_rate": 3.5122315266973146e-06, "loss": 0.11532819271087646, "step": 3147 }, { "epoch": 0.8172377985462098, "grad_norm": 3.1963682174682617, "learning_rate": 3.511341387923032e-06, "loss": 0.06216800585389137, "step": 3148 }, { "epoch": 0.8174974039460021, "grad_norm": 4.110545635223389, "learning_rate": 3.5104511145610255e-06, "loss": 0.13542546331882477, "step": 3149 }, { "epoch": 0.8177570093457944, "grad_norm": 4.4871506690979, "learning_rate": 3.509560706768674e-06, "loss": 0.10526230931282043, "step": 3150 }, { "epoch": 0.8180166147455867, "grad_norm": 1.5719740390777588, "learning_rate": 3.508670164703382e-06, "loss": 0.11016601324081421, "step": 3151 }, { "epoch": 0.818276220145379, "grad_norm": 2.4939358234405518, "learning_rate": 3.5077794885225747e-06, "loss": 0.10241538286209106, "step": 3152 }, { "epoch": 0.8185358255451713, "grad_norm": 2.8070552349090576, "learning_rate": 3.5068886783837047e-06, "loss": 0.09658431261777878, "step": 3153 }, { "epoch": 0.8187954309449637, "grad_norm": 3.7469279766082764, "learning_rate": 3.505997734444248e-06, "loss": 0.05642658472061157, "step": 3154 }, { "epoch": 0.819055036344756, "grad_norm": 4.241229057312012, "learning_rate": 3.505106656861701e-06, "loss": 0.08410907536745071, "step": 3155 }, { "epoch": 0.8193146417445483, "grad_norm": 2.703064203262329, "learning_rate": 3.5042154457935867e-06, "loss": 0.0960293710231781, "step": 3156 }, { "epoch": 0.8195742471443406, "grad_norm": 3.3968350887298584, "learning_rate": 3.5033241013974512e-06, "loss": 0.05996100232005119, "step": 3157 }, { "epoch": 0.8198338525441329, "grad_norm": 3.094583511352539, "learning_rate": 3.502432623830863e-06, "loss": 0.1204279214143753, "step": 3158 }, { "epoch": 0.8200934579439252, "grad_norm": 3.6827447414398193, "learning_rate": 3.5015410132514153e-06, "loss": 0.07073773443698883, "step": 3159 }, { "epoch": 0.8203530633437176, "grad_norm": 1.2092262506484985, "learning_rate": 3.5006492698167256e-06, "loss": 0.1098293885588646, "step": 3160 }, { "epoch": 0.8206126687435099, "grad_norm": 2.205807685852051, "learning_rate": 3.499757393684431e-06, "loss": 0.09025605022907257, "step": 3161 }, { "epoch": 0.8208722741433022, "grad_norm": 1.421751618385315, "learning_rate": 3.4988653850121975e-06, "loss": 0.1128188818693161, "step": 3162 }, { "epoch": 0.8211318795430945, "grad_norm": 1.196070909500122, "learning_rate": 3.4979732439577103e-06, "loss": 0.1161685436964035, "step": 3163 }, { "epoch": 0.8213914849428868, "grad_norm": 2.6414272785186768, "learning_rate": 3.4970809706786802e-06, "loss": 0.1283348798751831, "step": 3164 }, { "epoch": 0.8216510903426791, "grad_norm": 2.0413174629211426, "learning_rate": 3.49618856533284e-06, "loss": 0.08769761025905609, "step": 3165 }, { "epoch": 0.8219106957424714, "grad_norm": 3.6112866401672363, "learning_rate": 3.4952960280779463e-06, "loss": 0.0675768032670021, "step": 3166 }, { "epoch": 0.8221703011422637, "grad_norm": 1.6076034307479858, "learning_rate": 3.494403359071781e-06, "loss": 0.10378862172365189, "step": 3167 }, { "epoch": 0.822429906542056, "grad_norm": 5.000674724578857, "learning_rate": 3.4935105584721455e-06, "loss": 0.12210887670516968, "step": 3168 }, { "epoch": 0.8226895119418484, "grad_norm": 2.9109082221984863, "learning_rate": 3.4926176264368683e-06, "loss": 0.09008283913135529, "step": 3169 }, { "epoch": 0.8229491173416407, "grad_norm": 4.045159816741943, "learning_rate": 3.491724563123798e-06, "loss": 0.06533791124820709, "step": 3170 }, { "epoch": 0.823208722741433, "grad_norm": 1.2668087482452393, "learning_rate": 3.4908313686908074e-06, "loss": 0.10976213216781616, "step": 3171 }, { "epoch": 0.8234683281412254, "grad_norm": 1.7362091541290283, "learning_rate": 3.4899380432957948e-06, "loss": 0.10298481583595276, "step": 3172 }, { "epoch": 0.8237279335410177, "grad_norm": 2.0393600463867188, "learning_rate": 3.4890445870966783e-06, "loss": 0.10555922985076904, "step": 3173 }, { "epoch": 0.82398753894081, "grad_norm": 3.464128255844116, "learning_rate": 3.4881510002514e-06, "loss": 0.06463884562253952, "step": 3174 }, { "epoch": 0.8242471443406023, "grad_norm": 0.9902530908584595, "learning_rate": 3.487257282917927e-06, "loss": 0.1048530861735344, "step": 3175 }, { "epoch": 0.8245067497403946, "grad_norm": 1.4879450798034668, "learning_rate": 3.4863634352542465e-06, "loss": 0.10334643721580505, "step": 3176 }, { "epoch": 0.8247663551401869, "grad_norm": 2.6632885932922363, "learning_rate": 3.4854694574183716e-06, "loss": 0.08625948429107666, "step": 3177 }, { "epoch": 0.8250259605399792, "grad_norm": 2.601147413253784, "learning_rate": 3.4845753495683366e-06, "loss": 0.09439734369516373, "step": 3178 }, { "epoch": 0.8252855659397715, "grad_norm": 1.7607964277267456, "learning_rate": 3.4836811118621995e-06, "loss": 0.11130322515964508, "step": 3179 }, { "epoch": 0.8255451713395638, "grad_norm": 3.083505153656006, "learning_rate": 3.482786744458041e-06, "loss": 0.06640174984931946, "step": 3180 }, { "epoch": 0.8258047767393562, "grad_norm": 1.987746238708496, "learning_rate": 3.4818922475139643e-06, "loss": 0.09508831799030304, "step": 3181 }, { "epoch": 0.8260643821391485, "grad_norm": 3.420213222503662, "learning_rate": 3.4809976211880963e-06, "loss": 0.09379607439041138, "step": 3182 }, { "epoch": 0.8263239875389408, "grad_norm": 1.6662242412567139, "learning_rate": 3.4801028656385867e-06, "loss": 0.09600570797920227, "step": 3183 }, { "epoch": 0.8265835929387332, "grad_norm": 4.112314701080322, "learning_rate": 3.4792079810236078e-06, "loss": 0.07842263579368591, "step": 3184 }, { "epoch": 0.8268431983385255, "grad_norm": 1.758486032485962, "learning_rate": 3.4783129675013545e-06, "loss": 0.09953585267066956, "step": 3185 }, { "epoch": 0.8271028037383178, "grad_norm": 1.862841248512268, "learning_rate": 3.4774178252300445e-06, "loss": 0.11027784645557404, "step": 3186 }, { "epoch": 0.8273624091381101, "grad_norm": 1.9874886274337769, "learning_rate": 3.476522554367918e-06, "loss": 0.11407982558012009, "step": 3187 }, { "epoch": 0.8276220145379024, "grad_norm": 3.4903571605682373, "learning_rate": 3.4756271550732397e-06, "loss": 0.09232089668512344, "step": 3188 }, { "epoch": 0.8278816199376947, "grad_norm": 1.59999680519104, "learning_rate": 3.474731627504296e-06, "loss": 0.09941224753856659, "step": 3189 }, { "epoch": 0.828141225337487, "grad_norm": 3.3069353103637695, "learning_rate": 3.4738359718193932e-06, "loss": 0.07760922610759735, "step": 3190 }, { "epoch": 0.8284008307372793, "grad_norm": 1.7322760820388794, "learning_rate": 3.4729401881768655e-06, "loss": 0.10185371339321136, "step": 3191 }, { "epoch": 0.8286604361370716, "grad_norm": 1.7809146642684937, "learning_rate": 3.4720442767350637e-06, "loss": 0.10061755031347275, "step": 3192 }, { "epoch": 0.8289200415368639, "grad_norm": 3.3413496017456055, "learning_rate": 3.4711482376523687e-06, "loss": 0.08746225386857986, "step": 3193 }, { "epoch": 0.8291796469366562, "grad_norm": 2.4500725269317627, "learning_rate": 3.4702520710871757e-06, "loss": 0.09654439985752106, "step": 3194 }, { "epoch": 0.8294392523364486, "grad_norm": 2.048347234725952, "learning_rate": 3.469355777197909e-06, "loss": 0.10693778842687607, "step": 3195 }, { "epoch": 0.829698857736241, "grad_norm": 2.236107587814331, "learning_rate": 3.468459356143012e-06, "loss": 0.10085071623325348, "step": 3196 }, { "epoch": 0.8299584631360333, "grad_norm": 1.3640494346618652, "learning_rate": 3.4675628080809508e-06, "loss": 0.11254099756479263, "step": 3197 }, { "epoch": 0.8302180685358256, "grad_norm": 2.3787336349487305, "learning_rate": 3.466666133170215e-06, "loss": 0.09205961227416992, "step": 3198 }, { "epoch": 0.8304776739356179, "grad_norm": 3.1469950675964355, "learning_rate": 3.4657693315693177e-06, "loss": 0.07336366176605225, "step": 3199 }, { "epoch": 0.8307372793354102, "grad_norm": 3.69081711769104, "learning_rate": 3.4648724034367903e-06, "loss": 0.07044047117233276, "step": 3200 }, { "epoch": 0.8309968847352025, "grad_norm": 3.889024496078491, "learning_rate": 3.4639753489311905e-06, "loss": 0.055554233491420746, "step": 3201 }, { "epoch": 0.8312564901349948, "grad_norm": 3.8649983406066895, "learning_rate": 3.4630781682110975e-06, "loss": 0.056577034294605255, "step": 3202 }, { "epoch": 0.8315160955347871, "grad_norm": 3.8682098388671875, "learning_rate": 3.462180861435111e-06, "loss": 0.07431219518184662, "step": 3203 }, { "epoch": 0.8317757009345794, "grad_norm": 3.327859878540039, "learning_rate": 3.4612834287618556e-06, "loss": 0.09422308206558228, "step": 3204 }, { "epoch": 0.8320353063343717, "grad_norm": 1.1924868822097778, "learning_rate": 3.4603858703499747e-06, "loss": 0.11402273178100586, "step": 3205 }, { "epoch": 0.832294911734164, "grad_norm": 1.6565929651260376, "learning_rate": 3.4594881863581397e-06, "loss": 0.11482619494199753, "step": 3206 }, { "epoch": 0.8325545171339563, "grad_norm": 1.2619915008544922, "learning_rate": 3.4585903769450372e-06, "loss": 0.10790901631116867, "step": 3207 }, { "epoch": 0.8328141225337488, "grad_norm": 3.134413003921509, "learning_rate": 3.4576924422693813e-06, "loss": 0.10051864385604858, "step": 3208 }, { "epoch": 0.8330737279335411, "grad_norm": 3.7520508766174316, "learning_rate": 3.4567943824899055e-06, "loss": 0.05937284231185913, "step": 3209 }, { "epoch": 0.8333333333333334, "grad_norm": 1.71026611328125, "learning_rate": 3.455896197765366e-06, "loss": 0.09484432637691498, "step": 3210 }, { "epoch": 0.8335929387331257, "grad_norm": 3.1961073875427246, "learning_rate": 3.454997888254541e-06, "loss": 0.09675239026546478, "step": 3211 }, { "epoch": 0.833852544132918, "grad_norm": 3.913480043411255, "learning_rate": 3.4540994541162333e-06, "loss": 0.05442925915122032, "step": 3212 }, { "epoch": 0.8341121495327103, "grad_norm": 3.608778238296509, "learning_rate": 3.4532008955092627e-06, "loss": 0.04632272198796272, "step": 3213 }, { "epoch": 0.8343717549325026, "grad_norm": 2.947089195251465, "learning_rate": 3.452302212592475e-06, "loss": 0.10831563174724579, "step": 3214 }, { "epoch": 0.8346313603322949, "grad_norm": 2.128103256225586, "learning_rate": 3.4514034055247366e-06, "loss": 0.09531442821025848, "step": 3215 }, { "epoch": 0.8348909657320872, "grad_norm": 2.8673532009124756, "learning_rate": 3.450504474464937e-06, "loss": 0.05387824773788452, "step": 3216 }, { "epoch": 0.8351505711318795, "grad_norm": 3.215245485305786, "learning_rate": 3.4496054195719847e-06, "loss": 0.10444444417953491, "step": 3217 }, { "epoch": 0.8354101765316718, "grad_norm": 3.623549222946167, "learning_rate": 3.4487062410048135e-06, "loss": 0.10061635076999664, "step": 3218 }, { "epoch": 0.8356697819314641, "grad_norm": 3.1634292602539062, "learning_rate": 3.447806938922377e-06, "loss": 0.08498744666576385, "step": 3219 }, { "epoch": 0.8359293873312564, "grad_norm": 3.6336002349853516, "learning_rate": 3.4469075134836516e-06, "loss": 0.1292828619480133, "step": 3220 }, { "epoch": 0.8361889927310489, "grad_norm": 3.1887853145599365, "learning_rate": 3.4460079648476354e-06, "loss": 0.09886769950389862, "step": 3221 }, { "epoch": 0.8364485981308412, "grad_norm": 3.0472681522369385, "learning_rate": 3.4451082931733463e-06, "loss": 0.06200874224305153, "step": 3222 }, { "epoch": 0.8367082035306335, "grad_norm": 2.909597635269165, "learning_rate": 3.444208498619827e-06, "loss": 0.09383311122655869, "step": 3223 }, { "epoch": 0.8369678089304258, "grad_norm": 3.308584451675415, "learning_rate": 3.4433085813461407e-06, "loss": 0.11115659773349762, "step": 3224 }, { "epoch": 0.8372274143302181, "grad_norm": 2.6823251247406006, "learning_rate": 3.4424085415113722e-06, "loss": 0.0953298956155777, "step": 3225 }, { "epoch": 0.8374870197300104, "grad_norm": 0.9933691024780273, "learning_rate": 3.4415083792746284e-06, "loss": 0.10272541642189026, "step": 3226 }, { "epoch": 0.8377466251298027, "grad_norm": 1.958242416381836, "learning_rate": 3.440608094795035e-06, "loss": 0.09514497965574265, "step": 3227 }, { "epoch": 0.838006230529595, "grad_norm": 2.538421154022217, "learning_rate": 3.4397076882317438e-06, "loss": 0.06369639188051224, "step": 3228 }, { "epoch": 0.8382658359293873, "grad_norm": 2.310917854309082, "learning_rate": 3.438807159743926e-06, "loss": 0.08330902457237244, "step": 3229 }, { "epoch": 0.8385254413291796, "grad_norm": 2.5897672176361084, "learning_rate": 3.4379065094907734e-06, "loss": 0.107622429728508, "step": 3230 }, { "epoch": 0.8387850467289719, "grad_norm": 1.4829919338226318, "learning_rate": 3.4370057376315e-06, "loss": 0.12243475764989853, "step": 3231 }, { "epoch": 0.8390446521287642, "grad_norm": 3.102478265762329, "learning_rate": 3.4361048443253443e-06, "loss": 0.08164924383163452, "step": 3232 }, { "epoch": 0.8393042575285566, "grad_norm": 3.253021240234375, "learning_rate": 3.4352038297315616e-06, "loss": 0.08944068849086761, "step": 3233 }, { "epoch": 0.839563862928349, "grad_norm": 3.2331812381744385, "learning_rate": 3.4343026940094297e-06, "loss": 0.0795150101184845, "step": 3234 }, { "epoch": 0.8398234683281413, "grad_norm": 3.992926597595215, "learning_rate": 3.433401437318251e-06, "loss": 0.13797099888324738, "step": 3235 }, { "epoch": 0.8400830737279336, "grad_norm": 1.729379653930664, "learning_rate": 3.4325000598173454e-06, "loss": 0.09451854228973389, "step": 3236 }, { "epoch": 0.8403426791277259, "grad_norm": 0.9562795162200928, "learning_rate": 3.4315985616660572e-06, "loss": 0.10918404906988144, "step": 3237 }, { "epoch": 0.8406022845275182, "grad_norm": 3.425522565841675, "learning_rate": 3.430696943023749e-06, "loss": 0.09876430034637451, "step": 3238 }, { "epoch": 0.8408618899273105, "grad_norm": 2.9523098468780518, "learning_rate": 3.429795204049808e-06, "loss": 0.10235343873500824, "step": 3239 }, { "epoch": 0.8411214953271028, "grad_norm": 3.595292091369629, "learning_rate": 3.428893344903639e-06, "loss": 0.08256131410598755, "step": 3240 }, { "epoch": 0.8413811007268951, "grad_norm": 3.423874616622925, "learning_rate": 3.427991365744672e-06, "loss": 0.05413984879851341, "step": 3241 }, { "epoch": 0.8416407061266874, "grad_norm": 2.7797186374664307, "learning_rate": 3.4270892667323553e-06, "loss": 0.09607769548892975, "step": 3242 }, { "epoch": 0.8419003115264797, "grad_norm": 3.2491767406463623, "learning_rate": 3.42618704802616e-06, "loss": 0.09354943037033081, "step": 3243 }, { "epoch": 0.842159916926272, "grad_norm": 2.658266067504883, "learning_rate": 3.4252847097855764e-06, "loss": 0.10528764128684998, "step": 3244 }, { "epoch": 0.8424195223260644, "grad_norm": 3.6334927082061768, "learning_rate": 3.4243822521701183e-06, "loss": 0.07189391553401947, "step": 3245 }, { "epoch": 0.8426791277258567, "grad_norm": 1.7779943943023682, "learning_rate": 3.4234796753393197e-06, "loss": 0.09795734286308289, "step": 3246 }, { "epoch": 0.842938733125649, "grad_norm": 2.0305848121643066, "learning_rate": 3.4225769794527354e-06, "loss": 0.09176211059093475, "step": 3247 }, { "epoch": 0.8431983385254413, "grad_norm": 2.9404585361480713, "learning_rate": 3.4216741646699396e-06, "loss": 0.09422975033521652, "step": 3248 }, { "epoch": 0.8434579439252337, "grad_norm": 4.20827054977417, "learning_rate": 3.420771231150532e-06, "loss": 0.04966890066862106, "step": 3249 }, { "epoch": 0.843717549325026, "grad_norm": 2.646178722381592, "learning_rate": 3.4198681790541296e-06, "loss": 0.07673788070678711, "step": 3250 }, { "epoch": 0.8439771547248183, "grad_norm": 3.388763427734375, "learning_rate": 3.418965008540371e-06, "loss": 0.08916642516851425, "step": 3251 }, { "epoch": 0.8442367601246106, "grad_norm": 1.820388674736023, "learning_rate": 3.418061719768916e-06, "loss": 0.08790456503629684, "step": 3252 }, { "epoch": 0.8444963655244029, "grad_norm": 3.6092820167541504, "learning_rate": 3.4171583128994453e-06, "loss": 0.08567577600479126, "step": 3253 }, { "epoch": 0.8447559709241952, "grad_norm": 2.7330398559570312, "learning_rate": 3.41625478809166e-06, "loss": 0.1137879490852356, "step": 3254 }, { "epoch": 0.8450155763239875, "grad_norm": 3.2030372619628906, "learning_rate": 3.4153511455052844e-06, "loss": 0.11406273394823074, "step": 3255 }, { "epoch": 0.8452751817237798, "grad_norm": 1.8132195472717285, "learning_rate": 3.414447385300061e-06, "loss": 0.09675782918930054, "step": 3256 }, { "epoch": 0.8455347871235722, "grad_norm": 3.6000068187713623, "learning_rate": 3.4135435076357537e-06, "loss": 0.08419261872768402, "step": 3257 }, { "epoch": 0.8457943925233645, "grad_norm": 2.834402084350586, "learning_rate": 3.412639512672147e-06, "loss": 0.058718908578157425, "step": 3258 }, { "epoch": 0.8460539979231568, "grad_norm": 2.8022890090942383, "learning_rate": 3.411735400569046e-06, "loss": 0.10538727045059204, "step": 3259 }, { "epoch": 0.8463136033229491, "grad_norm": 3.064936637878418, "learning_rate": 3.41083117148628e-06, "loss": 0.11471932381391525, "step": 3260 }, { "epoch": 0.8465732087227414, "grad_norm": 3.602874517440796, "learning_rate": 3.409926825583692e-06, "loss": 0.08584973216056824, "step": 3261 }, { "epoch": 0.8468328141225338, "grad_norm": 2.323155403137207, "learning_rate": 3.4090223630211523e-06, "loss": 0.0586712583899498, "step": 3262 }, { "epoch": 0.8470924195223261, "grad_norm": 1.078328013420105, "learning_rate": 3.4081177839585485e-06, "loss": 0.10928191989660263, "step": 3263 }, { "epoch": 0.8473520249221184, "grad_norm": 1.7934846878051758, "learning_rate": 3.407213088555789e-06, "loss": 0.0946338027715683, "step": 3264 }, { "epoch": 0.8476116303219107, "grad_norm": 2.7738215923309326, "learning_rate": 3.4063082769728045e-06, "loss": 0.06527368724346161, "step": 3265 }, { "epoch": 0.847871235721703, "grad_norm": 4.018083572387695, "learning_rate": 3.405403349369544e-06, "loss": 0.06766581535339355, "step": 3266 }, { "epoch": 0.8481308411214953, "grad_norm": 3.2676050662994385, "learning_rate": 3.4044983059059766e-06, "loss": 0.07313887774944305, "step": 3267 }, { "epoch": 0.8483904465212876, "grad_norm": 1.876236915588379, "learning_rate": 3.4035931467420964e-06, "loss": 0.11883509159088135, "step": 3268 }, { "epoch": 0.84865005192108, "grad_norm": 2.8113763332366943, "learning_rate": 3.402687872037912e-06, "loss": 0.09075891226530075, "step": 3269 }, { "epoch": 0.8489096573208723, "grad_norm": 4.929562091827393, "learning_rate": 3.401782481953457e-06, "loss": 0.05689570680260658, "step": 3270 }, { "epoch": 0.8491692627206646, "grad_norm": 2.722412347793579, "learning_rate": 3.4008769766487826e-06, "loss": 0.10247932374477386, "step": 3271 }, { "epoch": 0.8494288681204569, "grad_norm": 1.2672629356384277, "learning_rate": 3.399971356283963e-06, "loss": 0.10473863780498505, "step": 3272 }, { "epoch": 0.8496884735202492, "grad_norm": 4.146210193634033, "learning_rate": 3.399065621019089e-06, "loss": 0.057955265045166016, "step": 3273 }, { "epoch": 0.8499480789200415, "grad_norm": 2.3867604732513428, "learning_rate": 3.3981597710142757e-06, "loss": 0.08499515056610107, "step": 3274 }, { "epoch": 0.8502076843198338, "grad_norm": 1.4394584894180298, "learning_rate": 3.3972538064296545e-06, "loss": 0.09808111935853958, "step": 3275 }, { "epoch": 0.8504672897196262, "grad_norm": 2.022594451904297, "learning_rate": 3.396347727425381e-06, "loss": 0.10609080642461777, "step": 3276 }, { "epoch": 0.8507268951194185, "grad_norm": 3.2172231674194336, "learning_rate": 3.3954415341616285e-06, "loss": 0.11303733289241791, "step": 3277 }, { "epoch": 0.8509865005192108, "grad_norm": 3.771385431289673, "learning_rate": 3.394535226798592e-06, "loss": 0.0609101802110672, "step": 3278 }, { "epoch": 0.8512461059190031, "grad_norm": 3.502891778945923, "learning_rate": 3.393628805496485e-06, "loss": 0.07903899252414703, "step": 3279 }, { "epoch": 0.8515057113187954, "grad_norm": 1.31452476978302, "learning_rate": 3.3927222704155415e-06, "loss": 0.11323224008083344, "step": 3280 }, { "epoch": 0.8517653167185878, "grad_norm": 1.5911428928375244, "learning_rate": 3.3918156217160182e-06, "loss": 0.08553754538297653, "step": 3281 }, { "epoch": 0.8520249221183801, "grad_norm": 3.0781614780426025, "learning_rate": 3.3909088595581884e-06, "loss": 0.11194103956222534, "step": 3282 }, { "epoch": 0.8522845275181724, "grad_norm": 2.952219247817993, "learning_rate": 3.3900019841023476e-06, "loss": 0.08902988582849503, "step": 3283 }, { "epoch": 0.8525441329179647, "grad_norm": 3.4117813110351562, "learning_rate": 3.389094995508809e-06, "loss": 0.09517225623130798, "step": 3284 }, { "epoch": 0.852803738317757, "grad_norm": 4.598580837249756, "learning_rate": 3.3881878939379086e-06, "loss": 0.0673670694231987, "step": 3285 }, { "epoch": 0.8530633437175493, "grad_norm": 3.4107131958007812, "learning_rate": 3.3872806795500014e-06, "loss": 0.05941835045814514, "step": 3286 }, { "epoch": 0.8533229491173416, "grad_norm": 4.036614894866943, "learning_rate": 3.386373352505462e-06, "loss": 0.09982068836688995, "step": 3287 }, { "epoch": 0.8535825545171339, "grad_norm": 2.6347739696502686, "learning_rate": 3.385465912964685e-06, "loss": 0.12063336372375488, "step": 3288 }, { "epoch": 0.8538421599169262, "grad_norm": 4.1044511795043945, "learning_rate": 3.3845583610880844e-06, "loss": 0.08558182418346405, "step": 3289 }, { "epoch": 0.8541017653167186, "grad_norm": 3.5441079139709473, "learning_rate": 3.383650697036095e-06, "loss": 0.08236090838909149, "step": 3290 }, { "epoch": 0.8543613707165109, "grad_norm": 1.1442368030548096, "learning_rate": 3.3827429209691722e-06, "loss": 0.10442593693733215, "step": 3291 }, { "epoch": 0.8546209761163032, "grad_norm": 1.9250521659851074, "learning_rate": 3.3818350330477874e-06, "loss": 0.10629092156887054, "step": 3292 }, { "epoch": 0.8548805815160956, "grad_norm": 2.143454074859619, "learning_rate": 3.3809270334324364e-06, "loss": 0.11197120696306229, "step": 3293 }, { "epoch": 0.8551401869158879, "grad_norm": 2.0886454582214355, "learning_rate": 3.3800189222836316e-06, "loss": 0.08842878043651581, "step": 3294 }, { "epoch": 0.8553997923156802, "grad_norm": 1.9465221166610718, "learning_rate": 3.3791106997619078e-06, "loss": 0.0825609341263771, "step": 3295 }, { "epoch": 0.8556593977154725, "grad_norm": 2.0844063758850098, "learning_rate": 3.378202366027817e-06, "loss": 0.10414554178714752, "step": 3296 }, { "epoch": 0.8559190031152648, "grad_norm": 2.467334508895874, "learning_rate": 3.3772939212419314e-06, "loss": 0.11004841327667236, "step": 3297 }, { "epoch": 0.8561786085150571, "grad_norm": 2.7897863388061523, "learning_rate": 3.3763853655648436e-06, "loss": 0.07330596446990967, "step": 3298 }, { "epoch": 0.8564382139148494, "grad_norm": 3.201282024383545, "learning_rate": 3.3754766991571657e-06, "loss": 0.049056168645620346, "step": 3299 }, { "epoch": 0.8566978193146417, "grad_norm": 3.343759298324585, "learning_rate": 3.3745679221795284e-06, "loss": 0.08229372650384903, "step": 3300 }, { "epoch": 0.856957424714434, "grad_norm": 1.3195141553878784, "learning_rate": 3.3736590347925842e-06, "loss": 0.0992126315832138, "step": 3301 }, { "epoch": 0.8572170301142263, "grad_norm": 1.790713906288147, "learning_rate": 3.3727500371570014e-06, "loss": 0.08827859908342361, "step": 3302 }, { "epoch": 0.8574766355140186, "grad_norm": 1.7931877374649048, "learning_rate": 3.371840929433472e-06, "loss": 0.13591597974300385, "step": 3303 }, { "epoch": 0.857736240913811, "grad_norm": 2.0205376148223877, "learning_rate": 3.370931711782704e-06, "loss": 0.09319691359996796, "step": 3304 }, { "epoch": 0.8579958463136034, "grad_norm": 1.9821419715881348, "learning_rate": 3.370022384365427e-06, "loss": 0.09644585102796555, "step": 3305 }, { "epoch": 0.8582554517133957, "grad_norm": 1.4464858770370483, "learning_rate": 3.3691129473423895e-06, "loss": 0.096053346991539, "step": 3306 }, { "epoch": 0.858515057113188, "grad_norm": 3.257643461227417, "learning_rate": 3.368203400874358e-06, "loss": 0.07644183933734894, "step": 3307 }, { "epoch": 0.8587746625129803, "grad_norm": 1.7354676723480225, "learning_rate": 3.3672937451221206e-06, "loss": 0.08541695773601532, "step": 3308 }, { "epoch": 0.8590342679127726, "grad_norm": 4.440587520599365, "learning_rate": 3.366383980246483e-06, "loss": 0.059323713183403015, "step": 3309 }, { "epoch": 0.8592938733125649, "grad_norm": 5.235567092895508, "learning_rate": 3.3654741064082712e-06, "loss": 0.0629788488149643, "step": 3310 }, { "epoch": 0.8595534787123572, "grad_norm": 1.670559287071228, "learning_rate": 3.3645641237683295e-06, "loss": 0.10251232981681824, "step": 3311 }, { "epoch": 0.8598130841121495, "grad_norm": 2.6244633197784424, "learning_rate": 3.363654032487523e-06, "loss": 0.09092941135168076, "step": 3312 }, { "epoch": 0.8600726895119418, "grad_norm": 1.4347273111343384, "learning_rate": 3.362743832726735e-06, "loss": 0.09790033102035522, "step": 3313 }, { "epoch": 0.8603322949117341, "grad_norm": 3.9705810546875, "learning_rate": 3.3618335246468657e-06, "loss": 0.04964379221200943, "step": 3314 }, { "epoch": 0.8605919003115264, "grad_norm": 3.5748422145843506, "learning_rate": 3.360923108408839e-06, "loss": 0.06593935936689377, "step": 3315 }, { "epoch": 0.8608515057113187, "grad_norm": 1.387990117073059, "learning_rate": 3.3600125841735955e-06, "loss": 0.1151074767112732, "step": 3316 }, { "epoch": 0.8611111111111112, "grad_norm": 2.0935168266296387, "learning_rate": 3.359101952102094e-06, "loss": 0.10297118127346039, "step": 3317 }, { "epoch": 0.8613707165109035, "grad_norm": 4.015507698059082, "learning_rate": 3.3581912123553144e-06, "loss": 0.06397296488285065, "step": 3318 }, { "epoch": 0.8616303219106958, "grad_norm": 3.4163312911987305, "learning_rate": 3.357280365094254e-06, "loss": 0.1196485087275505, "step": 3319 }, { "epoch": 0.8618899273104881, "grad_norm": 1.691624641418457, "learning_rate": 3.3563694104799292e-06, "loss": 0.1078580915927887, "step": 3320 }, { "epoch": 0.8621495327102804, "grad_norm": 3.622541666030884, "learning_rate": 3.3554583486733776e-06, "loss": 0.09339050203561783, "step": 3321 }, { "epoch": 0.8624091381100727, "grad_norm": 1.7345528602600098, "learning_rate": 3.3545471798356526e-06, "loss": 0.08157406747341156, "step": 3322 }, { "epoch": 0.862668743509865, "grad_norm": 3.7873213291168213, "learning_rate": 3.353635904127829e-06, "loss": 0.08054625988006592, "step": 3323 }, { "epoch": 0.8629283489096573, "grad_norm": 2.1359856128692627, "learning_rate": 3.3527245217109986e-06, "loss": 0.09350833296775818, "step": 3324 }, { "epoch": 0.8631879543094496, "grad_norm": 1.9019092321395874, "learning_rate": 3.3518130327462734e-06, "loss": 0.10339827090501785, "step": 3325 }, { "epoch": 0.8634475597092419, "grad_norm": 1.7788519859313965, "learning_rate": 3.3509014373947834e-06, "loss": 0.10635080188512802, "step": 3326 }, { "epoch": 0.8637071651090342, "grad_norm": 1.8669676780700684, "learning_rate": 3.3499897358176793e-06, "loss": 0.10905998945236206, "step": 3327 }, { "epoch": 0.8639667705088265, "grad_norm": 2.553861379623413, "learning_rate": 3.349077928176126e-06, "loss": 0.07983998954296112, "step": 3328 }, { "epoch": 0.864226375908619, "grad_norm": 3.1916344165802, "learning_rate": 3.3481660146313133e-06, "loss": 0.062252070754766464, "step": 3329 }, { "epoch": 0.8644859813084113, "grad_norm": 3.0965237617492676, "learning_rate": 3.3472539953444453e-06, "loss": 0.06621208786964417, "step": 3330 }, { "epoch": 0.8647455867082036, "grad_norm": 2.9216716289520264, "learning_rate": 3.346341870476746e-06, "loss": 0.04229151830077171, "step": 3331 }, { "epoch": 0.8650051921079959, "grad_norm": 2.87282657623291, "learning_rate": 3.3454296401894585e-06, "loss": 0.08245134353637695, "step": 3332 }, { "epoch": 0.8652647975077882, "grad_norm": 3.4287030696868896, "learning_rate": 3.344517304643844e-06, "loss": 0.09318184852600098, "step": 3333 }, { "epoch": 0.8655244029075805, "grad_norm": 4.1314697265625, "learning_rate": 3.3436048640011832e-06, "loss": 0.08626794815063477, "step": 3334 }, { "epoch": 0.8657840083073728, "grad_norm": 2.3672385215759277, "learning_rate": 3.342692318422773e-06, "loss": 0.11152516305446625, "step": 3335 }, { "epoch": 0.8660436137071651, "grad_norm": 1.4796980619430542, "learning_rate": 3.3417796680699334e-06, "loss": 0.1052401065826416, "step": 3336 }, { "epoch": 0.8663032191069574, "grad_norm": 3.4529683589935303, "learning_rate": 3.3408669131039972e-06, "loss": 0.0705564022064209, "step": 3337 }, { "epoch": 0.8665628245067497, "grad_norm": 3.3859755992889404, "learning_rate": 3.33995405368632e-06, "loss": 0.10080547630786896, "step": 3338 }, { "epoch": 0.866822429906542, "grad_norm": 2.685678482055664, "learning_rate": 3.3390410899782748e-06, "loss": 0.10998944938182831, "step": 3339 }, { "epoch": 0.8670820353063343, "grad_norm": 1.6386808156967163, "learning_rate": 3.3381280221412517e-06, "loss": 0.11067318916320801, "step": 3340 }, { "epoch": 0.8673416407061267, "grad_norm": 3.7356200218200684, "learning_rate": 3.3372148503366607e-06, "loss": 0.03840690851211548, "step": 3341 }, { "epoch": 0.867601246105919, "grad_norm": 1.960430383682251, "learning_rate": 3.3363015747259294e-06, "loss": 0.0965406596660614, "step": 3342 }, { "epoch": 0.8678608515057114, "grad_norm": 2.6263046264648438, "learning_rate": 3.3353881954705047e-06, "loss": 0.08934906870126724, "step": 3343 }, { "epoch": 0.8681204569055037, "grad_norm": 2.620455503463745, "learning_rate": 3.3344747127318506e-06, "loss": 0.1011648178100586, "step": 3344 }, { "epoch": 0.868380062305296, "grad_norm": 4.645274639129639, "learning_rate": 3.33356112667145e-06, "loss": 0.07896436750888824, "step": 3345 }, { "epoch": 0.8686396677050883, "grad_norm": 3.389012336730957, "learning_rate": 3.332647437450803e-06, "loss": 0.0610642172396183, "step": 3346 }, { "epoch": 0.8688992731048806, "grad_norm": 3.5375235080718994, "learning_rate": 3.3317336452314308e-06, "loss": 0.08358864486217499, "step": 3347 }, { "epoch": 0.8691588785046729, "grad_norm": 1.267446756362915, "learning_rate": 3.33081975017487e-06, "loss": 0.10271179676055908, "step": 3348 }, { "epoch": 0.8694184839044652, "grad_norm": 1.330583930015564, "learning_rate": 3.329905752442677e-06, "loss": 0.10314375162124634, "step": 3349 }, { "epoch": 0.8696780893042575, "grad_norm": 2.356259346008301, "learning_rate": 3.328991652196424e-06, "loss": 0.10640566796064377, "step": 3350 }, { "epoch": 0.8699376947040498, "grad_norm": 2.040921688079834, "learning_rate": 3.3280774495977048e-06, "loss": 0.09255120158195496, "step": 3351 }, { "epoch": 0.8701973001038421, "grad_norm": 2.501138925552368, "learning_rate": 3.3271631448081277e-06, "loss": 0.047334060072898865, "step": 3352 }, { "epoch": 0.8704569055036345, "grad_norm": 1.9004250764846802, "learning_rate": 3.3262487379893227e-06, "loss": 0.09191928803920746, "step": 3353 }, { "epoch": 0.8707165109034268, "grad_norm": 1.686693787574768, "learning_rate": 3.325334229302935e-06, "loss": 0.10288675129413605, "step": 3354 }, { "epoch": 0.8709761163032191, "grad_norm": 2.1763715744018555, "learning_rate": 3.3244196189106287e-06, "loss": 0.09635553508996964, "step": 3355 }, { "epoch": 0.8712357217030114, "grad_norm": 3.7433862686157227, "learning_rate": 3.3235049069740864e-06, "loss": 0.08654168248176575, "step": 3356 }, { "epoch": 0.8714953271028038, "grad_norm": 2.7247354984283447, "learning_rate": 3.3225900936550094e-06, "loss": 0.06371830403804779, "step": 3357 }, { "epoch": 0.8717549325025961, "grad_norm": 1.9285638332366943, "learning_rate": 3.3216751791151124e-06, "loss": 0.10041020065546036, "step": 3358 }, { "epoch": 0.8720145379023884, "grad_norm": 4.23439884185791, "learning_rate": 3.320760163516134e-06, "loss": 0.07061935216188431, "step": 3359 }, { "epoch": 0.8722741433021807, "grad_norm": 1.8768967390060425, "learning_rate": 3.3198450470198273e-06, "loss": 0.11022195219993591, "step": 3360 }, { "epoch": 0.872533748701973, "grad_norm": 1.93241548538208, "learning_rate": 3.3189298297879653e-06, "loss": 0.09351777285337448, "step": 3361 }, { "epoch": 0.8727933541017653, "grad_norm": 2.1835029125213623, "learning_rate": 3.3180145119823347e-06, "loss": 0.08308494091033936, "step": 3362 }, { "epoch": 0.8730529595015576, "grad_norm": 4.193514347076416, "learning_rate": 3.3170990937647445e-06, "loss": 0.07734818756580353, "step": 3363 }, { "epoch": 0.8733125649013499, "grad_norm": 1.5904866456985474, "learning_rate": 3.316183575297019e-06, "loss": 0.0969255194067955, "step": 3364 }, { "epoch": 0.8735721703011423, "grad_norm": 1.2982982397079468, "learning_rate": 3.3152679567410023e-06, "loss": 0.10853497684001923, "step": 3365 }, { "epoch": 0.8738317757009346, "grad_norm": 3.231865167617798, "learning_rate": 3.3143522382585536e-06, "loss": 0.08925461769104004, "step": 3366 }, { "epoch": 0.8740913811007269, "grad_norm": 2.539698362350464, "learning_rate": 3.3134364200115504e-06, "loss": 0.09661641716957092, "step": 3367 }, { "epoch": 0.8743509865005192, "grad_norm": 1.9134913682937622, "learning_rate": 3.3125205021618893e-06, "loss": 0.10621781647205353, "step": 3368 }, { "epoch": 0.8746105919003115, "grad_norm": 2.6365110874176025, "learning_rate": 3.3116044848714833e-06, "loss": 0.0572948232293129, "step": 3369 }, { "epoch": 0.8748701973001038, "grad_norm": 2.8366661071777344, "learning_rate": 3.3106883683022646e-06, "loss": 0.07628294825553894, "step": 3370 }, { "epoch": 0.8751298026998962, "grad_norm": 1.3731508255004883, "learning_rate": 3.309772152616179e-06, "loss": 0.10551363229751587, "step": 3371 }, { "epoch": 0.8753894080996885, "grad_norm": 1.0010462999343872, "learning_rate": 3.3088558379751945e-06, "loss": 0.10510018467903137, "step": 3372 }, { "epoch": 0.8756490134994808, "grad_norm": 1.6034468412399292, "learning_rate": 3.307939424541294e-06, "loss": 0.10525018721818924, "step": 3373 }, { "epoch": 0.8759086188992731, "grad_norm": 3.114576816558838, "learning_rate": 3.3070229124764784e-06, "loss": 0.06305424869060516, "step": 3374 }, { "epoch": 0.8761682242990654, "grad_norm": 3.035982608795166, "learning_rate": 3.3061063019427653e-06, "loss": 0.12413087487220764, "step": 3375 }, { "epoch": 0.8764278296988577, "grad_norm": 0.9493362307548523, "learning_rate": 3.305189593102192e-06, "loss": 0.10620502382516861, "step": 3376 }, { "epoch": 0.8766874350986501, "grad_norm": 1.7249287366867065, "learning_rate": 3.30427278611681e-06, "loss": 0.09296741336584091, "step": 3377 }, { "epoch": 0.8769470404984424, "grad_norm": 2.6248674392700195, "learning_rate": 3.3033558811486915e-06, "loss": 0.08903172612190247, "step": 3378 }, { "epoch": 0.8772066458982347, "grad_norm": 3.7404115200042725, "learning_rate": 3.302438878359923e-06, "loss": 0.08472387492656708, "step": 3379 }, { "epoch": 0.877466251298027, "grad_norm": 2.045409679412842, "learning_rate": 3.30152177791261e-06, "loss": 0.07993816584348679, "step": 3380 }, { "epoch": 0.8777258566978193, "grad_norm": 1.4336037635803223, "learning_rate": 3.300604579968875e-06, "loss": 0.09698298573493958, "step": 3381 }, { "epoch": 0.8779854620976116, "grad_norm": 1.1251333951950073, "learning_rate": 3.2996872846908574e-06, "loss": 0.10181634873151779, "step": 3382 }, { "epoch": 0.8782450674974039, "grad_norm": 3.360536813735962, "learning_rate": 3.298769892240713e-06, "loss": 0.055083900690078735, "step": 3383 }, { "epoch": 0.8785046728971962, "grad_norm": 4.285340785980225, "learning_rate": 3.2978524027806185e-06, "loss": 0.06110847368836403, "step": 3384 }, { "epoch": 0.8787642782969886, "grad_norm": 1.409030795097351, "learning_rate": 3.2969348164727624e-06, "loss": 0.0894411951303482, "step": 3385 }, { "epoch": 0.8790238836967809, "grad_norm": 3.1648240089416504, "learning_rate": 3.296017133479354e-06, "loss": 0.04942311346530914, "step": 3386 }, { "epoch": 0.8792834890965732, "grad_norm": 2.793962001800537, "learning_rate": 3.295099353962618e-06, "loss": 0.10733391344547272, "step": 3387 }, { "epoch": 0.8795430944963655, "grad_norm": 2.395385980606079, "learning_rate": 3.2941814780847987e-06, "loss": 0.09700915962457657, "step": 3388 }, { "epoch": 0.8798026998961579, "grad_norm": 4.22196102142334, "learning_rate": 3.293263506008153e-06, "loss": 0.05417408049106598, "step": 3389 }, { "epoch": 0.8800623052959502, "grad_norm": 4.259340286254883, "learning_rate": 3.2923454378949587e-06, "loss": 0.0888698399066925, "step": 3390 }, { "epoch": 0.8803219106957425, "grad_norm": 2.319600820541382, "learning_rate": 3.291427273907509e-06, "loss": 0.087367944419384, "step": 3391 }, { "epoch": 0.8805815160955348, "grad_norm": 3.5011839866638184, "learning_rate": 3.2905090142081147e-06, "loss": 0.0966377854347229, "step": 3392 }, { "epoch": 0.8808411214953271, "grad_norm": 2.586557149887085, "learning_rate": 3.289590658959103e-06, "loss": 0.08864893764257431, "step": 3393 }, { "epoch": 0.8811007268951194, "grad_norm": 3.14225435256958, "learning_rate": 3.2886722083228177e-06, "loss": 0.05396828427910805, "step": 3394 }, { "epoch": 0.8813603322949117, "grad_norm": 3.3717141151428223, "learning_rate": 3.2877536624616198e-06, "loss": 0.0913383886218071, "step": 3395 }, { "epoch": 0.881619937694704, "grad_norm": 3.788015127182007, "learning_rate": 3.286835021537886e-06, "loss": 0.04712148383259773, "step": 3396 }, { "epoch": 0.8818795430944963, "grad_norm": 2.410845994949341, "learning_rate": 3.2859162857140143e-06, "loss": 0.11542536318302155, "step": 3397 }, { "epoch": 0.8821391484942886, "grad_norm": 3.3739731311798096, "learning_rate": 3.2849974551524135e-06, "loss": 0.05603986233472824, "step": 3398 }, { "epoch": 0.882398753894081, "grad_norm": 2.9592700004577637, "learning_rate": 3.284078530015512e-06, "loss": 0.10297076404094696, "step": 3399 }, { "epoch": 0.8826583592938733, "grad_norm": 2.5961763858795166, "learning_rate": 3.2831595104657547e-06, "loss": 0.07375909388065338, "step": 3400 }, { "epoch": 0.8829179646936657, "grad_norm": 4.605088710784912, "learning_rate": 3.2822403966656045e-06, "loss": 0.06920693814754486, "step": 3401 }, { "epoch": 0.883177570093458, "grad_norm": 1.317557692527771, "learning_rate": 3.2813211887775386e-06, "loss": 0.10531601309776306, "step": 3402 }, { "epoch": 0.8834371754932503, "grad_norm": 2.159895658493042, "learning_rate": 3.2804018869640525e-06, "loss": 0.09555398672819138, "step": 3403 }, { "epoch": 0.8836967808930426, "grad_norm": 3.0047671794891357, "learning_rate": 3.279482491387657e-06, "loss": 0.10268673300743103, "step": 3404 }, { "epoch": 0.8839563862928349, "grad_norm": 3.328965663909912, "learning_rate": 3.2785630022108804e-06, "loss": 0.1209707111120224, "step": 3405 }, { "epoch": 0.8842159916926272, "grad_norm": 2.787163734436035, "learning_rate": 3.2776434195962682e-06, "loss": 0.06650852411985397, "step": 3406 }, { "epoch": 0.8844755970924195, "grad_norm": 1.8154314756393433, "learning_rate": 3.2767237437063803e-06, "loss": 0.10978635400533676, "step": 3407 }, { "epoch": 0.8847352024922118, "grad_norm": 3.1532647609710693, "learning_rate": 3.2758039747037944e-06, "loss": 0.062036383897066116, "step": 3408 }, { "epoch": 0.8849948078920041, "grad_norm": 1.3130556344985962, "learning_rate": 3.274884112751106e-06, "loss": 0.10214865207672119, "step": 3409 }, { "epoch": 0.8852544132917964, "grad_norm": 3.560602903366089, "learning_rate": 3.2739641580109243e-06, "loss": 0.07875264436006546, "step": 3410 }, { "epoch": 0.8855140186915887, "grad_norm": 4.249600410461426, "learning_rate": 3.273044110645878e-06, "loss": 0.10837551951408386, "step": 3411 }, { "epoch": 0.885773624091381, "grad_norm": 3.4762327671051025, "learning_rate": 3.2721239708186077e-06, "loss": 0.06934276968240738, "step": 3412 }, { "epoch": 0.8860332294911735, "grad_norm": 2.397857189178467, "learning_rate": 3.271203738691774e-06, "loss": 0.08392444252967834, "step": 3413 }, { "epoch": 0.8862928348909658, "grad_norm": 1.5310426950454712, "learning_rate": 3.2702834144280536e-06, "loss": 0.10980413854122162, "step": 3414 }, { "epoch": 0.8865524402907581, "grad_norm": 1.4039372205734253, "learning_rate": 3.2693629981901392e-06, "loss": 0.10057205706834793, "step": 3415 }, { "epoch": 0.8868120456905504, "grad_norm": 1.450006365776062, "learning_rate": 3.268442490140738e-06, "loss": 0.09564439952373505, "step": 3416 }, { "epoch": 0.8870716510903427, "grad_norm": 1.0916552543640137, "learning_rate": 3.2675218904425748e-06, "loss": 0.09365823864936829, "step": 3417 }, { "epoch": 0.887331256490135, "grad_norm": 2.314638137817383, "learning_rate": 3.266601199258391e-06, "loss": 0.0874297097325325, "step": 3418 }, { "epoch": 0.8875908618899273, "grad_norm": 2.3991880416870117, "learning_rate": 3.2656804167509456e-06, "loss": 0.07411012053489685, "step": 3419 }, { "epoch": 0.8878504672897196, "grad_norm": 1.3586848974227905, "learning_rate": 3.264759543083009e-06, "loss": 0.09730817377567291, "step": 3420 }, { "epoch": 0.8881100726895119, "grad_norm": 2.895165205001831, "learning_rate": 3.2638385784173704e-06, "loss": 0.08003706485033035, "step": 3421 }, { "epoch": 0.8883696780893042, "grad_norm": 1.7942966222763062, "learning_rate": 3.262917522916838e-06, "loss": 0.11149954795837402, "step": 3422 }, { "epoch": 0.8886292834890965, "grad_norm": 2.921905517578125, "learning_rate": 3.2619963767442314e-06, "loss": 0.06156671792268753, "step": 3423 }, { "epoch": 0.8888888888888888, "grad_norm": 4.624691963195801, "learning_rate": 3.2610751400623887e-06, "loss": 0.07232597470283508, "step": 3424 }, { "epoch": 0.8891484942886813, "grad_norm": 5.111496448516846, "learning_rate": 3.2601538130341627e-06, "loss": 0.0668654590845108, "step": 3425 }, { "epoch": 0.8894080996884736, "grad_norm": 2.2355141639709473, "learning_rate": 3.2592323958224236e-06, "loss": 0.0813874825835228, "step": 3426 }, { "epoch": 0.8896677050882659, "grad_norm": 3.639638900756836, "learning_rate": 3.258310888590058e-06, "loss": 0.045657265931367874, "step": 3427 }, { "epoch": 0.8899273104880582, "grad_norm": 1.4520198106765747, "learning_rate": 3.257389291499965e-06, "loss": 0.1079626977443695, "step": 3428 }, { "epoch": 0.8901869158878505, "grad_norm": 2.445294141769409, "learning_rate": 3.256467604715062e-06, "loss": 0.11596646159887314, "step": 3429 }, { "epoch": 0.8904465212876428, "grad_norm": 1.9640729427337646, "learning_rate": 3.255545828398285e-06, "loss": 0.09605339169502258, "step": 3430 }, { "epoch": 0.8907061266874351, "grad_norm": 3.605832576751709, "learning_rate": 3.2546239627125796e-06, "loss": 0.09984704852104187, "step": 3431 }, { "epoch": 0.8909657320872274, "grad_norm": 3.8469502925872803, "learning_rate": 3.2537020078209126e-06, "loss": 0.05420738458633423, "step": 3432 }, { "epoch": 0.8912253374870197, "grad_norm": 1.7112760543823242, "learning_rate": 3.252779963886264e-06, "loss": 0.09257195144891739, "step": 3433 }, { "epoch": 0.891484942886812, "grad_norm": 3.057739734649658, "learning_rate": 3.25185783107163e-06, "loss": 0.07109027355909348, "step": 3434 }, { "epoch": 0.8917445482866043, "grad_norm": 3.6049137115478516, "learning_rate": 3.250935609540021e-06, "loss": 0.11239074170589447, "step": 3435 }, { "epoch": 0.8920041536863966, "grad_norm": 3.7235934734344482, "learning_rate": 3.250013299454468e-06, "loss": 0.07790619134902954, "step": 3436 }, { "epoch": 0.892263759086189, "grad_norm": 2.435032367706299, "learning_rate": 3.2490909009780117e-06, "loss": 0.10376279056072235, "step": 3437 }, { "epoch": 0.8925233644859814, "grad_norm": 1.5237401723861694, "learning_rate": 3.248168414273712e-06, "loss": 0.09900452196598053, "step": 3438 }, { "epoch": 0.8927829698857737, "grad_norm": 4.014402866363525, "learning_rate": 3.2472458395046423e-06, "loss": 0.07106226682662964, "step": 3439 }, { "epoch": 0.893042575285566, "grad_norm": 3.27894926071167, "learning_rate": 3.2463231768338954e-06, "loss": 0.11854828894138336, "step": 3440 }, { "epoch": 0.8933021806853583, "grad_norm": 2.1606171131134033, "learning_rate": 3.2454004264245735e-06, "loss": 0.08625137805938721, "step": 3441 }, { "epoch": 0.8935617860851506, "grad_norm": 2.6519734859466553, "learning_rate": 3.2444775884398e-06, "loss": 0.0691198855638504, "step": 3442 }, { "epoch": 0.8938213914849429, "grad_norm": 1.8812357187271118, "learning_rate": 3.2435546630427105e-06, "loss": 0.09238085895776749, "step": 3443 }, { "epoch": 0.8940809968847352, "grad_norm": 1.9962352514266968, "learning_rate": 3.2426316503964583e-06, "loss": 0.07317288964986801, "step": 3444 }, { "epoch": 0.8943406022845275, "grad_norm": 2.06166672706604, "learning_rate": 3.2417085506642095e-06, "loss": 0.10489257425069809, "step": 3445 }, { "epoch": 0.8946002076843198, "grad_norm": 3.12717342376709, "learning_rate": 3.240785364009148e-06, "loss": 0.09666489064693451, "step": 3446 }, { "epoch": 0.8948598130841121, "grad_norm": 1.7129021883010864, "learning_rate": 3.239862090594471e-06, "loss": 0.09652353823184967, "step": 3447 }, { "epoch": 0.8951194184839044, "grad_norm": 1.2779885530471802, "learning_rate": 3.2389387305833932e-06, "loss": 0.09879914671182632, "step": 3448 }, { "epoch": 0.8953790238836968, "grad_norm": 2.5178987979888916, "learning_rate": 3.238015284139143e-06, "loss": 0.07298637926578522, "step": 3449 }, { "epoch": 0.8956386292834891, "grad_norm": 2.45034122467041, "learning_rate": 3.2370917514249646e-06, "loss": 0.09853577613830566, "step": 3450 }, { "epoch": 0.8958982346832814, "grad_norm": 1.5055650472640991, "learning_rate": 3.2361681326041167e-06, "loss": 0.08711904287338257, "step": 3451 }, { "epoch": 0.8961578400830738, "grad_norm": 2.766376256942749, "learning_rate": 3.2352444278398754e-06, "loss": 0.07822659611701965, "step": 3452 }, { "epoch": 0.8964174454828661, "grad_norm": 2.8868255615234375, "learning_rate": 3.2343206372955305e-06, "loss": 0.0661822110414505, "step": 3453 }, { "epoch": 0.8966770508826584, "grad_norm": 2.424391031265259, "learning_rate": 3.2333967611343854e-06, "loss": 0.10852464288473129, "step": 3454 }, { "epoch": 0.8969366562824507, "grad_norm": 2.6693367958068848, "learning_rate": 3.232472799519762e-06, "loss": 0.11274591833353043, "step": 3455 }, { "epoch": 0.897196261682243, "grad_norm": 3.2108731269836426, "learning_rate": 3.231548752614994e-06, "loss": 0.07692048698663712, "step": 3456 }, { "epoch": 0.8974558670820353, "grad_norm": 3.693120241165161, "learning_rate": 3.2306246205834333e-06, "loss": 0.08613869547843933, "step": 3457 }, { "epoch": 0.8977154724818276, "grad_norm": 2.314185857772827, "learning_rate": 3.229700403588444e-06, "loss": 0.1068665087223053, "step": 3458 }, { "epoch": 0.8979750778816199, "grad_norm": 3.378552198410034, "learning_rate": 3.2287761017934076e-06, "loss": 0.08566536009311676, "step": 3459 }, { "epoch": 0.8982346832814122, "grad_norm": 6.1034746170043945, "learning_rate": 3.2278517153617186e-06, "loss": 0.09708546102046967, "step": 3460 }, { "epoch": 0.8984942886812046, "grad_norm": 2.428002119064331, "learning_rate": 3.226927244456788e-06, "loss": 0.0586358942091465, "step": 3461 }, { "epoch": 0.8987538940809969, "grad_norm": 2.057166814804077, "learning_rate": 3.2260026892420397e-06, "loss": 0.10097317397594452, "step": 3462 }, { "epoch": 0.8990134994807892, "grad_norm": 4.649993419647217, "learning_rate": 3.2250780498809157e-06, "loss": 0.06772969663143158, "step": 3463 }, { "epoch": 0.8992731048805815, "grad_norm": 2.669048547744751, "learning_rate": 3.22415332653687e-06, "loss": 0.09895916283130646, "step": 3464 }, { "epoch": 0.8995327102803738, "grad_norm": 1.3149738311767578, "learning_rate": 3.2232285193733735e-06, "loss": 0.10259298980236053, "step": 3465 }, { "epoch": 0.8997923156801662, "grad_norm": 2.1675801277160645, "learning_rate": 3.2223036285539092e-06, "loss": 0.08622972667217255, "step": 3466 }, { "epoch": 0.9000519210799585, "grad_norm": 2.997011184692383, "learning_rate": 3.2213786542419796e-06, "loss": 0.06468494236469269, "step": 3467 }, { "epoch": 0.9003115264797508, "grad_norm": 2.3107409477233887, "learning_rate": 3.220453596601096e-06, "loss": 0.10102999955415726, "step": 3468 }, { "epoch": 0.9005711318795431, "grad_norm": 1.9424185752868652, "learning_rate": 3.219528455794788e-06, "loss": 0.1020001545548439, "step": 3469 }, { "epoch": 0.9008307372793354, "grad_norm": 2.801330089569092, "learning_rate": 3.2186032319866e-06, "loss": 0.0735401064157486, "step": 3470 }, { "epoch": 0.9010903426791277, "grad_norm": 2.656104564666748, "learning_rate": 3.217677925340091e-06, "loss": 0.05397491529583931, "step": 3471 }, { "epoch": 0.90134994807892, "grad_norm": 1.1860929727554321, "learning_rate": 3.2167525360188317e-06, "loss": 0.09733740985393524, "step": 3472 }, { "epoch": 0.9016095534787124, "grad_norm": 2.1627204418182373, "learning_rate": 3.2158270641864126e-06, "loss": 0.08960548043251038, "step": 3473 }, { "epoch": 0.9018691588785047, "grad_norm": 2.832646131515503, "learning_rate": 3.2149015100064333e-06, "loss": 0.09995979815721512, "step": 3474 }, { "epoch": 0.902128764278297, "grad_norm": 3.3013722896575928, "learning_rate": 3.213975873642513e-06, "loss": 0.09155522286891937, "step": 3475 }, { "epoch": 0.9023883696780893, "grad_norm": 4.285091876983643, "learning_rate": 3.2130501552582823e-06, "loss": 0.0662025660276413, "step": 3476 }, { "epoch": 0.9026479750778816, "grad_norm": 3.049381971359253, "learning_rate": 3.2121243550173854e-06, "loss": 0.11278975009918213, "step": 3477 }, { "epoch": 0.9029075804776739, "grad_norm": 1.5807814598083496, "learning_rate": 3.2111984730834837e-06, "loss": 0.08725182712078094, "step": 3478 }, { "epoch": 0.9031671858774662, "grad_norm": 1.0502691268920898, "learning_rate": 3.2102725096202527e-06, "loss": 0.0983998104929924, "step": 3479 }, { "epoch": 0.9034267912772586, "grad_norm": 2.627398729324341, "learning_rate": 3.2093464647913815e-06, "loss": 0.0843883752822876, "step": 3480 }, { "epoch": 0.9036863966770509, "grad_norm": 2.333570718765259, "learning_rate": 3.2084203387605716e-06, "loss": 0.11800917237997055, "step": 3481 }, { "epoch": 0.9039460020768432, "grad_norm": 2.812394857406616, "learning_rate": 3.207494131691542e-06, "loss": 0.10199877619743347, "step": 3482 }, { "epoch": 0.9042056074766355, "grad_norm": 3.208777666091919, "learning_rate": 3.2065678437480265e-06, "loss": 0.09001582860946655, "step": 3483 }, { "epoch": 0.9044652128764278, "grad_norm": 3.4315764904022217, "learning_rate": 3.2056414750937696e-06, "loss": 0.05963603034615517, "step": 3484 }, { "epoch": 0.9047248182762202, "grad_norm": 3.9210071563720703, "learning_rate": 3.2047150258925324e-06, "loss": 0.07096835970878601, "step": 3485 }, { "epoch": 0.9049844236760125, "grad_norm": 2.5122573375701904, "learning_rate": 3.203788496308091e-06, "loss": 0.08311720937490463, "step": 3486 }, { "epoch": 0.9052440290758048, "grad_norm": 1.9596307277679443, "learning_rate": 3.2028618865042327e-06, "loss": 0.10371610522270203, "step": 3487 }, { "epoch": 0.9055036344755971, "grad_norm": 2.6820902824401855, "learning_rate": 3.201935196644763e-06, "loss": 0.05965109169483185, "step": 3488 }, { "epoch": 0.9057632398753894, "grad_norm": 1.5862200260162354, "learning_rate": 3.2010084268934984e-06, "loss": 0.1075955331325531, "step": 3489 }, { "epoch": 0.9060228452751817, "grad_norm": 2.311527729034424, "learning_rate": 3.2000815774142703e-06, "loss": 0.09723986685276031, "step": 3490 }, { "epoch": 0.906282450674974, "grad_norm": 4.79542350769043, "learning_rate": 3.199154648370925e-06, "loss": 0.08935460448265076, "step": 3491 }, { "epoch": 0.9065420560747663, "grad_norm": 3.1330196857452393, "learning_rate": 3.198227639927322e-06, "loss": 0.04616154730319977, "step": 3492 }, { "epoch": 0.9068016614745587, "grad_norm": 3.563318967819214, "learning_rate": 3.1973005522473355e-06, "loss": 0.05086298659443855, "step": 3493 }, { "epoch": 0.907061266874351, "grad_norm": 3.6348748207092285, "learning_rate": 3.196373385494853e-06, "loss": 0.07386009395122528, "step": 3494 }, { "epoch": 0.9073208722741433, "grad_norm": 1.6102243661880493, "learning_rate": 3.1954461398337766e-06, "loss": 0.11649571359157562, "step": 3495 }, { "epoch": 0.9075804776739356, "grad_norm": 2.0223240852355957, "learning_rate": 3.194518815428022e-06, "loss": 0.08973757177591324, "step": 3496 }, { "epoch": 0.907840083073728, "grad_norm": 1.8386452198028564, "learning_rate": 3.193591412441519e-06, "loss": 0.1166750118136406, "step": 3497 }, { "epoch": 0.9080996884735203, "grad_norm": 1.3625764846801758, "learning_rate": 3.1926639310382114e-06, "loss": 0.09272871911525726, "step": 3498 }, { "epoch": 0.9083592938733126, "grad_norm": 2.889202117919922, "learning_rate": 3.1917363713820565e-06, "loss": 0.06265819072723389, "step": 3499 }, { "epoch": 0.9086188992731049, "grad_norm": 2.1640095710754395, "learning_rate": 3.1908087336370245e-06, "loss": 0.11204766482114792, "step": 3500 }, { "epoch": 0.9088785046728972, "grad_norm": 3.5341508388519287, "learning_rate": 3.189881017967103e-06, "loss": 0.09234894812107086, "step": 3501 }, { "epoch": 0.9091381100726895, "grad_norm": 2.697584629058838, "learning_rate": 3.188953224536289e-06, "loss": 0.0980486273765564, "step": 3502 }, { "epoch": 0.9093977154724818, "grad_norm": 2.9599838256835938, "learning_rate": 3.188025353508595e-06, "loss": 0.06506117433309555, "step": 3503 }, { "epoch": 0.9096573208722741, "grad_norm": 3.879413366317749, "learning_rate": 3.187097405048049e-06, "loss": 0.101982980966568, "step": 3504 }, { "epoch": 0.9099169262720664, "grad_norm": 4.0299296379089355, "learning_rate": 3.18616937931869e-06, "loss": 0.06190481036901474, "step": 3505 }, { "epoch": 0.9101765316718587, "grad_norm": 2.667015790939331, "learning_rate": 3.1852412764845704e-06, "loss": 0.1167021244764328, "step": 3506 }, { "epoch": 0.910436137071651, "grad_norm": 2.480057716369629, "learning_rate": 3.1843130967097596e-06, "loss": 0.06078986078500748, "step": 3507 }, { "epoch": 0.9106957424714434, "grad_norm": 2.2583720684051514, "learning_rate": 3.183384840158338e-06, "loss": 0.0757041722536087, "step": 3508 }, { "epoch": 0.9109553478712358, "grad_norm": 2.686589479446411, "learning_rate": 3.1824565069943993e-06, "loss": 0.07320142537355423, "step": 3509 }, { "epoch": 0.9112149532710281, "grad_norm": 1.5576609373092651, "learning_rate": 3.181528097382052e-06, "loss": 0.0947377160191536, "step": 3510 }, { "epoch": 0.9114745586708204, "grad_norm": 2.8132545948028564, "learning_rate": 3.180599611485418e-06, "loss": 0.06342639774084091, "step": 3511 }, { "epoch": 0.9117341640706127, "grad_norm": 2.3912410736083984, "learning_rate": 3.179671049468632e-06, "loss": 0.09842196106910706, "step": 3512 }, { "epoch": 0.911993769470405, "grad_norm": 2.624298095703125, "learning_rate": 3.1787424114958433e-06, "loss": 0.08337509632110596, "step": 3513 }, { "epoch": 0.9122533748701973, "grad_norm": 1.9872084856033325, "learning_rate": 3.1778136977312125e-06, "loss": 0.10482316464185715, "step": 3514 }, { "epoch": 0.9125129802699896, "grad_norm": 1.7411481142044067, "learning_rate": 3.1768849083389164e-06, "loss": 0.0939578264951706, "step": 3515 }, { "epoch": 0.9127725856697819, "grad_norm": 3.7934141159057617, "learning_rate": 3.175956043483142e-06, "loss": 0.08895735442638397, "step": 3516 }, { "epoch": 0.9130321910695742, "grad_norm": 2.9225831031799316, "learning_rate": 3.1750271033280925e-06, "loss": 0.11157446354627609, "step": 3517 }, { "epoch": 0.9132917964693665, "grad_norm": 2.0712621212005615, "learning_rate": 3.1740980880379823e-06, "loss": 0.07993082702159882, "step": 3518 }, { "epoch": 0.9135514018691588, "grad_norm": 2.734586477279663, "learning_rate": 3.1731689977770414e-06, "loss": 0.08146625012159348, "step": 3519 }, { "epoch": 0.9138110072689511, "grad_norm": 1.1626825332641602, "learning_rate": 3.172239832709511e-06, "loss": 0.10661596804857254, "step": 3520 }, { "epoch": 0.9140706126687436, "grad_norm": 1.921417474746704, "learning_rate": 3.171310592999646e-06, "loss": 0.09098741412162781, "step": 3521 }, { "epoch": 0.9143302180685359, "grad_norm": 4.195058345794678, "learning_rate": 3.1703812788117157e-06, "loss": 0.11274462193250656, "step": 3522 }, { "epoch": 0.9145898234683282, "grad_norm": 3.8317923545837402, "learning_rate": 3.1694518903099993e-06, "loss": 0.0643266960978508, "step": 3523 }, { "epoch": 0.9148494288681205, "grad_norm": 3.8467092514038086, "learning_rate": 3.168522427658794e-06, "loss": 0.05170054733753204, "step": 3524 }, { "epoch": 0.9151090342679128, "grad_norm": 1.673613429069519, "learning_rate": 3.1675928910224063e-06, "loss": 0.10613183677196503, "step": 3525 }, { "epoch": 0.9153686396677051, "grad_norm": 3.4195547103881836, "learning_rate": 3.1666632805651564e-06, "loss": 0.0751216784119606, "step": 3526 }, { "epoch": 0.9156282450674974, "grad_norm": 3.1236422061920166, "learning_rate": 3.165733596451379e-06, "loss": 0.10521095991134644, "step": 3527 }, { "epoch": 0.9158878504672897, "grad_norm": 2.339232921600342, "learning_rate": 3.164803838845421e-06, "loss": 0.10094958543777466, "step": 3528 }, { "epoch": 0.916147455867082, "grad_norm": 1.8346209526062012, "learning_rate": 3.1638740079116427e-06, "loss": 0.1060708612203598, "step": 3529 }, { "epoch": 0.9164070612668743, "grad_norm": 2.3196091651916504, "learning_rate": 3.1629441038144152e-06, "loss": 0.09845760464668274, "step": 3530 }, { "epoch": 0.9166666666666666, "grad_norm": 3.7640626430511475, "learning_rate": 3.1620141267181253e-06, "loss": 0.09427410364151001, "step": 3531 }, { "epoch": 0.9169262720664589, "grad_norm": 3.369494676589966, "learning_rate": 3.1610840767871726e-06, "loss": 0.0763261541724205, "step": 3532 }, { "epoch": 0.9171858774662512, "grad_norm": 1.8780791759490967, "learning_rate": 3.160153954185967e-06, "loss": 0.09870312362909317, "step": 3533 }, { "epoch": 0.9174454828660437, "grad_norm": 2.4844534397125244, "learning_rate": 3.159223759078934e-06, "loss": 0.10176412761211395, "step": 3534 }, { "epoch": 0.917705088265836, "grad_norm": 2.8823556900024414, "learning_rate": 3.15829349163051e-06, "loss": 0.04870777577161789, "step": 3535 }, { "epoch": 0.9179646936656283, "grad_norm": 2.9276199340820312, "learning_rate": 3.157363152005145e-06, "loss": 0.09418143332004547, "step": 3536 }, { "epoch": 0.9182242990654206, "grad_norm": 4.144491195678711, "learning_rate": 3.1564327403673024e-06, "loss": 0.0876661017537117, "step": 3537 }, { "epoch": 0.9184839044652129, "grad_norm": 3.9265806674957275, "learning_rate": 3.155502256881458e-06, "loss": 0.06433944404125214, "step": 3538 }, { "epoch": 0.9187435098650052, "grad_norm": 1.1044800281524658, "learning_rate": 3.1545717017120993e-06, "loss": 0.09395261108875275, "step": 3539 }, { "epoch": 0.9190031152647975, "grad_norm": 3.5318310260772705, "learning_rate": 3.153641075023726e-06, "loss": 0.08322370052337646, "step": 3540 }, { "epoch": 0.9192627206645898, "grad_norm": 3.6896448135375977, "learning_rate": 3.152710376980854e-06, "loss": 0.062469400465488434, "step": 3541 }, { "epoch": 0.9195223260643821, "grad_norm": 2.0932462215423584, "learning_rate": 3.151779607748008e-06, "loss": 0.10051462054252625, "step": 3542 }, { "epoch": 0.9197819314641744, "grad_norm": 1.8776081800460815, "learning_rate": 3.150848767489726e-06, "loss": 0.09622339904308319, "step": 3543 }, { "epoch": 0.9200415368639667, "grad_norm": 1.4352591037750244, "learning_rate": 3.1499178563705605e-06, "loss": 0.10184371471405029, "step": 3544 }, { "epoch": 0.920301142263759, "grad_norm": 2.5907418727874756, "learning_rate": 3.1489868745550754e-06, "loss": 0.08987732231616974, "step": 3545 }, { "epoch": 0.9205607476635514, "grad_norm": 2.0778911113739014, "learning_rate": 3.1480558222078467e-06, "loss": 0.10388903319835663, "step": 3546 }, { "epoch": 0.9208203530633438, "grad_norm": 2.2021844387054443, "learning_rate": 3.147124699493462e-06, "loss": 0.10628572106361389, "step": 3547 }, { "epoch": 0.9210799584631361, "grad_norm": 1.0688259601593018, "learning_rate": 3.146193506576524e-06, "loss": 0.0987844318151474, "step": 3548 }, { "epoch": 0.9213395638629284, "grad_norm": 2.6007237434387207, "learning_rate": 3.1452622436216452e-06, "loss": 0.07207131385803223, "step": 3549 }, { "epoch": 0.9215991692627207, "grad_norm": 1.2000703811645508, "learning_rate": 3.144330910793453e-06, "loss": 0.10804711282253265, "step": 3550 }, { "epoch": 0.921858774662513, "grad_norm": 3.576164960861206, "learning_rate": 3.143399508256584e-06, "loss": 0.05447113513946533, "step": 3551 }, { "epoch": 0.9221183800623053, "grad_norm": 2.694854259490967, "learning_rate": 3.14246803617569e-06, "loss": 0.08260681480169296, "step": 3552 }, { "epoch": 0.9223779854620976, "grad_norm": 2.5790135860443115, "learning_rate": 3.1415364947154323e-06, "loss": 0.11030316352844238, "step": 3553 }, { "epoch": 0.9226375908618899, "grad_norm": 2.696810245513916, "learning_rate": 3.1406048840404886e-06, "loss": 0.09299397468566895, "step": 3554 }, { "epoch": 0.9228971962616822, "grad_norm": 3.3488028049468994, "learning_rate": 3.1396732043155447e-06, "loss": 0.04543329030275345, "step": 3555 }, { "epoch": 0.9231568016614745, "grad_norm": 3.9822165966033936, "learning_rate": 3.1387414557053013e-06, "loss": 0.07294310629367828, "step": 3556 }, { "epoch": 0.9234164070612668, "grad_norm": 1.3592053651809692, "learning_rate": 3.1378096383744685e-06, "loss": 0.10197578370571136, "step": 3557 }, { "epoch": 0.9236760124610592, "grad_norm": 3.688030481338501, "learning_rate": 3.1368777524877713e-06, "loss": 0.08712348341941833, "step": 3558 }, { "epoch": 0.9239356178608515, "grad_norm": 3.208214044570923, "learning_rate": 3.1359457982099467e-06, "loss": 0.11219029873609543, "step": 3559 }, { "epoch": 0.9241952232606438, "grad_norm": 2.3355724811553955, "learning_rate": 3.1350137757057412e-06, "loss": 0.07888687402009964, "step": 3560 }, { "epoch": 0.9244548286604362, "grad_norm": 1.4065415859222412, "learning_rate": 3.1340816851399164e-06, "loss": 0.09812767803668976, "step": 3561 }, { "epoch": 0.9247144340602285, "grad_norm": 2.445178270339966, "learning_rate": 3.133149526677243e-06, "loss": 0.10333714634180069, "step": 3562 }, { "epoch": 0.9249740394600208, "grad_norm": 2.2113900184631348, "learning_rate": 3.1322173004825076e-06, "loss": 0.11144132167100906, "step": 3563 }, { "epoch": 0.9252336448598131, "grad_norm": 3.650486707687378, "learning_rate": 3.1312850067205036e-06, "loss": 0.09404852986335754, "step": 3564 }, { "epoch": 0.9254932502596054, "grad_norm": 3.475795269012451, "learning_rate": 3.130352645556042e-06, "loss": 0.05486001819372177, "step": 3565 }, { "epoch": 0.9257528556593977, "grad_norm": 1.5376776456832886, "learning_rate": 3.1294202171539406e-06, "loss": 0.08376917243003845, "step": 3566 }, { "epoch": 0.92601246105919, "grad_norm": 1.3458110094070435, "learning_rate": 3.128487721679033e-06, "loss": 0.10390675067901611, "step": 3567 }, { "epoch": 0.9262720664589823, "grad_norm": 1.5617811679840088, "learning_rate": 3.1275551592961617e-06, "loss": 0.08680014312267303, "step": 3568 }, { "epoch": 0.9265316718587746, "grad_norm": 2.37766170501709, "learning_rate": 3.1266225301701842e-06, "loss": 0.09664551913738251, "step": 3569 }, { "epoch": 0.926791277258567, "grad_norm": 5.5613532066345215, "learning_rate": 3.1256898344659657e-06, "loss": 0.08430549502372742, "step": 3570 }, { "epoch": 0.9270508826583593, "grad_norm": 3.2104361057281494, "learning_rate": 3.124757072348388e-06, "loss": 0.07606418430805206, "step": 3571 }, { "epoch": 0.9273104880581516, "grad_norm": 1.9645119905471802, "learning_rate": 3.1238242439823393e-06, "loss": 0.09977243840694427, "step": 3572 }, { "epoch": 0.927570093457944, "grad_norm": 3.075758218765259, "learning_rate": 3.1228913495327245e-06, "loss": 0.061811402440071106, "step": 3573 }, { "epoch": 0.9278296988577363, "grad_norm": 3.0078468322753906, "learning_rate": 3.1219583891644574e-06, "loss": 0.06234640255570412, "step": 3574 }, { "epoch": 0.9280893042575286, "grad_norm": 2.278426170349121, "learning_rate": 3.121025363042463e-06, "loss": 0.07910848408937454, "step": 3575 }, { "epoch": 0.9283489096573209, "grad_norm": 3.4754576683044434, "learning_rate": 3.12009227133168e-06, "loss": 0.0880737453699112, "step": 3576 }, { "epoch": 0.9286085150571132, "grad_norm": 5.156193256378174, "learning_rate": 3.119159114197058e-06, "loss": 0.06992721557617188, "step": 3577 }, { "epoch": 0.9288681204569055, "grad_norm": 3.900515079498291, "learning_rate": 3.118225891803556e-06, "loss": 0.06118912994861603, "step": 3578 }, { "epoch": 0.9291277258566978, "grad_norm": 2.0138533115386963, "learning_rate": 3.1172926043161496e-06, "loss": 0.0852862298488617, "step": 3579 }, { "epoch": 0.9293873312564901, "grad_norm": 2.914792060852051, "learning_rate": 3.116359251899819e-06, "loss": 0.060045864433050156, "step": 3580 }, { "epoch": 0.9296469366562824, "grad_norm": 1.4498915672302246, "learning_rate": 3.1154258347195614e-06, "loss": 0.10936366766691208, "step": 3581 }, { "epoch": 0.9299065420560748, "grad_norm": 1.8513206243515015, "learning_rate": 3.114492352940384e-06, "loss": 0.10552434623241425, "step": 3582 }, { "epoch": 0.9301661474558671, "grad_norm": 4.523972034454346, "learning_rate": 3.1135588067273044e-06, "loss": 0.08593358844518661, "step": 3583 }, { "epoch": 0.9304257528556594, "grad_norm": 2.146735906600952, "learning_rate": 3.1126251962453507e-06, "loss": 0.09291474521160126, "step": 3584 }, { "epoch": 0.9306853582554517, "grad_norm": 2.0893263816833496, "learning_rate": 3.111691521659567e-06, "loss": 0.08896977454423904, "step": 3585 }, { "epoch": 0.930944963655244, "grad_norm": 1.4163066148757935, "learning_rate": 3.1107577831350033e-06, "loss": 0.10312919318675995, "step": 3586 }, { "epoch": 0.9312045690550363, "grad_norm": 3.6854991912841797, "learning_rate": 3.1098239808367237e-06, "loss": 0.0508161261677742, "step": 3587 }, { "epoch": 0.9314641744548287, "grad_norm": 3.6477231979370117, "learning_rate": 3.108890114929803e-06, "loss": 0.0659763514995575, "step": 3588 }, { "epoch": 0.931723779854621, "grad_norm": 4.745521068572998, "learning_rate": 3.1079561855793277e-06, "loss": 0.11842793226242065, "step": 3589 }, { "epoch": 0.9319833852544133, "grad_norm": 3.624663829803467, "learning_rate": 3.107022192950395e-06, "loss": 0.028311073780059814, "step": 3590 }, { "epoch": 0.9322429906542056, "grad_norm": 2.8887360095977783, "learning_rate": 3.106088137208113e-06, "loss": 0.08906810730695724, "step": 3591 }, { "epoch": 0.9325025960539979, "grad_norm": 3.3420944213867188, "learning_rate": 3.105154018517602e-06, "loss": 0.09428063035011292, "step": 3592 }, { "epoch": 0.9327622014537902, "grad_norm": 3.414698600769043, "learning_rate": 3.1042198370439925e-06, "loss": 0.09241271018981934, "step": 3593 }, { "epoch": 0.9330218068535826, "grad_norm": 1.3904565572738647, "learning_rate": 3.103285592952427e-06, "loss": 0.1056395024061203, "step": 3594 }, { "epoch": 0.9332814122533749, "grad_norm": 2.4168879985809326, "learning_rate": 3.1023512864080563e-06, "loss": 0.07047528028488159, "step": 3595 }, { "epoch": 0.9335410176531672, "grad_norm": 2.424689531326294, "learning_rate": 3.101416917576047e-06, "loss": 0.07052461802959442, "step": 3596 }, { "epoch": 0.9338006230529595, "grad_norm": 2.6272342205047607, "learning_rate": 3.1004824866215723e-06, "loss": 0.10094346851110458, "step": 3597 }, { "epoch": 0.9340602284527518, "grad_norm": 2.7140953540802, "learning_rate": 3.09954799370982e-06, "loss": 0.0775444507598877, "step": 3598 }, { "epoch": 0.9343198338525441, "grad_norm": 3.6280922889709473, "learning_rate": 3.0986134390059847e-06, "loss": 0.07248266786336899, "step": 3599 }, { "epoch": 0.9345794392523364, "grad_norm": 1.8416401147842407, "learning_rate": 3.0976788226752767e-06, "loss": 0.10441946983337402, "step": 3600 }, { "epoch": 0.9348390446521287, "grad_norm": 2.0816779136657715, "learning_rate": 3.0967441448829132e-06, "loss": 0.11238531768321991, "step": 3601 }, { "epoch": 0.935098650051921, "grad_norm": 3.034031867980957, "learning_rate": 3.0958094057941245e-06, "loss": 0.09917302429676056, "step": 3602 }, { "epoch": 0.9353582554517134, "grad_norm": 1.2624565362930298, "learning_rate": 3.09487460557415e-06, "loss": 0.0913548469543457, "step": 3603 }, { "epoch": 0.9356178608515057, "grad_norm": 3.813518524169922, "learning_rate": 3.093939744388243e-06, "loss": 0.057840753346681595, "step": 3604 }, { "epoch": 0.935877466251298, "grad_norm": 3.5660383701324463, "learning_rate": 3.093004822401664e-06, "loss": 0.06962384283542633, "step": 3605 }, { "epoch": 0.9361370716510904, "grad_norm": 1.9227368831634521, "learning_rate": 3.092069839779686e-06, "loss": 0.07553812861442566, "step": 3606 }, { "epoch": 0.9363966770508827, "grad_norm": 3.7283456325531006, "learning_rate": 3.0911347966875925e-06, "loss": 0.06731721758842468, "step": 3607 }, { "epoch": 0.936656282450675, "grad_norm": 3.543833017349243, "learning_rate": 3.090199693290678e-06, "loss": 0.07006724178791046, "step": 3608 }, { "epoch": 0.9369158878504673, "grad_norm": 4.043896675109863, "learning_rate": 3.0892645297542475e-06, "loss": 0.05990707874298096, "step": 3609 }, { "epoch": 0.9371754932502596, "grad_norm": 2.631091833114624, "learning_rate": 3.088329306243616e-06, "loss": 0.08672875165939331, "step": 3610 }, { "epoch": 0.9374350986500519, "grad_norm": 2.8447766304016113, "learning_rate": 3.0873940229241095e-06, "loss": 0.0674453154206276, "step": 3611 }, { "epoch": 0.9376947040498442, "grad_norm": 3.1545214653015137, "learning_rate": 3.086458679961066e-06, "loss": 0.03380049020051956, "step": 3612 }, { "epoch": 0.9379543094496365, "grad_norm": 4.118220329284668, "learning_rate": 3.0855232775198317e-06, "loss": 0.0530940480530262, "step": 3613 }, { "epoch": 0.9382139148494288, "grad_norm": 2.913048267364502, "learning_rate": 3.084587815765764e-06, "loss": 0.059843748807907104, "step": 3614 }, { "epoch": 0.9384735202492211, "grad_norm": 4.170830249786377, "learning_rate": 3.0836522948642317e-06, "loss": 0.06874063611030579, "step": 3615 }, { "epoch": 0.9387331256490135, "grad_norm": 2.0943734645843506, "learning_rate": 3.0827167149806126e-06, "loss": 0.08811832219362259, "step": 3616 }, { "epoch": 0.9389927310488058, "grad_norm": 1.65646231174469, "learning_rate": 3.0817810762802974e-06, "loss": 0.09195375442504883, "step": 3617 }, { "epoch": 0.9392523364485982, "grad_norm": 1.6727577447891235, "learning_rate": 3.0808453789286842e-06, "loss": 0.11793515086174011, "step": 3618 }, { "epoch": 0.9395119418483905, "grad_norm": 2.5075109004974365, "learning_rate": 3.079909623091183e-06, "loss": 0.08444958925247192, "step": 3619 }, { "epoch": 0.9397715472481828, "grad_norm": 3.9944534301757812, "learning_rate": 3.078973808933215e-06, "loss": 0.08047191798686981, "step": 3620 }, { "epoch": 0.9400311526479751, "grad_norm": 2.9286537170410156, "learning_rate": 3.0780379366202106e-06, "loss": 0.05311274155974388, "step": 3621 }, { "epoch": 0.9402907580477674, "grad_norm": 2.6691129207611084, "learning_rate": 3.077102006317609e-06, "loss": 0.05924198776483536, "step": 3622 }, { "epoch": 0.9405503634475597, "grad_norm": 3.455125570297241, "learning_rate": 3.0761660181908624e-06, "loss": 0.09915336966514587, "step": 3623 }, { "epoch": 0.940809968847352, "grad_norm": 3.5644993782043457, "learning_rate": 3.0752299724054324e-06, "loss": 0.056683316826820374, "step": 3624 }, { "epoch": 0.9410695742471443, "grad_norm": 2.567620038986206, "learning_rate": 3.0742938691267903e-06, "loss": 0.08265090733766556, "step": 3625 }, { "epoch": 0.9413291796469366, "grad_norm": 3.805586338043213, "learning_rate": 3.073357708520417e-06, "loss": 0.05268148332834244, "step": 3626 }, { "epoch": 0.9415887850467289, "grad_norm": 1.6490998268127441, "learning_rate": 3.072421490751805e-06, "loss": 0.09495775401592255, "step": 3627 }, { "epoch": 0.9418483904465212, "grad_norm": 2.2026076316833496, "learning_rate": 3.071485215986455e-06, "loss": 0.07608284801244736, "step": 3628 }, { "epoch": 0.9421079958463136, "grad_norm": 2.7181556224823, "learning_rate": 3.0705488843898816e-06, "loss": 0.0632459819316864, "step": 3629 }, { "epoch": 0.942367601246106, "grad_norm": 2.8106000423431396, "learning_rate": 3.0696124961276036e-06, "loss": 0.076058030128479, "step": 3630 }, { "epoch": 0.9426272066458983, "grad_norm": 4.530228137969971, "learning_rate": 3.0686760513651555e-06, "loss": 0.05466419458389282, "step": 3631 }, { "epoch": 0.9428868120456906, "grad_norm": 3.473780870437622, "learning_rate": 3.0677395502680774e-06, "loss": 0.05271844193339348, "step": 3632 }, { "epoch": 0.9431464174454829, "grad_norm": 2.1627142429351807, "learning_rate": 3.066802993001922e-06, "loss": 0.08175012469291687, "step": 3633 }, { "epoch": 0.9434060228452752, "grad_norm": 3.5175387859344482, "learning_rate": 3.0658663797322518e-06, "loss": 0.10682190954685211, "step": 3634 }, { "epoch": 0.9436656282450675, "grad_norm": 3.3935203552246094, "learning_rate": 3.0649297106246386e-06, "loss": 0.10184940695762634, "step": 3635 }, { "epoch": 0.9439252336448598, "grad_norm": 1.7010568380355835, "learning_rate": 3.0639929858446626e-06, "loss": 0.08181488513946533, "step": 3636 }, { "epoch": 0.9441848390446521, "grad_norm": 1.4674314260482788, "learning_rate": 3.0630562055579154e-06, "loss": 0.08802691847085953, "step": 3637 }, { "epoch": 0.9444444444444444, "grad_norm": 1.5642653703689575, "learning_rate": 3.0621193699300004e-06, "loss": 0.09242713451385498, "step": 3638 }, { "epoch": 0.9447040498442367, "grad_norm": 2.151108980178833, "learning_rate": 3.0611824791265278e-06, "loss": 0.09657721221446991, "step": 3639 }, { "epoch": 0.944963655244029, "grad_norm": 3.3417646884918213, "learning_rate": 3.060245533313117e-06, "loss": 0.10285492241382599, "step": 3640 }, { "epoch": 0.9452232606438213, "grad_norm": 3.3865315914154053, "learning_rate": 3.0593085326553995e-06, "loss": 0.07238836586475372, "step": 3641 }, { "epoch": 0.9454828660436138, "grad_norm": 2.0344746112823486, "learning_rate": 3.058371477319016e-06, "loss": 0.0980326384305954, "step": 3642 }, { "epoch": 0.9457424714434061, "grad_norm": 4.644151210784912, "learning_rate": 3.0574343674696162e-06, "loss": 0.07620994001626968, "step": 3643 }, { "epoch": 0.9460020768431984, "grad_norm": 4.441461563110352, "learning_rate": 3.05649720327286e-06, "loss": 0.06729917228221893, "step": 3644 }, { "epoch": 0.9462616822429907, "grad_norm": 3.160829544067383, "learning_rate": 3.0555599848944152e-06, "loss": 0.11689523607492447, "step": 3645 }, { "epoch": 0.946521287642783, "grad_norm": 2.768444538116455, "learning_rate": 3.054622712499962e-06, "loss": 0.1021183654665947, "step": 3646 }, { "epoch": 0.9467808930425753, "grad_norm": 2.381300210952759, "learning_rate": 3.0536853862551875e-06, "loss": 0.09975363314151764, "step": 3647 }, { "epoch": 0.9470404984423676, "grad_norm": 3.671311378479004, "learning_rate": 3.0527480063257908e-06, "loss": 0.05893195420503616, "step": 3648 }, { "epoch": 0.9473001038421599, "grad_norm": 2.6174986362457275, "learning_rate": 3.0518105728774782e-06, "loss": 0.08508177846670151, "step": 3649 }, { "epoch": 0.9475597092419522, "grad_norm": 3.8778598308563232, "learning_rate": 3.0508730860759673e-06, "loss": 0.0721292793750763, "step": 3650 }, { "epoch": 0.9478193146417445, "grad_norm": 3.349337339401245, "learning_rate": 3.0499355460869833e-06, "loss": 0.09149685502052307, "step": 3651 }, { "epoch": 0.9480789200415368, "grad_norm": 4.4770307540893555, "learning_rate": 3.048997953076263e-06, "loss": 0.04113920032978058, "step": 3652 }, { "epoch": 0.9483385254413291, "grad_norm": 5.544017314910889, "learning_rate": 3.0480603072095507e-06, "loss": 0.09050415456295013, "step": 3653 }, { "epoch": 0.9485981308411215, "grad_norm": 1.7356473207473755, "learning_rate": 3.0471226086526e-06, "loss": 0.10430008172988892, "step": 3654 }, { "epoch": 0.9488577362409139, "grad_norm": 3.53974986076355, "learning_rate": 3.046184857571176e-06, "loss": 0.042245540767908096, "step": 3655 }, { "epoch": 0.9491173416407062, "grad_norm": 1.373363971710205, "learning_rate": 3.0452470541310507e-06, "loss": 0.09532134234905243, "step": 3656 }, { "epoch": 0.9493769470404985, "grad_norm": 4.005951881408691, "learning_rate": 3.0443091984980066e-06, "loss": 0.05660445988178253, "step": 3657 }, { "epoch": 0.9496365524402908, "grad_norm": 2.33392071723938, "learning_rate": 3.0433712908378356e-06, "loss": 0.08275019377470016, "step": 3658 }, { "epoch": 0.9498961578400831, "grad_norm": 3.164034128189087, "learning_rate": 3.042433331316336e-06, "loss": 0.057718679308891296, "step": 3659 }, { "epoch": 0.9501557632398754, "grad_norm": 2.9710874557495117, "learning_rate": 3.04149532009932e-06, "loss": 0.09054244309663773, "step": 3660 }, { "epoch": 0.9504153686396677, "grad_norm": 4.036672115325928, "learning_rate": 3.0405572573526067e-06, "loss": 0.06495906412601471, "step": 3661 }, { "epoch": 0.95067497403946, "grad_norm": 4.128302574157715, "learning_rate": 3.0396191432420225e-06, "loss": 0.07622602581977844, "step": 3662 }, { "epoch": 0.9509345794392523, "grad_norm": 2.2587850093841553, "learning_rate": 3.0386809779334055e-06, "loss": 0.0895543098449707, "step": 3663 }, { "epoch": 0.9511941848390446, "grad_norm": 2.865107297897339, "learning_rate": 3.037742761592601e-06, "loss": 0.09397847950458527, "step": 3664 }, { "epoch": 0.9514537902388369, "grad_norm": 3.839154005050659, "learning_rate": 3.0368044943854647e-06, "loss": 0.07361748814582825, "step": 3665 }, { "epoch": 0.9517133956386293, "grad_norm": 3.2870874404907227, "learning_rate": 3.0358661764778614e-06, "loss": 0.09302496165037155, "step": 3666 }, { "epoch": 0.9519730010384216, "grad_norm": 1.7043402194976807, "learning_rate": 3.0349278080356626e-06, "loss": 0.09537641704082489, "step": 3667 }, { "epoch": 0.952232606438214, "grad_norm": 3.6592955589294434, "learning_rate": 3.0339893892247518e-06, "loss": 0.061870940029621124, "step": 3668 }, { "epoch": 0.9524922118380063, "grad_norm": 3.15263295173645, "learning_rate": 3.0330509202110197e-06, "loss": 0.0907943993806839, "step": 3669 }, { "epoch": 0.9527518172377986, "grad_norm": 3.0732030868530273, "learning_rate": 3.032112401160366e-06, "loss": 0.09015735983848572, "step": 3670 }, { "epoch": 0.9530114226375909, "grad_norm": 4.374019622802734, "learning_rate": 3.0311738322386996e-06, "loss": 0.06326737254858017, "step": 3671 }, { "epoch": 0.9532710280373832, "grad_norm": 1.674330472946167, "learning_rate": 3.0302352136119364e-06, "loss": 0.11870540678501129, "step": 3672 }, { "epoch": 0.9535306334371755, "grad_norm": 2.2428927421569824, "learning_rate": 3.029296545446005e-06, "loss": 0.07993535697460175, "step": 3673 }, { "epoch": 0.9537902388369678, "grad_norm": 2.2794573307037354, "learning_rate": 3.0283578279068387e-06, "loss": 0.059828050434589386, "step": 3674 }, { "epoch": 0.9540498442367601, "grad_norm": 3.8553414344787598, "learning_rate": 3.027419061160383e-06, "loss": 0.08514700829982758, "step": 3675 }, { "epoch": 0.9543094496365524, "grad_norm": 4.299561023712158, "learning_rate": 3.026480245372588e-06, "loss": 0.06502852588891983, "step": 3676 }, { "epoch": 0.9545690550363447, "grad_norm": 4.789578914642334, "learning_rate": 3.025541380709417e-06, "loss": 0.07542859017848969, "step": 3677 }, { "epoch": 0.9548286604361371, "grad_norm": 4.6288604736328125, "learning_rate": 3.024602467336839e-06, "loss": 0.088480643928051, "step": 3678 }, { "epoch": 0.9550882658359294, "grad_norm": 2.0514185428619385, "learning_rate": 3.0236635054208322e-06, "loss": 0.09858736395835876, "step": 3679 }, { "epoch": 0.9553478712357217, "grad_norm": 5.759708404541016, "learning_rate": 3.022724495127384e-06, "loss": 0.06961704790592194, "step": 3680 }, { "epoch": 0.955607476635514, "grad_norm": 1.6668126583099365, "learning_rate": 3.0217854366224885e-06, "loss": 0.10294584184885025, "step": 3681 }, { "epoch": 0.9558670820353063, "grad_norm": 3.0363097190856934, "learning_rate": 3.020846330072152e-06, "loss": 0.06767910718917847, "step": 3682 }, { "epoch": 0.9561266874350987, "grad_norm": 3.7893388271331787, "learning_rate": 3.019907175642386e-06, "loss": 0.05704499036073685, "step": 3683 }, { "epoch": 0.956386292834891, "grad_norm": 4.311222076416016, "learning_rate": 3.0189679734992107e-06, "loss": 0.04909363016486168, "step": 3684 }, { "epoch": 0.9566458982346833, "grad_norm": 4.152881145477295, "learning_rate": 3.0180287238086563e-06, "loss": 0.05345311760902405, "step": 3685 }, { "epoch": 0.9569055036344756, "grad_norm": 1.5759681463241577, "learning_rate": 3.0170894267367605e-06, "loss": 0.10045653581619263, "step": 3686 }, { "epoch": 0.9571651090342679, "grad_norm": 5.450930118560791, "learning_rate": 3.0161500824495704e-06, "loss": 0.07635097205638885, "step": 3687 }, { "epoch": 0.9574247144340602, "grad_norm": 3.3697969913482666, "learning_rate": 3.015210691113139e-06, "loss": 0.045638613402843475, "step": 3688 }, { "epoch": 0.9576843198338525, "grad_norm": 2.128580331802368, "learning_rate": 3.014271252893531e-06, "loss": 0.09923327714204788, "step": 3689 }, { "epoch": 0.9579439252336449, "grad_norm": 3.937572956085205, "learning_rate": 3.0133317679568145e-06, "loss": 0.0524103119969368, "step": 3690 }, { "epoch": 0.9582035306334372, "grad_norm": 1.7739629745483398, "learning_rate": 3.012392236469073e-06, "loss": 0.08933213353157043, "step": 3691 }, { "epoch": 0.9584631360332295, "grad_norm": 1.6752456426620483, "learning_rate": 3.0114526585963903e-06, "loss": 0.08350002765655518, "step": 3692 }, { "epoch": 0.9587227414330218, "grad_norm": 1.934113621711731, "learning_rate": 3.0105130345048654e-06, "loss": 0.08179901540279388, "step": 3693 }, { "epoch": 0.9589823468328141, "grad_norm": 2.003202199935913, "learning_rate": 3.0095733643606005e-06, "loss": 0.08911026269197464, "step": 3694 }, { "epoch": 0.9592419522326064, "grad_norm": 2.3629114627838135, "learning_rate": 3.0086336483297084e-06, "loss": 0.07884903252124786, "step": 3695 }, { "epoch": 0.9595015576323987, "grad_norm": 1.6599822044372559, "learning_rate": 3.007693886578309e-06, "loss": 0.09817411005496979, "step": 3696 }, { "epoch": 0.9597611630321911, "grad_norm": 3.555387258529663, "learning_rate": 3.0067540792725304e-06, "loss": 0.10362543165683746, "step": 3697 }, { "epoch": 0.9600207684319834, "grad_norm": 4.860383987426758, "learning_rate": 3.00581422657851e-06, "loss": 0.0566069521009922, "step": 3698 }, { "epoch": 0.9602803738317757, "grad_norm": 1.5938124656677246, "learning_rate": 3.004874328662391e-06, "loss": 0.11259271204471588, "step": 3699 }, { "epoch": 0.960539979231568, "grad_norm": 3.243175506591797, "learning_rate": 3.003934385690328e-06, "loss": 0.03356234356760979, "step": 3700 }, { "epoch": 0.9607995846313603, "grad_norm": 5.597649574279785, "learning_rate": 3.002994397828479e-06, "loss": 0.05255626142024994, "step": 3701 }, { "epoch": 0.9610591900311527, "grad_norm": 3.901491165161133, "learning_rate": 3.002054365243014e-06, "loss": 0.05499144643545151, "step": 3702 }, { "epoch": 0.961318795430945, "grad_norm": 2.36871600151062, "learning_rate": 3.001114288100107e-06, "loss": 0.09224960207939148, "step": 3703 }, { "epoch": 0.9615784008307373, "grad_norm": 3.3684096336364746, "learning_rate": 3.0001741665659444e-06, "loss": 0.0476701557636261, "step": 3704 }, { "epoch": 0.9618380062305296, "grad_norm": 2.5881881713867188, "learning_rate": 2.9992340008067177e-06, "loss": 0.08723937720060349, "step": 3705 }, { "epoch": 0.9620976116303219, "grad_norm": 2.2089297771453857, "learning_rate": 2.9982937909886267e-06, "loss": 0.07888585329055786, "step": 3706 }, { "epoch": 0.9623572170301142, "grad_norm": 3.4774763584136963, "learning_rate": 2.997353537277878e-06, "loss": 0.06857387721538544, "step": 3707 }, { "epoch": 0.9626168224299065, "grad_norm": 3.20133638381958, "learning_rate": 2.996413239840687e-06, "loss": 0.0966297835111618, "step": 3708 }, { "epoch": 0.9628764278296988, "grad_norm": 1.7519145011901855, "learning_rate": 2.995472898843278e-06, "loss": 0.08594135195016861, "step": 3709 }, { "epoch": 0.9631360332294912, "grad_norm": 3.5651094913482666, "learning_rate": 2.994532514451881e-06, "loss": 0.061971358954906464, "step": 3710 }, { "epoch": 0.9633956386292835, "grad_norm": 3.57389497756958, "learning_rate": 2.9935920868327345e-06, "loss": 0.05544668808579445, "step": 3711 }, { "epoch": 0.9636552440290758, "grad_norm": 3.9634902477264404, "learning_rate": 2.9926516161520848e-06, "loss": 0.08177411556243896, "step": 3712 }, { "epoch": 0.9639148494288681, "grad_norm": 1.9644436836242676, "learning_rate": 2.9917111025761848e-06, "loss": 0.0962480828166008, "step": 3713 }, { "epoch": 0.9641744548286605, "grad_norm": 3.3868701457977295, "learning_rate": 2.9907705462712974e-06, "loss": 0.05977747589349747, "step": 3714 }, { "epoch": 0.9644340602284528, "grad_norm": 4.048116207122803, "learning_rate": 2.989829947403689e-06, "loss": 0.06588780879974365, "step": 3715 }, { "epoch": 0.9646936656282451, "grad_norm": 1.8529466390609741, "learning_rate": 2.9888893061396375e-06, "loss": 0.10148105770349503, "step": 3716 }, { "epoch": 0.9649532710280374, "grad_norm": 4.010188579559326, "learning_rate": 2.987948622645427e-06, "loss": 0.031642328947782516, "step": 3717 }, { "epoch": 0.9652128764278297, "grad_norm": 1.651265025138855, "learning_rate": 2.9870078970873483e-06, "loss": 0.0944724828004837, "step": 3718 }, { "epoch": 0.965472481827622, "grad_norm": 3.1945271492004395, "learning_rate": 2.9860671296317e-06, "loss": 0.08914631605148315, "step": 3719 }, { "epoch": 0.9657320872274143, "grad_norm": 1.83596932888031, "learning_rate": 2.985126320444789e-06, "loss": 0.09508201479911804, "step": 3720 }, { "epoch": 0.9659916926272066, "grad_norm": 2.1360955238342285, "learning_rate": 2.9841854696929266e-06, "loss": 0.09695896506309509, "step": 3721 }, { "epoch": 0.9662512980269989, "grad_norm": 3.9556589126586914, "learning_rate": 2.983244577542437e-06, "loss": 0.08969169855117798, "step": 3722 }, { "epoch": 0.9665109034267912, "grad_norm": 3.8058273792266846, "learning_rate": 2.9823036441596465e-06, "loss": 0.04570002108812332, "step": 3723 }, { "epoch": 0.9667705088265836, "grad_norm": 2.0212321281433105, "learning_rate": 2.98136266971089e-06, "loss": 0.09459122270345688, "step": 3724 }, { "epoch": 0.9670301142263759, "grad_norm": 2.2409908771514893, "learning_rate": 2.980421654362511e-06, "loss": 0.12198961526155472, "step": 3725 }, { "epoch": 0.9672897196261683, "grad_norm": 3.9856488704681396, "learning_rate": 2.97948059828086e-06, "loss": 0.06883585453033447, "step": 3726 }, { "epoch": 0.9675493250259606, "grad_norm": 3.14955735206604, "learning_rate": 2.978539501632293e-06, "loss": 0.0763464868068695, "step": 3727 }, { "epoch": 0.9678089304257529, "grad_norm": 4.288901329040527, "learning_rate": 2.977598364583176e-06, "loss": 0.06029623746871948, "step": 3728 }, { "epoch": 0.9680685358255452, "grad_norm": 2.722208261489868, "learning_rate": 2.9766571872998777e-06, "loss": 0.07933486998081207, "step": 3729 }, { "epoch": 0.9683281412253375, "grad_norm": 3.4451956748962402, "learning_rate": 2.9757159699487798e-06, "loss": 0.07066041976213455, "step": 3730 }, { "epoch": 0.9685877466251298, "grad_norm": 4.01724100112915, "learning_rate": 2.9747747126962666e-06, "loss": 0.06786338239908218, "step": 3731 }, { "epoch": 0.9688473520249221, "grad_norm": 2.571070432662964, "learning_rate": 2.9738334157087307e-06, "loss": 0.10430273413658142, "step": 3732 }, { "epoch": 0.9691069574247144, "grad_norm": 3.272552490234375, "learning_rate": 2.9728920791525724e-06, "loss": 0.08961574733257294, "step": 3733 }, { "epoch": 0.9693665628245067, "grad_norm": 3.8489370346069336, "learning_rate": 2.971950703194197e-06, "loss": 0.07373034209012985, "step": 3734 }, { "epoch": 0.969626168224299, "grad_norm": 4.289632797241211, "learning_rate": 2.97100928800002e-06, "loss": 0.04352794587612152, "step": 3735 }, { "epoch": 0.9698857736240913, "grad_norm": 4.034949779510498, "learning_rate": 2.9700678337364615e-06, "loss": 0.07847610116004944, "step": 3736 }, { "epoch": 0.9701453790238836, "grad_norm": 3.3900258541107178, "learning_rate": 2.9691263405699485e-06, "loss": 0.05765172466635704, "step": 3737 }, { "epoch": 0.9704049844236761, "grad_norm": 3.7113218307495117, "learning_rate": 2.968184808666916e-06, "loss": 0.05996687710285187, "step": 3738 }, { "epoch": 0.9706645898234684, "grad_norm": 3.597993850708008, "learning_rate": 2.967243238193805e-06, "loss": 0.06012851372361183, "step": 3739 }, { "epoch": 0.9709241952232607, "grad_norm": 2.6158459186553955, "learning_rate": 2.9663016293170632e-06, "loss": 0.09305056184530258, "step": 3740 }, { "epoch": 0.971183800623053, "grad_norm": 3.1381163597106934, "learning_rate": 2.965359982203147e-06, "loss": 0.04356885328888893, "step": 3741 }, { "epoch": 0.9714434060228453, "grad_norm": 3.76541805267334, "learning_rate": 2.9644182970185163e-06, "loss": 0.0480472594499588, "step": 3742 }, { "epoch": 0.9717030114226376, "grad_norm": 3.140334367752075, "learning_rate": 2.96347657392964e-06, "loss": 0.05294357240200043, "step": 3743 }, { "epoch": 0.9719626168224299, "grad_norm": 1.2765618562698364, "learning_rate": 2.962534813102994e-06, "loss": 0.12143637984991074, "step": 3744 }, { "epoch": 0.9722222222222222, "grad_norm": 2.2182130813598633, "learning_rate": 2.9615930147050597e-06, "loss": 0.09345324337482452, "step": 3745 }, { "epoch": 0.9724818276220145, "grad_norm": 2.262674570083618, "learning_rate": 2.9606511789023247e-06, "loss": 0.08134831488132477, "step": 3746 }, { "epoch": 0.9727414330218068, "grad_norm": 4.058751583099365, "learning_rate": 2.9597093058612843e-06, "loss": 0.05141162872314453, "step": 3747 }, { "epoch": 0.9730010384215991, "grad_norm": 3.284709930419922, "learning_rate": 2.958767395748441e-06, "loss": 0.08291293680667877, "step": 3748 }, { "epoch": 0.9732606438213914, "grad_norm": 3.2027692794799805, "learning_rate": 2.957825448730302e-06, "loss": 0.09665380418300629, "step": 3749 }, { "epoch": 0.9735202492211839, "grad_norm": 1.854649305343628, "learning_rate": 2.9568834649733825e-06, "loss": 0.10162712633609772, "step": 3750 }, { "epoch": 0.9737798546209762, "grad_norm": 3.8596436977386475, "learning_rate": 2.9559414446442024e-06, "loss": 0.05738899111747742, "step": 3751 }, { "epoch": 0.9740394600207685, "grad_norm": 3.119643449783325, "learning_rate": 2.9549993879092912e-06, "loss": 0.044160146266222, "step": 3752 }, { "epoch": 0.9742990654205608, "grad_norm": 1.764150619506836, "learning_rate": 2.954057294935182e-06, "loss": 0.08207149803638458, "step": 3753 }, { "epoch": 0.9745586708203531, "grad_norm": 2.795138359069824, "learning_rate": 2.9531151658884156e-06, "loss": 0.07680969685316086, "step": 3754 }, { "epoch": 0.9748182762201454, "grad_norm": 4.140651226043701, "learning_rate": 2.9521730009355383e-06, "loss": 0.075322724878788, "step": 3755 }, { "epoch": 0.9750778816199377, "grad_norm": 3.4456727504730225, "learning_rate": 2.951230800243103e-06, "loss": 0.06370905041694641, "step": 3756 }, { "epoch": 0.97533748701973, "grad_norm": 3.9694912433624268, "learning_rate": 2.95028856397767e-06, "loss": 0.05878719314932823, "step": 3757 }, { "epoch": 0.9755970924195223, "grad_norm": 5.772354602813721, "learning_rate": 2.9493462923058057e-06, "loss": 0.05351804569363594, "step": 3758 }, { "epoch": 0.9758566978193146, "grad_norm": 3.177046775817871, "learning_rate": 2.948403985394081e-06, "loss": 0.08853171765804291, "step": 3759 }, { "epoch": 0.9761163032191069, "grad_norm": 2.934659242630005, "learning_rate": 2.9474616434090737e-06, "loss": 0.0939771756529808, "step": 3760 }, { "epoch": 0.9763759086188992, "grad_norm": 2.9095892906188965, "learning_rate": 2.9465192665173697e-06, "loss": 0.10345730185508728, "step": 3761 }, { "epoch": 0.9766355140186916, "grad_norm": 3.6295909881591797, "learning_rate": 2.9455768548855597e-06, "loss": 0.05711677297949791, "step": 3762 }, { "epoch": 0.976895119418484, "grad_norm": 1.5515005588531494, "learning_rate": 2.944634408680239e-06, "loss": 0.10416745394468307, "step": 3763 }, { "epoch": 0.9771547248182763, "grad_norm": 1.9922655820846558, "learning_rate": 2.9436919280680115e-06, "loss": 0.10472021251916885, "step": 3764 }, { "epoch": 0.9774143302180686, "grad_norm": 2.209766149520874, "learning_rate": 2.942749413215486e-06, "loss": 0.09403250366449356, "step": 3765 }, { "epoch": 0.9776739356178609, "grad_norm": 1.3219958543777466, "learning_rate": 2.941806864289279e-06, "loss": 0.10314483940601349, "step": 3766 }, { "epoch": 0.9779335410176532, "grad_norm": 2.346670389175415, "learning_rate": 2.9408642814560082e-06, "loss": 0.07970486581325531, "step": 3767 }, { "epoch": 0.9781931464174455, "grad_norm": 3.9056787490844727, "learning_rate": 2.939921664882304e-06, "loss": 0.04717036709189415, "step": 3768 }, { "epoch": 0.9784527518172378, "grad_norm": 4.395453929901123, "learning_rate": 2.9389790147347973e-06, "loss": 0.07124593108892441, "step": 3769 }, { "epoch": 0.9787123572170301, "grad_norm": 4.222437858581543, "learning_rate": 2.9380363311801286e-06, "loss": 0.05269584059715271, "step": 3770 }, { "epoch": 0.9789719626168224, "grad_norm": 3.397653102874756, "learning_rate": 2.9370936143849415e-06, "loss": 0.05015336722135544, "step": 3771 }, { "epoch": 0.9792315680166147, "grad_norm": 2.5071089267730713, "learning_rate": 2.9361508645158875e-06, "loss": 0.08952329307794571, "step": 3772 }, { "epoch": 0.979491173416407, "grad_norm": 3.4593276977539062, "learning_rate": 2.9352080817396227e-06, "loss": 0.08520950376987457, "step": 3773 }, { "epoch": 0.9797507788161994, "grad_norm": 3.7504801750183105, "learning_rate": 2.93426526622281e-06, "loss": 0.04293660819530487, "step": 3774 }, { "epoch": 0.9800103842159917, "grad_norm": 1.6174352169036865, "learning_rate": 2.933322418132117e-06, "loss": 0.11194588243961334, "step": 3775 }, { "epoch": 0.980269989615784, "grad_norm": 2.137787103652954, "learning_rate": 2.932379537634218e-06, "loss": 0.06280428171157837, "step": 3776 }, { "epoch": 0.9805295950155763, "grad_norm": 3.5950026512145996, "learning_rate": 2.9314366248957927e-06, "loss": 0.06342744827270508, "step": 3777 }, { "epoch": 0.9807892004153687, "grad_norm": 3.1076154708862305, "learning_rate": 2.9304936800835264e-06, "loss": 0.06807491183280945, "step": 3778 }, { "epoch": 0.981048805815161, "grad_norm": 4.192346096038818, "learning_rate": 2.9295507033641105e-06, "loss": 0.053594812750816345, "step": 3779 }, { "epoch": 0.9813084112149533, "grad_norm": 1.940161943435669, "learning_rate": 2.9286076949042415e-06, "loss": 0.10307636111974716, "step": 3780 }, { "epoch": 0.9815680166147456, "grad_norm": 1.7476645708084106, "learning_rate": 2.9276646548706222e-06, "loss": 0.09697765111923218, "step": 3781 }, { "epoch": 0.9818276220145379, "grad_norm": 1.3177584409713745, "learning_rate": 2.9267215834299585e-06, "loss": 0.10432390123605728, "step": 3782 }, { "epoch": 0.9820872274143302, "grad_norm": 3.0882568359375, "learning_rate": 2.9257784807489663e-06, "loss": 0.039057787507772446, "step": 3783 }, { "epoch": 0.9823468328141225, "grad_norm": 4.662946701049805, "learning_rate": 2.924835346994363e-06, "loss": 0.03141526132822037, "step": 3784 }, { "epoch": 0.9826064382139148, "grad_norm": 2.042773723602295, "learning_rate": 2.9238921823328737e-06, "loss": 0.10338152199983597, "step": 3785 }, { "epoch": 0.9828660436137072, "grad_norm": 1.7302265167236328, "learning_rate": 2.9229489869312276e-06, "loss": 0.09231288731098175, "step": 3786 }, { "epoch": 0.9831256490134995, "grad_norm": 2.101046323776245, "learning_rate": 2.9220057609561616e-06, "loss": 0.07117001712322235, "step": 3787 }, { "epoch": 0.9833852544132918, "grad_norm": 3.482602834701538, "learning_rate": 2.9210625045744143e-06, "loss": 0.07715369760990143, "step": 3788 }, { "epoch": 0.9836448598130841, "grad_norm": 2.0727665424346924, "learning_rate": 2.920119217952735e-06, "loss": 0.09416355192661285, "step": 3789 }, { "epoch": 0.9839044652128764, "grad_norm": 4.350922107696533, "learning_rate": 2.9191759012578712e-06, "loss": 0.05803317949175835, "step": 3790 }, { "epoch": 0.9841640706126688, "grad_norm": 1.4818902015686035, "learning_rate": 2.9182325546565822e-06, "loss": 0.10344406962394714, "step": 3791 }, { "epoch": 0.9844236760124611, "grad_norm": 1.7615128755569458, "learning_rate": 2.91728917831563e-06, "loss": 0.1023402214050293, "step": 3792 }, { "epoch": 0.9846832814122534, "grad_norm": 4.419196605682373, "learning_rate": 2.916345772401781e-06, "loss": 0.09274981915950775, "step": 3793 }, { "epoch": 0.9849428868120457, "grad_norm": 1.2007758617401123, "learning_rate": 2.9154023370818085e-06, "loss": 0.11810126155614853, "step": 3794 }, { "epoch": 0.985202492211838, "grad_norm": 2.671468496322632, "learning_rate": 2.91445887252249e-06, "loss": 0.06927206367254257, "step": 3795 }, { "epoch": 0.9854620976116303, "grad_norm": 2.9628818035125732, "learning_rate": 2.9135153788906084e-06, "loss": 0.10534881055355072, "step": 3796 }, { "epoch": 0.9857217030114226, "grad_norm": 2.2991974353790283, "learning_rate": 2.912571856352952e-06, "loss": 0.0963401049375534, "step": 3797 }, { "epoch": 0.985981308411215, "grad_norm": 3.793158769607544, "learning_rate": 2.9116283050763137e-06, "loss": 0.058929748833179474, "step": 3798 }, { "epoch": 0.9862409138110073, "grad_norm": 3.4577996730804443, "learning_rate": 2.910684725227492e-06, "loss": 0.09410711377859116, "step": 3799 }, { "epoch": 0.9865005192107996, "grad_norm": 2.6787848472595215, "learning_rate": 2.909741116973289e-06, "loss": 0.0887911319732666, "step": 3800 }, { "epoch": 0.9867601246105919, "grad_norm": 3.350346326828003, "learning_rate": 2.908797480480515e-06, "loss": 0.05716901645064354, "step": 3801 }, { "epoch": 0.9870197300103842, "grad_norm": 4.440993785858154, "learning_rate": 2.9078538159159823e-06, "loss": 0.06707073003053665, "step": 3802 }, { "epoch": 0.9872793354101765, "grad_norm": 3.063890218734741, "learning_rate": 2.90691012344651e-06, "loss": 0.03313381224870682, "step": 3803 }, { "epoch": 0.9875389408099688, "grad_norm": 4.716711044311523, "learning_rate": 2.90596640323892e-06, "loss": 0.09162501990795135, "step": 3804 }, { "epoch": 0.9877985462097612, "grad_norm": 3.1095876693725586, "learning_rate": 2.905022655460041e-06, "loss": 0.10095643997192383, "step": 3805 }, { "epoch": 0.9880581516095535, "grad_norm": 1.9011375904083252, "learning_rate": 2.9040788802767065e-06, "loss": 0.08848802745342255, "step": 3806 }, { "epoch": 0.9883177570093458, "grad_norm": 1.6128458976745605, "learning_rate": 2.903135077855754e-06, "loss": 0.10744383186101913, "step": 3807 }, { "epoch": 0.9885773624091381, "grad_norm": 3.473118782043457, "learning_rate": 2.902191248364025e-06, "loss": 0.05429825559258461, "step": 3808 }, { "epoch": 0.9888369678089304, "grad_norm": 3.2687273025512695, "learning_rate": 2.9012473919683686e-06, "loss": 0.10804633796215057, "step": 3809 }, { "epoch": 0.9890965732087228, "grad_norm": 2.4981470108032227, "learning_rate": 2.900303508835637e-06, "loss": 0.10241737961769104, "step": 3810 }, { "epoch": 0.9893561786085151, "grad_norm": 3.71797513961792, "learning_rate": 2.8993595991326854e-06, "loss": 0.03978119045495987, "step": 3811 }, { "epoch": 0.9896157840083074, "grad_norm": 3.616995096206665, "learning_rate": 2.8984156630263777e-06, "loss": 0.04249778389930725, "step": 3812 }, { "epoch": 0.9898753894080997, "grad_norm": 3.5524485111236572, "learning_rate": 2.8974717006835785e-06, "loss": 0.07295583188533783, "step": 3813 }, { "epoch": 0.990134994807892, "grad_norm": 4.163711071014404, "learning_rate": 2.8965277122711593e-06, "loss": 0.06427454203367233, "step": 3814 }, { "epoch": 0.9903946002076843, "grad_norm": 3.6929619312286377, "learning_rate": 2.895583697955995e-06, "loss": 0.0826290100812912, "step": 3815 }, { "epoch": 0.9906542056074766, "grad_norm": 2.4346764087677, "learning_rate": 2.8946396579049674e-06, "loss": 0.08912179619073868, "step": 3816 }, { "epoch": 0.9909138110072689, "grad_norm": 1.4274239540100098, "learning_rate": 2.8936955922849597e-06, "loss": 0.0948462188243866, "step": 3817 }, { "epoch": 0.9911734164070612, "grad_norm": 3.897770643234253, "learning_rate": 2.8927515012628616e-06, "loss": 0.06194593384861946, "step": 3818 }, { "epoch": 0.9914330218068536, "grad_norm": 3.268219470977783, "learning_rate": 2.8918073850055663e-06, "loss": 0.09082257002592087, "step": 3819 }, { "epoch": 0.9916926272066459, "grad_norm": 1.9677422046661377, "learning_rate": 2.8908632436799725e-06, "loss": 0.1037038266658783, "step": 3820 }, { "epoch": 0.9919522326064382, "grad_norm": 3.8055055141448975, "learning_rate": 2.8899190774529823e-06, "loss": 0.06763671338558197, "step": 3821 }, { "epoch": 0.9922118380062306, "grad_norm": 4.18775749206543, "learning_rate": 2.888974886491503e-06, "loss": 0.06990934908390045, "step": 3822 }, { "epoch": 0.9924714434060229, "grad_norm": 4.002546310424805, "learning_rate": 2.888030670962446e-06, "loss": 0.08070377260446548, "step": 3823 }, { "epoch": 0.9927310488058152, "grad_norm": 3.5735042095184326, "learning_rate": 2.8870864310327267e-06, "loss": 0.0453089214861393, "step": 3824 }, { "epoch": 0.9929906542056075, "grad_norm": 1.377545952796936, "learning_rate": 2.886142166869265e-06, "loss": 0.10349705815315247, "step": 3825 }, { "epoch": 0.9932502596053998, "grad_norm": 2.7581002712249756, "learning_rate": 2.885197878638986e-06, "loss": 0.07650306075811386, "step": 3826 }, { "epoch": 0.9935098650051921, "grad_norm": 3.702047109603882, "learning_rate": 2.8842535665088177e-06, "loss": 0.03672522306442261, "step": 3827 }, { "epoch": 0.9937694704049844, "grad_norm": 2.8396401405334473, "learning_rate": 2.8833092306456934e-06, "loss": 0.06446611881256104, "step": 3828 }, { "epoch": 0.9940290758047767, "grad_norm": 3.1943373680114746, "learning_rate": 2.882364871216548e-06, "loss": 0.05349871888756752, "step": 3829 }, { "epoch": 0.994288681204569, "grad_norm": 2.7462446689605713, "learning_rate": 2.881420488388325e-06, "loss": 0.10826031118631363, "step": 3830 }, { "epoch": 0.9945482866043613, "grad_norm": 2.3159689903259277, "learning_rate": 2.8804760823279687e-06, "loss": 0.09761229157447815, "step": 3831 }, { "epoch": 0.9948078920041536, "grad_norm": 3.9502034187316895, "learning_rate": 2.8795316532024287e-06, "loss": 0.07598866522312164, "step": 3832 }, { "epoch": 0.995067497403946, "grad_norm": 2.896371841430664, "learning_rate": 2.878587201178658e-06, "loss": 0.0989428162574768, "step": 3833 }, { "epoch": 0.9953271028037384, "grad_norm": 2.936708927154541, "learning_rate": 2.8776427264236147e-06, "loss": 0.07487104088068008, "step": 3834 }, { "epoch": 0.9955867082035307, "grad_norm": 3.7769761085510254, "learning_rate": 2.8766982291042593e-06, "loss": 0.05500185489654541, "step": 3835 }, { "epoch": 0.995846313603323, "grad_norm": 2.159705877304077, "learning_rate": 2.875753709387559e-06, "loss": 0.08390428125858307, "step": 3836 }, { "epoch": 0.9961059190031153, "grad_norm": 3.641385793685913, "learning_rate": 2.8748091674404816e-06, "loss": 0.06887105107307434, "step": 3837 }, { "epoch": 0.9963655244029076, "grad_norm": 4.632155895233154, "learning_rate": 2.8738646034300015e-06, "loss": 0.055250898003578186, "step": 3838 }, { "epoch": 0.9966251298026999, "grad_norm": 3.646052360534668, "learning_rate": 2.8729200175230955e-06, "loss": 0.07191421091556549, "step": 3839 }, { "epoch": 0.9968847352024922, "grad_norm": 4.201744556427002, "learning_rate": 2.8719754098867446e-06, "loss": 0.046204470098018646, "step": 3840 }, { "epoch": 0.9971443406022845, "grad_norm": 2.28869366645813, "learning_rate": 2.8710307806879343e-06, "loss": 0.06837010383605957, "step": 3841 }, { "epoch": 0.9974039460020768, "grad_norm": 5.133845329284668, "learning_rate": 2.870086130093653e-06, "loss": 0.050934381783008575, "step": 3842 }, { "epoch": 0.9976635514018691, "grad_norm": 3.78029465675354, "learning_rate": 2.8691414582708943e-06, "loss": 0.05182820186018944, "step": 3843 }, { "epoch": 0.9979231568016614, "grad_norm": 4.4122490882873535, "learning_rate": 2.8681967653866527e-06, "loss": 0.035583607852458954, "step": 3844 }, { "epoch": 0.9981827622014537, "grad_norm": 1.8228809833526611, "learning_rate": 2.86725205160793e-06, "loss": 0.10429354012012482, "step": 3845 }, { "epoch": 0.9984423676012462, "grad_norm": 2.9873130321502686, "learning_rate": 2.8663073171017284e-06, "loss": 0.05800313502550125, "step": 3846 }, { "epoch": 0.9987019730010385, "grad_norm": 3.822270154953003, "learning_rate": 2.865362562035058e-06, "loss": 0.04088886082172394, "step": 3847 }, { "epoch": 0.9989615784008308, "grad_norm": 3.297255754470825, "learning_rate": 2.8644177865749266e-06, "loss": 0.06220449134707451, "step": 3848 }, { "epoch": 0.9992211838006231, "grad_norm": 4.0427775382995605, "learning_rate": 2.8634729908883506e-06, "loss": 0.07457905262708664, "step": 3849 }, { "epoch": 0.9994807892004154, "grad_norm": 3.7315518856048584, "learning_rate": 2.8625281751423485e-06, "loss": 0.08598074316978455, "step": 3850 }, { "epoch": 0.9997403946002077, "grad_norm": 2.5912890434265137, "learning_rate": 2.861583339503941e-06, "loss": 0.07676048576831818, "step": 3851 }, { "epoch": 1.0, "grad_norm": 3.3661513328552246, "learning_rate": 2.8606384841401548e-06, "loss": 0.0556032657623291, "step": 3852 }, { "epoch": 1.0002596053997923, "grad_norm": 3.2063560485839844, "learning_rate": 2.8596936092180173e-06, "loss": 0.07013827562332153, "step": 3853 }, { "epoch": 1.0005192107995846, "grad_norm": 3.541959762573242, "learning_rate": 2.858748714904561e-06, "loss": 0.046537186950445175, "step": 3854 }, { "epoch": 1.000778816199377, "grad_norm": 1.9636648893356323, "learning_rate": 2.857803801366823e-06, "loss": 0.08789487183094025, "step": 3855 }, { "epoch": 1.0010384215991692, "grad_norm": 3.0053153038024902, "learning_rate": 2.85685886877184e-06, "loss": 0.039828211069107056, "step": 3856 }, { "epoch": 1.0012980269989615, "grad_norm": 3.8160979747772217, "learning_rate": 2.8559139172866563e-06, "loss": 0.027743253856897354, "step": 3857 }, { "epoch": 1.0015576323987538, "grad_norm": 2.6632330417633057, "learning_rate": 2.854968947078317e-06, "loss": 0.08319748938083649, "step": 3858 }, { "epoch": 1.0018172377985461, "grad_norm": 3.221773386001587, "learning_rate": 2.854023958313872e-06, "loss": 0.08997337520122528, "step": 3859 }, { "epoch": 1.0020768431983385, "grad_norm": 3.4035699367523193, "learning_rate": 2.853078951160373e-06, "loss": 0.022275583818554878, "step": 3860 }, { "epoch": 1.0023364485981308, "grad_norm": 2.0279476642608643, "learning_rate": 2.852133925784875e-06, "loss": 0.08317860960960388, "step": 3861 }, { "epoch": 1.002596053997923, "grad_norm": 1.8323724269866943, "learning_rate": 2.8511888823544376e-06, "loss": 0.0875239223241806, "step": 3862 }, { "epoch": 1.0028556593977154, "grad_norm": 5.001754283905029, "learning_rate": 2.850243821036123e-06, "loss": 0.04552220553159714, "step": 3863 }, { "epoch": 1.0031152647975077, "grad_norm": 2.062739133834839, "learning_rate": 2.849298741996996e-06, "loss": 0.08270089328289032, "step": 3864 }, { "epoch": 1.0033748701973002, "grad_norm": 3.7661566734313965, "learning_rate": 2.8483536454041245e-06, "loss": 0.023021593689918518, "step": 3865 }, { "epoch": 1.0036344755970925, "grad_norm": 1.6341543197631836, "learning_rate": 2.847408531424581e-06, "loss": 0.07400538772344589, "step": 3866 }, { "epoch": 1.0038940809968848, "grad_norm": 3.4578659534454346, "learning_rate": 2.8464634002254387e-06, "loss": 0.017155135050415993, "step": 3867 }, { "epoch": 1.0041536863966771, "grad_norm": 3.383413553237915, "learning_rate": 2.8455182519737774e-06, "loss": 0.05727564916014671, "step": 3868 }, { "epoch": 1.0044132917964694, "grad_norm": 3.584312915802002, "learning_rate": 2.844573086836675e-06, "loss": 0.03551732003688812, "step": 3869 }, { "epoch": 1.0046728971962617, "grad_norm": 3.077777862548828, "learning_rate": 2.8436279049812167e-06, "loss": 0.03079775534570217, "step": 3870 }, { "epoch": 1.004932502596054, "grad_norm": 4.054869651794434, "learning_rate": 2.8426827065744887e-06, "loss": 0.035153429955244064, "step": 3871 }, { "epoch": 1.0051921079958464, "grad_norm": 2.7471506595611572, "learning_rate": 2.8417374917835804e-06, "loss": 0.06278656423091888, "step": 3872 }, { "epoch": 1.0054517133956387, "grad_norm": 1.6551330089569092, "learning_rate": 2.8407922607755843e-06, "loss": 0.08901407569646835, "step": 3873 }, { "epoch": 1.005711318795431, "grad_norm": 1.8037744760513306, "learning_rate": 2.839847013717595e-06, "loss": 0.07957002520561218, "step": 3874 }, { "epoch": 1.0059709241952233, "grad_norm": 3.318236827850342, "learning_rate": 2.838901750776711e-06, "loss": 0.02462673932313919, "step": 3875 }, { "epoch": 1.0062305295950156, "grad_norm": 3.0864038467407227, "learning_rate": 2.8379564721200333e-06, "loss": 0.049985144287347794, "step": 3876 }, { "epoch": 1.0064901349948079, "grad_norm": 1.7689943313598633, "learning_rate": 2.8370111779146646e-06, "loss": 0.10421743988990784, "step": 3877 }, { "epoch": 1.0067497403946002, "grad_norm": 1.9958431720733643, "learning_rate": 2.8360658683277125e-06, "loss": 0.08271698653697968, "step": 3878 }, { "epoch": 1.0070093457943925, "grad_norm": 2.8583855628967285, "learning_rate": 2.835120543526285e-06, "loss": 0.042725831270217896, "step": 3879 }, { "epoch": 1.0072689511941848, "grad_norm": 3.6256258487701416, "learning_rate": 2.8341752036774945e-06, "loss": 0.04033851623535156, "step": 3880 }, { "epoch": 1.007528556593977, "grad_norm": 2.6031668186187744, "learning_rate": 2.8332298489484555e-06, "loss": 0.04100828990340233, "step": 3881 }, { "epoch": 1.0077881619937694, "grad_norm": 2.308138608932495, "learning_rate": 2.832284479506286e-06, "loss": 0.060905035585165024, "step": 3882 }, { "epoch": 1.0080477673935617, "grad_norm": 3.557830572128296, "learning_rate": 2.8313390955181025e-06, "loss": 0.03271614760160446, "step": 3883 }, { "epoch": 1.008307372793354, "grad_norm": 2.0777978897094727, "learning_rate": 2.830393697151031e-06, "loss": 0.1057397648692131, "step": 3884 }, { "epoch": 1.0085669781931463, "grad_norm": 2.9042623043060303, "learning_rate": 2.8294482845721934e-06, "loss": 0.06537553668022156, "step": 3885 }, { "epoch": 1.0088265835929386, "grad_norm": 2.2774300575256348, "learning_rate": 2.828502857948719e-06, "loss": 0.06570485979318619, "step": 3886 }, { "epoch": 1.009086188992731, "grad_norm": 2.3489601612091064, "learning_rate": 2.8275574174477372e-06, "loss": 0.05610572546720505, "step": 3887 }, { "epoch": 1.0093457943925233, "grad_norm": 1.9092215299606323, "learning_rate": 2.8266119632363786e-06, "loss": 0.07582561671733856, "step": 3888 }, { "epoch": 1.0096053997923158, "grad_norm": 2.054839849472046, "learning_rate": 2.8256664954817805e-06, "loss": 0.08760957419872284, "step": 3889 }, { "epoch": 1.009865005192108, "grad_norm": 1.7160168886184692, "learning_rate": 2.8247210143510774e-06, "loss": 0.08151420205831528, "step": 3890 }, { "epoch": 1.0101246105919004, "grad_norm": 4.533924102783203, "learning_rate": 2.8237755200114105e-06, "loss": 0.05472264066338539, "step": 3891 }, { "epoch": 1.0103842159916927, "grad_norm": 4.00726318359375, "learning_rate": 2.82283001262992e-06, "loss": 0.02946130931377411, "step": 3892 }, { "epoch": 1.010643821391485, "grad_norm": 2.747621536254883, "learning_rate": 2.8218844923737517e-06, "loss": 0.06393659114837646, "step": 3893 }, { "epoch": 1.0109034267912773, "grad_norm": 1.9437428712844849, "learning_rate": 2.8209389594100507e-06, "loss": 0.0892542153596878, "step": 3894 }, { "epoch": 1.0111630321910696, "grad_norm": 2.389765501022339, "learning_rate": 2.8199934139059664e-06, "loss": 0.06118575111031532, "step": 3895 }, { "epoch": 1.011422637590862, "grad_norm": 3.218963861465454, "learning_rate": 2.8190478560286488e-06, "loss": 0.04428906738758087, "step": 3896 }, { "epoch": 1.0116822429906542, "grad_norm": 3.836575508117676, "learning_rate": 2.818102285945251e-06, "loss": 0.045919790863990784, "step": 3897 }, { "epoch": 1.0119418483904465, "grad_norm": 2.5592286586761475, "learning_rate": 2.8171567038229286e-06, "loss": 0.06973408162593842, "step": 3898 }, { "epoch": 1.0122014537902388, "grad_norm": 2.9722163677215576, "learning_rate": 2.8162111098288393e-06, "loss": 0.01808508113026619, "step": 3899 }, { "epoch": 1.0124610591900312, "grad_norm": 3.6369409561157227, "learning_rate": 2.8152655041301413e-06, "loss": 0.09853056073188782, "step": 3900 }, { "epoch": 1.0127206645898235, "grad_norm": 3.220689058303833, "learning_rate": 2.8143198868939957e-06, "loss": 0.041452035307884216, "step": 3901 }, { "epoch": 1.0129802699896158, "grad_norm": 3.4944827556610107, "learning_rate": 2.8133742582875677e-06, "loss": 0.04444236308336258, "step": 3902 }, { "epoch": 1.013239875389408, "grad_norm": 2.7527847290039062, "learning_rate": 2.8124286184780224e-06, "loss": 0.03108736127614975, "step": 3903 }, { "epoch": 1.0134994807892004, "grad_norm": 1.9040634632110596, "learning_rate": 2.8114829676325263e-06, "loss": 0.07954153418540955, "step": 3904 }, { "epoch": 1.0137590861889927, "grad_norm": 2.6860427856445312, "learning_rate": 2.810537305918249e-06, "loss": 0.058224018663167953, "step": 3905 }, { "epoch": 1.014018691588785, "grad_norm": 2.859191656112671, "learning_rate": 2.809591633502362e-06, "loss": 0.018792152404785156, "step": 3906 }, { "epoch": 1.0142782969885773, "grad_norm": 2.090925455093384, "learning_rate": 2.808645950552039e-06, "loss": 0.0675475150346756, "step": 3907 }, { "epoch": 1.0145379023883696, "grad_norm": 5.373064994812012, "learning_rate": 2.807700257234455e-06, "loss": 0.05576936900615692, "step": 3908 }, { "epoch": 1.014797507788162, "grad_norm": 3.5207135677337646, "learning_rate": 2.8067545537167874e-06, "loss": 0.03597695380449295, "step": 3909 }, { "epoch": 1.0150571131879542, "grad_norm": 3.8331351280212402, "learning_rate": 2.805808840166212e-06, "loss": 0.05264636129140854, "step": 3910 }, { "epoch": 1.0153167185877465, "grad_norm": 3.3011608123779297, "learning_rate": 2.804863116749913e-06, "loss": 0.09375524520874023, "step": 3911 }, { "epoch": 1.0155763239875388, "grad_norm": 2.4309656620025635, "learning_rate": 2.8039173836350714e-06, "loss": 0.03203817084431648, "step": 3912 }, { "epoch": 1.0158359293873314, "grad_norm": 2.068115472793579, "learning_rate": 2.80297164098887e-06, "loss": 0.07463070750236511, "step": 3913 }, { "epoch": 1.0160955347871237, "grad_norm": 3.429947853088379, "learning_rate": 2.802025888978496e-06, "loss": 0.03171161562204361, "step": 3914 }, { "epoch": 1.016355140186916, "grad_norm": 3.254769802093506, "learning_rate": 2.801080127771136e-06, "loss": 0.035803649574518204, "step": 3915 }, { "epoch": 1.0166147455867083, "grad_norm": 3.6888442039489746, "learning_rate": 2.800134357533979e-06, "loss": 0.02109329029917717, "step": 3916 }, { "epoch": 1.0168743509865006, "grad_norm": 6.187073230743408, "learning_rate": 2.7991885784342154e-06, "loss": 0.030412061139941216, "step": 3917 }, { "epoch": 1.017133956386293, "grad_norm": 2.45902943611145, "learning_rate": 2.7982427906390374e-06, "loss": 0.027069196105003357, "step": 3918 }, { "epoch": 1.0173935617860852, "grad_norm": 3.04081130027771, "learning_rate": 2.797296994315638e-06, "loss": 0.07950553297996521, "step": 3919 }, { "epoch": 1.0176531671858775, "grad_norm": 3.1618194580078125, "learning_rate": 2.796351189631214e-06, "loss": 0.032857030630111694, "step": 3920 }, { "epoch": 1.0179127725856698, "grad_norm": 3.3438541889190674, "learning_rate": 2.795405376752961e-06, "loss": 0.05120880529284477, "step": 3921 }, { "epoch": 1.0181723779854621, "grad_norm": 2.091973304748535, "learning_rate": 2.7944595558480765e-06, "loss": 0.07290150225162506, "step": 3922 }, { "epoch": 1.0184319833852544, "grad_norm": 3.3158366680145264, "learning_rate": 2.7935137270837605e-06, "loss": 0.04861030355095863, "step": 3923 }, { "epoch": 1.0186915887850467, "grad_norm": 2.167131185531616, "learning_rate": 2.7925678906272143e-06, "loss": 0.07025744020938873, "step": 3924 }, { "epoch": 1.018951194184839, "grad_norm": 2.7703561782836914, "learning_rate": 2.7916220466456384e-06, "loss": 0.07401028275489807, "step": 3925 }, { "epoch": 1.0192107995846313, "grad_norm": 3.2555994987487793, "learning_rate": 2.7906761953062396e-06, "loss": 0.06687062233686447, "step": 3926 }, { "epoch": 1.0194704049844237, "grad_norm": 2.8754982948303223, "learning_rate": 2.78973033677622e-06, "loss": 0.038459744304418564, "step": 3927 }, { "epoch": 1.019730010384216, "grad_norm": 4.7718658447265625, "learning_rate": 2.788784471222786e-06, "loss": 0.03358754143118858, "step": 3928 }, { "epoch": 1.0199896157840083, "grad_norm": 1.8454240560531616, "learning_rate": 2.7878385988131455e-06, "loss": 0.08436235785484314, "step": 3929 }, { "epoch": 1.0202492211838006, "grad_norm": 2.170412063598633, "learning_rate": 2.7868927197145086e-06, "loss": 0.08049328625202179, "step": 3930 }, { "epoch": 1.0205088265835929, "grad_norm": 4.553791522979736, "learning_rate": 2.785946834094082e-06, "loss": 0.014669383876025677, "step": 3931 }, { "epoch": 1.0207684319833852, "grad_norm": 1.9112900495529175, "learning_rate": 2.7850009421190787e-06, "loss": 0.07182155549526215, "step": 3932 }, { "epoch": 1.0210280373831775, "grad_norm": 3.2286882400512695, "learning_rate": 2.7840550439567104e-06, "loss": 0.016280293464660645, "step": 3933 }, { "epoch": 1.0212876427829698, "grad_norm": 2.681680202484131, "learning_rate": 2.7831091397741905e-06, "loss": 0.04468443989753723, "step": 3934 }, { "epoch": 1.021547248182762, "grad_norm": 2.954122543334961, "learning_rate": 2.782163229738732e-06, "loss": 0.0324317067861557, "step": 3935 }, { "epoch": 1.0218068535825544, "grad_norm": 1.5810056924819946, "learning_rate": 2.7812173140175506e-06, "loss": 0.10074564069509506, "step": 3936 }, { "epoch": 1.022066458982347, "grad_norm": 2.8501038551330566, "learning_rate": 2.780271392777864e-06, "loss": 0.011042816564440727, "step": 3937 }, { "epoch": 1.0223260643821392, "grad_norm": 2.9886176586151123, "learning_rate": 2.7793254661868877e-06, "loss": 0.05792722851037979, "step": 3938 }, { "epoch": 1.0225856697819315, "grad_norm": 3.4789915084838867, "learning_rate": 2.778379534411841e-06, "loss": 0.05790736526250839, "step": 3939 }, { "epoch": 1.0228452751817239, "grad_norm": 2.3654587268829346, "learning_rate": 2.7774335976199422e-06, "loss": 0.07938128709793091, "step": 3940 }, { "epoch": 1.0231048805815162, "grad_norm": 3.33566951751709, "learning_rate": 2.7764876559784105e-06, "loss": 0.025173399597406387, "step": 3941 }, { "epoch": 1.0233644859813085, "grad_norm": 3.9931790828704834, "learning_rate": 2.775541709654469e-06, "loss": 0.06810196489095688, "step": 3942 }, { "epoch": 1.0236240913811008, "grad_norm": 5.378033638000488, "learning_rate": 2.7745957588153376e-06, "loss": 0.0366281196475029, "step": 3943 }, { "epoch": 1.023883696780893, "grad_norm": 2.4398446083068848, "learning_rate": 2.7736498036282388e-06, "loss": 0.07306565344333649, "step": 3944 }, { "epoch": 1.0241433021806854, "grad_norm": 3.355494499206543, "learning_rate": 2.7727038442603967e-06, "loss": 0.04160017520189285, "step": 3945 }, { "epoch": 1.0244029075804777, "grad_norm": 1.6872546672821045, "learning_rate": 2.7717578808790342e-06, "loss": 0.09437252581119537, "step": 3946 }, { "epoch": 1.02466251298027, "grad_norm": 2.774301052093506, "learning_rate": 2.7708119136513778e-06, "loss": 0.09313520044088364, "step": 3947 }, { "epoch": 1.0249221183800623, "grad_norm": 2.8013803958892822, "learning_rate": 2.7698659427446507e-06, "loss": 0.028467539697885513, "step": 3948 }, { "epoch": 1.0251817237798546, "grad_norm": 3.2589571475982666, "learning_rate": 2.768919968326079e-06, "loss": 0.08882827311754227, "step": 3949 }, { "epoch": 1.025441329179647, "grad_norm": 3.251757860183716, "learning_rate": 2.767973990562891e-06, "loss": 0.0628509372472763, "step": 3950 }, { "epoch": 1.0257009345794392, "grad_norm": 4.933777332305908, "learning_rate": 2.7670280096223136e-06, "loss": 0.03679857775568962, "step": 3951 }, { "epoch": 1.0259605399792315, "grad_norm": 3.571631908416748, "learning_rate": 2.766082025671573e-06, "loss": 0.033624909818172455, "step": 3952 }, { "epoch": 1.0262201453790238, "grad_norm": 3.658522129058838, "learning_rate": 2.7651360388778993e-06, "loss": 0.04080890864133835, "step": 3953 }, { "epoch": 1.0264797507788161, "grad_norm": 3.3197412490844727, "learning_rate": 2.7641900494085193e-06, "loss": 0.10075497627258301, "step": 3954 }, { "epoch": 1.0267393561786085, "grad_norm": 3.119878053665161, "learning_rate": 2.763244057430665e-06, "loss": 0.014888737350702286, "step": 3955 }, { "epoch": 1.0269989615784008, "grad_norm": 3.23665714263916, "learning_rate": 2.762298063111563e-06, "loss": 0.021692287176847458, "step": 3956 }, { "epoch": 1.027258566978193, "grad_norm": 3.9624247550964355, "learning_rate": 2.7613520666184466e-06, "loss": 0.03594984486699104, "step": 3957 }, { "epoch": 1.0275181723779854, "grad_norm": 3.063751220703125, "learning_rate": 2.7604060681185442e-06, "loss": 0.03642912209033966, "step": 3958 }, { "epoch": 1.0277777777777777, "grad_norm": 3.3675293922424316, "learning_rate": 2.7594600677790866e-06, "loss": 0.04448112100362778, "step": 3959 }, { "epoch": 1.02803738317757, "grad_norm": 3.4840402603149414, "learning_rate": 2.758514065767306e-06, "loss": 0.010673515498638153, "step": 3960 }, { "epoch": 1.0282969885773625, "grad_norm": 2.1266772747039795, "learning_rate": 2.7575680622504342e-06, "loss": 0.08428482711315155, "step": 3961 }, { "epoch": 1.0285565939771548, "grad_norm": 2.985944986343384, "learning_rate": 2.7566220573957015e-06, "loss": 0.053605347871780396, "step": 3962 }, { "epoch": 1.0288161993769471, "grad_norm": 3.7176434993743896, "learning_rate": 2.755676051370341e-06, "loss": 0.029964881017804146, "step": 3963 }, { "epoch": 1.0290758047767394, "grad_norm": 2.2355029582977295, "learning_rate": 2.7547300443415836e-06, "loss": 0.07785502821207047, "step": 3964 }, { "epoch": 1.0293354101765317, "grad_norm": 1.760502815246582, "learning_rate": 2.753784036476664e-06, "loss": 0.09823530167341232, "step": 3965 }, { "epoch": 1.029595015576324, "grad_norm": 2.4884588718414307, "learning_rate": 2.752838027942813e-06, "loss": 0.025610901415348053, "step": 3966 }, { "epoch": 1.0298546209761164, "grad_norm": 1.6001075506210327, "learning_rate": 2.751892018907263e-06, "loss": 0.09452100098133087, "step": 3967 }, { "epoch": 1.0301142263759087, "grad_norm": 2.133378505706787, "learning_rate": 2.7509460095372477e-06, "loss": 0.09108789265155792, "step": 3968 }, { "epoch": 1.030373831775701, "grad_norm": 1.3462049961090088, "learning_rate": 2.7500000000000004e-06, "loss": 0.09271818399429321, "step": 3969 }, { "epoch": 1.0306334371754933, "grad_norm": 1.9328131675720215, "learning_rate": 2.749053990462752e-06, "loss": 0.06768110394477844, "step": 3970 }, { "epoch": 1.0308930425752856, "grad_norm": 1.5114989280700684, "learning_rate": 2.7481079810927374e-06, "loss": 0.09596692025661469, "step": 3971 }, { "epoch": 1.0311526479750779, "grad_norm": 2.273009777069092, "learning_rate": 2.747161972057188e-06, "loss": 0.0890234038233757, "step": 3972 }, { "epoch": 1.0314122533748702, "grad_norm": 4.276823997497559, "learning_rate": 2.7462159635233372e-06, "loss": 0.03630002588033676, "step": 3973 }, { "epoch": 1.0316718587746625, "grad_norm": 3.5725677013397217, "learning_rate": 2.7452699556584167e-06, "loss": 0.03285381942987442, "step": 3974 }, { "epoch": 1.0319314641744548, "grad_norm": 2.538224220275879, "learning_rate": 2.7443239486296603e-06, "loss": 0.04864082112908363, "step": 3975 }, { "epoch": 1.032191069574247, "grad_norm": 3.5465610027313232, "learning_rate": 2.7433779426042997e-06, "loss": 0.06239720433950424, "step": 3976 }, { "epoch": 1.0324506749740394, "grad_norm": 1.759753704071045, "learning_rate": 2.742431937749567e-06, "loss": 0.07999344170093536, "step": 3977 }, { "epoch": 1.0327102803738317, "grad_norm": 3.2817747592926025, "learning_rate": 2.7414859342326943e-06, "loss": 0.04516077786684036, "step": 3978 }, { "epoch": 1.032969885773624, "grad_norm": 1.7762044668197632, "learning_rate": 2.740539932220915e-06, "loss": 0.09940318763256073, "step": 3979 }, { "epoch": 1.0332294911734163, "grad_norm": 2.902984142303467, "learning_rate": 2.739593931881457e-06, "loss": 0.025287741795182228, "step": 3980 }, { "epoch": 1.0334890965732086, "grad_norm": 2.6723411083221436, "learning_rate": 2.738647933381554e-06, "loss": 0.09049010276794434, "step": 3981 }, { "epoch": 1.033748701973001, "grad_norm": 2.9680566787719727, "learning_rate": 2.7377019368884367e-06, "loss": 0.03429662808775902, "step": 3982 }, { "epoch": 1.0340083073727933, "grad_norm": 1.8910688161849976, "learning_rate": 2.7367559425693364e-06, "loss": 0.0901520624756813, "step": 3983 }, { "epoch": 1.0342679127725856, "grad_norm": 2.0224897861480713, "learning_rate": 2.7358099505914814e-06, "loss": 0.08609843254089355, "step": 3984 }, { "epoch": 1.034527518172378, "grad_norm": 3.551502227783203, "learning_rate": 2.7348639611221014e-06, "loss": 0.025669440627098083, "step": 3985 }, { "epoch": 1.0347871235721704, "grad_norm": 4.635325908660889, "learning_rate": 2.7339179743284273e-06, "loss": 0.03975334018468857, "step": 3986 }, { "epoch": 1.0350467289719627, "grad_norm": 4.215904235839844, "learning_rate": 2.732971990377688e-06, "loss": 0.043205827474594116, "step": 3987 }, { "epoch": 1.035306334371755, "grad_norm": 2.4922287464141846, "learning_rate": 2.7320260094371097e-06, "loss": 0.015884116291999817, "step": 3988 }, { "epoch": 1.0355659397715473, "grad_norm": 2.4090723991394043, "learning_rate": 2.7310800316739215e-06, "loss": 0.03785035014152527, "step": 3989 }, { "epoch": 1.0358255451713396, "grad_norm": 3.309817314147949, "learning_rate": 2.7301340572553505e-06, "loss": 0.04056236892938614, "step": 3990 }, { "epoch": 1.036085150571132, "grad_norm": 3.783815622329712, "learning_rate": 2.729188086348624e-06, "loss": 0.05238870158791542, "step": 3991 }, { "epoch": 1.0363447559709242, "grad_norm": 3.245744228363037, "learning_rate": 2.728242119120967e-06, "loss": 0.052873771637678146, "step": 3992 }, { "epoch": 1.0366043613707165, "grad_norm": 1.9728519916534424, "learning_rate": 2.7272961557396044e-06, "loss": 0.08447548002004623, "step": 3993 }, { "epoch": 1.0368639667705088, "grad_norm": 4.854185581207275, "learning_rate": 2.726350196371762e-06, "loss": 0.025858009234070778, "step": 3994 }, { "epoch": 1.0371235721703012, "grad_norm": 2.260701894760132, "learning_rate": 2.7254042411846636e-06, "loss": 0.02567683905363083, "step": 3995 }, { "epoch": 1.0373831775700935, "grad_norm": 2.479146718978882, "learning_rate": 2.724458290345532e-06, "loss": 0.05712328851222992, "step": 3996 }, { "epoch": 1.0376427829698858, "grad_norm": 3.1200106143951416, "learning_rate": 2.7235123440215898e-06, "loss": 0.0336885005235672, "step": 3997 }, { "epoch": 1.037902388369678, "grad_norm": 2.511526584625244, "learning_rate": 2.7225664023800597e-06, "loss": 0.08186586201190948, "step": 3998 }, { "epoch": 1.0381619937694704, "grad_norm": 3.1589367389678955, "learning_rate": 2.72162046558816e-06, "loss": 0.032001931220293045, "step": 3999 }, { "epoch": 1.0384215991692627, "grad_norm": 1.5951275825500488, "learning_rate": 2.720674533813113e-06, "loss": 0.08981513977050781, "step": 4000 }, { "epoch": 1.038681204569055, "grad_norm": 3.1293818950653076, "learning_rate": 2.719728607222136e-06, "loss": 0.02420969307422638, "step": 4001 }, { "epoch": 1.0389408099688473, "grad_norm": 1.831788420677185, "learning_rate": 2.7187826859824497e-06, "loss": 0.075126051902771, "step": 4002 }, { "epoch": 1.0392004153686396, "grad_norm": 3.8777353763580322, "learning_rate": 2.717836770261269e-06, "loss": 0.039854470640420914, "step": 4003 }, { "epoch": 1.039460020768432, "grad_norm": 3.5263779163360596, "learning_rate": 2.7168908602258106e-06, "loss": 0.03651867061853409, "step": 4004 }, { "epoch": 1.0397196261682242, "grad_norm": 3.7151596546173096, "learning_rate": 2.7159449560432903e-06, "loss": 0.04969275742769241, "step": 4005 }, { "epoch": 1.0399792315680165, "grad_norm": 3.8368940353393555, "learning_rate": 2.714999057880922e-06, "loss": 0.023361487314105034, "step": 4006 }, { "epoch": 1.0402388369678088, "grad_norm": 2.531277894973755, "learning_rate": 2.714053165905919e-06, "loss": 0.06963320076465607, "step": 4007 }, { "epoch": 1.0404984423676011, "grad_norm": 1.7394086122512817, "learning_rate": 2.7131072802854925e-06, "loss": 0.09414003789424896, "step": 4008 }, { "epoch": 1.0407580477673937, "grad_norm": 2.3654589653015137, "learning_rate": 2.7121614011868547e-06, "loss": 0.03781434893608093, "step": 4009 }, { "epoch": 1.041017653167186, "grad_norm": 2.91645884513855, "learning_rate": 2.711215528777216e-06, "loss": 0.041578441858291626, "step": 4010 }, { "epoch": 1.0412772585669783, "grad_norm": 2.3294413089752197, "learning_rate": 2.7102696632237813e-06, "loss": 0.075728639960289, "step": 4011 }, { "epoch": 1.0415368639667706, "grad_norm": 3.8036506175994873, "learning_rate": 2.7093238046937615e-06, "loss": 0.024693390354514122, "step": 4012 }, { "epoch": 1.041796469366563, "grad_norm": 5.280838966369629, "learning_rate": 2.708377953354362e-06, "loss": 0.037866514176130295, "step": 4013 }, { "epoch": 1.0420560747663552, "grad_norm": 2.8498480319976807, "learning_rate": 2.7074321093727877e-06, "loss": 0.017936361953616142, "step": 4014 }, { "epoch": 1.0423156801661475, "grad_norm": 3.177910804748535, "learning_rate": 2.706486272916241e-06, "loss": 0.09373216331005096, "step": 4015 }, { "epoch": 1.0425752855659398, "grad_norm": 2.3077616691589355, "learning_rate": 2.7055404441519246e-06, "loss": 0.0963074117898941, "step": 4016 }, { "epoch": 1.0428348909657321, "grad_norm": 3.801748037338257, "learning_rate": 2.70459462324704e-06, "loss": 0.05792248994112015, "step": 4017 }, { "epoch": 1.0430944963655244, "grad_norm": 2.4661707878112793, "learning_rate": 2.7036488103687868e-06, "loss": 0.07548107206821442, "step": 4018 }, { "epoch": 1.0433541017653167, "grad_norm": 2.046013116836548, "learning_rate": 2.702703005684363e-06, "loss": 0.07556694746017456, "step": 4019 }, { "epoch": 1.043613707165109, "grad_norm": 2.232010841369629, "learning_rate": 2.7017572093609633e-06, "loss": 0.08612027764320374, "step": 4020 }, { "epoch": 1.0438733125649013, "grad_norm": 1.491335391998291, "learning_rate": 2.700811421565785e-06, "loss": 0.09667648375034332, "step": 4021 }, { "epoch": 1.0441329179646937, "grad_norm": 2.1560680866241455, "learning_rate": 2.6998656424660226e-06, "loss": 0.07990775257349014, "step": 4022 }, { "epoch": 1.044392523364486, "grad_norm": 2.930351972579956, "learning_rate": 2.698919872228865e-06, "loss": 0.060089536011219025, "step": 4023 }, { "epoch": 1.0446521287642783, "grad_norm": 3.1282482147216797, "learning_rate": 2.6979741110215045e-06, "loss": 0.045443616807460785, "step": 4024 }, { "epoch": 1.0449117341640706, "grad_norm": 5.288837432861328, "learning_rate": 2.6970283590111307e-06, "loss": 0.03479637950658798, "step": 4025 }, { "epoch": 1.0451713395638629, "grad_norm": 1.737447738647461, "learning_rate": 2.6960826163649297e-06, "loss": 0.08779361844062805, "step": 4026 }, { "epoch": 1.0454309449636552, "grad_norm": 3.1465773582458496, "learning_rate": 2.695136883250088e-06, "loss": 0.02610514685511589, "step": 4027 }, { "epoch": 1.0456905503634475, "grad_norm": 3.4966461658477783, "learning_rate": 2.6941911598337885e-06, "loss": 0.02776087448000908, "step": 4028 }, { "epoch": 1.0459501557632398, "grad_norm": 1.6501665115356445, "learning_rate": 2.6932454462832146e-06, "loss": 0.09335151314735413, "step": 4029 }, { "epoch": 1.046209761163032, "grad_norm": 3.257983684539795, "learning_rate": 2.6922997427655456e-06, "loss": 0.028428563848137856, "step": 4030 }, { "epoch": 1.0464693665628244, "grad_norm": 2.7948880195617676, "learning_rate": 2.6913540494479613e-06, "loss": 0.06739668548107147, "step": 4031 }, { "epoch": 1.0467289719626167, "grad_norm": 3.609778642654419, "learning_rate": 2.6904083664976384e-06, "loss": 0.04969782382249832, "step": 4032 }, { "epoch": 1.0469885773624092, "grad_norm": 3.3297743797302246, "learning_rate": 2.689462694081752e-06, "loss": 0.060633860528469086, "step": 4033 }, { "epoch": 1.0472481827622016, "grad_norm": 2.744783401489258, "learning_rate": 2.6885170323674748e-06, "loss": 0.0603218749165535, "step": 4034 }, { "epoch": 1.0475077881619939, "grad_norm": 2.893209934234619, "learning_rate": 2.6875713815219783e-06, "loss": 0.026581794023513794, "step": 4035 }, { "epoch": 1.0477673935617862, "grad_norm": 2.1846837997436523, "learning_rate": 2.6866257417124326e-06, "loss": 0.0812418982386589, "step": 4036 }, { "epoch": 1.0480269989615785, "grad_norm": 2.313544750213623, "learning_rate": 2.6856801131060046e-06, "loss": 0.06877495348453522, "step": 4037 }, { "epoch": 1.0482866043613708, "grad_norm": 3.0866281986236572, "learning_rate": 2.6847344958698594e-06, "loss": 0.040242910385131836, "step": 4038 }, { "epoch": 1.048546209761163, "grad_norm": 3.0212464332580566, "learning_rate": 2.683788890171162e-06, "loss": 0.027264898642897606, "step": 4039 }, { "epoch": 1.0488058151609554, "grad_norm": 2.715059518814087, "learning_rate": 2.6828432961770717e-06, "loss": 0.06571222841739655, "step": 4040 }, { "epoch": 1.0490654205607477, "grad_norm": 3.342026710510254, "learning_rate": 2.68189771405475e-06, "loss": 0.04009197652339935, "step": 4041 }, { "epoch": 1.04932502596054, "grad_norm": 1.9805035591125488, "learning_rate": 2.680952143971352e-06, "loss": 0.07410116493701935, "step": 4042 }, { "epoch": 1.0495846313603323, "grad_norm": 2.204152822494507, "learning_rate": 2.6800065860940343e-06, "loss": 0.0819171890616417, "step": 4043 }, { "epoch": 1.0498442367601246, "grad_norm": 1.905158519744873, "learning_rate": 2.6790610405899496e-06, "loss": 0.008619414642453194, "step": 4044 }, { "epoch": 1.050103842159917, "grad_norm": 3.4357659816741943, "learning_rate": 2.6781155076262494e-06, "loss": 0.05004049092531204, "step": 4045 }, { "epoch": 1.0503634475597092, "grad_norm": 3.2255632877349854, "learning_rate": 2.6771699873700806e-06, "loss": 0.0508282445371151, "step": 4046 }, { "epoch": 1.0506230529595015, "grad_norm": 2.6866745948791504, "learning_rate": 2.6762244799885906e-06, "loss": 0.06435270607471466, "step": 4047 }, { "epoch": 1.0508826583592938, "grad_norm": 3.435046911239624, "learning_rate": 2.675278985648923e-06, "loss": 0.054948415607213974, "step": 4048 }, { "epoch": 1.0511422637590861, "grad_norm": 1.8560371398925781, "learning_rate": 2.674333504518221e-06, "loss": 0.08786888420581818, "step": 4049 }, { "epoch": 1.0514018691588785, "grad_norm": 2.9239070415496826, "learning_rate": 2.673388036763622e-06, "loss": 0.0777704268693924, "step": 4050 }, { "epoch": 1.0516614745586708, "grad_norm": 3.4385251998901367, "learning_rate": 2.672442582552264e-06, "loss": 0.020594488829374313, "step": 4051 }, { "epoch": 1.051921079958463, "grad_norm": 3.624168634414673, "learning_rate": 2.671497142051282e-06, "loss": 0.028984786942601204, "step": 4052 }, { "epoch": 1.0521806853582554, "grad_norm": 2.5251657962799072, "learning_rate": 2.6705517154278073e-06, "loss": 0.06386086344718933, "step": 4053 }, { "epoch": 1.0524402907580477, "grad_norm": 5.647150039672852, "learning_rate": 2.66960630284897e-06, "loss": 0.034949421882629395, "step": 4054 }, { "epoch": 1.05269989615784, "grad_norm": 2.937410831451416, "learning_rate": 2.668660904481898e-06, "loss": 0.04759399965405464, "step": 4055 }, { "epoch": 1.0529595015576323, "grad_norm": 2.118805170059204, "learning_rate": 2.667715520493716e-06, "loss": 0.06828363984823227, "step": 4056 }, { "epoch": 1.0532191069574246, "grad_norm": 1.8489184379577637, "learning_rate": 2.666770151051545e-06, "loss": 0.10568312555551529, "step": 4057 }, { "epoch": 1.0534787123572171, "grad_norm": 3.7133076190948486, "learning_rate": 2.665824796322506e-06, "loss": 0.02498379349708557, "step": 4058 }, { "epoch": 1.0537383177570094, "grad_norm": 3.245755910873413, "learning_rate": 2.6648794564737156e-06, "loss": 0.026107728481292725, "step": 4059 }, { "epoch": 1.0539979231568017, "grad_norm": 3.8106768131256104, "learning_rate": 2.663934131672289e-06, "loss": 0.05288104712963104, "step": 4060 }, { "epoch": 1.054257528556594, "grad_norm": 2.6178085803985596, "learning_rate": 2.6629888220853365e-06, "loss": 0.04815945774316788, "step": 4061 }, { "epoch": 1.0545171339563864, "grad_norm": 2.7192389965057373, "learning_rate": 2.662043527879968e-06, "loss": 0.04161861911416054, "step": 4062 }, { "epoch": 1.0547767393561787, "grad_norm": 3.2946033477783203, "learning_rate": 2.66109824922329e-06, "loss": 0.0729231908917427, "step": 4063 }, { "epoch": 1.055036344755971, "grad_norm": 2.8047420978546143, "learning_rate": 2.660152986282406e-06, "loss": 0.0837753564119339, "step": 4064 }, { "epoch": 1.0552959501557633, "grad_norm": 2.7884681224823, "learning_rate": 2.659207739224417e-06, "loss": 0.06223863363265991, "step": 4065 }, { "epoch": 1.0555555555555556, "grad_norm": 3.1087610721588135, "learning_rate": 2.65826250821642e-06, "loss": 0.07007569819688797, "step": 4066 }, { "epoch": 1.0558151609553479, "grad_norm": 3.2138776779174805, "learning_rate": 2.6573172934255116e-06, "loss": 0.01920379512012005, "step": 4067 }, { "epoch": 1.0560747663551402, "grad_norm": 2.866109848022461, "learning_rate": 2.656372095018784e-06, "loss": 0.06251922249794006, "step": 4068 }, { "epoch": 1.0563343717549325, "grad_norm": 4.215813636779785, "learning_rate": 2.6554269131633254e-06, "loss": 0.041543133556842804, "step": 4069 }, { "epoch": 1.0565939771547248, "grad_norm": 3.8499953746795654, "learning_rate": 2.6544817480262237e-06, "loss": 0.040379174053668976, "step": 4070 }, { "epoch": 1.0568535825545171, "grad_norm": 2.467910051345825, "learning_rate": 2.6535365997745616e-06, "loss": 0.07868600636720657, "step": 4071 }, { "epoch": 1.0571131879543094, "grad_norm": 2.5164976119995117, "learning_rate": 2.65259146857542e-06, "loss": 0.09622485935688019, "step": 4072 }, { "epoch": 1.0573727933541017, "grad_norm": 2.220778465270996, "learning_rate": 2.6516463545958766e-06, "loss": 0.0646764263510704, "step": 4073 }, { "epoch": 1.057632398753894, "grad_norm": 4.51074743270874, "learning_rate": 2.650701258003005e-06, "loss": 0.04865938425064087, "step": 4074 }, { "epoch": 1.0578920041536863, "grad_norm": 2.5633432865142822, "learning_rate": 2.6497561789638783e-06, "loss": 0.06997641175985336, "step": 4075 }, { "epoch": 1.0581516095534786, "grad_norm": 2.9522933959960938, "learning_rate": 2.648811117645564e-06, "loss": 0.10029196739196777, "step": 4076 }, { "epoch": 1.058411214953271, "grad_norm": 3.1136767864227295, "learning_rate": 2.6478660742151257e-06, "loss": 0.037380725145339966, "step": 4077 }, { "epoch": 1.0586708203530633, "grad_norm": 3.234226942062378, "learning_rate": 2.646921048839628e-06, "loss": 0.04217613860964775, "step": 4078 }, { "epoch": 1.0589304257528556, "grad_norm": 2.4587929248809814, "learning_rate": 2.6459760416861293e-06, "loss": 0.05958116799592972, "step": 4079 }, { "epoch": 1.0591900311526479, "grad_norm": 3.2956409454345703, "learning_rate": 2.6450310529216837e-06, "loss": 0.051195576786994934, "step": 4080 }, { "epoch": 1.0594496365524404, "grad_norm": 2.867765426635742, "learning_rate": 2.644086082713345e-06, "loss": 0.060341790318489075, "step": 4081 }, { "epoch": 1.0597092419522327, "grad_norm": 2.7278475761413574, "learning_rate": 2.64314113122816e-06, "loss": 0.03474373370409012, "step": 4082 }, { "epoch": 1.059968847352025, "grad_norm": 2.1691346168518066, "learning_rate": 2.6421961986331786e-06, "loss": 0.09581257402896881, "step": 4083 }, { "epoch": 1.0602284527518173, "grad_norm": 3.450749397277832, "learning_rate": 2.6412512850954397e-06, "loss": 0.028394149616360664, "step": 4084 }, { "epoch": 1.0604880581516096, "grad_norm": 2.856306314468384, "learning_rate": 2.640306390781984e-06, "loss": 0.0317179299890995, "step": 4085 }, { "epoch": 1.060747663551402, "grad_norm": 2.876201868057251, "learning_rate": 2.6393615158598464e-06, "loss": 0.03449677303433418, "step": 4086 }, { "epoch": 1.0610072689511942, "grad_norm": 4.270005226135254, "learning_rate": 2.6384166604960596e-06, "loss": 0.02451493591070175, "step": 4087 }, { "epoch": 1.0612668743509865, "grad_norm": 2.837146520614624, "learning_rate": 2.6374718248576526e-06, "loss": 0.07466146349906921, "step": 4088 }, { "epoch": 1.0615264797507789, "grad_norm": 4.781948089599609, "learning_rate": 2.6365270091116497e-06, "loss": 0.018437491729855537, "step": 4089 }, { "epoch": 1.0617860851505712, "grad_norm": 1.9217652082443237, "learning_rate": 2.635582213425074e-06, "loss": 0.09631191194057465, "step": 4090 }, { "epoch": 1.0620456905503635, "grad_norm": 2.342876434326172, "learning_rate": 2.6346374379649432e-06, "loss": 0.07586844265460968, "step": 4091 }, { "epoch": 1.0623052959501558, "grad_norm": 1.8273851871490479, "learning_rate": 2.633692682898272e-06, "loss": 0.09244643151760101, "step": 4092 }, { "epoch": 1.062564901349948, "grad_norm": 2.7063610553741455, "learning_rate": 2.632747948392071e-06, "loss": 0.016824983060359955, "step": 4093 }, { "epoch": 1.0628245067497404, "grad_norm": 4.026025772094727, "learning_rate": 2.6318032346133476e-06, "loss": 0.029773438349366188, "step": 4094 }, { "epoch": 1.0630841121495327, "grad_norm": 2.1349406242370605, "learning_rate": 2.6308585417291073e-06, "loss": 0.09101177006959915, "step": 4095 }, { "epoch": 1.063343717549325, "grad_norm": 4.51036262512207, "learning_rate": 2.6299138699063475e-06, "loss": 0.0409284271299839, "step": 4096 }, { "epoch": 1.0636033229491173, "grad_norm": 2.3252899646759033, "learning_rate": 2.6289692193120664e-06, "loss": 0.10318531095981598, "step": 4097 }, { "epoch": 1.0638629283489096, "grad_norm": 2.172527551651001, "learning_rate": 2.628024590113256e-06, "loss": 0.07565195858478546, "step": 4098 }, { "epoch": 1.064122533748702, "grad_norm": 4.857824802398682, "learning_rate": 2.6270799824769056e-06, "loss": 0.02060641348361969, "step": 4099 }, { "epoch": 1.0643821391484942, "grad_norm": 2.457974910736084, "learning_rate": 2.626135396569999e-06, "loss": 0.06354911625385284, "step": 4100 }, { "epoch": 1.0646417445482865, "grad_norm": 2.0882065296173096, "learning_rate": 2.6251908325595187e-06, "loss": 0.08087871223688126, "step": 4101 }, { "epoch": 1.0649013499480788, "grad_norm": 4.03876256942749, "learning_rate": 2.6242462906124415e-06, "loss": 0.01907522976398468, "step": 4102 }, { "epoch": 1.0651609553478711, "grad_norm": 3.1313469409942627, "learning_rate": 2.623301770895741e-06, "loss": 0.025722023099660873, "step": 4103 }, { "epoch": 1.0654205607476634, "grad_norm": 3.035834789276123, "learning_rate": 2.6223572735763864e-06, "loss": 0.065513014793396, "step": 4104 }, { "epoch": 1.0656801661474558, "grad_norm": 2.820040702819824, "learning_rate": 2.621412798821343e-06, "loss": 0.04032588005065918, "step": 4105 }, { "epoch": 1.0659397715472483, "grad_norm": 2.7582027912139893, "learning_rate": 2.6204683467975716e-06, "loss": 0.018445879220962524, "step": 4106 }, { "epoch": 1.0661993769470406, "grad_norm": 2.142514705657959, "learning_rate": 2.619523917672033e-06, "loss": 0.08687391877174377, "step": 4107 }, { "epoch": 1.066458982346833, "grad_norm": 3.6016786098480225, "learning_rate": 2.6185795116116756e-06, "loss": 0.055049702525138855, "step": 4108 }, { "epoch": 1.0667185877466252, "grad_norm": 3.315617084503174, "learning_rate": 2.617635128783452e-06, "loss": 0.016482092440128326, "step": 4109 }, { "epoch": 1.0669781931464175, "grad_norm": 3.7105114459991455, "learning_rate": 2.6166907693543085e-06, "loss": 0.017674580216407776, "step": 4110 }, { "epoch": 1.0672377985462098, "grad_norm": 3.3579792976379395, "learning_rate": 2.615746433491183e-06, "loss": 0.04140206426382065, "step": 4111 }, { "epoch": 1.0674974039460021, "grad_norm": 3.4958486557006836, "learning_rate": 2.6148021213610147e-06, "loss": 0.03930862993001938, "step": 4112 }, { "epoch": 1.0677570093457944, "grad_norm": 2.5213234424591064, "learning_rate": 2.613857833130735e-06, "loss": 0.10976296663284302, "step": 4113 }, { "epoch": 1.0680166147455867, "grad_norm": 2.9244191646575928, "learning_rate": 2.6129135689672745e-06, "loss": 0.0928773432970047, "step": 4114 }, { "epoch": 1.068276220145379, "grad_norm": 1.6687524318695068, "learning_rate": 2.6119693290375555e-06, "loss": 0.08406703174114227, "step": 4115 }, { "epoch": 1.0685358255451713, "grad_norm": 2.2848308086395264, "learning_rate": 2.611025113508498e-06, "loss": 0.08579618483781815, "step": 4116 }, { "epoch": 1.0687954309449637, "grad_norm": 1.9956953525543213, "learning_rate": 2.610080922547019e-06, "loss": 0.06482107937335968, "step": 4117 }, { "epoch": 1.069055036344756, "grad_norm": 3.7883658409118652, "learning_rate": 2.6091367563200294e-06, "loss": 0.06009000167250633, "step": 4118 }, { "epoch": 1.0693146417445483, "grad_norm": 2.48042368888855, "learning_rate": 2.608192614994435e-06, "loss": 0.09597110748291016, "step": 4119 }, { "epoch": 1.0695742471443406, "grad_norm": 1.8163118362426758, "learning_rate": 2.6072484987371395e-06, "loss": 0.09995703399181366, "step": 4120 }, { "epoch": 1.0698338525441329, "grad_norm": 2.615267276763916, "learning_rate": 2.6063044077150414e-06, "loss": 0.03424980118870735, "step": 4121 }, { "epoch": 1.0700934579439252, "grad_norm": 3.268772602081299, "learning_rate": 2.6053603420950337e-06, "loss": 0.029869254678487778, "step": 4122 }, { "epoch": 1.0703530633437175, "grad_norm": 3.516369342803955, "learning_rate": 2.6044163020440055e-06, "loss": 0.06271620094776154, "step": 4123 }, { "epoch": 1.0706126687435098, "grad_norm": 2.8686771392822266, "learning_rate": 2.6034722877288415e-06, "loss": 0.009134409949183464, "step": 4124 }, { "epoch": 1.070872274143302, "grad_norm": 2.5811164379119873, "learning_rate": 2.6025282993164226e-06, "loss": 0.017059538513422012, "step": 4125 }, { "epoch": 1.0711318795430944, "grad_norm": 2.1262552738189697, "learning_rate": 2.6015843369736234e-06, "loss": 0.07701572775840759, "step": 4126 }, { "epoch": 1.0713914849428867, "grad_norm": 3.070462942123413, "learning_rate": 2.6006404008673153e-06, "loss": 0.05151132121682167, "step": 4127 }, { "epoch": 1.071651090342679, "grad_norm": 3.7102999687194824, "learning_rate": 2.599696491164364e-06, "loss": 0.0449959933757782, "step": 4128 }, { "epoch": 1.0719106957424716, "grad_norm": 5.460855484008789, "learning_rate": 2.5987526080316316e-06, "loss": 0.03827857971191406, "step": 4129 }, { "epoch": 1.0721703011422639, "grad_norm": 2.9158873558044434, "learning_rate": 2.5978087516359756e-06, "loss": 0.08623823523521423, "step": 4130 }, { "epoch": 1.0724299065420562, "grad_norm": 4.167246341705322, "learning_rate": 2.5968649221442473e-06, "loss": 0.038742270320653915, "step": 4131 }, { "epoch": 1.0726895119418485, "grad_norm": 3.431067705154419, "learning_rate": 2.595921119723294e-06, "loss": 0.03421344980597496, "step": 4132 }, { "epoch": 1.0729491173416408, "grad_norm": 3.4400274753570557, "learning_rate": 2.5949773445399595e-06, "loss": 0.018437735736370087, "step": 4133 }, { "epoch": 1.073208722741433, "grad_norm": 2.3637447357177734, "learning_rate": 2.5940335967610807e-06, "loss": 0.07234475016593933, "step": 4134 }, { "epoch": 1.0734683281412254, "grad_norm": 3.839799165725708, "learning_rate": 2.5930898765534906e-06, "loss": 0.04908950999379158, "step": 4135 }, { "epoch": 1.0737279335410177, "grad_norm": 3.3646607398986816, "learning_rate": 2.592146184084018e-06, "loss": 0.06086526811122894, "step": 4136 }, { "epoch": 1.07398753894081, "grad_norm": 3.405148983001709, "learning_rate": 2.5912025195194855e-06, "loss": 0.06758642196655273, "step": 4137 }, { "epoch": 1.0742471443406023, "grad_norm": 2.1219277381896973, "learning_rate": 2.5902588830267116e-06, "loss": 0.08207084238529205, "step": 4138 }, { "epoch": 1.0745067497403946, "grad_norm": 3.5634684562683105, "learning_rate": 2.5893152747725094e-06, "loss": 0.060161374509334564, "step": 4139 }, { "epoch": 1.074766355140187, "grad_norm": 4.407021522521973, "learning_rate": 2.588371694923687e-06, "loss": 0.02233179099857807, "step": 4140 }, { "epoch": 1.0750259605399792, "grad_norm": 2.9922969341278076, "learning_rate": 2.5874281436470494e-06, "loss": 0.020647861063480377, "step": 4141 }, { "epoch": 1.0752855659397715, "grad_norm": 2.1621174812316895, "learning_rate": 2.586484621109393e-06, "loss": 0.0695314109325409, "step": 4142 }, { "epoch": 1.0755451713395638, "grad_norm": 3.2135629653930664, "learning_rate": 2.585541127477511e-06, "loss": 0.05633968114852905, "step": 4143 }, { "epoch": 1.0758047767393562, "grad_norm": 2.28671932220459, "learning_rate": 2.5845976629181917e-06, "loss": 0.08342716097831726, "step": 4144 }, { "epoch": 1.0760643821391485, "grad_norm": 2.7608373165130615, "learning_rate": 2.58365422759822e-06, "loss": 0.01832013577222824, "step": 4145 }, { "epoch": 1.0763239875389408, "grad_norm": 2.348543167114258, "learning_rate": 2.582710821684371e-06, "loss": 0.07322230935096741, "step": 4146 }, { "epoch": 1.076583592938733, "grad_norm": 3.5748326778411865, "learning_rate": 2.5817674453434185e-06, "loss": 0.027262020856142044, "step": 4147 }, { "epoch": 1.0768431983385254, "grad_norm": 2.786862373352051, "learning_rate": 2.5808240987421295e-06, "loss": 0.07327282428741455, "step": 4148 }, { "epoch": 1.0771028037383177, "grad_norm": 3.000938892364502, "learning_rate": 2.579880782047267e-06, "loss": 0.036297596991062164, "step": 4149 }, { "epoch": 1.07736240913811, "grad_norm": 3.7977542877197266, "learning_rate": 2.5789374954255864e-06, "loss": 0.05287381261587143, "step": 4150 }, { "epoch": 1.0776220145379023, "grad_norm": 3.239888906478882, "learning_rate": 2.5779942390438396e-06, "loss": 0.05418406054377556, "step": 4151 }, { "epoch": 1.0778816199376946, "grad_norm": 3.647951602935791, "learning_rate": 2.5770510130687726e-06, "loss": 0.02356434427201748, "step": 4152 }, { "epoch": 1.078141225337487, "grad_norm": 4.190105438232422, "learning_rate": 2.5761078176671275e-06, "loss": 0.017912548035383224, "step": 4153 }, { "epoch": 1.0784008307372794, "grad_norm": 2.6467201709747314, "learning_rate": 2.5751646530056378e-06, "loss": 0.033173661679029465, "step": 4154 }, { "epoch": 1.0786604361370717, "grad_norm": 3.2750368118286133, "learning_rate": 2.574221519251035e-06, "loss": 0.04287215322256088, "step": 4155 }, { "epoch": 1.078920041536864, "grad_norm": 4.325124740600586, "learning_rate": 2.5732784165700418e-06, "loss": 0.03234044462442398, "step": 4156 }, { "epoch": 1.0791796469366564, "grad_norm": 3.782944679260254, "learning_rate": 2.5723353451293793e-06, "loss": 0.04605870693922043, "step": 4157 }, { "epoch": 1.0794392523364487, "grad_norm": 2.4267337322235107, "learning_rate": 2.571392305095759e-06, "loss": 0.02349887788295746, "step": 4158 }, { "epoch": 1.079698857736241, "grad_norm": 2.3068227767944336, "learning_rate": 2.5704492966358897e-06, "loss": 0.08489439636468887, "step": 4159 }, { "epoch": 1.0799584631360333, "grad_norm": 2.8971614837646484, "learning_rate": 2.569506319916474e-06, "loss": 0.026069946587085724, "step": 4160 }, { "epoch": 1.0802180685358256, "grad_norm": 3.4087772369384766, "learning_rate": 2.5685633751042084e-06, "loss": 0.025185322389006615, "step": 4161 }, { "epoch": 1.0804776739356179, "grad_norm": 2.5936167240142822, "learning_rate": 2.5676204623657824e-06, "loss": 0.08782884478569031, "step": 4162 }, { "epoch": 1.0807372793354102, "grad_norm": 2.9708101749420166, "learning_rate": 2.566677581867884e-06, "loss": 0.07115718722343445, "step": 4163 }, { "epoch": 1.0809968847352025, "grad_norm": 4.158768177032471, "learning_rate": 2.565734733777192e-06, "loss": 0.05404911935329437, "step": 4164 }, { "epoch": 1.0812564901349948, "grad_norm": 3.2556724548339844, "learning_rate": 2.564791918260378e-06, "loss": 0.05014851316809654, "step": 4165 }, { "epoch": 1.0815160955347871, "grad_norm": 2.715909719467163, "learning_rate": 2.563849135484113e-06, "loss": 0.05367853492498398, "step": 4166 }, { "epoch": 1.0817757009345794, "grad_norm": 2.76627254486084, "learning_rate": 2.562906385615059e-06, "loss": 0.05437590926885605, "step": 4167 }, { "epoch": 1.0820353063343717, "grad_norm": 3.8611130714416504, "learning_rate": 2.561963668819873e-06, "loss": 0.04443785175681114, "step": 4168 }, { "epoch": 1.082294911734164, "grad_norm": 3.6980907917022705, "learning_rate": 2.5610209852652034e-06, "loss": 0.03754841536283493, "step": 4169 }, { "epoch": 1.0825545171339563, "grad_norm": 2.1577911376953125, "learning_rate": 2.5600783351176966e-06, "loss": 0.07636244595050812, "step": 4170 }, { "epoch": 1.0828141225337486, "grad_norm": 2.474963903427124, "learning_rate": 2.559135718543992e-06, "loss": 0.04609137773513794, "step": 4171 }, { "epoch": 1.083073727933541, "grad_norm": 3.1735551357269287, "learning_rate": 2.5581931357107226e-06, "loss": 0.03191487863659859, "step": 4172 }, { "epoch": 1.0833333333333333, "grad_norm": 1.7498791217803955, "learning_rate": 2.5572505867845145e-06, "loss": 0.08587925136089325, "step": 4173 }, { "epoch": 1.0835929387331256, "grad_norm": 2.0740833282470703, "learning_rate": 2.5563080719319892e-06, "loss": 0.08208750188350677, "step": 4174 }, { "epoch": 1.0838525441329179, "grad_norm": 2.435178756713867, "learning_rate": 2.5553655913197612e-06, "loss": 0.05824267864227295, "step": 4175 }, { "epoch": 1.0841121495327102, "grad_norm": 2.1634228229522705, "learning_rate": 2.554423145114442e-06, "loss": 0.07247179746627808, "step": 4176 }, { "epoch": 1.0843717549325027, "grad_norm": 2.8829586505889893, "learning_rate": 2.553480733482631e-06, "loss": 0.06768226623535156, "step": 4177 }, { "epoch": 1.084631360332295, "grad_norm": 3.9848790168762207, "learning_rate": 2.5525383565909274e-06, "loss": 0.028331506997346878, "step": 4178 }, { "epoch": 1.0848909657320873, "grad_norm": 4.528716087341309, "learning_rate": 2.5515960146059204e-06, "loss": 0.028841756284236908, "step": 4179 }, { "epoch": 1.0851505711318796, "grad_norm": 3.514479160308838, "learning_rate": 2.5506537076941955e-06, "loss": 0.03261406719684601, "step": 4180 }, { "epoch": 1.085410176531672, "grad_norm": 3.879781484603882, "learning_rate": 2.5497114360223306e-06, "loss": 0.02678348682820797, "step": 4181 }, { "epoch": 1.0856697819314642, "grad_norm": 1.946588397026062, "learning_rate": 2.548769199756898e-06, "loss": 0.0900593250989914, "step": 4182 }, { "epoch": 1.0859293873312565, "grad_norm": 2.597478151321411, "learning_rate": 2.547826999064463e-06, "loss": 0.08246200531721115, "step": 4183 }, { "epoch": 1.0861889927310489, "grad_norm": 3.7206742763519287, "learning_rate": 2.546884834111586e-06, "loss": 0.05950665846467018, "step": 4184 }, { "epoch": 1.0864485981308412, "grad_norm": 3.2211546897888184, "learning_rate": 2.545942705064819e-06, "loss": 0.0551319345831871, "step": 4185 }, { "epoch": 1.0867082035306335, "grad_norm": 2.5914459228515625, "learning_rate": 2.5450006120907095e-06, "loss": 0.06404261291027069, "step": 4186 }, { "epoch": 1.0869678089304258, "grad_norm": 2.979377508163452, "learning_rate": 2.544058555355798e-06, "loss": 0.03083353489637375, "step": 4187 }, { "epoch": 1.087227414330218, "grad_norm": 2.3785290718078613, "learning_rate": 2.543116535026619e-06, "loss": 0.01628034934401512, "step": 4188 }, { "epoch": 1.0874870197300104, "grad_norm": 2.0782885551452637, "learning_rate": 2.5421745512696988e-06, "loss": 0.08936161547899246, "step": 4189 }, { "epoch": 1.0877466251298027, "grad_norm": 2.711773633956909, "learning_rate": 2.54123260425156e-06, "loss": 0.08535651117563248, "step": 4190 }, { "epoch": 1.088006230529595, "grad_norm": 2.6770942211151123, "learning_rate": 2.540290694138717e-06, "loss": 0.05028248205780983, "step": 4191 }, { "epoch": 1.0882658359293873, "grad_norm": 2.4567553997039795, "learning_rate": 2.5393488210976764e-06, "loss": 0.061039239168167114, "step": 4192 }, { "epoch": 1.0885254413291796, "grad_norm": 3.2900617122650146, "learning_rate": 2.5384069852949415e-06, "loss": 0.024193882942199707, "step": 4193 }, { "epoch": 1.088785046728972, "grad_norm": 3.1814393997192383, "learning_rate": 2.537465186897006e-06, "loss": 0.03781253099441528, "step": 4194 }, { "epoch": 1.0890446521287642, "grad_norm": 2.579256296157837, "learning_rate": 2.5365234260703607e-06, "loss": 0.04606865718960762, "step": 4195 }, { "epoch": 1.0893042575285565, "grad_norm": 3.5097296237945557, "learning_rate": 2.5355817029814844e-06, "loss": 0.03701651096343994, "step": 4196 }, { "epoch": 1.0895638629283488, "grad_norm": 3.314798355102539, "learning_rate": 2.534640017796854e-06, "loss": 0.06543281674385071, "step": 4197 }, { "epoch": 1.0898234683281411, "grad_norm": 4.691206455230713, "learning_rate": 2.533698370682937e-06, "loss": 0.03936227783560753, "step": 4198 }, { "epoch": 1.0900830737279334, "grad_norm": 3.046159029006958, "learning_rate": 2.5327567618061965e-06, "loss": 0.06407850980758667, "step": 4199 }, { "epoch": 1.0903426791277258, "grad_norm": 2.198378801345825, "learning_rate": 2.5318151913330857e-06, "loss": 0.09998823702335358, "step": 4200 }, { "epoch": 1.090602284527518, "grad_norm": 3.867802381515503, "learning_rate": 2.5308736594300522e-06, "loss": 0.047979775816202164, "step": 4201 }, { "epoch": 1.0908618899273106, "grad_norm": 4.838170528411865, "learning_rate": 2.529932166263539e-06, "loss": 0.05919615179300308, "step": 4202 }, { "epoch": 1.091121495327103, "grad_norm": 2.889779806137085, "learning_rate": 2.5289907119999814e-06, "loss": 0.06701487302780151, "step": 4203 }, { "epoch": 1.0913811007268952, "grad_norm": 3.3167200088500977, "learning_rate": 2.528049296805804e-06, "loss": 0.0520576611161232, "step": 4204 }, { "epoch": 1.0916407061266875, "grad_norm": 4.402966022491455, "learning_rate": 2.5271079208474288e-06, "loss": 0.05545617267489433, "step": 4205 }, { "epoch": 1.0919003115264798, "grad_norm": 2.75323224067688, "learning_rate": 2.5261665842912696e-06, "loss": 0.01936066895723343, "step": 4206 }, { "epoch": 1.0921599169262721, "grad_norm": 3.417262315750122, "learning_rate": 2.525225287303735e-06, "loss": 0.027509812265634537, "step": 4207 }, { "epoch": 1.0924195223260644, "grad_norm": 2.4863522052764893, "learning_rate": 2.5242840300512205e-06, "loss": 0.025005269795656204, "step": 4208 }, { "epoch": 1.0926791277258567, "grad_norm": 2.9430956840515137, "learning_rate": 2.5233428127001226e-06, "loss": 0.04701828211545944, "step": 4209 }, { "epoch": 1.092938733125649, "grad_norm": 3.3077914714813232, "learning_rate": 2.522401635416825e-06, "loss": 0.02777433767914772, "step": 4210 }, { "epoch": 1.0931983385254413, "grad_norm": 2.911144971847534, "learning_rate": 2.5214604983677084e-06, "loss": 0.06220022216439247, "step": 4211 }, { "epoch": 1.0934579439252337, "grad_norm": 4.316479206085205, "learning_rate": 2.5205194017191414e-06, "loss": 0.06529544293880463, "step": 4212 }, { "epoch": 1.093717549325026, "grad_norm": 3.9026222229003906, "learning_rate": 2.5195783456374902e-06, "loss": 0.03089253604412079, "step": 4213 }, { "epoch": 1.0939771547248183, "grad_norm": 3.5120441913604736, "learning_rate": 2.518637330289111e-06, "loss": 0.04620634764432907, "step": 4214 }, { "epoch": 1.0942367601246106, "grad_norm": 4.076213836669922, "learning_rate": 2.517696355840355e-06, "loss": 0.045387186110019684, "step": 4215 }, { "epoch": 1.0944963655244029, "grad_norm": 2.0018651485443115, "learning_rate": 2.5167554224575643e-06, "loss": 0.07751122862100601, "step": 4216 }, { "epoch": 1.0947559709241952, "grad_norm": 2.949112892150879, "learning_rate": 2.5158145303070737e-06, "loss": 0.05983557179570198, "step": 4217 }, { "epoch": 1.0950155763239875, "grad_norm": 6.725052833557129, "learning_rate": 2.5148736795552125e-06, "loss": 0.027313917875289917, "step": 4218 }, { "epoch": 1.0952751817237798, "grad_norm": 4.3805389404296875, "learning_rate": 2.5139328703683007e-06, "loss": 0.08256751298904419, "step": 4219 }, { "epoch": 1.095534787123572, "grad_norm": 2.774414539337158, "learning_rate": 2.5129921029126524e-06, "loss": 0.07818637788295746, "step": 4220 }, { "epoch": 1.0957943925233644, "grad_norm": 3.4905078411102295, "learning_rate": 2.5120513773545737e-06, "loss": 0.02987533062696457, "step": 4221 }, { "epoch": 1.0960539979231567, "grad_norm": 2.292273759841919, "learning_rate": 2.511110693860363e-06, "loss": 0.04147430509328842, "step": 4222 }, { "epoch": 1.096313603322949, "grad_norm": 2.6922574043273926, "learning_rate": 2.510170052596312e-06, "loss": 0.01805470511317253, "step": 4223 }, { "epoch": 1.0965732087227413, "grad_norm": 2.7975943088531494, "learning_rate": 2.5092294537287038e-06, "loss": 0.073690265417099, "step": 4224 }, { "epoch": 1.0968328141225339, "grad_norm": 2.291734457015991, "learning_rate": 2.5082888974238155e-06, "loss": 0.06843860447406769, "step": 4225 }, { "epoch": 1.0970924195223262, "grad_norm": 3.7507753372192383, "learning_rate": 2.5073483838479164e-06, "loss": 0.04147392138838768, "step": 4226 }, { "epoch": 1.0973520249221185, "grad_norm": 3.0016672611236572, "learning_rate": 2.5064079131672658e-06, "loss": 0.014168732799589634, "step": 4227 }, { "epoch": 1.0976116303219108, "grad_norm": 2.1921658515930176, "learning_rate": 2.5054674855481198e-06, "loss": 0.011132724583148956, "step": 4228 }, { "epoch": 1.097871235721703, "grad_norm": 2.4075767993927, "learning_rate": 2.5045271011567225e-06, "loss": 0.028353478759527206, "step": 4229 }, { "epoch": 1.0981308411214954, "grad_norm": 3.0760865211486816, "learning_rate": 2.5035867601593145e-06, "loss": 0.061654843389987946, "step": 4230 }, { "epoch": 1.0983904465212877, "grad_norm": 4.234135627746582, "learning_rate": 2.5026464627221233e-06, "loss": 0.023704983294010162, "step": 4231 }, { "epoch": 1.09865005192108, "grad_norm": 3.242741584777832, "learning_rate": 2.501706209011374e-06, "loss": 0.011827593669295311, "step": 4232 }, { "epoch": 1.0989096573208723, "grad_norm": 3.9325668811798096, "learning_rate": 2.5007659991932826e-06, "loss": 0.027841530740261078, "step": 4233 }, { "epoch": 1.0991692627206646, "grad_norm": 3.773897409439087, "learning_rate": 2.4998258334340563e-06, "loss": 0.03478047251701355, "step": 4234 }, { "epoch": 1.099428868120457, "grad_norm": 2.862250566482544, "learning_rate": 2.498885711899894e-06, "loss": 0.03083062544465065, "step": 4235 }, { "epoch": 1.0996884735202492, "grad_norm": 2.955289840698242, "learning_rate": 2.497945634756987e-06, "loss": 0.06484346091747284, "step": 4236 }, { "epoch": 1.0999480789200415, "grad_norm": 1.7206475734710693, "learning_rate": 2.497005602171522e-06, "loss": 0.09508936107158661, "step": 4237 }, { "epoch": 1.1002076843198338, "grad_norm": 4.081972599029541, "learning_rate": 2.4960656143096732e-06, "loss": 0.04522094130516052, "step": 4238 }, { "epoch": 1.1004672897196262, "grad_norm": 2.8316335678100586, "learning_rate": 2.4951256713376092e-06, "loss": 0.020286858081817627, "step": 4239 }, { "epoch": 1.1007268951194185, "grad_norm": 1.9572186470031738, "learning_rate": 2.494185773421491e-06, "loss": 0.10504257678985596, "step": 4240 }, { "epoch": 1.1009865005192108, "grad_norm": 3.958068370819092, "learning_rate": 2.4932459207274694e-06, "loss": 0.019984949380159378, "step": 4241 }, { "epoch": 1.101246105919003, "grad_norm": 3.2427048683166504, "learning_rate": 2.4923061134216927e-06, "loss": 0.0675538182258606, "step": 4242 }, { "epoch": 1.1015057113187954, "grad_norm": 4.7433085441589355, "learning_rate": 2.4913663516702927e-06, "loss": 0.031259872019290924, "step": 4243 }, { "epoch": 1.1017653167185877, "grad_norm": 3.0454494953155518, "learning_rate": 2.4904266356394003e-06, "loss": 0.03049481473863125, "step": 4244 }, { "epoch": 1.10202492211838, "grad_norm": 2.5264079570770264, "learning_rate": 2.4894869654951353e-06, "loss": 0.045044317841529846, "step": 4245 }, { "epoch": 1.1022845275181723, "grad_norm": 3.783684730529785, "learning_rate": 2.48854734140361e-06, "loss": 0.021899782121181488, "step": 4246 }, { "epoch": 1.1025441329179646, "grad_norm": 2.6616952419281006, "learning_rate": 2.4876077635309283e-06, "loss": 0.048995859920978546, "step": 4247 }, { "epoch": 1.102803738317757, "grad_norm": 3.5319294929504395, "learning_rate": 2.4866682320431858e-06, "loss": 0.04754868894815445, "step": 4248 }, { "epoch": 1.1030633437175492, "grad_norm": 2.0098884105682373, "learning_rate": 2.4857287471064707e-06, "loss": 0.09060564637184143, "step": 4249 }, { "epoch": 1.1033229491173417, "grad_norm": 3.464723825454712, "learning_rate": 2.484789308886862e-06, "loss": 0.012838151305913925, "step": 4250 }, { "epoch": 1.103582554517134, "grad_norm": 2.828640937805176, "learning_rate": 2.4838499175504303e-06, "loss": 0.024210389703512192, "step": 4251 }, { "epoch": 1.1038421599169264, "grad_norm": 2.6333932876586914, "learning_rate": 2.4829105732632397e-06, "loss": 0.05752280727028847, "step": 4252 }, { "epoch": 1.1041017653167187, "grad_norm": 3.437946319580078, "learning_rate": 2.4819712761913444e-06, "loss": 0.041954368352890015, "step": 4253 }, { "epoch": 1.104361370716511, "grad_norm": 3.544273853302002, "learning_rate": 2.48103202650079e-06, "loss": 0.03585594892501831, "step": 4254 }, { "epoch": 1.1046209761163033, "grad_norm": 2.8462908267974854, "learning_rate": 2.480092824357615e-06, "loss": 0.02990187332034111, "step": 4255 }, { "epoch": 1.1048805815160956, "grad_norm": 3.50667405128479, "learning_rate": 2.4791536699278487e-06, "loss": 0.022055961191654205, "step": 4256 }, { "epoch": 1.105140186915888, "grad_norm": 3.2066659927368164, "learning_rate": 2.478214563377512e-06, "loss": 0.03914918377995491, "step": 4257 }, { "epoch": 1.1053997923156802, "grad_norm": 1.905247449874878, "learning_rate": 2.4772755048726172e-06, "loss": 0.08464878797531128, "step": 4258 }, { "epoch": 1.1056593977154725, "grad_norm": 3.1830692291259766, "learning_rate": 2.4763364945791685e-06, "loss": 0.060246434062719345, "step": 4259 }, { "epoch": 1.1059190031152648, "grad_norm": 2.902618169784546, "learning_rate": 2.475397532663161e-06, "loss": 0.053408198058605194, "step": 4260 }, { "epoch": 1.1061786085150571, "grad_norm": 3.4320666790008545, "learning_rate": 2.474458619290584e-06, "loss": 0.03005310334265232, "step": 4261 }, { "epoch": 1.1064382139148494, "grad_norm": 3.1990673542022705, "learning_rate": 2.4735197546274125e-06, "loss": 0.021636446937918663, "step": 4262 }, { "epoch": 1.1066978193146417, "grad_norm": 2.0616610050201416, "learning_rate": 2.4725809388396175e-06, "loss": 0.007161751855164766, "step": 4263 }, { "epoch": 1.106957424714434, "grad_norm": 3.2181012630462646, "learning_rate": 2.4716421720931616e-06, "loss": 0.01843133568763733, "step": 4264 }, { "epoch": 1.1072170301142263, "grad_norm": 1.6310735940933228, "learning_rate": 2.4707034545539963e-06, "loss": 0.07951532304286957, "step": 4265 }, { "epoch": 1.1074766355140186, "grad_norm": 2.3154914379119873, "learning_rate": 2.4697647863880648e-06, "loss": 0.06805701553821564, "step": 4266 }, { "epoch": 1.107736240913811, "grad_norm": 2.358858823776245, "learning_rate": 2.468826167761301e-06, "loss": 0.08359664678573608, "step": 4267 }, { "epoch": 1.1079958463136033, "grad_norm": 1.7748008966445923, "learning_rate": 2.4678875988396344e-06, "loss": 0.0813605785369873, "step": 4268 }, { "epoch": 1.1082554517133956, "grad_norm": 1.7597419023513794, "learning_rate": 2.466949079788981e-06, "loss": 0.09027935564517975, "step": 4269 }, { "epoch": 1.1085150571131879, "grad_norm": 3.0763626098632812, "learning_rate": 2.4660106107752485e-06, "loss": 0.03189988061785698, "step": 4270 }, { "epoch": 1.1087746625129802, "grad_norm": 4.8994293212890625, "learning_rate": 2.4650721919643376e-06, "loss": 0.048844411969184875, "step": 4271 }, { "epoch": 1.1090342679127725, "grad_norm": 2.7781591415405273, "learning_rate": 2.46413382352214e-06, "loss": 0.025331862270832062, "step": 4272 }, { "epoch": 1.109293873312565, "grad_norm": 3.659700393676758, "learning_rate": 2.463195505614536e-06, "loss": 0.038655176758766174, "step": 4273 }, { "epoch": 1.1095534787123573, "grad_norm": 3.5369582176208496, "learning_rate": 2.4622572384074e-06, "loss": 0.025078561156988144, "step": 4274 }, { "epoch": 1.1098130841121496, "grad_norm": 2.025895357131958, "learning_rate": 2.4613190220665957e-06, "loss": 0.09306420385837555, "step": 4275 }, { "epoch": 1.110072689511942, "grad_norm": 1.7375737428665161, "learning_rate": 2.4603808567579787e-06, "loss": 0.0877901017665863, "step": 4276 }, { "epoch": 1.1103322949117342, "grad_norm": 3.131901979446411, "learning_rate": 2.4594427426473944e-06, "loss": 0.04144278168678284, "step": 4277 }, { "epoch": 1.1105919003115265, "grad_norm": 3.1480166912078857, "learning_rate": 2.45850467990068e-06, "loss": 0.05982111021876335, "step": 4278 }, { "epoch": 1.1108515057113189, "grad_norm": 2.165738344192505, "learning_rate": 2.457566668683664e-06, "loss": 0.0872306227684021, "step": 4279 }, { "epoch": 1.1111111111111112, "grad_norm": 3.568843126296997, "learning_rate": 2.456628709162166e-06, "loss": 0.028144007548689842, "step": 4280 }, { "epoch": 1.1113707165109035, "grad_norm": 2.690596103668213, "learning_rate": 2.455690801501994e-06, "loss": 0.09287773072719574, "step": 4281 }, { "epoch": 1.1116303219106958, "grad_norm": 3.6936349868774414, "learning_rate": 2.4547529458689496e-06, "loss": 0.028114188462495804, "step": 4282 }, { "epoch": 1.111889927310488, "grad_norm": 3.1550936698913574, "learning_rate": 2.4538151424288246e-06, "loss": 0.03952398523688316, "step": 4283 }, { "epoch": 1.1121495327102804, "grad_norm": 3.542382001876831, "learning_rate": 2.452877391347401e-06, "loss": 0.012155924923717976, "step": 4284 }, { "epoch": 1.1124091381100727, "grad_norm": 2.568701982498169, "learning_rate": 2.4519396927904505e-06, "loss": 0.0727752149105072, "step": 4285 }, { "epoch": 1.112668743509865, "grad_norm": 3.18323016166687, "learning_rate": 2.4510020469237376e-06, "loss": 0.03517339751124382, "step": 4286 }, { "epoch": 1.1129283489096573, "grad_norm": 2.4686083793640137, "learning_rate": 2.450064453913017e-06, "loss": 0.07967326790094376, "step": 4287 }, { "epoch": 1.1131879543094496, "grad_norm": 2.5964391231536865, "learning_rate": 2.449126913924034e-06, "loss": 0.07328793406486511, "step": 4288 }, { "epoch": 1.113447559709242, "grad_norm": 3.955275774002075, "learning_rate": 2.4481894271225225e-06, "loss": 0.05801249295473099, "step": 4289 }, { "epoch": 1.1137071651090342, "grad_norm": 4.201114654541016, "learning_rate": 2.44725199367421e-06, "loss": 0.03652132302522659, "step": 4290 }, { "epoch": 1.1139667705088265, "grad_norm": 2.9074273109436035, "learning_rate": 2.4463146137448127e-06, "loss": 0.07175081968307495, "step": 4291 }, { "epoch": 1.1142263759086188, "grad_norm": 4.13562536239624, "learning_rate": 2.4453772875000398e-06, "loss": 0.05516012758016586, "step": 4292 }, { "epoch": 1.1144859813084111, "grad_norm": 2.7097198963165283, "learning_rate": 2.444440015105586e-06, "loss": 0.011002326384186745, "step": 4293 }, { "epoch": 1.1147455867082035, "grad_norm": 2.3301329612731934, "learning_rate": 2.4435027967271413e-06, "loss": 0.08455710113048553, "step": 4294 }, { "epoch": 1.1150051921079958, "grad_norm": 4.303420066833496, "learning_rate": 2.442565632530384e-06, "loss": 0.04460037499666214, "step": 4295 }, { "epoch": 1.115264797507788, "grad_norm": 2.4170682430267334, "learning_rate": 2.441628522680985e-06, "loss": 0.01089124009013176, "step": 4296 }, { "epoch": 1.1155244029075804, "grad_norm": 2.7633488178253174, "learning_rate": 2.4406914673446012e-06, "loss": 0.05969291925430298, "step": 4297 }, { "epoch": 1.1157840083073727, "grad_norm": 1.8746864795684814, "learning_rate": 2.4397544666868837e-06, "loss": 0.08885471522808075, "step": 4298 }, { "epoch": 1.1160436137071652, "grad_norm": 2.317294120788574, "learning_rate": 2.438817520873474e-06, "loss": 0.01337455864995718, "step": 4299 }, { "epoch": 1.1163032191069575, "grad_norm": 3.636087417602539, "learning_rate": 2.4378806300700003e-06, "loss": 0.05395478010177612, "step": 4300 }, { "epoch": 1.1165628245067498, "grad_norm": 2.740133047103882, "learning_rate": 2.436943794442085e-06, "loss": 0.05367694050073624, "step": 4301 }, { "epoch": 1.1168224299065421, "grad_norm": 3.1253395080566406, "learning_rate": 2.436007014155339e-06, "loss": 0.05907442420721054, "step": 4302 }, { "epoch": 1.1170820353063344, "grad_norm": 4.522801876068115, "learning_rate": 2.435070289375363e-06, "loss": 0.020485224202275276, "step": 4303 }, { "epoch": 1.1173416407061267, "grad_norm": 2.2622134685516357, "learning_rate": 2.434133620267749e-06, "loss": 0.06455487012863159, "step": 4304 }, { "epoch": 1.117601246105919, "grad_norm": 1.9723423719406128, "learning_rate": 2.433197006998078e-06, "loss": 0.08289206027984619, "step": 4305 }, { "epoch": 1.1178608515057114, "grad_norm": 2.9732253551483154, "learning_rate": 2.432260449731923e-06, "loss": 0.022655578330159187, "step": 4306 }, { "epoch": 1.1181204569055037, "grad_norm": 3.3635308742523193, "learning_rate": 2.431323948634846e-06, "loss": 0.05494218319654465, "step": 4307 }, { "epoch": 1.118380062305296, "grad_norm": 3.521688222885132, "learning_rate": 2.4303875038723975e-06, "loss": 0.022311924025416374, "step": 4308 }, { "epoch": 1.1186396677050883, "grad_norm": 2.5077121257781982, "learning_rate": 2.429451115610119e-06, "loss": 0.020786425098776817, "step": 4309 }, { "epoch": 1.1188992731048806, "grad_norm": 2.341643810272217, "learning_rate": 2.428514784013545e-06, "loss": 0.0872948169708252, "step": 4310 }, { "epoch": 1.1191588785046729, "grad_norm": 5.079035758972168, "learning_rate": 2.4275785092481965e-06, "loss": 0.01791956275701523, "step": 4311 }, { "epoch": 1.1194184839044652, "grad_norm": 2.652740716934204, "learning_rate": 2.426642291479584e-06, "loss": 0.04770214483141899, "step": 4312 }, { "epoch": 1.1196780893042575, "grad_norm": 7.103214263916016, "learning_rate": 2.4257061308732104e-06, "loss": 0.046000003814697266, "step": 4313 }, { "epoch": 1.1199376947040498, "grad_norm": 2.4728121757507324, "learning_rate": 2.424770027594568e-06, "loss": 0.04421994835138321, "step": 4314 }, { "epoch": 1.120197300103842, "grad_norm": 4.3038740158081055, "learning_rate": 2.4238339818091383e-06, "loss": 0.028405115008354187, "step": 4315 }, { "epoch": 1.1204569055036344, "grad_norm": 3.3136677742004395, "learning_rate": 2.422897993682392e-06, "loss": 0.07686799764633179, "step": 4316 }, { "epoch": 1.1207165109034267, "grad_norm": 3.9570250511169434, "learning_rate": 2.421962063379791e-06, "loss": 0.06578241288661957, "step": 4317 }, { "epoch": 1.120976116303219, "grad_norm": 4.388457775115967, "learning_rate": 2.421026191066785e-06, "loss": 0.04187189042568207, "step": 4318 }, { "epoch": 1.1212357217030113, "grad_norm": 3.728149175643921, "learning_rate": 2.4200903769088174e-06, "loss": 0.0242605023086071, "step": 4319 }, { "epoch": 1.1214953271028036, "grad_norm": 3.164118766784668, "learning_rate": 2.419154621071317e-06, "loss": 0.04176989942789078, "step": 4320 }, { "epoch": 1.1217549325025962, "grad_norm": 3.1719000339508057, "learning_rate": 2.4182189237197033e-06, "loss": 0.04182036966085434, "step": 4321 }, { "epoch": 1.1220145379023885, "grad_norm": 2.5900509357452393, "learning_rate": 2.4172832850193877e-06, "loss": 0.08854497224092484, "step": 4322 }, { "epoch": 1.1222741433021808, "grad_norm": 2.43104887008667, "learning_rate": 2.41634770513577e-06, "loss": 0.06426753848791122, "step": 4323 }, { "epoch": 1.122533748701973, "grad_norm": 1.8203285932540894, "learning_rate": 2.4154121842342363e-06, "loss": 0.08215510845184326, "step": 4324 }, { "epoch": 1.1227933541017654, "grad_norm": 3.5993380546569824, "learning_rate": 2.414476722480169e-06, "loss": 0.040479984134435654, "step": 4325 }, { "epoch": 1.1230529595015577, "grad_norm": 3.818331718444824, "learning_rate": 2.413541320038934e-06, "loss": 0.032872363924980164, "step": 4326 }, { "epoch": 1.12331256490135, "grad_norm": 2.9701693058013916, "learning_rate": 2.4126059770758907e-06, "loss": 0.052225276827812195, "step": 4327 }, { "epoch": 1.1235721703011423, "grad_norm": 1.8247283697128296, "learning_rate": 2.4116706937563848e-06, "loss": 0.10295413434505463, "step": 4328 }, { "epoch": 1.1238317757009346, "grad_norm": 2.9534199237823486, "learning_rate": 2.4107354702457524e-06, "loss": 0.052734166383743286, "step": 4329 }, { "epoch": 1.124091381100727, "grad_norm": 3.619391918182373, "learning_rate": 2.409800306709323e-06, "loss": 0.06749270856380463, "step": 4330 }, { "epoch": 1.1243509865005192, "grad_norm": 2.581566333770752, "learning_rate": 2.4088652033124086e-06, "loss": 0.049853578209877014, "step": 4331 }, { "epoch": 1.1246105919003115, "grad_norm": 1.779829740524292, "learning_rate": 2.407930160220315e-06, "loss": 0.09954208135604858, "step": 4332 }, { "epoch": 1.1248701973001038, "grad_norm": 1.3320660591125488, "learning_rate": 2.406995177598337e-06, "loss": 0.013631505891680717, "step": 4333 }, { "epoch": 1.1251298026998962, "grad_norm": 1.7431713342666626, "learning_rate": 2.4060602556117578e-06, "loss": 0.08400818705558777, "step": 4334 }, { "epoch": 1.1253894080996885, "grad_norm": 2.0144569873809814, "learning_rate": 2.40512539442585e-06, "loss": 0.006275423802435398, "step": 4335 }, { "epoch": 1.1256490134994808, "grad_norm": 2.3205065727233887, "learning_rate": 2.404190594205876e-06, "loss": 0.07122312486171722, "step": 4336 }, { "epoch": 1.125908618899273, "grad_norm": 2.237473964691162, "learning_rate": 2.4032558551170875e-06, "loss": 0.08203814178705215, "step": 4337 }, { "epoch": 1.1261682242990654, "grad_norm": 3.5383450984954834, "learning_rate": 2.402321177324724e-06, "loss": 0.014232847839593887, "step": 4338 }, { "epoch": 1.1264278296988577, "grad_norm": 3.5009913444519043, "learning_rate": 2.4013865609940156e-06, "loss": 0.05439642444252968, "step": 4339 }, { "epoch": 1.12668743509865, "grad_norm": 3.1426374912261963, "learning_rate": 2.400452006290181e-06, "loss": 0.04437888413667679, "step": 4340 }, { "epoch": 1.1269470404984423, "grad_norm": 2.9613847732543945, "learning_rate": 2.399517513378428e-06, "loss": 0.02787579409778118, "step": 4341 }, { "epoch": 1.1272066458982346, "grad_norm": 2.976123809814453, "learning_rate": 2.398583082423954e-06, "loss": 0.045925840735435486, "step": 4342 }, { "epoch": 1.127466251298027, "grad_norm": 2.6536357402801514, "learning_rate": 2.3976487135919444e-06, "loss": 0.05449530482292175, "step": 4343 }, { "epoch": 1.1277258566978192, "grad_norm": 3.9768898487091064, "learning_rate": 2.3967144070475746e-06, "loss": 0.029447153210639954, "step": 4344 }, { "epoch": 1.1279854620976115, "grad_norm": 2.7927355766296387, "learning_rate": 2.395780162956008e-06, "loss": 0.01731717586517334, "step": 4345 }, { "epoch": 1.1282450674974038, "grad_norm": 3.542130470275879, "learning_rate": 2.3948459814823986e-06, "loss": 0.015967058017849922, "step": 4346 }, { "epoch": 1.1285046728971961, "grad_norm": 3.907862901687622, "learning_rate": 2.3939118627918874e-06, "loss": 0.032944656908512115, "step": 4347 }, { "epoch": 1.1287642782969887, "grad_norm": 4.002291679382324, "learning_rate": 2.3929778070496055e-06, "loss": 0.03270501643419266, "step": 4348 }, { "epoch": 1.129023883696781, "grad_norm": 3.8242273330688477, "learning_rate": 2.3920438144206726e-06, "loss": 0.02252109907567501, "step": 4349 }, { "epoch": 1.1292834890965733, "grad_norm": 2.6996490955352783, "learning_rate": 2.3911098850701975e-06, "loss": 0.07188991457223892, "step": 4350 }, { "epoch": 1.1295430944963656, "grad_norm": 4.478633403778076, "learning_rate": 2.3901760191632774e-06, "loss": 0.028331313282251358, "step": 4351 }, { "epoch": 1.129802699896158, "grad_norm": 2.6731157302856445, "learning_rate": 2.389242216864997e-06, "loss": 0.016792239621281624, "step": 4352 }, { "epoch": 1.1300623052959502, "grad_norm": 2.269451379776001, "learning_rate": 2.3883084783404337e-06, "loss": 0.09907244890928268, "step": 4353 }, { "epoch": 1.1303219106957425, "grad_norm": 1.8550344705581665, "learning_rate": 2.38737480375465e-06, "loss": 0.10039110481739044, "step": 4354 }, { "epoch": 1.1305815160955348, "grad_norm": 3.2662742137908936, "learning_rate": 2.3864411932726967e-06, "loss": 0.029509814456105232, "step": 4355 }, { "epoch": 1.1308411214953271, "grad_norm": 5.403820037841797, "learning_rate": 2.3855076470596165e-06, "loss": 0.02569306455552578, "step": 4356 }, { "epoch": 1.1311007268951194, "grad_norm": 3.1233131885528564, "learning_rate": 2.3845741652804393e-06, "loss": 0.013724287040531635, "step": 4357 }, { "epoch": 1.1313603322949117, "grad_norm": 3.080320358276367, "learning_rate": 2.3836407481001823e-06, "loss": 0.07174926996231079, "step": 4358 }, { "epoch": 1.131619937694704, "grad_norm": 3.033345937728882, "learning_rate": 2.3827073956838515e-06, "loss": 0.014190874062478542, "step": 4359 }, { "epoch": 1.1318795430944963, "grad_norm": 2.628650188446045, "learning_rate": 2.3817741081964433e-06, "loss": 0.07496717572212219, "step": 4360 }, { "epoch": 1.1321391484942886, "grad_norm": 3.0472753047943115, "learning_rate": 2.3808408858029437e-06, "loss": 0.06122760847210884, "step": 4361 }, { "epoch": 1.132398753894081, "grad_norm": 2.1151788234710693, "learning_rate": 2.379907728668321e-06, "loss": 0.019591927528381348, "step": 4362 }, { "epoch": 1.1326583592938733, "grad_norm": 2.8952553272247314, "learning_rate": 2.378974636957538e-06, "loss": 0.01832878217101097, "step": 4363 }, { "epoch": 1.1329179646936656, "grad_norm": 2.7662439346313477, "learning_rate": 2.3780416108355433e-06, "loss": 0.07277297973632812, "step": 4364 }, { "epoch": 1.1331775700934579, "grad_norm": 2.85111665725708, "learning_rate": 2.3771086504672762e-06, "loss": 0.04664165526628494, "step": 4365 }, { "epoch": 1.1334371754932502, "grad_norm": 3.794923782348633, "learning_rate": 2.3761757560176614e-06, "loss": 0.019511792808771133, "step": 4366 }, { "epoch": 1.1336967808930425, "grad_norm": 3.111198663711548, "learning_rate": 2.3752429276516132e-06, "loss": 0.05271743983030319, "step": 4367 }, { "epoch": 1.1339563862928348, "grad_norm": 2.289430856704712, "learning_rate": 2.3743101655340346e-06, "loss": 0.08162638545036316, "step": 4368 }, { "epoch": 1.1342159916926273, "grad_norm": 1.8397103548049927, "learning_rate": 2.3733774698298173e-06, "loss": 0.0943191722035408, "step": 4369 }, { "epoch": 1.1344755970924196, "grad_norm": 3.5873117446899414, "learning_rate": 2.3724448407038386e-06, "loss": 0.01676715910434723, "step": 4370 }, { "epoch": 1.134735202492212, "grad_norm": 4.4215874671936035, "learning_rate": 2.371512278320968e-06, "loss": 0.03781845420598984, "step": 4371 }, { "epoch": 1.1349948078920042, "grad_norm": 4.57398796081543, "learning_rate": 2.3705797828460596e-06, "loss": 0.044183555990457535, "step": 4372 }, { "epoch": 1.1352544132917965, "grad_norm": 4.542795658111572, "learning_rate": 2.369647354443959e-06, "loss": 0.06341858208179474, "step": 4373 }, { "epoch": 1.1355140186915889, "grad_norm": 3.511702537536621, "learning_rate": 2.3687149932794967e-06, "loss": 0.03284189850091934, "step": 4374 }, { "epoch": 1.1357736240913812, "grad_norm": 2.7969489097595215, "learning_rate": 2.3677826995174935e-06, "loss": 0.06708353757858276, "step": 4375 }, { "epoch": 1.1360332294911735, "grad_norm": 4.035797595977783, "learning_rate": 2.366850473322757e-06, "loss": 0.016015931963920593, "step": 4376 }, { "epoch": 1.1362928348909658, "grad_norm": 2.607215166091919, "learning_rate": 2.3659183148600847e-06, "loss": 0.06816243380308151, "step": 4377 }, { "epoch": 1.136552440290758, "grad_norm": 3.5981571674346924, "learning_rate": 2.3649862242942595e-06, "loss": 0.10061906278133392, "step": 4378 }, { "epoch": 1.1368120456905504, "grad_norm": 2.200770854949951, "learning_rate": 2.364054201790054e-06, "loss": 0.07065980136394501, "step": 4379 }, { "epoch": 1.1370716510903427, "grad_norm": 2.6960349082946777, "learning_rate": 2.363122247512229e-06, "loss": 0.09217904508113861, "step": 4380 }, { "epoch": 1.137331256490135, "grad_norm": 4.3342509269714355, "learning_rate": 2.3621903616255322e-06, "loss": 0.01798785850405693, "step": 4381 }, { "epoch": 1.1375908618899273, "grad_norm": 2.8655030727386475, "learning_rate": 2.3612585442946994e-06, "loss": 0.008298450149595737, "step": 4382 }, { "epoch": 1.1378504672897196, "grad_norm": 3.215604543685913, "learning_rate": 2.3603267956844555e-06, "loss": 0.06332967430353165, "step": 4383 }, { "epoch": 1.138110072689512, "grad_norm": 2.326786518096924, "learning_rate": 2.3593951159595125e-06, "loss": 0.06166511029005051, "step": 4384 }, { "epoch": 1.1383696780893042, "grad_norm": 2.4310686588287354, "learning_rate": 2.358463505284568e-06, "loss": 0.070094034075737, "step": 4385 }, { "epoch": 1.1386292834890965, "grad_norm": 2.886749505996704, "learning_rate": 2.3575319638243103e-06, "loss": 0.0548379048705101, "step": 4386 }, { "epoch": 1.1388888888888888, "grad_norm": 4.441597938537598, "learning_rate": 2.3566004917434165e-06, "loss": 0.028086543083190918, "step": 4387 }, { "epoch": 1.1391484942886811, "grad_norm": 5.568615436553955, "learning_rate": 2.3556690892065487e-06, "loss": 0.024476192891597748, "step": 4388 }, { "epoch": 1.1394080996884735, "grad_norm": 2.1532716751098633, "learning_rate": 2.354737756378355e-06, "loss": 0.08189177513122559, "step": 4389 }, { "epoch": 1.1396677050882658, "grad_norm": 1.9032803773880005, "learning_rate": 2.353806493423476e-06, "loss": 0.08418654650449753, "step": 4390 }, { "epoch": 1.139927310488058, "grad_norm": 3.426947832107544, "learning_rate": 2.352875300506538e-06, "loss": 0.03181132674217224, "step": 4391 }, { "epoch": 1.1401869158878504, "grad_norm": 2.730710744857788, "learning_rate": 2.351944177792155e-06, "loss": 0.06931319832801819, "step": 4392 }, { "epoch": 1.1404465212876427, "grad_norm": 1.5583479404449463, "learning_rate": 2.3510131254449253e-06, "loss": 0.09296537935733795, "step": 4393 }, { "epoch": 1.140706126687435, "grad_norm": 1.8918310403823853, "learning_rate": 2.35008214362944e-06, "loss": 0.08049718290567398, "step": 4394 }, { "epoch": 1.1409657320872273, "grad_norm": 2.312731981277466, "learning_rate": 2.349151232510274e-06, "loss": 0.011671346612274647, "step": 4395 }, { "epoch": 1.1412253374870198, "grad_norm": 3.1298303604125977, "learning_rate": 2.3482203922519937e-06, "loss": 0.06497252732515335, "step": 4396 }, { "epoch": 1.1414849428868121, "grad_norm": 2.5270869731903076, "learning_rate": 2.3472896230191473e-06, "loss": 0.05807407200336456, "step": 4397 }, { "epoch": 1.1417445482866044, "grad_norm": 3.255312204360962, "learning_rate": 2.3463589249762747e-06, "loss": 0.012673809193074703, "step": 4398 }, { "epoch": 1.1420041536863967, "grad_norm": 1.6357429027557373, "learning_rate": 2.345428298287902e-06, "loss": 0.09305479377508163, "step": 4399 }, { "epoch": 1.142263759086189, "grad_norm": 2.3614156246185303, "learning_rate": 2.3444977431185432e-06, "loss": 0.05117900297045708, "step": 4400 }, { "epoch": 1.1425233644859814, "grad_norm": 3.237196683883667, "learning_rate": 2.3435672596326987e-06, "loss": 0.051445573568344116, "step": 4401 }, { "epoch": 1.1427829698857737, "grad_norm": 2.6753525733947754, "learning_rate": 2.3426368479948556e-06, "loss": 0.0507480651140213, "step": 4402 }, { "epoch": 1.143042575285566, "grad_norm": 3.4643819332122803, "learning_rate": 2.3417065083694913e-06, "loss": 0.012290555983781815, "step": 4403 }, { "epoch": 1.1433021806853583, "grad_norm": 3.2361881732940674, "learning_rate": 2.3407762409210677e-06, "loss": 0.04553801193833351, "step": 4404 }, { "epoch": 1.1435617860851506, "grad_norm": 2.7610371112823486, "learning_rate": 2.339846045814034e-06, "loss": 0.02301124483346939, "step": 4405 }, { "epoch": 1.1438213914849429, "grad_norm": 4.031012058258057, "learning_rate": 2.3389159232128285e-06, "loss": 0.04648642987012863, "step": 4406 }, { "epoch": 1.1440809968847352, "grad_norm": 4.070212364196777, "learning_rate": 2.337985873281875e-06, "loss": 0.053973231464624405, "step": 4407 }, { "epoch": 1.1443406022845275, "grad_norm": 1.9487826824188232, "learning_rate": 2.337055896185586e-06, "loss": 0.08020325005054474, "step": 4408 }, { "epoch": 1.1446002076843198, "grad_norm": 4.211673259735107, "learning_rate": 2.336125992088359e-06, "loss": 0.03487209230661392, "step": 4409 }, { "epoch": 1.144859813084112, "grad_norm": 2.2984840869903564, "learning_rate": 2.3351961611545798e-06, "loss": 0.05317167565226555, "step": 4410 }, { "epoch": 1.1451194184839044, "grad_norm": 2.7941741943359375, "learning_rate": 2.3342664035486216e-06, "loss": 0.051430147141218185, "step": 4411 }, { "epoch": 1.1453790238836967, "grad_norm": 1.9828773736953735, "learning_rate": 2.3333367194348443e-06, "loss": 0.07703027129173279, "step": 4412 }, { "epoch": 1.145638629283489, "grad_norm": 1.988938331604004, "learning_rate": 2.3324071089775953e-06, "loss": 0.09329831600189209, "step": 4413 }, { "epoch": 1.1458982346832813, "grad_norm": 2.4566519260406494, "learning_rate": 2.3314775723412065e-06, "loss": 0.072905033826828, "step": 4414 }, { "epoch": 1.1461578400830736, "grad_norm": 3.7562570571899414, "learning_rate": 2.330548109690002e-06, "loss": 0.052370402961969376, "step": 4415 }, { "epoch": 1.146417445482866, "grad_norm": 2.9452309608459473, "learning_rate": 2.3296187211882854e-06, "loss": 0.05061729997396469, "step": 4416 }, { "epoch": 1.1466770508826585, "grad_norm": 3.2032177448272705, "learning_rate": 2.328689407000354e-06, "loss": 0.05886406451463699, "step": 4417 }, { "epoch": 1.1469366562824508, "grad_norm": 2.948499917984009, "learning_rate": 2.327760167290489e-06, "loss": 0.018541086465120316, "step": 4418 }, { "epoch": 1.147196261682243, "grad_norm": 4.356044292449951, "learning_rate": 2.3268310022229597e-06, "loss": 0.03186982870101929, "step": 4419 }, { "epoch": 1.1474558670820354, "grad_norm": 3.6969218254089355, "learning_rate": 2.3259019119620184e-06, "loss": 0.038329944014549255, "step": 4420 }, { "epoch": 1.1477154724818277, "grad_norm": 2.371605157852173, "learning_rate": 2.324972896671908e-06, "loss": 0.07171037793159485, "step": 4421 }, { "epoch": 1.14797507788162, "grad_norm": 4.305217742919922, "learning_rate": 2.324043956516859e-06, "loss": 0.030558915808796883, "step": 4422 }, { "epoch": 1.1482346832814123, "grad_norm": 2.994316816329956, "learning_rate": 2.323115091661085e-06, "loss": 0.06678441166877747, "step": 4423 }, { "epoch": 1.1484942886812046, "grad_norm": 2.221827983856201, "learning_rate": 2.3221863022687877e-06, "loss": 0.08470557630062103, "step": 4424 }, { "epoch": 1.148753894080997, "grad_norm": 2.7212061882019043, "learning_rate": 2.321257588504158e-06, "loss": 0.03373374789953232, "step": 4425 }, { "epoch": 1.1490134994807892, "grad_norm": 3.007896900177002, "learning_rate": 2.3203289505313677e-06, "loss": 0.02078387327492237, "step": 4426 }, { "epoch": 1.1492731048805815, "grad_norm": 3.7565598487854004, "learning_rate": 2.3194003885145826e-06, "loss": 0.030653107911348343, "step": 4427 }, { "epoch": 1.1495327102803738, "grad_norm": 2.9563276767730713, "learning_rate": 2.3184719026179485e-06, "loss": 0.07803945243358612, "step": 4428 }, { "epoch": 1.1497923156801662, "grad_norm": 3.705918312072754, "learning_rate": 2.317543493005602e-06, "loss": 0.058858923614025116, "step": 4429 }, { "epoch": 1.1500519210799585, "grad_norm": 4.515199661254883, "learning_rate": 2.316615159841663e-06, "loss": 0.02264375425875187, "step": 4430 }, { "epoch": 1.1503115264797508, "grad_norm": 4.704576015472412, "learning_rate": 2.315686903290242e-06, "loss": 0.027563493698835373, "step": 4431 }, { "epoch": 1.150571131879543, "grad_norm": 3.440669059753418, "learning_rate": 2.3147587235154307e-06, "loss": 0.055973879992961884, "step": 4432 }, { "epoch": 1.1508307372793354, "grad_norm": 3.340047836303711, "learning_rate": 2.3138306206813117e-06, "loss": 0.019612271338701248, "step": 4433 }, { "epoch": 1.1510903426791277, "grad_norm": 3.4099795818328857, "learning_rate": 2.3129025949519517e-06, "loss": 0.028510307893157005, "step": 4434 }, { "epoch": 1.15134994807892, "grad_norm": 2.345857620239258, "learning_rate": 2.3119746464914055e-06, "loss": 0.0648767277598381, "step": 4435 }, { "epoch": 1.1516095534787123, "grad_norm": 2.755704641342163, "learning_rate": 2.3110467754637115e-06, "loss": 0.02018565684556961, "step": 4436 }, { "epoch": 1.1518691588785046, "grad_norm": 2.6134140491485596, "learning_rate": 2.3101189820328977e-06, "loss": 0.08115772902965546, "step": 4437 }, { "epoch": 1.152128764278297, "grad_norm": 3.890268564224243, "learning_rate": 2.3091912663629758e-06, "loss": 0.024187222123146057, "step": 4438 }, { "epoch": 1.1523883696780892, "grad_norm": 4.8096184730529785, "learning_rate": 2.3082636286179446e-06, "loss": 0.03832067921757698, "step": 4439 }, { "epoch": 1.1526479750778815, "grad_norm": 3.0466601848602295, "learning_rate": 2.3073360689617893e-06, "loss": 0.039586715400218964, "step": 4440 }, { "epoch": 1.1529075804776738, "grad_norm": 2.2373127937316895, "learning_rate": 2.3064085875584814e-06, "loss": 0.017887089401483536, "step": 4441 }, { "epoch": 1.1531671858774661, "grad_norm": 5.897799968719482, "learning_rate": 2.3054811845719786e-06, "loss": 0.02594168856739998, "step": 4442 }, { "epoch": 1.1534267912772584, "grad_norm": 3.7425737380981445, "learning_rate": 2.304553860166224e-06, "loss": 0.03327739238739014, "step": 4443 }, { "epoch": 1.153686396677051, "grad_norm": 2.8263866901397705, "learning_rate": 2.303626614505147e-06, "loss": 0.01319904811680317, "step": 4444 }, { "epoch": 1.1539460020768433, "grad_norm": 2.013911008834839, "learning_rate": 2.302699447752665e-06, "loss": 0.07676240801811218, "step": 4445 }, { "epoch": 1.1542056074766356, "grad_norm": 2.2841031551361084, "learning_rate": 2.3017723600726795e-06, "loss": 0.0714891254901886, "step": 4446 }, { "epoch": 1.154465212876428, "grad_norm": 2.654554605484009, "learning_rate": 2.3008453516290756e-06, "loss": 0.07674131542444229, "step": 4447 }, { "epoch": 1.1547248182762202, "grad_norm": 2.2191925048828125, "learning_rate": 2.29991842258573e-06, "loss": 0.0653316006064415, "step": 4448 }, { "epoch": 1.1549844236760125, "grad_norm": 2.390897035598755, "learning_rate": 2.2989915731065023e-06, "loss": 0.058184631168842316, "step": 4449 }, { "epoch": 1.1552440290758048, "grad_norm": 2.2712998390197754, "learning_rate": 2.2980648033552384e-06, "loss": 0.07860117405653, "step": 4450 }, { "epoch": 1.1555036344755971, "grad_norm": 4.204117774963379, "learning_rate": 2.297138113495768e-06, "loss": 0.02369306981563568, "step": 4451 }, { "epoch": 1.1557632398753894, "grad_norm": 2.776029586791992, "learning_rate": 2.2962115036919095e-06, "loss": 0.015183592215180397, "step": 4452 }, { "epoch": 1.1560228452751817, "grad_norm": 3.9894864559173584, "learning_rate": 2.295284974107468e-06, "loss": 0.04655511677265167, "step": 4453 }, { "epoch": 1.156282450674974, "grad_norm": 2.344750165939331, "learning_rate": 2.2943585249062316e-06, "loss": 0.0277310311794281, "step": 4454 }, { "epoch": 1.1565420560747663, "grad_norm": 2.1164863109588623, "learning_rate": 2.2934321562519746e-06, "loss": 0.08147585391998291, "step": 4455 }, { "epoch": 1.1568016614745587, "grad_norm": 3.1256229877471924, "learning_rate": 2.2925058683084582e-06, "loss": 0.06340473145246506, "step": 4456 }, { "epoch": 1.157061266874351, "grad_norm": 3.1706182956695557, "learning_rate": 2.291579661239429e-06, "loss": 0.030322827398777008, "step": 4457 }, { "epoch": 1.1573208722741433, "grad_norm": 3.038520097732544, "learning_rate": 2.2906535352086205e-06, "loss": 0.043515294790267944, "step": 4458 }, { "epoch": 1.1575804776739356, "grad_norm": 3.525665521621704, "learning_rate": 2.2897274903797484e-06, "loss": 0.04260371997952461, "step": 4459 }, { "epoch": 1.1578400830737279, "grad_norm": 1.5745519399642944, "learning_rate": 2.2888015269165166e-06, "loss": 0.08533485233783722, "step": 4460 }, { "epoch": 1.1580996884735202, "grad_norm": 2.087332248687744, "learning_rate": 2.2878756449826157e-06, "loss": 0.09092198312282562, "step": 4461 }, { "epoch": 1.1583592938733125, "grad_norm": 3.647822618484497, "learning_rate": 2.2869498447417193e-06, "loss": 0.03191101551055908, "step": 4462 }, { "epoch": 1.1586188992731048, "grad_norm": 3.008387565612793, "learning_rate": 2.286024126357488e-06, "loss": 0.01853969134390354, "step": 4463 }, { "epoch": 1.158878504672897, "grad_norm": 3.716867446899414, "learning_rate": 2.285098489993567e-06, "loss": 0.025738980621099472, "step": 4464 }, { "epoch": 1.1591381100726896, "grad_norm": 3.265855073928833, "learning_rate": 2.2841729358135885e-06, "loss": 0.03707259148359299, "step": 4465 }, { "epoch": 1.159397715472482, "grad_norm": 2.0920934677124023, "learning_rate": 2.2832474639811686e-06, "loss": 0.08420956134796143, "step": 4466 }, { "epoch": 1.1596573208722742, "grad_norm": 2.685943126678467, "learning_rate": 2.28232207465991e-06, "loss": 0.05328736454248428, "step": 4467 }, { "epoch": 1.1599169262720666, "grad_norm": 2.5881359577178955, "learning_rate": 2.281396768013401e-06, "loss": 0.014338669367134571, "step": 4468 }, { "epoch": 1.1601765316718589, "grad_norm": 1.927399754524231, "learning_rate": 2.280471544205213e-06, "loss": 0.07904684543609619, "step": 4469 }, { "epoch": 1.1604361370716512, "grad_norm": 2.437859535217285, "learning_rate": 2.279546403398905e-06, "loss": 0.051973748952150345, "step": 4470 }, { "epoch": 1.1606957424714435, "grad_norm": 2.371802568435669, "learning_rate": 2.2786213457580216e-06, "loss": 0.05746118351817131, "step": 4471 }, { "epoch": 1.1609553478712358, "grad_norm": 2.296238660812378, "learning_rate": 2.27769637144609e-06, "loss": 0.08462539315223694, "step": 4472 }, { "epoch": 1.161214953271028, "grad_norm": 2.8232836723327637, "learning_rate": 2.276771480626627e-06, "loss": 0.04802243039011955, "step": 4473 }, { "epoch": 1.1614745586708204, "grad_norm": 2.8759374618530273, "learning_rate": 2.27584667346313e-06, "loss": 0.0706041008234024, "step": 4474 }, { "epoch": 1.1617341640706127, "grad_norm": 3.2092785835266113, "learning_rate": 2.2749219501190846e-06, "loss": 0.04667139798402786, "step": 4475 }, { "epoch": 1.161993769470405, "grad_norm": 2.6461682319641113, "learning_rate": 2.2739973107579606e-06, "loss": 0.0754832997918129, "step": 4476 }, { "epoch": 1.1622533748701973, "grad_norm": 4.6263427734375, "learning_rate": 2.2730727555432135e-06, "loss": 0.047704316675662994, "step": 4477 }, { "epoch": 1.1625129802699896, "grad_norm": 2.807192802429199, "learning_rate": 2.272148284638282e-06, "loss": 0.08355361968278885, "step": 4478 }, { "epoch": 1.162772585669782, "grad_norm": 2.2177975177764893, "learning_rate": 2.271223898206593e-06, "loss": 0.0574372261762619, "step": 4479 }, { "epoch": 1.1630321910695742, "grad_norm": 2.0684585571289062, "learning_rate": 2.270299596411556e-06, "loss": 0.08176542818546295, "step": 4480 }, { "epoch": 1.1632917964693665, "grad_norm": 3.9405975341796875, "learning_rate": 2.269375379416568e-06, "loss": 0.044969938695430756, "step": 4481 }, { "epoch": 1.1635514018691588, "grad_norm": 2.2596540451049805, "learning_rate": 2.268451247385007e-06, "loss": 0.08041926473379135, "step": 4482 }, { "epoch": 1.1638110072689511, "grad_norm": 2.39083194732666, "learning_rate": 2.2675272004802383e-06, "loss": 0.07444576174020767, "step": 4483 }, { "epoch": 1.1640706126687435, "grad_norm": 1.8671376705169678, "learning_rate": 2.2666032388656144e-06, "loss": 0.014909441582858562, "step": 4484 }, { "epoch": 1.1643302180685358, "grad_norm": 4.381900787353516, "learning_rate": 2.265679362704471e-06, "loss": 0.04014618322253227, "step": 4485 }, { "epoch": 1.164589823468328, "grad_norm": 3.226823091506958, "learning_rate": 2.264755572160125e-06, "loss": 0.028473984450101852, "step": 4486 }, { "epoch": 1.1648494288681204, "grad_norm": 2.9973273277282715, "learning_rate": 2.2638318673958836e-06, "loss": 0.011462656781077385, "step": 4487 }, { "epoch": 1.1651090342679127, "grad_norm": 2.680384635925293, "learning_rate": 2.262908248575036e-06, "loss": 0.03139760345220566, "step": 4488 }, { "epoch": 1.165368639667705, "grad_norm": 4.698833465576172, "learning_rate": 2.261984715860858e-06, "loss": 0.030291613191366196, "step": 4489 }, { "epoch": 1.1656282450674973, "grad_norm": 2.341064214706421, "learning_rate": 2.261061269416608e-06, "loss": 0.08408170938491821, "step": 4490 }, { "epoch": 1.1658878504672896, "grad_norm": 2.5857391357421875, "learning_rate": 2.2601379094055297e-06, "loss": 0.01405596174299717, "step": 4491 }, { "epoch": 1.1661474558670821, "grad_norm": 2.0637214183807373, "learning_rate": 2.2592146359908534e-06, "loss": 0.05818536505103111, "step": 4492 }, { "epoch": 1.1664070612668744, "grad_norm": 2.8020026683807373, "learning_rate": 2.2582914493357912e-06, "loss": 0.06626610457897186, "step": 4493 }, { "epoch": 1.1666666666666667, "grad_norm": 2.7358052730560303, "learning_rate": 2.2573683496035424e-06, "loss": 0.042554207146167755, "step": 4494 }, { "epoch": 1.166926272066459, "grad_norm": 1.8767274618148804, "learning_rate": 2.2564453369572898e-06, "loss": 0.07952620089054108, "step": 4495 }, { "epoch": 1.1671858774662514, "grad_norm": 2.3556761741638184, "learning_rate": 2.255522411560201e-06, "loss": 0.07018589973449707, "step": 4496 }, { "epoch": 1.1674454828660437, "grad_norm": 2.4877047538757324, "learning_rate": 2.254599573575427e-06, "loss": 0.009985895827412605, "step": 4497 }, { "epoch": 1.167705088265836, "grad_norm": 2.2872538566589355, "learning_rate": 2.2536768231661053e-06, "loss": 0.02892816625535488, "step": 4498 }, { "epoch": 1.1679646936656283, "grad_norm": 2.7543728351593018, "learning_rate": 2.252754160495357e-06, "loss": 0.03127003461122513, "step": 4499 }, { "epoch": 1.1682242990654206, "grad_norm": 4.039117336273193, "learning_rate": 2.251831585726289e-06, "loss": 0.03269239887595177, "step": 4500 }, { "epoch": 1.1684839044652129, "grad_norm": 2.3115596771240234, "learning_rate": 2.250909099021989e-06, "loss": 0.06026250869035721, "step": 4501 }, { "epoch": 1.1687435098650052, "grad_norm": 2.139618396759033, "learning_rate": 2.249986700545533e-06, "loss": 0.07195413112640381, "step": 4502 }, { "epoch": 1.1690031152647975, "grad_norm": 3.3174774646759033, "learning_rate": 2.2490643904599787e-06, "loss": 0.042282093316316605, "step": 4503 }, { "epoch": 1.1692627206645898, "grad_norm": 2.993777275085449, "learning_rate": 2.248142168928371e-06, "loss": 0.06464832276105881, "step": 4504 }, { "epoch": 1.1695223260643821, "grad_norm": 3.7307417392730713, "learning_rate": 2.247220036113737e-06, "loss": 0.035316988825798035, "step": 4505 }, { "epoch": 1.1697819314641744, "grad_norm": 4.217886447906494, "learning_rate": 2.2462979921790877e-06, "loss": 0.04935617372393608, "step": 4506 }, { "epoch": 1.1700415368639667, "grad_norm": 3.510629653930664, "learning_rate": 2.2453760372874206e-06, "loss": 0.04093783721327782, "step": 4507 }, { "epoch": 1.170301142263759, "grad_norm": 1.9352786540985107, "learning_rate": 2.2444541716017167e-06, "loss": 0.007603546604514122, "step": 4508 }, { "epoch": 1.1705607476635513, "grad_norm": 1.9282722473144531, "learning_rate": 2.243532395284938e-06, "loss": 0.06944561004638672, "step": 4509 }, { "epoch": 1.1708203530633436, "grad_norm": 2.4355838298797607, "learning_rate": 2.242610708500036e-06, "loss": 0.015735303983092308, "step": 4510 }, { "epoch": 1.171079958463136, "grad_norm": 3.7979722023010254, "learning_rate": 2.2416891114099433e-06, "loss": 0.055483706295490265, "step": 4511 }, { "epoch": 1.1713395638629283, "grad_norm": 3.361171007156372, "learning_rate": 2.240767604177577e-06, "loss": 0.055319372564554214, "step": 4512 }, { "epoch": 1.1715991692627208, "grad_norm": 2.5850253105163574, "learning_rate": 2.2398461869658384e-06, "loss": 0.0648145079612732, "step": 4513 }, { "epoch": 1.171858774662513, "grad_norm": 3.2867748737335205, "learning_rate": 2.2389248599376125e-06, "loss": 0.030720196664333344, "step": 4514 }, { "epoch": 1.1721183800623054, "grad_norm": 3.2380142211914062, "learning_rate": 2.238003623255769e-06, "loss": 0.03282453492283821, "step": 4515 }, { "epoch": 1.1723779854620977, "grad_norm": 2.7324461936950684, "learning_rate": 2.2370824770831634e-06, "loss": 0.009287801571190357, "step": 4516 }, { "epoch": 1.17263759086189, "grad_norm": 3.8523457050323486, "learning_rate": 2.23616142158263e-06, "loss": 0.044677603989839554, "step": 4517 }, { "epoch": 1.1728971962616823, "grad_norm": 2.2621538639068604, "learning_rate": 2.2352404569169924e-06, "loss": 0.08593153953552246, "step": 4518 }, { "epoch": 1.1731568016614746, "grad_norm": 2.5003840923309326, "learning_rate": 2.234319583249056e-06, "loss": 0.04798021912574768, "step": 4519 }, { "epoch": 1.173416407061267, "grad_norm": 3.5307695865631104, "learning_rate": 2.233398800741609e-06, "loss": 0.02706829458475113, "step": 4520 }, { "epoch": 1.1736760124610592, "grad_norm": 3.7921712398529053, "learning_rate": 2.2324781095574255e-06, "loss": 0.03224397450685501, "step": 4521 }, { "epoch": 1.1739356178608515, "grad_norm": 3.7066125869750977, "learning_rate": 2.2315575098592627e-06, "loss": 0.04353439435362816, "step": 4522 }, { "epoch": 1.1741952232606438, "grad_norm": 3.4816319942474365, "learning_rate": 2.2306370018098623e-06, "loss": 0.06321541965007782, "step": 4523 }, { "epoch": 1.1744548286604362, "grad_norm": 2.7382078170776367, "learning_rate": 2.229716585571947e-06, "loss": 0.019636183977127075, "step": 4524 }, { "epoch": 1.1747144340602285, "grad_norm": 3.9539482593536377, "learning_rate": 2.228796261308227e-06, "loss": 0.0282999686896801, "step": 4525 }, { "epoch": 1.1749740394600208, "grad_norm": 3.6133575439453125, "learning_rate": 2.2278760291813934e-06, "loss": 0.02680250257253647, "step": 4526 }, { "epoch": 1.175233644859813, "grad_norm": 3.711164712905884, "learning_rate": 2.2269558893541237e-06, "loss": 0.024243663996458054, "step": 4527 }, { "epoch": 1.1754932502596054, "grad_norm": 3.8971033096313477, "learning_rate": 2.226035841989076e-06, "loss": 0.015171146020293236, "step": 4528 }, { "epoch": 1.1757528556593977, "grad_norm": 3.930399179458618, "learning_rate": 2.225115887248894e-06, "loss": 0.05865546315908432, "step": 4529 }, { "epoch": 1.17601246105919, "grad_norm": 2.824730157852173, "learning_rate": 2.2241960252962054e-06, "loss": 0.06818237900733948, "step": 4530 }, { "epoch": 1.1762720664589823, "grad_norm": 3.654942512512207, "learning_rate": 2.2232762562936204e-06, "loss": 0.03966495394706726, "step": 4531 }, { "epoch": 1.1765316718587746, "grad_norm": 1.809503436088562, "learning_rate": 2.222356580403733e-06, "loss": 0.08220353722572327, "step": 4532 }, { "epoch": 1.176791277258567, "grad_norm": 1.9341459274291992, "learning_rate": 2.22143699778912e-06, "loss": 0.08533615618944168, "step": 4533 }, { "epoch": 1.1770508826583592, "grad_norm": 2.505742073059082, "learning_rate": 2.2205175086123436e-06, "loss": 0.057092517614364624, "step": 4534 }, { "epoch": 1.1773104880581515, "grad_norm": 3.2426600456237793, "learning_rate": 2.2195981130359482e-06, "loss": 0.04182085767388344, "step": 4535 }, { "epoch": 1.1775700934579438, "grad_norm": 3.9999465942382812, "learning_rate": 2.218678811222462e-06, "loss": 0.01872595213353634, "step": 4536 }, { "epoch": 1.1778296988577361, "grad_norm": 2.516772508621216, "learning_rate": 2.217759603334396e-06, "loss": 0.025328299030661583, "step": 4537 }, { "epoch": 1.1780893042575284, "grad_norm": 3.070708990097046, "learning_rate": 2.2168404895342456e-06, "loss": 0.05274658277630806, "step": 4538 }, { "epoch": 1.1783489096573208, "grad_norm": 3.2803025245666504, "learning_rate": 2.2159214699844895e-06, "loss": 0.021259864792227745, "step": 4539 }, { "epoch": 1.1786085150571133, "grad_norm": 1.832465648651123, "learning_rate": 2.2150025448475876e-06, "loss": 0.01048965472728014, "step": 4540 }, { "epoch": 1.1788681204569056, "grad_norm": 2.8314266204833984, "learning_rate": 2.214083714285987e-06, "loss": 0.05172732099890709, "step": 4541 }, { "epoch": 1.179127725856698, "grad_norm": 3.040485382080078, "learning_rate": 2.213164978462114e-06, "loss": 0.06014919653534889, "step": 4542 }, { "epoch": 1.1793873312564902, "grad_norm": 3.2230215072631836, "learning_rate": 2.212246337538382e-06, "loss": 0.030332107096910477, "step": 4543 }, { "epoch": 1.1796469366562825, "grad_norm": 3.2542686462402344, "learning_rate": 2.211327791677183e-06, "loss": 0.046941157430410385, "step": 4544 }, { "epoch": 1.1799065420560748, "grad_norm": 4.319819450378418, "learning_rate": 2.2104093410408972e-06, "loss": 0.06291685253381729, "step": 4545 }, { "epoch": 1.1801661474558671, "grad_norm": 3.2220468521118164, "learning_rate": 2.209490985791886e-06, "loss": 0.06109564006328583, "step": 4546 }, { "epoch": 1.1804257528556594, "grad_norm": 3.5775041580200195, "learning_rate": 2.2085727260924916e-06, "loss": 0.07164603471755981, "step": 4547 }, { "epoch": 1.1806853582554517, "grad_norm": 4.059387683868408, "learning_rate": 2.2076545621050416e-06, "loss": 0.04194263368844986, "step": 4548 }, { "epoch": 1.180944963655244, "grad_norm": 2.9126076698303223, "learning_rate": 2.2067364939918473e-06, "loss": 0.041970428079366684, "step": 4549 }, { "epoch": 1.1812045690550363, "grad_norm": 2.488351345062256, "learning_rate": 2.205818521915203e-06, "loss": 0.09207381308078766, "step": 4550 }, { "epoch": 1.1814641744548287, "grad_norm": 2.2852025032043457, "learning_rate": 2.2049006460373824e-06, "loss": 0.0993281900882721, "step": 4551 }, { "epoch": 1.181723779854621, "grad_norm": 3.936722755432129, "learning_rate": 2.203982866520647e-06, "loss": 0.03910240903496742, "step": 4552 }, { "epoch": 1.1819833852544133, "grad_norm": 2.3917579650878906, "learning_rate": 2.2030651835272383e-06, "loss": 0.0741109549999237, "step": 4553 }, { "epoch": 1.1822429906542056, "grad_norm": 3.0544393062591553, "learning_rate": 2.202147597219383e-06, "loss": 0.03938274085521698, "step": 4554 }, { "epoch": 1.1825025960539979, "grad_norm": 2.54766845703125, "learning_rate": 2.2012301077592874e-06, "loss": 0.012821391224861145, "step": 4555 }, { "epoch": 1.1827622014537902, "grad_norm": 3.1283397674560547, "learning_rate": 2.2003127153091437e-06, "loss": 0.027732573449611664, "step": 4556 }, { "epoch": 1.1830218068535825, "grad_norm": 3.521991491317749, "learning_rate": 2.199395420031126e-06, "loss": 0.020428668707609177, "step": 4557 }, { "epoch": 1.1832814122533748, "grad_norm": 2.3055579662323, "learning_rate": 2.198478222087391e-06, "loss": 0.06954541802406311, "step": 4558 }, { "epoch": 1.183541017653167, "grad_norm": 3.6224536895751953, "learning_rate": 2.1975611216400776e-06, "loss": 0.022913828492164612, "step": 4559 }, { "epoch": 1.1838006230529594, "grad_norm": 1.7502357959747314, "learning_rate": 2.196644118851309e-06, "loss": 0.009090780280530453, "step": 4560 }, { "epoch": 1.184060228452752, "grad_norm": 2.8066296577453613, "learning_rate": 2.19572721388319e-06, "loss": 0.10407833755016327, "step": 4561 }, { "epoch": 1.1843198338525442, "grad_norm": 2.627497911453247, "learning_rate": 2.194810406897809e-06, "loss": 0.028598081320524216, "step": 4562 }, { "epoch": 1.1845794392523366, "grad_norm": 2.761880874633789, "learning_rate": 2.193893698057235e-06, "loss": 0.030624669045209885, "step": 4563 }, { "epoch": 1.1848390446521289, "grad_norm": 3.82369327545166, "learning_rate": 2.1929770875235227e-06, "loss": 0.05675991624593735, "step": 4564 }, { "epoch": 1.1850986500519212, "grad_norm": 3.172461748123169, "learning_rate": 2.1920605754587073e-06, "loss": 0.01649491861462593, "step": 4565 }, { "epoch": 1.1853582554517135, "grad_norm": 3.3904170989990234, "learning_rate": 2.1911441620248066e-06, "loss": 0.03333000838756561, "step": 4566 }, { "epoch": 1.1856178608515058, "grad_norm": 3.7068822383880615, "learning_rate": 2.1902278473838217e-06, "loss": 0.02074027992784977, "step": 4567 }, { "epoch": 1.185877466251298, "grad_norm": 2.691845178604126, "learning_rate": 2.1893116316977365e-06, "loss": 0.010314678773283958, "step": 4568 }, { "epoch": 1.1861370716510904, "grad_norm": 1.7568490505218506, "learning_rate": 2.188395515128517e-06, "loss": 0.07311321794986725, "step": 4569 }, { "epoch": 1.1863966770508827, "grad_norm": 3.8075103759765625, "learning_rate": 2.1874794978381114e-06, "loss": 0.008374836295843124, "step": 4570 }, { "epoch": 1.186656282450675, "grad_norm": 1.7106986045837402, "learning_rate": 2.1865635799884507e-06, "loss": 0.0690668523311615, "step": 4571 }, { "epoch": 1.1869158878504673, "grad_norm": 3.6090824604034424, "learning_rate": 2.185647761741447e-06, "loss": 0.030636869370937347, "step": 4572 }, { "epoch": 1.1871754932502596, "grad_norm": 2.4058494567871094, "learning_rate": 2.184732043258998e-06, "loss": 0.021849066019058228, "step": 4573 }, { "epoch": 1.187435098650052, "grad_norm": 4.583136081695557, "learning_rate": 2.183816424702982e-06, "loss": 0.02609003148972988, "step": 4574 }, { "epoch": 1.1876947040498442, "grad_norm": 3.047398328781128, "learning_rate": 2.1829009062352558e-06, "loss": 0.02200957201421261, "step": 4575 }, { "epoch": 1.1879543094496365, "grad_norm": 1.7845977544784546, "learning_rate": 2.181985488017666e-06, "loss": 0.006542894523590803, "step": 4576 }, { "epoch": 1.1882139148494288, "grad_norm": 3.9719719886779785, "learning_rate": 2.1810701702120367e-06, "loss": 0.04094889014959335, "step": 4577 }, { "epoch": 1.1884735202492211, "grad_norm": 4.017845630645752, "learning_rate": 2.180154952980173e-06, "loss": 0.03365502506494522, "step": 4578 }, { "epoch": 1.1887331256490135, "grad_norm": 3.565451145172119, "learning_rate": 2.179239836483866e-06, "loss": 0.049741700291633606, "step": 4579 }, { "epoch": 1.1889927310488058, "grad_norm": 3.7799675464630127, "learning_rate": 2.178324820884888e-06, "loss": 0.024014757946133614, "step": 4580 }, { "epoch": 1.189252336448598, "grad_norm": 4.45000696182251, "learning_rate": 2.177409906344992e-06, "loss": 0.03649711608886719, "step": 4581 }, { "epoch": 1.1895119418483904, "grad_norm": 2.8912057876586914, "learning_rate": 2.1764950930259143e-06, "loss": 0.06983962655067444, "step": 4582 }, { "epoch": 1.1897715472481827, "grad_norm": 5.353612899780273, "learning_rate": 2.1755803810893716e-06, "loss": 0.030497010797262192, "step": 4583 }, { "epoch": 1.190031152647975, "grad_norm": 3.186729907989502, "learning_rate": 2.174665770697066e-06, "loss": 0.049246784299612045, "step": 4584 }, { "epoch": 1.1902907580477673, "grad_norm": 3.9544739723205566, "learning_rate": 2.1737512620106784e-06, "loss": 0.055373214185237885, "step": 4585 }, { "epoch": 1.1905503634475596, "grad_norm": 2.5055646896362305, "learning_rate": 2.1728368551918726e-06, "loss": 0.06725423783063889, "step": 4586 }, { "epoch": 1.190809968847352, "grad_norm": 3.1145570278167725, "learning_rate": 2.1719225504022968e-06, "loss": 0.022900138050317764, "step": 4587 }, { "epoch": 1.1910695742471444, "grad_norm": 3.436864137649536, "learning_rate": 2.1710083478035763e-06, "loss": 0.04368378221988678, "step": 4588 }, { "epoch": 1.1913291796469367, "grad_norm": 2.5564887523651123, "learning_rate": 2.1700942475573246e-06, "loss": 0.03894292935729027, "step": 4589 }, { "epoch": 1.191588785046729, "grad_norm": 3.374762535095215, "learning_rate": 2.169180249825131e-06, "loss": 0.06541788578033447, "step": 4590 }, { "epoch": 1.1918483904465214, "grad_norm": 2.2797250747680664, "learning_rate": 2.1682663547685695e-06, "loss": 0.01462529320269823, "step": 4591 }, { "epoch": 1.1921079958463137, "grad_norm": 3.2579033374786377, "learning_rate": 2.167352562549197e-06, "loss": 0.05069754272699356, "step": 4592 }, { "epoch": 1.192367601246106, "grad_norm": 3.7031729221343994, "learning_rate": 2.1664388733285514e-06, "loss": 0.04152808338403702, "step": 4593 }, { "epoch": 1.1926272066458983, "grad_norm": 3.8858954906463623, "learning_rate": 2.1655252872681505e-06, "loss": 0.032040417194366455, "step": 4594 }, { "epoch": 1.1928868120456906, "grad_norm": 1.8304866552352905, "learning_rate": 2.164611804529496e-06, "loss": 0.08305656909942627, "step": 4595 }, { "epoch": 1.1931464174454829, "grad_norm": 3.0392796993255615, "learning_rate": 2.1636984252740705e-06, "loss": 0.019996900111436844, "step": 4596 }, { "epoch": 1.1934060228452752, "grad_norm": 3.4654881954193115, "learning_rate": 2.16278514966334e-06, "loss": 0.02129683457314968, "step": 4597 }, { "epoch": 1.1936656282450675, "grad_norm": 2.8744022846221924, "learning_rate": 2.161871977858749e-06, "loss": 0.06516749411821365, "step": 4598 }, { "epoch": 1.1939252336448598, "grad_norm": 2.4922614097595215, "learning_rate": 2.160958910021726e-06, "loss": 0.0705065056681633, "step": 4599 }, { "epoch": 1.1941848390446521, "grad_norm": 1.8472185134887695, "learning_rate": 2.16004594631368e-06, "loss": 0.08014459908008575, "step": 4600 }, { "epoch": 1.1944444444444444, "grad_norm": 3.48687744140625, "learning_rate": 2.1591330868960035e-06, "loss": 0.0565713606774807, "step": 4601 }, { "epoch": 1.1947040498442367, "grad_norm": 3.5680994987487793, "learning_rate": 2.1582203319300677e-06, "loss": 0.08030609786510468, "step": 4602 }, { "epoch": 1.194963655244029, "grad_norm": 4.552304267883301, "learning_rate": 2.157307681577227e-06, "loss": 0.03309495002031326, "step": 4603 }, { "epoch": 1.1952232606438213, "grad_norm": 2.470759630203247, "learning_rate": 2.1563951359988183e-06, "loss": 0.08706341683864594, "step": 4604 }, { "epoch": 1.1954828660436136, "grad_norm": 2.0277628898620605, "learning_rate": 2.155482695356157e-06, "loss": 0.08069238811731339, "step": 4605 }, { "epoch": 1.195742471443406, "grad_norm": 2.803269386291504, "learning_rate": 2.154570359810542e-06, "loss": 0.017682570964097977, "step": 4606 }, { "epoch": 1.1960020768431983, "grad_norm": 3.667281150817871, "learning_rate": 2.1536581295232545e-06, "loss": 0.02389104850590229, "step": 4607 }, { "epoch": 1.1962616822429906, "grad_norm": 2.639784574508667, "learning_rate": 2.152746004655556e-06, "loss": 0.0653407871723175, "step": 4608 }, { "epoch": 1.196521287642783, "grad_norm": 3.939415216445923, "learning_rate": 2.1518339853686874e-06, "loss": 0.04019417241215706, "step": 4609 }, { "epoch": 1.1967808930425754, "grad_norm": 2.480598211288452, "learning_rate": 2.1509220718238744e-06, "loss": 0.08796578645706177, "step": 4610 }, { "epoch": 1.1970404984423677, "grad_norm": 3.7538182735443115, "learning_rate": 2.150010264182322e-06, "loss": 0.030811304226517677, "step": 4611 }, { "epoch": 1.19730010384216, "grad_norm": 2.0403850078582764, "learning_rate": 2.1490985626052173e-06, "loss": 0.10123113542795181, "step": 4612 }, { "epoch": 1.1975597092419523, "grad_norm": 3.4288573265075684, "learning_rate": 2.1481869672537277e-06, "loss": 0.014315614476799965, "step": 4613 }, { "epoch": 1.1978193146417446, "grad_norm": 3.2656216621398926, "learning_rate": 2.147275478289002e-06, "loss": 0.026770738884806633, "step": 4614 }, { "epoch": 1.198078920041537, "grad_norm": 3.5035488605499268, "learning_rate": 2.1463640958721714e-06, "loss": 0.02267238311469555, "step": 4615 }, { "epoch": 1.1983385254413292, "grad_norm": 2.886690855026245, "learning_rate": 2.145452820164348e-06, "loss": 0.023263394832611084, "step": 4616 }, { "epoch": 1.1985981308411215, "grad_norm": 2.579080820083618, "learning_rate": 2.144541651326623e-06, "loss": 0.044429961591959, "step": 4617 }, { "epoch": 1.1988577362409139, "grad_norm": 2.085698366165161, "learning_rate": 2.1436305895200715e-06, "loss": 0.07075107097625732, "step": 4618 }, { "epoch": 1.1991173416407062, "grad_norm": 3.0175812244415283, "learning_rate": 2.142719634905747e-06, "loss": 0.03415268659591675, "step": 4619 }, { "epoch": 1.1993769470404985, "grad_norm": 2.13029146194458, "learning_rate": 2.1418087876446867e-06, "loss": 0.09531019628047943, "step": 4620 }, { "epoch": 1.1996365524402908, "grad_norm": 3.7163689136505127, "learning_rate": 2.140898047897907e-06, "loss": 0.038833778351545334, "step": 4621 }, { "epoch": 1.199896157840083, "grad_norm": 1.6863912343978882, "learning_rate": 2.139987415826405e-06, "loss": 0.07715078443288803, "step": 4622 }, { "epoch": 1.2001557632398754, "grad_norm": 2.64332914352417, "learning_rate": 2.1390768915911614e-06, "loss": 0.03965909406542778, "step": 4623 }, { "epoch": 1.2004153686396677, "grad_norm": 2.4963529109954834, "learning_rate": 2.138166475353135e-06, "loss": 0.028355378657579422, "step": 4624 }, { "epoch": 1.20067497403946, "grad_norm": 2.5387673377990723, "learning_rate": 2.137256167273266e-06, "loss": 0.07002110779285431, "step": 4625 }, { "epoch": 1.2009345794392523, "grad_norm": 2.448838472366333, "learning_rate": 2.136345967512477e-06, "loss": 0.06540712714195251, "step": 4626 }, { "epoch": 1.2011941848390446, "grad_norm": 2.654792070388794, "learning_rate": 2.1354358762316703e-06, "loss": 0.06029568240046501, "step": 4627 }, { "epoch": 1.201453790238837, "grad_norm": 2.965733289718628, "learning_rate": 2.134525893591729e-06, "loss": 0.039940547198057175, "step": 4628 }, { "epoch": 1.2017133956386292, "grad_norm": 2.4975972175598145, "learning_rate": 2.133616019753517e-06, "loss": 0.07427572458982468, "step": 4629 }, { "epoch": 1.2019730010384215, "grad_norm": 2.4099230766296387, "learning_rate": 2.1327062548778797e-06, "loss": 0.06222376227378845, "step": 4630 }, { "epoch": 1.2022326064382138, "grad_norm": 3.045698642730713, "learning_rate": 2.131796599125643e-06, "loss": 0.020918095484375954, "step": 4631 }, { "epoch": 1.2024922118380061, "grad_norm": 3.344759702682495, "learning_rate": 2.1308870526576112e-06, "loss": 0.02995271049439907, "step": 4632 }, { "epoch": 1.2027518172377984, "grad_norm": 3.1667847633361816, "learning_rate": 2.129977615634573e-06, "loss": 0.07225004583597183, "step": 4633 }, { "epoch": 1.2030114226375908, "grad_norm": 4.719912528991699, "learning_rate": 2.129068288217296e-06, "loss": 0.038538068532943726, "step": 4634 }, { "epoch": 1.203271028037383, "grad_norm": 3.4285762310028076, "learning_rate": 2.128159070566529e-06, "loss": 0.023818861693143845, "step": 4635 }, { "epoch": 1.2035306334371756, "grad_norm": 3.0743868350982666, "learning_rate": 2.1272499628429993e-06, "loss": 0.035040415823459625, "step": 4636 }, { "epoch": 1.203790238836968, "grad_norm": 2.208468437194824, "learning_rate": 2.1263409652074165e-06, "loss": 0.06856332719326019, "step": 4637 }, { "epoch": 1.2040498442367602, "grad_norm": 3.3824591636657715, "learning_rate": 2.125432077820471e-06, "loss": 0.026795629411935806, "step": 4638 }, { "epoch": 1.2043094496365525, "grad_norm": 2.0020854473114014, "learning_rate": 2.1245233008428355e-06, "loss": 0.01226736605167389, "step": 4639 }, { "epoch": 1.2045690550363448, "grad_norm": 2.3842697143554688, "learning_rate": 2.123614634435158e-06, "loss": 0.08350586146116257, "step": 4640 }, { "epoch": 1.2048286604361371, "grad_norm": 3.0761654376983643, "learning_rate": 2.1227060787580693e-06, "loss": 0.061422981321811676, "step": 4641 }, { "epoch": 1.2050882658359294, "grad_norm": 3.1941921710968018, "learning_rate": 2.1217976339721837e-06, "loss": 0.045014917850494385, "step": 4642 }, { "epoch": 1.2053478712357217, "grad_norm": 4.729218006134033, "learning_rate": 2.120889300238093e-06, "loss": 0.030927076935768127, "step": 4643 }, { "epoch": 1.205607476635514, "grad_norm": 2.5793707370758057, "learning_rate": 2.1199810777163687e-06, "loss": 0.04562150314450264, "step": 4644 }, { "epoch": 1.2058670820353063, "grad_norm": 4.364323616027832, "learning_rate": 2.1190729665675647e-06, "loss": 0.026652663946151733, "step": 4645 }, { "epoch": 1.2061266874350987, "grad_norm": 1.9230010509490967, "learning_rate": 2.1181649669522138e-06, "loss": 0.08285690099000931, "step": 4646 }, { "epoch": 1.206386292834891, "grad_norm": 3.351991653442383, "learning_rate": 2.1172570790308293e-06, "loss": 0.03886517882347107, "step": 4647 }, { "epoch": 1.2066458982346833, "grad_norm": 3.9104232788085938, "learning_rate": 2.1163493029639056e-06, "loss": 0.030564555898308754, "step": 4648 }, { "epoch": 1.2069055036344756, "grad_norm": 3.4700822830200195, "learning_rate": 2.1154416389119163e-06, "loss": 0.03374768793582916, "step": 4649 }, { "epoch": 1.2071651090342679, "grad_norm": 3.3518688678741455, "learning_rate": 2.114534087035316e-06, "loss": 0.03497789427638054, "step": 4650 }, { "epoch": 1.2074247144340602, "grad_norm": 2.3503708839416504, "learning_rate": 2.1136266474945387e-06, "loss": 0.08378276973962784, "step": 4651 }, { "epoch": 1.2076843198338525, "grad_norm": 3.4487016201019287, "learning_rate": 2.112719320449999e-06, "loss": 0.009267288260161877, "step": 4652 }, { "epoch": 1.2079439252336448, "grad_norm": 3.0582761764526367, "learning_rate": 2.111812106062092e-06, "loss": 0.061868201941251755, "step": 4653 }, { "epoch": 1.208203530633437, "grad_norm": 3.8716940879821777, "learning_rate": 2.1109050044911922e-06, "loss": 0.046329207718372345, "step": 4654 }, { "epoch": 1.2084631360332294, "grad_norm": 3.460245370864868, "learning_rate": 2.109998015897654e-06, "loss": 0.029585421085357666, "step": 4655 }, { "epoch": 1.2087227414330217, "grad_norm": 3.9877676963806152, "learning_rate": 2.1090911404418123e-06, "loss": 0.0209265798330307, "step": 4656 }, { "epoch": 1.208982346832814, "grad_norm": 3.679166793823242, "learning_rate": 2.108184378283982e-06, "loss": 0.017580699175596237, "step": 4657 }, { "epoch": 1.2092419522326066, "grad_norm": 3.741865396499634, "learning_rate": 2.107277729584459e-06, "loss": 0.028796762228012085, "step": 4658 }, { "epoch": 1.2095015576323989, "grad_norm": 2.375457525253296, "learning_rate": 2.106371194503516e-06, "loss": 0.09299987554550171, "step": 4659 }, { "epoch": 1.2097611630321912, "grad_norm": 3.532837152481079, "learning_rate": 2.1054647732014088e-06, "loss": 0.03360908478498459, "step": 4660 }, { "epoch": 1.2100207684319835, "grad_norm": 3.3797547817230225, "learning_rate": 2.1045584658383718e-06, "loss": 0.06607216596603394, "step": 4661 }, { "epoch": 1.2102803738317758, "grad_norm": 2.5026137828826904, "learning_rate": 2.1036522725746207e-06, "loss": 0.03631557524204254, "step": 4662 }, { "epoch": 1.210539979231568, "grad_norm": 1.8805259466171265, "learning_rate": 2.102746193570346e-06, "loss": 0.07074787467718124, "step": 4663 }, { "epoch": 1.2107995846313604, "grad_norm": 4.607743740081787, "learning_rate": 2.1018402289857254e-06, "loss": 0.02823711559176445, "step": 4664 }, { "epoch": 1.2110591900311527, "grad_norm": 3.358664035797119, "learning_rate": 2.1009343789809116e-06, "loss": 0.03671613335609436, "step": 4665 }, { "epoch": 1.211318795430945, "grad_norm": 3.0151607990264893, "learning_rate": 2.1000286437160385e-06, "loss": 0.036657869815826416, "step": 4666 }, { "epoch": 1.2115784008307373, "grad_norm": 2.050858974456787, "learning_rate": 2.0991230233512177e-06, "loss": 0.06324160099029541, "step": 4667 }, { "epoch": 1.2118380062305296, "grad_norm": 3.3921360969543457, "learning_rate": 2.098217518046543e-06, "loss": 0.01746555231511593, "step": 4668 }, { "epoch": 1.212097611630322, "grad_norm": 2.553285598754883, "learning_rate": 2.097312127962088e-06, "loss": 0.06203678995370865, "step": 4669 }, { "epoch": 1.2123572170301142, "grad_norm": 1.1914259195327759, "learning_rate": 2.0964068532579047e-06, "loss": 0.005008588545024395, "step": 4670 }, { "epoch": 1.2126168224299065, "grad_norm": 2.0384092330932617, "learning_rate": 2.095501694094024e-06, "loss": 0.08389470726251602, "step": 4671 }, { "epoch": 1.2128764278296988, "grad_norm": 2.7666304111480713, "learning_rate": 2.094596650630457e-06, "loss": 0.05466676875948906, "step": 4672 }, { "epoch": 1.2131360332294912, "grad_norm": 3.8312981128692627, "learning_rate": 2.093691723027196e-06, "loss": 0.04415527731180191, "step": 4673 }, { "epoch": 1.2133956386292835, "grad_norm": 3.405071496963501, "learning_rate": 2.0927869114442118e-06, "loss": 0.04818575829267502, "step": 4674 }, { "epoch": 1.2136552440290758, "grad_norm": 3.2249886989593506, "learning_rate": 2.091882216041452e-06, "loss": 0.018262768164277077, "step": 4675 }, { "epoch": 1.213914849428868, "grad_norm": 2.095621347427368, "learning_rate": 2.090977636978848e-06, "loss": 0.0969960018992424, "step": 4676 }, { "epoch": 1.2141744548286604, "grad_norm": 3.574854850769043, "learning_rate": 2.090073174416308e-06, "loss": 0.04166784882545471, "step": 4677 }, { "epoch": 1.2144340602284527, "grad_norm": 2.4894261360168457, "learning_rate": 2.089168828513722e-06, "loss": 0.05439288169145584, "step": 4678 }, { "epoch": 1.214693665628245, "grad_norm": 2.6670424938201904, "learning_rate": 2.0882645994309543e-06, "loss": 0.0827186182141304, "step": 4679 }, { "epoch": 1.2149532710280373, "grad_norm": 3.8996145725250244, "learning_rate": 2.0873604873278543e-06, "loss": 0.02729393169283867, "step": 4680 }, { "epoch": 1.2152128764278296, "grad_norm": 2.6271843910217285, "learning_rate": 2.086456492364247e-06, "loss": 0.04907097667455673, "step": 4681 }, { "epoch": 1.215472481827622, "grad_norm": 3.0136523246765137, "learning_rate": 2.08555261469994e-06, "loss": 0.012590830214321613, "step": 4682 }, { "epoch": 1.2157320872274142, "grad_norm": 3.289062261581421, "learning_rate": 2.084648854494716e-06, "loss": 0.016356119886040688, "step": 4683 }, { "epoch": 1.2159916926272065, "grad_norm": 2.5384035110473633, "learning_rate": 2.08374521190834e-06, "loss": 0.0820583775639534, "step": 4684 }, { "epoch": 1.216251298026999, "grad_norm": 2.8925564289093018, "learning_rate": 2.082841687100556e-06, "loss": 0.032805945724248886, "step": 4685 }, { "epoch": 1.2165109034267914, "grad_norm": 1.925325632095337, "learning_rate": 2.0819382802310857e-06, "loss": 0.08433543890714645, "step": 4686 }, { "epoch": 1.2167705088265837, "grad_norm": 2.3702051639556885, "learning_rate": 2.08103499145963e-06, "loss": 0.07177669554948807, "step": 4687 }, { "epoch": 1.217030114226376, "grad_norm": 2.412827253341675, "learning_rate": 2.080131820945871e-06, "loss": 0.018510840833187103, "step": 4688 }, { "epoch": 1.2172897196261683, "grad_norm": 3.163238525390625, "learning_rate": 2.0792287688494683e-06, "loss": 0.029035136103630066, "step": 4689 }, { "epoch": 1.2175493250259606, "grad_norm": 3.1316511631011963, "learning_rate": 2.0783258353300607e-06, "loss": 0.023712242022156715, "step": 4690 }, { "epoch": 1.217808930425753, "grad_norm": 2.382267713546753, "learning_rate": 2.0774230205472657e-06, "loss": 0.050275735557079315, "step": 4691 }, { "epoch": 1.2180685358255452, "grad_norm": 2.004988193511963, "learning_rate": 2.0765203246606806e-06, "loss": 0.017145227640867233, "step": 4692 }, { "epoch": 1.2183281412253375, "grad_norm": 3.1766934394836426, "learning_rate": 2.0756177478298824e-06, "loss": 0.05350115895271301, "step": 4693 }, { "epoch": 1.2185877466251298, "grad_norm": 2.3407795429229736, "learning_rate": 2.0747152902144243e-06, "loss": 0.018558405339717865, "step": 4694 }, { "epoch": 1.2188473520249221, "grad_norm": 2.7947542667388916, "learning_rate": 2.073812951973841e-06, "loss": 0.0523083433508873, "step": 4695 }, { "epoch": 1.2191069574247144, "grad_norm": 3.3673787117004395, "learning_rate": 2.072910733267645e-06, "loss": 0.013108128681778908, "step": 4696 }, { "epoch": 1.2193665628245067, "grad_norm": 2.8329083919525146, "learning_rate": 2.072008634255329e-06, "loss": 0.024204859510064125, "step": 4697 }, { "epoch": 1.219626168224299, "grad_norm": 3.1113948822021484, "learning_rate": 2.0711066550963615e-06, "loss": 0.03726591914892197, "step": 4698 }, { "epoch": 1.2198857736240913, "grad_norm": 3.374159336090088, "learning_rate": 2.070204795950193e-06, "loss": 0.03600292652845383, "step": 4699 }, { "epoch": 1.2201453790238836, "grad_norm": 3.0922155380249023, "learning_rate": 2.069303056976251e-06, "loss": 0.08074604719877243, "step": 4700 }, { "epoch": 1.220404984423676, "grad_norm": 3.0306966304779053, "learning_rate": 2.0684014383339447e-06, "loss": 0.01907735876739025, "step": 4701 }, { "epoch": 1.2206645898234683, "grad_norm": 3.215867280960083, "learning_rate": 2.0674999401826558e-06, "loss": 0.04202662408351898, "step": 4702 }, { "epoch": 1.2209241952232606, "grad_norm": 2.412081480026245, "learning_rate": 2.0665985626817496e-06, "loss": 0.08663436770439148, "step": 4703 }, { "epoch": 1.2211838006230529, "grad_norm": 2.962691068649292, "learning_rate": 2.06569730599057e-06, "loss": 0.015325428918004036, "step": 4704 }, { "epoch": 1.2214434060228452, "grad_norm": 2.2994744777679443, "learning_rate": 2.06479617026844e-06, "loss": 0.08255556970834732, "step": 4705 }, { "epoch": 1.2217030114226377, "grad_norm": 2.7170469760894775, "learning_rate": 2.063895155674657e-06, "loss": 0.05569467693567276, "step": 4706 }, { "epoch": 1.22196261682243, "grad_norm": 3.197040319442749, "learning_rate": 2.0629942623685e-06, "loss": 0.02461932599544525, "step": 4707 }, { "epoch": 1.2222222222222223, "grad_norm": 3.9272515773773193, "learning_rate": 2.0620934905092273e-06, "loss": 0.039732426404953, "step": 4708 }, { "epoch": 1.2224818276220146, "grad_norm": 3.8335893154144287, "learning_rate": 2.0611928402560756e-06, "loss": 0.039347678422927856, "step": 4709 }, { "epoch": 1.222741433021807, "grad_norm": 3.221261501312256, "learning_rate": 2.0602923117682573e-06, "loss": 0.04678436741232872, "step": 4710 }, { "epoch": 1.2230010384215992, "grad_norm": 3.660651922225952, "learning_rate": 2.059391905204966e-06, "loss": 0.03579594939947128, "step": 4711 }, { "epoch": 1.2232606438213915, "grad_norm": 2.4739439487457275, "learning_rate": 2.0584916207253736e-06, "loss": 0.07923705130815506, "step": 4712 }, { "epoch": 1.2235202492211839, "grad_norm": 2.4851300716400146, "learning_rate": 2.0575914584886285e-06, "loss": 0.06954441219568253, "step": 4713 }, { "epoch": 1.2237798546209762, "grad_norm": 3.135596752166748, "learning_rate": 2.0566914186538596e-06, "loss": 0.04316004365682602, "step": 4714 }, { "epoch": 1.2240394600207685, "grad_norm": 4.408258438110352, "learning_rate": 2.0557915013801732e-06, "loss": 0.022533034905791283, "step": 4715 }, { "epoch": 1.2242990654205608, "grad_norm": 2.8301448822021484, "learning_rate": 2.054891706826655e-06, "loss": 0.07119442522525787, "step": 4716 }, { "epoch": 1.224558670820353, "grad_norm": 2.311661720275879, "learning_rate": 2.053992035152366e-06, "loss": 0.07133279740810394, "step": 4717 }, { "epoch": 1.2248182762201454, "grad_norm": 2.0840322971343994, "learning_rate": 2.053092486516349e-06, "loss": 0.06164877116680145, "step": 4718 }, { "epoch": 1.2250778816199377, "grad_norm": 2.840329647064209, "learning_rate": 2.0521930610776235e-06, "loss": 0.05218173936009407, "step": 4719 }, { "epoch": 1.22533748701973, "grad_norm": 2.9114348888397217, "learning_rate": 2.0512937589951877e-06, "loss": 0.05795588344335556, "step": 4720 }, { "epoch": 1.2255970924195223, "grad_norm": 3.6932668685913086, "learning_rate": 2.050394580428016e-06, "loss": 0.016393262892961502, "step": 4721 }, { "epoch": 1.2258566978193146, "grad_norm": 3.200561761856079, "learning_rate": 2.049495525535064e-06, "loss": 0.06887156516313553, "step": 4722 }, { "epoch": 1.226116303219107, "grad_norm": 2.764963150024414, "learning_rate": 2.048596594475264e-06, "loss": 0.040478453040122986, "step": 4723 }, { "epoch": 1.2263759086188992, "grad_norm": 2.038175582885742, "learning_rate": 2.047697787407526e-06, "loss": 0.07820408791303635, "step": 4724 }, { "epoch": 1.2266355140186915, "grad_norm": 3.4125783443450928, "learning_rate": 2.046799104490738e-06, "loss": 0.03164410591125488, "step": 4725 }, { "epoch": 1.2268951194184838, "grad_norm": 3.356975555419922, "learning_rate": 2.045900545883768e-06, "loss": 0.013582587242126465, "step": 4726 }, { "epoch": 1.2271547248182761, "grad_norm": 2.922931432723999, "learning_rate": 2.045002111745459e-06, "loss": 0.021110031753778458, "step": 4727 }, { "epoch": 1.2274143302180685, "grad_norm": 3.869948148727417, "learning_rate": 2.0441038022346354e-06, "loss": 0.016829222440719604, "step": 4728 }, { "epoch": 1.2276739356178608, "grad_norm": 3.6268882751464844, "learning_rate": 2.0432056175100957e-06, "loss": 0.035234831273555756, "step": 4729 }, { "epoch": 1.227933541017653, "grad_norm": 3.3104352951049805, "learning_rate": 2.0423075577306194e-06, "loss": 0.04046069458127022, "step": 4730 }, { "epoch": 1.2281931464174454, "grad_norm": 2.6509768962860107, "learning_rate": 2.041409623054963e-06, "loss": 0.04224199801683426, "step": 4731 }, { "epoch": 1.2284527518172377, "grad_norm": 3.4408979415893555, "learning_rate": 2.040511813641862e-06, "loss": 0.0725572258234024, "step": 4732 }, { "epoch": 1.2287123572170302, "grad_norm": 3.157762289047241, "learning_rate": 2.0396141296500256e-06, "loss": 0.026504606008529663, "step": 4733 }, { "epoch": 1.2289719626168225, "grad_norm": 2.033525228500366, "learning_rate": 2.038716571238145e-06, "loss": 0.008826675824820995, "step": 4734 }, { "epoch": 1.2292315680166148, "grad_norm": 2.012214183807373, "learning_rate": 2.037819138564889e-06, "loss": 0.10609519481658936, "step": 4735 }, { "epoch": 1.2294911734164071, "grad_norm": 2.149667978286743, "learning_rate": 2.036921831788904e-06, "loss": 0.07579277455806732, "step": 4736 }, { "epoch": 1.2297507788161994, "grad_norm": 2.5447230339050293, "learning_rate": 2.0360246510688102e-06, "loss": 0.034009236842393875, "step": 4737 }, { "epoch": 1.2300103842159917, "grad_norm": 3.741743803024292, "learning_rate": 2.0351275965632104e-06, "loss": 0.049501683562994, "step": 4738 }, { "epoch": 1.230269989615784, "grad_norm": 3.797334909439087, "learning_rate": 2.0342306684306834e-06, "loss": 0.029972290620207787, "step": 4739 }, { "epoch": 1.2305295950155763, "grad_norm": 2.604501962661743, "learning_rate": 2.033333866829785e-06, "loss": 0.026702605187892914, "step": 4740 }, { "epoch": 1.2307892004153687, "grad_norm": 2.728337287902832, "learning_rate": 2.03243719191905e-06, "loss": 0.013952254317700863, "step": 4741 }, { "epoch": 1.231048805815161, "grad_norm": 3.293118476867676, "learning_rate": 2.031540643856989e-06, "loss": 0.038106393069028854, "step": 4742 }, { "epoch": 1.2313084112149533, "grad_norm": 2.809081792831421, "learning_rate": 2.0306442228020917e-06, "loss": 0.01920364424586296, "step": 4743 }, { "epoch": 1.2315680166147456, "grad_norm": 3.875091552734375, "learning_rate": 2.0297479289128246e-06, "loss": 0.09923949092626572, "step": 4744 }, { "epoch": 1.2318276220145379, "grad_norm": 3.145307779312134, "learning_rate": 2.0288517623476324e-06, "loss": 0.0210138950496912, "step": 4745 }, { "epoch": 1.2320872274143302, "grad_norm": 3.0988070964813232, "learning_rate": 2.027955723264936e-06, "loss": 0.08100537210702896, "step": 4746 }, { "epoch": 1.2323468328141225, "grad_norm": 3.003678560256958, "learning_rate": 2.027059811823136e-06, "loss": 0.026387520134449005, "step": 4747 }, { "epoch": 1.2326064382139148, "grad_norm": 2.5740602016448975, "learning_rate": 2.026164028180608e-06, "loss": 0.09026255458593369, "step": 4748 }, { "epoch": 1.232866043613707, "grad_norm": 3.7814440727233887, "learning_rate": 2.0252683724957052e-06, "loss": 0.013272488489747047, "step": 4749 }, { "epoch": 1.2331256490134994, "grad_norm": 2.850982666015625, "learning_rate": 2.02437284492676e-06, "loss": 0.06092388555407524, "step": 4750 }, { "epoch": 1.2333852544132917, "grad_norm": 2.988253593444824, "learning_rate": 2.0234774456320823e-06, "loss": 0.030268702656030655, "step": 4751 }, { "epoch": 1.233644859813084, "grad_norm": 2.9470512866973877, "learning_rate": 2.0225821747699566e-06, "loss": 0.04805454984307289, "step": 4752 }, { "epoch": 1.2339044652128763, "grad_norm": 3.681673288345337, "learning_rate": 2.0216870324986466e-06, "loss": 0.039765872061252594, "step": 4753 }, { "epoch": 1.2341640706126689, "grad_norm": 2.9408957958221436, "learning_rate": 2.020792018976393e-06, "loss": 0.011160233989357948, "step": 4754 }, { "epoch": 1.2344236760124612, "grad_norm": 3.833458423614502, "learning_rate": 2.0198971343614136e-06, "loss": 0.057644687592983246, "step": 4755 }, { "epoch": 1.2346832814122535, "grad_norm": 2.9122848510742188, "learning_rate": 2.0190023788119044e-06, "loss": 0.05781884863972664, "step": 4756 }, { "epoch": 1.2349428868120458, "grad_norm": 4.5192484855651855, "learning_rate": 2.018107752486036e-06, "loss": 0.017331458628177643, "step": 4757 }, { "epoch": 1.235202492211838, "grad_norm": 3.4815874099731445, "learning_rate": 2.01721325554196e-06, "loss": 0.022002292796969414, "step": 4758 }, { "epoch": 1.2354620976116304, "grad_norm": 5.8741631507873535, "learning_rate": 2.0163188881378017e-06, "loss": 0.018859805539250374, "step": 4759 }, { "epoch": 1.2357217030114227, "grad_norm": 3.6342618465423584, "learning_rate": 2.0154246504316637e-06, "loss": 0.028337471187114716, "step": 4760 }, { "epoch": 1.235981308411215, "grad_norm": 3.230011463165283, "learning_rate": 2.0145305425816287e-06, "loss": 0.029743898659944534, "step": 4761 }, { "epoch": 1.2362409138110073, "grad_norm": 4.138111114501953, "learning_rate": 2.013636564745754e-06, "loss": 0.028053630143404007, "step": 4762 }, { "epoch": 1.2365005192107996, "grad_norm": 2.196150779724121, "learning_rate": 2.0127427170820746e-06, "loss": 0.06723829358816147, "step": 4763 }, { "epoch": 1.236760124610592, "grad_norm": 2.3920066356658936, "learning_rate": 2.0118489997486013e-06, "loss": 0.07916492968797684, "step": 4764 }, { "epoch": 1.2370197300103842, "grad_norm": 2.0228631496429443, "learning_rate": 2.0109554129033225e-06, "loss": 0.07980826497077942, "step": 4765 }, { "epoch": 1.2372793354101765, "grad_norm": 2.303861618041992, "learning_rate": 2.0100619567042068e-06, "loss": 0.011474519968032837, "step": 4766 }, { "epoch": 1.2375389408099688, "grad_norm": 2.0624966621398926, "learning_rate": 2.0091686313091933e-06, "loss": 0.01986558362841606, "step": 4767 }, { "epoch": 1.2377985462097612, "grad_norm": 2.195016622543335, "learning_rate": 2.0082754368762036e-06, "loss": 0.015597538091242313, "step": 4768 }, { "epoch": 1.2380581516095535, "grad_norm": 2.893096923828125, "learning_rate": 2.0073823735631324e-06, "loss": 0.009460010565817356, "step": 4769 }, { "epoch": 1.2383177570093458, "grad_norm": 2.6797163486480713, "learning_rate": 2.006489441527855e-06, "loss": 0.06717254221439362, "step": 4770 }, { "epoch": 1.238577362409138, "grad_norm": 3.0185177326202393, "learning_rate": 2.00559664092822e-06, "loss": 0.051669828593730927, "step": 4771 }, { "epoch": 1.2388369678089304, "grad_norm": 3.814150094985962, "learning_rate": 2.004703971922054e-06, "loss": 0.041208378970623016, "step": 4772 }, { "epoch": 1.2390965732087227, "grad_norm": 2.6385347843170166, "learning_rate": 2.0038114346671612e-06, "loss": 0.02568904682993889, "step": 4773 }, { "epoch": 1.239356178608515, "grad_norm": 2.5098698139190674, "learning_rate": 2.0029190293213213e-06, "loss": 0.05176851153373718, "step": 4774 }, { "epoch": 1.2396157840083073, "grad_norm": 3.8400542736053467, "learning_rate": 2.002026756042291e-06, "loss": 0.033060912042856216, "step": 4775 }, { "epoch": 1.2398753894080996, "grad_norm": 2.3716936111450195, "learning_rate": 2.0011346149878032e-06, "loss": 0.04795865714550018, "step": 4776 }, { "epoch": 1.240134994807892, "grad_norm": 3.3550267219543457, "learning_rate": 2.0002426063155694e-06, "loss": 0.06953698396682739, "step": 4777 }, { "epoch": 1.2403946002076842, "grad_norm": 3.6811139583587646, "learning_rate": 1.999350730183276e-06, "loss": 0.0153689906001091, "step": 4778 }, { "epoch": 1.2406542056074765, "grad_norm": 2.427060127258301, "learning_rate": 1.998458986748585e-06, "loss": 0.10235530883073807, "step": 4779 }, { "epoch": 1.2409138110072688, "grad_norm": 2.7888712882995605, "learning_rate": 1.997567376169137e-06, "loss": 0.02246561273932457, "step": 4780 }, { "epoch": 1.2411734164070614, "grad_norm": 2.716636896133423, "learning_rate": 1.99667589860255e-06, "loss": 0.016251137480139732, "step": 4781 }, { "epoch": 1.2414330218068537, "grad_norm": 2.444507122039795, "learning_rate": 1.995784554206414e-06, "loss": 0.017589706927537918, "step": 4782 }, { "epoch": 1.241692627206646, "grad_norm": 2.0908098220825195, "learning_rate": 1.9948933431383e-06, "loss": 0.02380288764834404, "step": 4783 }, { "epoch": 1.2419522326064383, "grad_norm": 2.561119556427002, "learning_rate": 1.9940022655557527e-06, "loss": 0.020799819380044937, "step": 4784 }, { "epoch": 1.2422118380062306, "grad_norm": 2.611887216567993, "learning_rate": 1.9931113216162956e-06, "loss": 0.03127727285027504, "step": 4785 }, { "epoch": 1.242471443406023, "grad_norm": 3.7737619876861572, "learning_rate": 1.992220511477426e-06, "loss": 0.03737322986125946, "step": 4786 }, { "epoch": 1.2427310488058152, "grad_norm": 3.9454174041748047, "learning_rate": 1.9913298352966193e-06, "loss": 0.049900125712156296, "step": 4787 }, { "epoch": 1.2429906542056075, "grad_norm": 2.9073381423950195, "learning_rate": 1.990439293231326e-06, "loss": 0.04411661624908447, "step": 4788 }, { "epoch": 1.2432502596053998, "grad_norm": 2.537804126739502, "learning_rate": 1.9895488854389744e-06, "loss": 0.02562827616930008, "step": 4789 }, { "epoch": 1.2435098650051921, "grad_norm": 2.9723806381225586, "learning_rate": 1.9886586120769685e-06, "loss": 0.015203777700662613, "step": 4790 }, { "epoch": 1.2437694704049844, "grad_norm": 3.335555076599121, "learning_rate": 1.9877684733026865e-06, "loss": 0.08992540836334229, "step": 4791 }, { "epoch": 1.2440290758047767, "grad_norm": 4.873503684997559, "learning_rate": 1.986878469273486e-06, "loss": 0.014580471441149712, "step": 4792 }, { "epoch": 1.244288681204569, "grad_norm": 2.115722417831421, "learning_rate": 1.9859886001466988e-06, "loss": 0.09623317420482635, "step": 4793 }, { "epoch": 1.2445482866043613, "grad_norm": 3.123239517211914, "learning_rate": 1.9850988660796344e-06, "loss": 0.012083631940186024, "step": 4794 }, { "epoch": 1.2448078920041536, "grad_norm": 2.388607978820801, "learning_rate": 1.984209267229576e-06, "loss": 0.08989386260509491, "step": 4795 }, { "epoch": 1.245067497403946, "grad_norm": 3.2425076961517334, "learning_rate": 1.9833198037537844e-06, "loss": 0.05591011047363281, "step": 4796 }, { "epoch": 1.2453271028037383, "grad_norm": 2.915480613708496, "learning_rate": 1.982430475809498e-06, "loss": 0.04902860149741173, "step": 4797 }, { "epoch": 1.2455867082035306, "grad_norm": 3.0474908351898193, "learning_rate": 1.9815412835539277e-06, "loss": 0.06341510266065598, "step": 4798 }, { "epoch": 1.2458463136033229, "grad_norm": 3.378873825073242, "learning_rate": 1.9806522271442624e-06, "loss": 0.02682919055223465, "step": 4799 }, { "epoch": 1.2461059190031152, "grad_norm": 2.1417782306671143, "learning_rate": 1.979763306737668e-06, "loss": 0.011808546259999275, "step": 4800 }, { "epoch": 1.2463655244029075, "grad_norm": 2.5217554569244385, "learning_rate": 1.978874522491286e-06, "loss": 0.05719747394323349, "step": 4801 }, { "epoch": 1.2466251298027, "grad_norm": 1.7471678256988525, "learning_rate": 1.977985874562231e-06, "loss": 0.008989866822957993, "step": 4802 }, { "epoch": 1.2468847352024923, "grad_norm": 3.221297264099121, "learning_rate": 1.977097363107597e-06, "loss": 0.025831298902630806, "step": 4803 }, { "epoch": 1.2471443406022846, "grad_norm": 3.1023714542388916, "learning_rate": 1.9762089882844517e-06, "loss": 0.04293616861104965, "step": 4804 }, { "epoch": 1.247403946002077, "grad_norm": 2.3327252864837646, "learning_rate": 1.97532075024984e-06, "loss": 0.08566273748874664, "step": 4805 }, { "epoch": 1.2476635514018692, "grad_norm": 4.195019721984863, "learning_rate": 1.974432649160782e-06, "loss": 0.017296070232987404, "step": 4806 }, { "epoch": 1.2479231568016615, "grad_norm": 2.480349540710449, "learning_rate": 1.973544685174273e-06, "loss": 0.01802472397685051, "step": 4807 }, { "epoch": 1.2481827622014539, "grad_norm": 3.016822576522827, "learning_rate": 1.972656858447285e-06, "loss": 0.03494022414088249, "step": 4808 }, { "epoch": 1.2484423676012462, "grad_norm": 0.9979987144470215, "learning_rate": 1.971769169136767e-06, "loss": 0.0031673996709287167, "step": 4809 }, { "epoch": 1.2487019730010385, "grad_norm": 2.3361973762512207, "learning_rate": 1.9708816173996392e-06, "loss": 0.08579801768064499, "step": 4810 }, { "epoch": 1.2489615784008308, "grad_norm": 2.6270124912261963, "learning_rate": 1.969994203392803e-06, "loss": 0.06004684790968895, "step": 4811 }, { "epoch": 1.249221183800623, "grad_norm": 3.0535311698913574, "learning_rate": 1.969106927273132e-06, "loss": 0.035373322665691376, "step": 4812 }, { "epoch": 1.2494807892004154, "grad_norm": 2.5949106216430664, "learning_rate": 1.9682197891974753e-06, "loss": 0.04365243762731552, "step": 4813 }, { "epoch": 1.2497403946002077, "grad_norm": 3.105125665664673, "learning_rate": 1.9673327893226603e-06, "loss": 0.030977975577116013, "step": 4814 }, { "epoch": 1.25, "grad_norm": 4.122777462005615, "learning_rate": 1.9664459278054866e-06, "loss": 0.035794682800769806, "step": 4815 }, { "epoch": 1.2502596053997923, "grad_norm": 2.4125924110412598, "learning_rate": 1.9655592048027324e-06, "loss": 0.017993813380599022, "step": 4816 }, { "epoch": 1.2505192107995846, "grad_norm": 2.5283288955688477, "learning_rate": 1.964672620471149e-06, "loss": 0.0232885479927063, "step": 4817 }, { "epoch": 1.250778816199377, "grad_norm": 2.106696367263794, "learning_rate": 1.963786174967465e-06, "loss": 0.01614300161600113, "step": 4818 }, { "epoch": 1.2510384215991692, "grad_norm": 3.4384162425994873, "learning_rate": 1.9628998684483835e-06, "loss": 0.02855704352259636, "step": 4819 }, { "epoch": 1.2512980269989615, "grad_norm": 4.615235805511475, "learning_rate": 1.962013701070583e-06, "loss": 0.01589379832148552, "step": 4820 }, { "epoch": 1.2515576323987538, "grad_norm": 3.382955312728882, "learning_rate": 1.9611276729907187e-06, "loss": 0.017192687839269638, "step": 4821 }, { "epoch": 1.2518172377985461, "grad_norm": 3.288264513015747, "learning_rate": 1.960241784365417e-06, "loss": 0.062367960810661316, "step": 4822 }, { "epoch": 1.2520768431983385, "grad_norm": 3.2448081970214844, "learning_rate": 1.959356035351286e-06, "loss": 0.056978605687618256, "step": 4823 }, { "epoch": 1.2523364485981308, "grad_norm": 2.2707719802856445, "learning_rate": 1.9584704261049047e-06, "loss": 0.008404447697103024, "step": 4824 }, { "epoch": 1.252596053997923, "grad_norm": 2.9530935287475586, "learning_rate": 1.957584956782828e-06, "loss": 0.04605146497488022, "step": 4825 }, { "epoch": 1.2528556593977154, "grad_norm": 3.0998404026031494, "learning_rate": 1.956699627541587e-06, "loss": 0.06202022358775139, "step": 4826 }, { "epoch": 1.2531152647975077, "grad_norm": 2.843611240386963, "learning_rate": 1.9558144385376877e-06, "loss": 0.0036993264220654964, "step": 4827 }, { "epoch": 1.2533748701973, "grad_norm": 4.3136796951293945, "learning_rate": 1.9549293899276115e-06, "loss": 0.015921438112854958, "step": 4828 }, { "epoch": 1.2536344755970923, "grad_norm": 3.3965156078338623, "learning_rate": 1.9540444818678146e-06, "loss": 0.04060167074203491, "step": 4829 }, { "epoch": 1.2538940809968846, "grad_norm": 2.642570734024048, "learning_rate": 1.953159714514728e-06, "loss": 0.08361341059207916, "step": 4830 }, { "epoch": 1.2541536863966771, "grad_norm": 3.1797358989715576, "learning_rate": 1.952275088024758e-06, "loss": 0.06135324761271477, "step": 4831 }, { "epoch": 1.2544132917964694, "grad_norm": 2.993849515914917, "learning_rate": 1.9513906025542883e-06, "loss": 0.03446725383400917, "step": 4832 }, { "epoch": 1.2546728971962617, "grad_norm": 2.068732738494873, "learning_rate": 1.9505062582596735e-06, "loss": 0.07783432304859161, "step": 4833 }, { "epoch": 1.254932502596054, "grad_norm": 3.3230390548706055, "learning_rate": 1.9496220552972467e-06, "loss": 0.03425154462456703, "step": 4834 }, { "epoch": 1.2551921079958464, "grad_norm": 3.0204875469207764, "learning_rate": 1.9487379938233136e-06, "loss": 0.03162379562854767, "step": 4835 }, { "epoch": 1.2554517133956387, "grad_norm": 2.179302215576172, "learning_rate": 1.9478540739941577e-06, "loss": 0.06886713206768036, "step": 4836 }, { "epoch": 1.255711318795431, "grad_norm": 4.004877090454102, "learning_rate": 1.946970295966034e-06, "loss": 0.02042827568948269, "step": 4837 }, { "epoch": 1.2559709241952233, "grad_norm": 4.082697868347168, "learning_rate": 1.9460866598951748e-06, "loss": 0.03886401280760765, "step": 4838 }, { "epoch": 1.2562305295950156, "grad_norm": 3.387270450592041, "learning_rate": 1.9452031659377874e-06, "loss": 0.026081949472427368, "step": 4839 }, { "epoch": 1.2564901349948079, "grad_norm": 3.470609426498413, "learning_rate": 1.9443198142500537e-06, "loss": 0.03276051953434944, "step": 4840 }, { "epoch": 1.2567497403946002, "grad_norm": 2.962270736694336, "learning_rate": 1.943436604988128e-06, "loss": 0.06711359322071075, "step": 4841 }, { "epoch": 1.2570093457943925, "grad_norm": 4.8461222648620605, "learning_rate": 1.9425535383081434e-06, "loss": 0.021550528705120087, "step": 4842 }, { "epoch": 1.2572689511941848, "grad_norm": 2.29661226272583, "learning_rate": 1.9416706143662044e-06, "loss": 0.06307964026927948, "step": 4843 }, { "epoch": 1.257528556593977, "grad_norm": 3.0466599464416504, "learning_rate": 1.9407878333183933e-06, "loss": 0.04420822858810425, "step": 4844 }, { "epoch": 1.2577881619937694, "grad_norm": 2.1674723625183105, "learning_rate": 1.9399051953207643e-06, "loss": 0.08164554834365845, "step": 4845 }, { "epoch": 1.2580477673935617, "grad_norm": 2.262096643447876, "learning_rate": 1.939022700529348e-06, "loss": 0.007650548592209816, "step": 4846 }, { "epoch": 1.258307372793354, "grad_norm": 3.01336669921875, "learning_rate": 1.9381403491001492e-06, "loss": 0.03479699045419693, "step": 4847 }, { "epoch": 1.2585669781931463, "grad_norm": 3.162670135498047, "learning_rate": 1.9372581411891485e-06, "loss": 0.04355982691049576, "step": 4848 }, { "epoch": 1.2588265835929389, "grad_norm": 3.4474613666534424, "learning_rate": 1.9363760769522982e-06, "loss": 0.04653241112828255, "step": 4849 }, { "epoch": 1.2590861889927312, "grad_norm": 3.594707727432251, "learning_rate": 1.935494156545528e-06, "loss": 0.06031472980976105, "step": 4850 }, { "epoch": 1.2593457943925235, "grad_norm": 3.9659411907196045, "learning_rate": 1.9346123801247422e-06, "loss": 0.01701323129236698, "step": 4851 }, { "epoch": 1.2596053997923158, "grad_norm": 2.7314870357513428, "learning_rate": 1.9337307478458167e-06, "loss": 0.005968628451228142, "step": 4852 }, { "epoch": 1.259865005192108, "grad_norm": 2.6055526733398438, "learning_rate": 1.9328492598646047e-06, "loss": 0.00978921726346016, "step": 4853 }, { "epoch": 1.2601246105919004, "grad_norm": 3.6953182220458984, "learning_rate": 1.9319679163369336e-06, "loss": 0.05048198997974396, "step": 4854 }, { "epoch": 1.2603842159916927, "grad_norm": 3.368267059326172, "learning_rate": 1.9310867174186054e-06, "loss": 0.03720863163471222, "step": 4855 }, { "epoch": 1.260643821391485, "grad_norm": 2.9694833755493164, "learning_rate": 1.9302056632653937e-06, "loss": 0.015762899070978165, "step": 4856 }, { "epoch": 1.2609034267912773, "grad_norm": 2.9749867916107178, "learning_rate": 1.92932475403305e-06, "loss": 0.007719440385699272, "step": 4857 }, { "epoch": 1.2611630321910696, "grad_norm": 2.871206760406494, "learning_rate": 1.9284439898772994e-06, "loss": 0.05271601304411888, "step": 4858 }, { "epoch": 1.261422637590862, "grad_norm": 3.7749900817871094, "learning_rate": 1.9275633709538403e-06, "loss": 0.05095303803682327, "step": 4859 }, { "epoch": 1.2616822429906542, "grad_norm": 3.1872193813323975, "learning_rate": 1.9266828974183456e-06, "loss": 0.02725951187312603, "step": 4860 }, { "epoch": 1.2619418483904465, "grad_norm": 3.9164648056030273, "learning_rate": 1.925802569426463e-06, "loss": 0.025185037404298782, "step": 4861 }, { "epoch": 1.2622014537902388, "grad_norm": 3.616948127746582, "learning_rate": 1.9249223871338142e-06, "loss": 0.01137557439506054, "step": 4862 }, { "epoch": 1.2624610591900312, "grad_norm": 3.413832902908325, "learning_rate": 1.924042350695997e-06, "loss": 0.025299586355686188, "step": 4863 }, { "epoch": 1.2627206645898235, "grad_norm": 2.2213237285614014, "learning_rate": 1.9231624602685787e-06, "loss": 0.07775674015283585, "step": 4864 }, { "epoch": 1.2629802699896158, "grad_norm": 3.5252304077148438, "learning_rate": 1.922282716007106e-06, "loss": 0.040615882724523544, "step": 4865 }, { "epoch": 1.263239875389408, "grad_norm": 4.078364849090576, "learning_rate": 1.9214031180670973e-06, "loss": 0.031280696392059326, "step": 4866 }, { "epoch": 1.2634994807892004, "grad_norm": 2.1394710540771484, "learning_rate": 1.9205236666040443e-06, "loss": 0.05547790974378586, "step": 4867 }, { "epoch": 1.2637590861889927, "grad_norm": 4.2824201583862305, "learning_rate": 1.919644361773415e-06, "loss": 0.01631828024983406, "step": 4868 }, { "epoch": 1.264018691588785, "grad_norm": 2.701674699783325, "learning_rate": 1.9187652037306496e-06, "loss": 0.08526937663555145, "step": 4869 }, { "epoch": 1.2642782969885773, "grad_norm": 2.329258918762207, "learning_rate": 1.917886192631163e-06, "loss": 0.03450462222099304, "step": 4870 }, { "epoch": 1.2645379023883696, "grad_norm": 2.580294609069824, "learning_rate": 1.9170073286303458e-06, "loss": 0.05696405470371246, "step": 4871 }, { "epoch": 1.264797507788162, "grad_norm": 2.161543369293213, "learning_rate": 1.9161286118835594e-06, "loss": 0.024120431393384933, "step": 4872 }, { "epoch": 1.2650571131879542, "grad_norm": 3.066707134246826, "learning_rate": 1.9152500425461414e-06, "loss": 0.04990384727716446, "step": 4873 }, { "epoch": 1.2653167185877465, "grad_norm": 3.407543420791626, "learning_rate": 1.914371620773402e-06, "loss": 0.01580190286040306, "step": 4874 }, { "epoch": 1.2655763239875388, "grad_norm": 2.893367290496826, "learning_rate": 1.9134933467206274e-06, "loss": 0.06143117696046829, "step": 4875 }, { "epoch": 1.2658359293873311, "grad_norm": 2.995018243789673, "learning_rate": 1.9126152205430754e-06, "loss": 0.060347285121679306, "step": 4876 }, { "epoch": 1.2660955347871234, "grad_norm": 1.9964088201522827, "learning_rate": 1.9117372423959786e-06, "loss": 0.08904176950454712, "step": 4877 }, { "epoch": 1.2663551401869158, "grad_norm": 1.9258607625961304, "learning_rate": 1.9108594124345444e-06, "loss": 0.07932164520025253, "step": 4878 }, { "epoch": 1.2666147455867083, "grad_norm": 4.026320457458496, "learning_rate": 1.9099817308139516e-06, "loss": 0.014549745246767998, "step": 4879 }, { "epoch": 1.2668743509865006, "grad_norm": 1.9295692443847656, "learning_rate": 1.9091041976893547e-06, "loss": 0.08722519874572754, "step": 4880 }, { "epoch": 1.267133956386293, "grad_norm": 3.564098358154297, "learning_rate": 1.908226813215882e-06, "loss": 0.023261791095137596, "step": 4881 }, { "epoch": 1.2673935617860852, "grad_norm": 3.6012778282165527, "learning_rate": 1.9073495775486356e-06, "loss": 0.049179792404174805, "step": 4882 }, { "epoch": 1.2676531671858775, "grad_norm": 3.9313580989837646, "learning_rate": 1.9064724908426885e-06, "loss": 0.056406110525131226, "step": 4883 }, { "epoch": 1.2679127725856698, "grad_norm": 2.8034536838531494, "learning_rate": 1.9055955532530912e-06, "loss": 0.06166290491819382, "step": 4884 }, { "epoch": 1.2681723779854621, "grad_norm": 3.2440946102142334, "learning_rate": 1.9047187649348652e-06, "loss": 0.0289761945605278, "step": 4885 }, { "epoch": 1.2684319833852544, "grad_norm": 2.802799701690674, "learning_rate": 1.903842126043009e-06, "loss": 0.029291676357388496, "step": 4886 }, { "epoch": 1.2686915887850467, "grad_norm": 4.087764263153076, "learning_rate": 1.902965636732489e-06, "loss": 0.020579155534505844, "step": 4887 }, { "epoch": 1.268951194184839, "grad_norm": 3.9105348587036133, "learning_rate": 1.9020892971582495e-06, "loss": 0.01989564299583435, "step": 4888 }, { "epoch": 1.2692107995846313, "grad_norm": 3.4173192977905273, "learning_rate": 1.9012131074752083e-06, "loss": 0.02580752968788147, "step": 4889 }, { "epoch": 1.2694704049844237, "grad_norm": 4.811099052429199, "learning_rate": 1.900337067838256e-06, "loss": 0.0253763310611248, "step": 4890 }, { "epoch": 1.269730010384216, "grad_norm": 3.5116288661956787, "learning_rate": 1.8994611784022542e-06, "loss": 0.05780169740319252, "step": 4891 }, { "epoch": 1.2699896157840083, "grad_norm": 3.459629774093628, "learning_rate": 1.8985854393220414e-06, "loss": 0.026022953912615776, "step": 4892 }, { "epoch": 1.2702492211838006, "grad_norm": 3.3599281311035156, "learning_rate": 1.8977098507524283e-06, "loss": 0.03934450447559357, "step": 4893 }, { "epoch": 1.2705088265835929, "grad_norm": 3.7253577709198, "learning_rate": 1.8968344128481992e-06, "loss": 0.04258469492197037, "step": 4894 }, { "epoch": 1.2707684319833852, "grad_norm": 4.209795951843262, "learning_rate": 1.8959591257641107e-06, "loss": 0.03832581639289856, "step": 4895 }, { "epoch": 1.2710280373831775, "grad_norm": 2.4834420680999756, "learning_rate": 1.8950839896548934e-06, "loss": 0.04416421428322792, "step": 4896 }, { "epoch": 1.27128764278297, "grad_norm": 2.169222354888916, "learning_rate": 1.894209004675252e-06, "loss": 0.07345268130302429, "step": 4897 }, { "epoch": 1.2715472481827623, "grad_norm": 3.8922221660614014, "learning_rate": 1.8933341709798642e-06, "loss": 0.043089039623737335, "step": 4898 }, { "epoch": 1.2718068535825546, "grad_norm": 2.138361930847168, "learning_rate": 1.8924594887233787e-06, "loss": 0.0812925174832344, "step": 4899 }, { "epoch": 1.272066458982347, "grad_norm": 2.023383378982544, "learning_rate": 1.8915849580604213e-06, "loss": 0.0785842090845108, "step": 4900 }, { "epoch": 1.2723260643821392, "grad_norm": 2.6329128742218018, "learning_rate": 1.8907105791455877e-06, "loss": 0.07128176093101501, "step": 4901 }, { "epoch": 1.2725856697819315, "grad_norm": 2.8712003231048584, "learning_rate": 1.8898363521334494e-06, "loss": 0.009664737619459629, "step": 4902 }, { "epoch": 1.2728452751817239, "grad_norm": 2.4825363159179688, "learning_rate": 1.888962277178548e-06, "loss": 0.016200125217437744, "step": 4903 }, { "epoch": 1.2731048805815162, "grad_norm": 2.9112086296081543, "learning_rate": 1.8880883544354017e-06, "loss": 0.042329661548137665, "step": 4904 }, { "epoch": 1.2733644859813085, "grad_norm": 3.4424593448638916, "learning_rate": 1.8872145840584987e-06, "loss": 0.043366849422454834, "step": 4905 }, { "epoch": 1.2736240913811008, "grad_norm": 2.078876256942749, "learning_rate": 1.8863409662023024e-06, "loss": 0.07850781828165054, "step": 4906 }, { "epoch": 1.273883696780893, "grad_norm": 2.296340227127075, "learning_rate": 1.8854675010212476e-06, "loss": 0.08658476173877716, "step": 4907 }, { "epoch": 1.2741433021806854, "grad_norm": 2.7491400241851807, "learning_rate": 1.8845941886697432e-06, "loss": 0.04346253350377083, "step": 4908 }, { "epoch": 1.2744029075804777, "grad_norm": 3.090869903564453, "learning_rate": 1.8837210293021724e-06, "loss": 0.05849669128656387, "step": 4909 }, { "epoch": 1.27466251298027, "grad_norm": 2.6972732543945312, "learning_rate": 1.8828480230728874e-06, "loss": 0.06378799676895142, "step": 4910 }, { "epoch": 1.2749221183800623, "grad_norm": 2.7923173904418945, "learning_rate": 1.8819751701362165e-06, "loss": 0.04017570987343788, "step": 4911 }, { "epoch": 1.2751817237798546, "grad_norm": 2.1797800064086914, "learning_rate": 1.8811024706464614e-06, "loss": 0.005216685123741627, "step": 4912 }, { "epoch": 1.275441329179647, "grad_norm": 3.005540132522583, "learning_rate": 1.8802299247578943e-06, "loss": 0.08441992104053497, "step": 4913 }, { "epoch": 1.2757009345794392, "grad_norm": 3.102532386779785, "learning_rate": 1.8793575326247603e-06, "loss": 0.01563584804534912, "step": 4914 }, { "epoch": 1.2759605399792315, "grad_norm": 3.057720899581909, "learning_rate": 1.8784852944012799e-06, "loss": 0.03835093230009079, "step": 4915 }, { "epoch": 1.2762201453790238, "grad_norm": 3.1494710445404053, "learning_rate": 1.8776132102416444e-06, "loss": 0.06769043207168579, "step": 4916 }, { "epoch": 1.2764797507788161, "grad_norm": 3.008105516433716, "learning_rate": 1.8767412803000195e-06, "loss": 0.03223644942045212, "step": 4917 }, { "epoch": 1.2767393561786085, "grad_norm": 2.4312307834625244, "learning_rate": 1.875869504730541e-06, "loss": 0.062257926911115646, "step": 4918 }, { "epoch": 1.2769989615784008, "grad_norm": 3.5247204303741455, "learning_rate": 1.8749978836873174e-06, "loss": 0.02773457020521164, "step": 4919 }, { "epoch": 1.277258566978193, "grad_norm": 5.537135124206543, "learning_rate": 1.874126417324434e-06, "loss": 0.029223475605249405, "step": 4920 }, { "epoch": 1.2775181723779854, "grad_norm": 2.1138319969177246, "learning_rate": 1.8732551057959459e-06, "loss": 0.06870293617248535, "step": 4921 }, { "epoch": 1.2777777777777777, "grad_norm": 3.6878700256347656, "learning_rate": 1.8723839492558793e-06, "loss": 0.03871045261621475, "step": 4922 }, { "epoch": 1.27803738317757, "grad_norm": 1.5616886615753174, "learning_rate": 1.8715129478582367e-06, "loss": 0.013410529121756554, "step": 4923 }, { "epoch": 1.2782969885773623, "grad_norm": 4.1607866287231445, "learning_rate": 1.8706421017569888e-06, "loss": 0.038477182388305664, "step": 4924 }, { "epoch": 1.2785565939771546, "grad_norm": 3.026214838027954, "learning_rate": 1.8697714111060843e-06, "loss": 0.06742770224809647, "step": 4925 }, { "epoch": 1.278816199376947, "grad_norm": 2.7900562286376953, "learning_rate": 1.868900876059439e-06, "loss": 0.06350702792406082, "step": 4926 }, { "epoch": 1.2790758047767394, "grad_norm": 2.131209373474121, "learning_rate": 1.868030496770944e-06, "loss": 0.07750554382801056, "step": 4927 }, { "epoch": 1.2793354101765317, "grad_norm": 3.5056209564208984, "learning_rate": 1.8671602733944633e-06, "loss": 0.0541527085006237, "step": 4928 }, { "epoch": 1.279595015576324, "grad_norm": 3.9397144317626953, "learning_rate": 1.8662902060838318e-06, "loss": 0.03029954805970192, "step": 4929 }, { "epoch": 1.2798546209761164, "grad_norm": 3.344693422317505, "learning_rate": 1.8654202949928574e-06, "loss": 0.0584937259554863, "step": 4930 }, { "epoch": 1.2801142263759087, "grad_norm": 4.072101593017578, "learning_rate": 1.8645505402753205e-06, "loss": 0.026394624263048172, "step": 4931 }, { "epoch": 1.280373831775701, "grad_norm": 3.049577236175537, "learning_rate": 1.8636809420849745e-06, "loss": 0.06345917284488678, "step": 4932 }, { "epoch": 1.2806334371754933, "grad_norm": 3.4770240783691406, "learning_rate": 1.8628115005755432e-06, "loss": 0.028729025274515152, "step": 4933 }, { "epoch": 1.2808930425752856, "grad_norm": 3.7067089080810547, "learning_rate": 1.861942215900725e-06, "loss": 0.05556758493185043, "step": 4934 }, { "epoch": 1.2811526479750779, "grad_norm": 2.9272420406341553, "learning_rate": 1.8610730882141893e-06, "loss": 0.030851595103740692, "step": 4935 }, { "epoch": 1.2814122533748702, "grad_norm": 2.4301722049713135, "learning_rate": 1.8602041176695779e-06, "loss": 0.0702294409275055, "step": 4936 }, { "epoch": 1.2816718587746625, "grad_norm": 1.955790638923645, "learning_rate": 1.8593353044205047e-06, "loss": 0.08413642644882202, "step": 4937 }, { "epoch": 1.2819314641744548, "grad_norm": 2.720801830291748, "learning_rate": 1.8584666486205567e-06, "loss": 0.07350397109985352, "step": 4938 }, { "epoch": 1.282191069574247, "grad_norm": 2.8441572189331055, "learning_rate": 1.857598150423291e-06, "loss": 0.014028172008693218, "step": 4939 }, { "epoch": 1.2824506749740394, "grad_norm": 2.482484817504883, "learning_rate": 1.8567298099822395e-06, "loss": 0.008485829457640648, "step": 4940 }, { "epoch": 1.2827102803738317, "grad_norm": 2.471980571746826, "learning_rate": 1.8558616274509046e-06, "loss": 0.07567140460014343, "step": 4941 }, { "epoch": 1.282969885773624, "grad_norm": 2.935559034347534, "learning_rate": 1.8549936029827603e-06, "loss": 0.03849451243877411, "step": 4942 }, { "epoch": 1.2832294911734163, "grad_norm": 2.600487470626831, "learning_rate": 1.8541257367312543e-06, "loss": 0.0793287605047226, "step": 4943 }, { "epoch": 1.2834890965732086, "grad_norm": 2.333817481994629, "learning_rate": 1.8532580288498065e-06, "loss": 0.06463340669870377, "step": 4944 }, { "epoch": 1.2837487019730012, "grad_norm": 3.1748929023742676, "learning_rate": 1.8523904794918046e-06, "loss": 0.06722459197044373, "step": 4945 }, { "epoch": 1.2840083073727935, "grad_norm": 3.655902624130249, "learning_rate": 1.851523088810614e-06, "loss": 0.01907014474272728, "step": 4946 }, { "epoch": 1.2842679127725858, "grad_norm": 4.954155445098877, "learning_rate": 1.850655856959569e-06, "loss": 0.02862156182527542, "step": 4947 }, { "epoch": 1.284527518172378, "grad_norm": 3.950437068939209, "learning_rate": 1.849788784091977e-06, "loss": 0.01418803445994854, "step": 4948 }, { "epoch": 1.2847871235721704, "grad_norm": 3.3948066234588623, "learning_rate": 1.848921870361115e-06, "loss": 0.056074678897857666, "step": 4949 }, { "epoch": 1.2850467289719627, "grad_norm": 3.695751190185547, "learning_rate": 1.8480551159202342e-06, "loss": 0.061372898519039154, "step": 4950 }, { "epoch": 1.285306334371755, "grad_norm": 4.0777177810668945, "learning_rate": 1.847188520922557e-06, "loss": 0.04205707460641861, "step": 4951 }, { "epoch": 1.2855659397715473, "grad_norm": 2.961235761642456, "learning_rate": 1.8463220855212789e-06, "loss": 0.056299787014722824, "step": 4952 }, { "epoch": 1.2858255451713396, "grad_norm": 2.3335392475128174, "learning_rate": 1.845455809869564e-06, "loss": 0.01361367292702198, "step": 4953 }, { "epoch": 1.286085150571132, "grad_norm": 3.1834206581115723, "learning_rate": 1.8445896941205503e-06, "loss": 0.04231078177690506, "step": 4954 }, { "epoch": 1.2863447559709242, "grad_norm": 3.105699300765991, "learning_rate": 1.8437237384273471e-06, "loss": 0.023995336145162582, "step": 4955 }, { "epoch": 1.2866043613707165, "grad_norm": 3.293813943862915, "learning_rate": 1.842857942943037e-06, "loss": 0.03537607192993164, "step": 4956 }, { "epoch": 1.2868639667705088, "grad_norm": 3.135199546813965, "learning_rate": 1.8419923078206714e-06, "loss": 0.031018463894724846, "step": 4957 }, { "epoch": 1.2871235721703012, "grad_norm": 2.8679006099700928, "learning_rate": 1.8411268332132747e-06, "loss": 0.03657654672861099, "step": 4958 }, { "epoch": 1.2873831775700935, "grad_norm": 2.809255361557007, "learning_rate": 1.8402615192738443e-06, "loss": 0.07340869307518005, "step": 4959 }, { "epoch": 1.2876427829698858, "grad_norm": 1.300000548362732, "learning_rate": 1.8393963661553466e-06, "loss": 0.012075855396687984, "step": 4960 }, { "epoch": 1.287902388369678, "grad_norm": 3.025174617767334, "learning_rate": 1.8385313740107208e-06, "loss": 0.009942315518856049, "step": 4961 }, { "epoch": 1.2881619937694704, "grad_norm": 2.4440910816192627, "learning_rate": 1.8376665429928786e-06, "loss": 0.08642219007015228, "step": 4962 }, { "epoch": 1.2884215991692627, "grad_norm": 4.823732852935791, "learning_rate": 1.8368018732547026e-06, "loss": 0.045290179550647736, "step": 4963 }, { "epoch": 1.288681204569055, "grad_norm": 3.295264482498169, "learning_rate": 1.8359373649490452e-06, "loss": 0.03073780983686447, "step": 4964 }, { "epoch": 1.2889408099688473, "grad_norm": 2.7885355949401855, "learning_rate": 1.8350730182287323e-06, "loss": 0.026122529059648514, "step": 4965 }, { "epoch": 1.2892004153686396, "grad_norm": 3.7658472061157227, "learning_rate": 1.8342088332465612e-06, "loss": 0.030061885714530945, "step": 4966 }, { "epoch": 1.289460020768432, "grad_norm": 2.4874489307403564, "learning_rate": 1.8333448101552993e-06, "loss": 0.015313347801566124, "step": 4967 }, { "epoch": 1.2897196261682242, "grad_norm": 2.183744430541992, "learning_rate": 1.8324809491076861e-06, "loss": 0.008229875937104225, "step": 4968 }, { "epoch": 1.2899792315680165, "grad_norm": 3.931527853012085, "learning_rate": 1.8316172502564334e-06, "loss": 0.04803554341197014, "step": 4969 }, { "epoch": 1.2902388369678088, "grad_norm": 2.581352949142456, "learning_rate": 1.8307537137542219e-06, "loss": 0.06698938459157944, "step": 4970 }, { "epoch": 1.2904984423676011, "grad_norm": 1.934927225112915, "learning_rate": 1.829890339753706e-06, "loss": 0.006112589035183191, "step": 4971 }, { "epoch": 1.2907580477673934, "grad_norm": 3.9586644172668457, "learning_rate": 1.8290271284075106e-06, "loss": 0.033015213906764984, "step": 4972 }, { "epoch": 1.2910176531671858, "grad_norm": 2.321354866027832, "learning_rate": 1.8281640798682316e-06, "loss": 0.02130243368446827, "step": 4973 }, { "epoch": 1.291277258566978, "grad_norm": 2.8214824199676514, "learning_rate": 1.8273011942884356e-06, "loss": 0.06586005538702011, "step": 4974 }, { "epoch": 1.2915368639667706, "grad_norm": 3.3344202041625977, "learning_rate": 1.8264384718206624e-06, "loss": 0.04851177707314491, "step": 4975 }, { "epoch": 1.291796469366563, "grad_norm": 4.015041351318359, "learning_rate": 1.8255759126174188e-06, "loss": 0.030612202361226082, "step": 4976 }, { "epoch": 1.2920560747663552, "grad_norm": 3.577166795730591, "learning_rate": 1.8247135168311885e-06, "loss": 0.017522061243653297, "step": 4977 }, { "epoch": 1.2923156801661475, "grad_norm": 3.879784345626831, "learning_rate": 1.823851284614422e-06, "loss": 0.043105218559503555, "step": 4978 }, { "epoch": 1.2925752855659398, "grad_norm": 2.607506513595581, "learning_rate": 1.8229892161195435e-06, "loss": 0.03037058189511299, "step": 4979 }, { "epoch": 1.2928348909657321, "grad_norm": 2.0644474029541016, "learning_rate": 1.8221273114989447e-06, "loss": 0.0763670951128006, "step": 4980 }, { "epoch": 1.2930944963655244, "grad_norm": 2.803344964981079, "learning_rate": 1.8212655709049908e-06, "loss": 0.028705816715955734, "step": 4981 }, { "epoch": 1.2933541017653167, "grad_norm": 2.825192451477051, "learning_rate": 1.8204039944900198e-06, "loss": 0.0594300776720047, "step": 4982 }, { "epoch": 1.293613707165109, "grad_norm": 2.53125262260437, "learning_rate": 1.8195425824063384e-06, "loss": 0.06635747849941254, "step": 4983 }, { "epoch": 1.2938733125649013, "grad_norm": 2.595956325531006, "learning_rate": 1.8186813348062232e-06, "loss": 0.06247662007808685, "step": 4984 }, { "epoch": 1.2941329179646937, "grad_norm": 2.8215017318725586, "learning_rate": 1.817820251841923e-06, "loss": 0.03562331572175026, "step": 4985 }, { "epoch": 1.294392523364486, "grad_norm": 2.282566785812378, "learning_rate": 1.8169593336656594e-06, "loss": 0.08338853716850281, "step": 4986 }, { "epoch": 1.2946521287642783, "grad_norm": 3.586383581161499, "learning_rate": 1.816098580429621e-06, "loss": 0.05267513543367386, "step": 4987 }, { "epoch": 1.2949117341640706, "grad_norm": 3.4679884910583496, "learning_rate": 1.8152379922859697e-06, "loss": 0.03458473086357117, "step": 4988 }, { "epoch": 1.2951713395638629, "grad_norm": 3.9152255058288574, "learning_rate": 1.8143775693868387e-06, "loss": 0.015485003590583801, "step": 4989 }, { "epoch": 1.2954309449636552, "grad_norm": 3.766515016555786, "learning_rate": 1.8135173118843313e-06, "loss": 0.042719073593616486, "step": 4990 }, { "epoch": 1.2956905503634475, "grad_norm": 2.866701126098633, "learning_rate": 1.8126572199305192e-06, "loss": 0.03389344736933708, "step": 4991 }, { "epoch": 1.2959501557632398, "grad_norm": 2.9304370880126953, "learning_rate": 1.811797293677448e-06, "loss": 0.054043225944042206, "step": 4992 }, { "epoch": 1.2962097611630323, "grad_norm": 2.4081225395202637, "learning_rate": 1.8109375332771346e-06, "loss": 0.015519732609391212, "step": 4993 }, { "epoch": 1.2964693665628246, "grad_norm": 2.9819016456604004, "learning_rate": 1.8100779388815631e-06, "loss": 0.06464722007513046, "step": 4994 }, { "epoch": 1.296728971962617, "grad_norm": 2.948733329772949, "learning_rate": 1.8092185106426913e-06, "loss": 0.0662677213549614, "step": 4995 }, { "epoch": 1.2969885773624092, "grad_norm": 2.5189526081085205, "learning_rate": 1.808359248712444e-06, "loss": 0.05663745105266571, "step": 4996 }, { "epoch": 1.2972481827622016, "grad_norm": 2.5681488513946533, "learning_rate": 1.8075001532427216e-06, "loss": 0.061747629195451736, "step": 4997 }, { "epoch": 1.2975077881619939, "grad_norm": 2.33260440826416, "learning_rate": 1.806641224385393e-06, "loss": 0.07296202331781387, "step": 4998 }, { "epoch": 1.2977673935617862, "grad_norm": 4.23645544052124, "learning_rate": 1.8057824622922948e-06, "loss": 0.024712076410651207, "step": 4999 }, { "epoch": 1.2980269989615785, "grad_norm": 3.5226328372955322, "learning_rate": 1.8049238671152381e-06, "loss": 0.025598429143428802, "step": 5000 }, { "epoch": 1.2982866043613708, "grad_norm": 3.577571392059326, "learning_rate": 1.804065439006001e-06, "loss": 0.020521564409136772, "step": 5001 }, { "epoch": 1.298546209761163, "grad_norm": 3.8016514778137207, "learning_rate": 1.8032071781163363e-06, "loss": 0.027459511533379555, "step": 5002 }, { "epoch": 1.2988058151609554, "grad_norm": 4.907610893249512, "learning_rate": 1.802349084597964e-06, "loss": 0.023771997541189194, "step": 5003 }, { "epoch": 1.2990654205607477, "grad_norm": 3.1685991287231445, "learning_rate": 1.8014911586025747e-06, "loss": 0.05544036626815796, "step": 5004 }, { "epoch": 1.29932502596054, "grad_norm": 3.764012575149536, "learning_rate": 1.800633400281831e-06, "loss": 0.051270145922899246, "step": 5005 }, { "epoch": 1.2995846313603323, "grad_norm": 2.462766408920288, "learning_rate": 1.7997758097873647e-06, "loss": 0.013261714950203896, "step": 5006 }, { "epoch": 1.2998442367601246, "grad_norm": 3.762561082839966, "learning_rate": 1.7989183872707767e-06, "loss": 0.02890198677778244, "step": 5007 }, { "epoch": 1.300103842159917, "grad_norm": 3.9707257747650146, "learning_rate": 1.7980611328836422e-06, "loss": 0.027187291532754898, "step": 5008 }, { "epoch": 1.3003634475597092, "grad_norm": 3.3170506954193115, "learning_rate": 1.7972040467775021e-06, "loss": 0.03442877531051636, "step": 5009 }, { "epoch": 1.3006230529595015, "grad_norm": 3.726137638092041, "learning_rate": 1.7963471291038708e-06, "loss": 0.051578640937805176, "step": 5010 }, { "epoch": 1.3008826583592938, "grad_norm": 3.2504372596740723, "learning_rate": 1.7954903800142321e-06, "loss": 0.02413800172507763, "step": 5011 }, { "epoch": 1.3011422637590861, "grad_norm": 2.8990304470062256, "learning_rate": 1.7946337996600371e-06, "loss": 0.014639725908637047, "step": 5012 }, { "epoch": 1.3014018691588785, "grad_norm": 4.244227409362793, "learning_rate": 1.793777388192713e-06, "loss": 0.04250927269458771, "step": 5013 }, { "epoch": 1.3016614745586708, "grad_norm": 3.2980196475982666, "learning_rate": 1.7929211457636514e-06, "loss": 0.05360410735011101, "step": 5014 }, { "epoch": 1.301921079958463, "grad_norm": 2.713608503341675, "learning_rate": 1.7920650725242167e-06, "loss": 0.07668359577655792, "step": 5015 }, { "epoch": 1.3021806853582554, "grad_norm": 2.5421574115753174, "learning_rate": 1.7912091686257438e-06, "loss": 0.07692626118659973, "step": 5016 }, { "epoch": 1.3024402907580477, "grad_norm": 4.227709770202637, "learning_rate": 1.7903534342195366e-06, "loss": 0.04486624896526337, "step": 5017 }, { "epoch": 1.30269989615784, "grad_norm": 4.636695384979248, "learning_rate": 1.7894978694568688e-06, "loss": 0.025566838681697845, "step": 5018 }, { "epoch": 1.3029595015576323, "grad_norm": 2.2132601737976074, "learning_rate": 1.7886424744889852e-06, "loss": 0.018467027693986893, "step": 5019 }, { "epoch": 1.3032191069574246, "grad_norm": 2.668966770172119, "learning_rate": 1.7877872494670996e-06, "loss": 0.020576883107423782, "step": 5020 }, { "epoch": 1.303478712357217, "grad_norm": 4.002267837524414, "learning_rate": 1.7869321945423985e-06, "loss": 0.017142560333013535, "step": 5021 }, { "epoch": 1.3037383177570092, "grad_norm": 3.0929133892059326, "learning_rate": 1.7860773098660316e-06, "loss": 0.024392884224653244, "step": 5022 }, { "epoch": 1.3039979231568017, "grad_norm": 2.712462902069092, "learning_rate": 1.7852225955891256e-06, "loss": 0.050000276416540146, "step": 5023 }, { "epoch": 1.304257528556594, "grad_norm": 3.584657907485962, "learning_rate": 1.7843680518627753e-06, "loss": 0.017265688627958298, "step": 5024 }, { "epoch": 1.3045171339563864, "grad_norm": 3.5311176776885986, "learning_rate": 1.7835136788380436e-06, "loss": 0.07258041948080063, "step": 5025 }, { "epoch": 1.3047767393561787, "grad_norm": 3.515592336654663, "learning_rate": 1.7826594766659638e-06, "loss": 0.03989119827747345, "step": 5026 }, { "epoch": 1.305036344755971, "grad_norm": 3.239520788192749, "learning_rate": 1.7818054454975387e-06, "loss": 0.023976149037480354, "step": 5027 }, { "epoch": 1.3052959501557633, "grad_norm": 3.260345458984375, "learning_rate": 1.7809515854837418e-06, "loss": 0.03368397802114487, "step": 5028 }, { "epoch": 1.3055555555555556, "grad_norm": 2.8766672611236572, "learning_rate": 1.7800978967755179e-06, "loss": 0.03555519878864288, "step": 5029 }, { "epoch": 1.3058151609553479, "grad_norm": 2.057464838027954, "learning_rate": 1.7792443795237764e-06, "loss": 0.07504107058048248, "step": 5030 }, { "epoch": 1.3060747663551402, "grad_norm": 3.145475149154663, "learning_rate": 1.7783910338794019e-06, "loss": 0.06597176939249039, "step": 5031 }, { "epoch": 1.3063343717549325, "grad_norm": 2.7981700897216797, "learning_rate": 1.7775378599932452e-06, "loss": 0.04805843532085419, "step": 5032 }, { "epoch": 1.3065939771547248, "grad_norm": 4.854965686798096, "learning_rate": 1.776684858016129e-06, "loss": 0.022738017141819, "step": 5033 }, { "epoch": 1.3068535825545171, "grad_norm": 3.0203187465667725, "learning_rate": 1.7758320280988444e-06, "loss": 0.0474601611495018, "step": 5034 }, { "epoch": 1.3071131879543094, "grad_norm": 2.344583034515381, "learning_rate": 1.774979370392151e-06, "loss": 0.009879683144390583, "step": 5035 }, { "epoch": 1.3073727933541017, "grad_norm": 2.428560733795166, "learning_rate": 1.7741268850467796e-06, "loss": 0.08510708063840866, "step": 5036 }, { "epoch": 1.307632398753894, "grad_norm": 2.3327279090881348, "learning_rate": 1.7732745722134315e-06, "loss": 0.01263302844017744, "step": 5037 }, { "epoch": 1.3078920041536863, "grad_norm": 3.1132640838623047, "learning_rate": 1.7724224320427741e-06, "loss": 0.061958178877830505, "step": 5038 }, { "epoch": 1.3081516095534786, "grad_norm": 3.8202130794525146, "learning_rate": 1.7715704646854476e-06, "loss": 0.03658550977706909, "step": 5039 }, { "epoch": 1.308411214953271, "grad_norm": 3.6893880367279053, "learning_rate": 1.7707186702920587e-06, "loss": 0.012670634314417839, "step": 5040 }, { "epoch": 1.3086708203530635, "grad_norm": 3.1818127632141113, "learning_rate": 1.7698670490131875e-06, "loss": 0.03006409853696823, "step": 5041 }, { "epoch": 1.3089304257528558, "grad_norm": 3.921358346939087, "learning_rate": 1.7690156009993803e-06, "loss": 0.021526340395212173, "step": 5042 }, { "epoch": 1.309190031152648, "grad_norm": 3.145219564437866, "learning_rate": 1.7681643264011523e-06, "loss": 0.01058365497738123, "step": 5043 }, { "epoch": 1.3094496365524404, "grad_norm": 3.69364595413208, "learning_rate": 1.7673132253689913e-06, "loss": 0.021074660122394562, "step": 5044 }, { "epoch": 1.3097092419522327, "grad_norm": 3.603943347930908, "learning_rate": 1.7664622980533508e-06, "loss": 0.06379405409097672, "step": 5045 }, { "epoch": 1.309968847352025, "grad_norm": 4.974455833435059, "learning_rate": 1.7656115446046558e-06, "loss": 0.029473932459950447, "step": 5046 }, { "epoch": 1.3102284527518173, "grad_norm": 3.8965697288513184, "learning_rate": 1.764760965173301e-06, "loss": 0.05075310543179512, "step": 5047 }, { "epoch": 1.3104880581516096, "grad_norm": 3.1304564476013184, "learning_rate": 1.7639105599096484e-06, "loss": 0.04772388935089111, "step": 5048 }, { "epoch": 1.310747663551402, "grad_norm": 3.2930493354797363, "learning_rate": 1.7630603289640298e-06, "loss": 0.02603595331311226, "step": 5049 }, { "epoch": 1.3110072689511942, "grad_norm": 3.0667712688446045, "learning_rate": 1.7622102724867473e-06, "loss": 0.017600875347852707, "step": 5050 }, { "epoch": 1.3112668743509865, "grad_norm": 3.7073614597320557, "learning_rate": 1.7613603906280708e-06, "loss": 0.027452049776911736, "step": 5051 }, { "epoch": 1.3115264797507789, "grad_norm": 2.657139778137207, "learning_rate": 1.7605106835382424e-06, "loss": 0.01183722261339426, "step": 5052 }, { "epoch": 1.3117860851505712, "grad_norm": 6.529397010803223, "learning_rate": 1.7596611513674667e-06, "loss": 0.017876042053103447, "step": 5053 }, { "epoch": 1.3120456905503635, "grad_norm": 3.899420976638794, "learning_rate": 1.7588117942659239e-06, "loss": 0.06507783383131027, "step": 5054 }, { "epoch": 1.3123052959501558, "grad_norm": 3.356207847595215, "learning_rate": 1.7579626123837612e-06, "loss": 0.0653575137257576, "step": 5055 }, { "epoch": 1.312564901349948, "grad_norm": 2.595304250717163, "learning_rate": 1.7571136058710942e-06, "loss": 0.08024287223815918, "step": 5056 }, { "epoch": 1.3128245067497404, "grad_norm": 1.8013362884521484, "learning_rate": 1.7562647748780076e-06, "loss": 0.09092533588409424, "step": 5057 }, { "epoch": 1.3130841121495327, "grad_norm": 2.616098642349243, "learning_rate": 1.7554161195545545e-06, "loss": 0.046188876032829285, "step": 5058 }, { "epoch": 1.313343717549325, "grad_norm": 3.194586992263794, "learning_rate": 1.754567640050759e-06, "loss": 0.012059744447469711, "step": 5059 }, { "epoch": 1.3136033229491173, "grad_norm": 2.7460451126098633, "learning_rate": 1.7537193365166133e-06, "loss": 0.08731140196323395, "step": 5060 }, { "epoch": 1.3138629283489096, "grad_norm": 2.6987106800079346, "learning_rate": 1.7528712091020756e-06, "loss": 0.013249605894088745, "step": 5061 }, { "epoch": 1.314122533748702, "grad_norm": 4.156635761260986, "learning_rate": 1.7520232579570768e-06, "loss": 0.043161310255527496, "step": 5062 }, { "epoch": 1.3143821391484942, "grad_norm": 2.7548160552978516, "learning_rate": 1.7511754832315165e-06, "loss": 0.05340975895524025, "step": 5063 }, { "epoch": 1.3146417445482865, "grad_norm": 2.995837688446045, "learning_rate": 1.7503278850752604e-06, "loss": 0.028262628242373466, "step": 5064 }, { "epoch": 1.3149013499480788, "grad_norm": 2.5408191680908203, "learning_rate": 1.7494804636381446e-06, "loss": 0.07352820038795471, "step": 5065 }, { "epoch": 1.3151609553478711, "grad_norm": 3.770552635192871, "learning_rate": 1.748633219069974e-06, "loss": 0.021312085911631584, "step": 5066 }, { "epoch": 1.3154205607476634, "grad_norm": 3.0078470706939697, "learning_rate": 1.7477861515205213e-06, "loss": 0.05429397523403168, "step": 5067 }, { "epoch": 1.3156801661474558, "grad_norm": 3.572131395339966, "learning_rate": 1.7469392611395309e-06, "loss": 0.026257172226905823, "step": 5068 }, { "epoch": 1.315939771547248, "grad_norm": 2.5232326984405518, "learning_rate": 1.7460925480767107e-06, "loss": 0.008964220061898232, "step": 5069 }, { "epoch": 1.3161993769470404, "grad_norm": 4.097848415374756, "learning_rate": 1.7452460124817424e-06, "loss": 0.042641229927539825, "step": 5070 }, { "epoch": 1.316458982346833, "grad_norm": 3.8913662433624268, "learning_rate": 1.7443996545042734e-06, "loss": 0.0491836741566658, "step": 5071 }, { "epoch": 1.3167185877466252, "grad_norm": 3.426485776901245, "learning_rate": 1.7435534742939193e-06, "loss": 0.021928369998931885, "step": 5072 }, { "epoch": 1.3169781931464175, "grad_norm": 3.022019386291504, "learning_rate": 1.7427074720002673e-06, "loss": 0.058648064732551575, "step": 5073 }, { "epoch": 1.3172377985462098, "grad_norm": 4.278599262237549, "learning_rate": 1.7418616477728694e-06, "loss": 0.05325569212436676, "step": 5074 }, { "epoch": 1.3174974039460021, "grad_norm": 3.3855369091033936, "learning_rate": 1.74101600176125e-06, "loss": 0.03716864064335823, "step": 5075 }, { "epoch": 1.3177570093457944, "grad_norm": 3.0074145793914795, "learning_rate": 1.740170534114898e-06, "loss": 0.02553921565413475, "step": 5076 }, { "epoch": 1.3180166147455867, "grad_norm": 4.131953716278076, "learning_rate": 1.7393252449832734e-06, "loss": 0.029841916635632515, "step": 5077 }, { "epoch": 1.318276220145379, "grad_norm": 2.839160203933716, "learning_rate": 1.7384801345158048e-06, "loss": 0.030162986367940903, "step": 5078 }, { "epoch": 1.3185358255451713, "grad_norm": 2.244511842727661, "learning_rate": 1.7376352028618877e-06, "loss": 0.006034381687641144, "step": 5079 }, { "epoch": 1.3187954309449637, "grad_norm": 2.2384519577026367, "learning_rate": 1.7367904501708856e-06, "loss": 0.0762028619647026, "step": 5080 }, { "epoch": 1.319055036344756, "grad_norm": 3.484822988510132, "learning_rate": 1.7359458765921334e-06, "loss": 0.015537532977759838, "step": 5081 }, { "epoch": 1.3193146417445483, "grad_norm": 2.740593194961548, "learning_rate": 1.7351014822749308e-06, "loss": 0.04887598380446434, "step": 5082 }, { "epoch": 1.3195742471443406, "grad_norm": 3.5552079677581787, "learning_rate": 1.734257267368549e-06, "loss": 0.011503279209136963, "step": 5083 }, { "epoch": 1.3198338525441329, "grad_norm": 4.527954578399658, "learning_rate": 1.7334132320222244e-06, "loss": 0.03770946338772774, "step": 5084 }, { "epoch": 1.3200934579439252, "grad_norm": 2.926265001296997, "learning_rate": 1.732569376385163e-06, "loss": 0.04071640223264694, "step": 5085 }, { "epoch": 1.3203530633437175, "grad_norm": 3.081554651260376, "learning_rate": 1.7317257006065407e-06, "loss": 0.049368005245923996, "step": 5086 }, { "epoch": 1.3206126687435098, "grad_norm": 2.5329253673553467, "learning_rate": 1.7308822048354996e-06, "loss": 0.011872520670294762, "step": 5087 }, { "epoch": 1.320872274143302, "grad_norm": 2.089766502380371, "learning_rate": 1.7300388892211483e-06, "loss": 0.09665839374065399, "step": 5088 }, { "epoch": 1.3211318795430946, "grad_norm": 2.2213001251220703, "learning_rate": 1.7291957539125686e-06, "loss": 0.07630874961614609, "step": 5089 }, { "epoch": 1.321391484942887, "grad_norm": 3.368185520172119, "learning_rate": 1.7283527990588055e-06, "loss": 0.03396303951740265, "step": 5090 }, { "epoch": 1.3216510903426792, "grad_norm": 2.7854371070861816, "learning_rate": 1.7275100248088766e-06, "loss": 0.008784161880612373, "step": 5091 }, { "epoch": 1.3219106957424716, "grad_norm": 5.071822166442871, "learning_rate": 1.7266674313117615e-06, "loss": 0.016042254865169525, "step": 5092 }, { "epoch": 1.3221703011422639, "grad_norm": 2.7950870990753174, "learning_rate": 1.7258250187164139e-06, "loss": 0.032351747155189514, "step": 5093 }, { "epoch": 1.3224299065420562, "grad_norm": 4.054747581481934, "learning_rate": 1.7249827871717526e-06, "loss": 0.034064579755067825, "step": 5094 }, { "epoch": 1.3226895119418485, "grad_norm": 3.303609848022461, "learning_rate": 1.7241407368266649e-06, "loss": 0.01984173059463501, "step": 5095 }, { "epoch": 1.3229491173416408, "grad_norm": 2.195988416671753, "learning_rate": 1.7232988678300062e-06, "loss": 0.08068567514419556, "step": 5096 }, { "epoch": 1.323208722741433, "grad_norm": 2.5781443119049072, "learning_rate": 1.7224571803305984e-06, "loss": 0.043495964258909225, "step": 5097 }, { "epoch": 1.3234683281412254, "grad_norm": 1.9733333587646484, "learning_rate": 1.721615674477234e-06, "loss": 0.07280915230512619, "step": 5098 }, { "epoch": 1.3237279335410177, "grad_norm": 2.1353886127471924, "learning_rate": 1.720774350418672e-06, "loss": 0.06603831797838211, "step": 5099 }, { "epoch": 1.32398753894081, "grad_norm": 2.689647674560547, "learning_rate": 1.7199332083036374e-06, "loss": 0.05042789876461029, "step": 5100 }, { "epoch": 1.3242471443406023, "grad_norm": 2.382297992706299, "learning_rate": 1.719092248280827e-06, "loss": 0.003011024557054043, "step": 5101 }, { "epoch": 1.3245067497403946, "grad_norm": 2.8191583156585693, "learning_rate": 1.7182514704989029e-06, "loss": 0.05376758798956871, "step": 5102 }, { "epoch": 1.324766355140187, "grad_norm": 2.5314133167266846, "learning_rate": 1.7174108751064939e-06, "loss": 0.016383761540055275, "step": 5103 }, { "epoch": 1.3250259605399792, "grad_norm": 2.528533935546875, "learning_rate": 1.7165704622521994e-06, "loss": 0.07473023980855942, "step": 5104 }, { "epoch": 1.3252855659397715, "grad_norm": 2.549607515335083, "learning_rate": 1.7157302320845848e-06, "loss": 0.07219646871089935, "step": 5105 }, { "epoch": 1.3255451713395638, "grad_norm": 2.9461171627044678, "learning_rate": 1.7148901847521837e-06, "loss": 0.030702706426382065, "step": 5106 }, { "epoch": 1.3258047767393562, "grad_norm": 2.3440253734588623, "learning_rate": 1.7140503204034973e-06, "loss": 0.0623757541179657, "step": 5107 }, { "epoch": 1.3260643821391485, "grad_norm": 2.7174551486968994, "learning_rate": 1.713210639186993e-06, "loss": 0.060846805572509766, "step": 5108 }, { "epoch": 1.3263239875389408, "grad_norm": 3.4275777339935303, "learning_rate": 1.7123711412511096e-06, "loss": 0.026311613619327545, "step": 5109 }, { "epoch": 1.326583592938733, "grad_norm": 2.5995872020721436, "learning_rate": 1.7115318267442496e-06, "loss": 0.057221848517656326, "step": 5110 }, { "epoch": 1.3268431983385254, "grad_norm": 3.692329168319702, "learning_rate": 1.7106926958147841e-06, "loss": 0.03419935703277588, "step": 5111 }, { "epoch": 1.3271028037383177, "grad_norm": 2.291666269302368, "learning_rate": 1.7098537486110539e-06, "loss": 0.07285919785499573, "step": 5112 }, { "epoch": 1.32736240913811, "grad_norm": 3.1783995628356934, "learning_rate": 1.709014985281364e-06, "loss": 0.03250672668218613, "step": 5113 }, { "epoch": 1.3276220145379023, "grad_norm": 2.690647840499878, "learning_rate": 1.70817640597399e-06, "loss": 0.07628589123487473, "step": 5114 }, { "epoch": 1.3278816199376946, "grad_norm": 3.005875825881958, "learning_rate": 1.7073380108371728e-06, "loss": 0.059551481157541275, "step": 5115 }, { "epoch": 1.328141225337487, "grad_norm": 4.503575325012207, "learning_rate": 1.7064998000191207e-06, "loss": 0.027278538793325424, "step": 5116 }, { "epoch": 1.3284008307372792, "grad_norm": 2.2271695137023926, "learning_rate": 1.705661773668012e-06, "loss": 0.05502280220389366, "step": 5117 }, { "epoch": 1.3286604361370715, "grad_norm": 3.0683445930480957, "learning_rate": 1.7048239319319894e-06, "loss": 0.026834405958652496, "step": 5118 }, { "epoch": 1.328920041536864, "grad_norm": 2.6955690383911133, "learning_rate": 1.7039862749591635e-06, "loss": 0.026278752833604813, "step": 5119 }, { "epoch": 1.3291796469366564, "grad_norm": 3.354785680770874, "learning_rate": 1.703148802897614e-06, "loss": 0.06912544369697571, "step": 5120 }, { "epoch": 1.3294392523364487, "grad_norm": 2.813710927963257, "learning_rate": 1.7023115158953862e-06, "loss": 0.05136127769947052, "step": 5121 }, { "epoch": 1.329698857736241, "grad_norm": 3.317195415496826, "learning_rate": 1.7014744141004948e-06, "loss": 0.014109517447650433, "step": 5122 }, { "epoch": 1.3299584631360333, "grad_norm": 1.9570306539535522, "learning_rate": 1.7006374976609174e-06, "loss": 0.011297551915049553, "step": 5123 }, { "epoch": 1.3302180685358256, "grad_norm": 2.1400325298309326, "learning_rate": 1.6998007667246037e-06, "loss": 0.07547225058078766, "step": 5124 }, { "epoch": 1.3304776739356179, "grad_norm": 2.70592999458313, "learning_rate": 1.698964221439469e-06, "loss": 0.05653621256351471, "step": 5125 }, { "epoch": 1.3307372793354102, "grad_norm": 4.147345542907715, "learning_rate": 1.698127861953393e-06, "loss": 0.06719651818275452, "step": 5126 }, { "epoch": 1.3309968847352025, "grad_norm": 3.2334647178649902, "learning_rate": 1.697291688414227e-06, "loss": 0.039321210235357285, "step": 5127 }, { "epoch": 1.3312564901349948, "grad_norm": 2.6190435886383057, "learning_rate": 1.6964557009697858e-06, "loss": 0.0196980107575655, "step": 5128 }, { "epoch": 1.3315160955347871, "grad_norm": 2.8567049503326416, "learning_rate": 1.6956198997678546e-06, "loss": 0.04687752574682236, "step": 5129 }, { "epoch": 1.3317757009345794, "grad_norm": 3.583239793777466, "learning_rate": 1.6947842849561835e-06, "loss": 0.010603571310639381, "step": 5130 }, { "epoch": 1.3320353063343717, "grad_norm": 3.8055930137634277, "learning_rate": 1.6939488566824885e-06, "loss": 0.04384146258234978, "step": 5131 }, { "epoch": 1.332294911734164, "grad_norm": 2.125312328338623, "learning_rate": 1.6931136150944561e-06, "loss": 0.00376348034478724, "step": 5132 }, { "epoch": 1.3325545171339563, "grad_norm": 4.054974555969238, "learning_rate": 1.692278560339737e-06, "loss": 0.01253633014857769, "step": 5133 }, { "epoch": 1.3328141225337486, "grad_norm": 4.88319206237793, "learning_rate": 1.6914436925659488e-06, "loss": 0.01282299030572176, "step": 5134 }, { "epoch": 1.333073727933541, "grad_norm": 2.8922438621520996, "learning_rate": 1.6906090119206792e-06, "loss": 0.07416320592164993, "step": 5135 }, { "epoch": 1.3333333333333333, "grad_norm": 3.406193733215332, "learning_rate": 1.689774518551478e-06, "loss": 0.043856181204319, "step": 5136 }, { "epoch": 1.3335929387331256, "grad_norm": 3.150209903717041, "learning_rate": 1.6889402126058674e-06, "loss": 0.06262329965829849, "step": 5137 }, { "epoch": 1.333852544132918, "grad_norm": 2.1338584423065186, "learning_rate": 1.6881060942313322e-06, "loss": 0.013265668414533138, "step": 5138 }, { "epoch": 1.3341121495327104, "grad_norm": 3.441638231277466, "learning_rate": 1.687272163575324e-06, "loss": 0.05543292313814163, "step": 5139 }, { "epoch": 1.3343717549325027, "grad_norm": 3.9054341316223145, "learning_rate": 1.6864384207852655e-06, "loss": 0.04661890119314194, "step": 5140 }, { "epoch": 1.334631360332295, "grad_norm": 4.172078609466553, "learning_rate": 1.6856048660085418e-06, "loss": 0.05115208029747009, "step": 5141 }, { "epoch": 1.3348909657320873, "grad_norm": 2.8709769248962402, "learning_rate": 1.6847714993925051e-06, "loss": 0.018431156873703003, "step": 5142 }, { "epoch": 1.3351505711318796, "grad_norm": 3.2645254135131836, "learning_rate": 1.6839383210844779e-06, "loss": 0.0465671569108963, "step": 5143 }, { "epoch": 1.335410176531672, "grad_norm": 2.7334022521972656, "learning_rate": 1.683105331231745e-06, "loss": 0.053823072463274, "step": 5144 }, { "epoch": 1.3356697819314642, "grad_norm": 3.1326141357421875, "learning_rate": 1.6822725299815618e-06, "loss": 0.05119115114212036, "step": 5145 }, { "epoch": 1.3359293873312565, "grad_norm": 1.6083683967590332, "learning_rate": 1.6814399174811475e-06, "loss": 0.0031712818890810013, "step": 5146 }, { "epoch": 1.3361889927310489, "grad_norm": 2.3048269748687744, "learning_rate": 1.6806074938776884e-06, "loss": 0.07431955635547638, "step": 5147 }, { "epoch": 1.3364485981308412, "grad_norm": 3.249485969543457, "learning_rate": 1.6797752593183393e-06, "loss": 0.02675403468310833, "step": 5148 }, { "epoch": 1.3367082035306335, "grad_norm": 1.2411067485809326, "learning_rate": 1.6789432139502192e-06, "loss": 0.0022177163045853376, "step": 5149 }, { "epoch": 1.3369678089304258, "grad_norm": 3.5073866844177246, "learning_rate": 1.6781113579204145e-06, "loss": 0.07432146370410919, "step": 5150 }, { "epoch": 1.337227414330218, "grad_norm": 3.5757081508636475, "learning_rate": 1.6772796913759792e-06, "loss": 0.032291118055582047, "step": 5151 }, { "epoch": 1.3374870197300104, "grad_norm": 4.483063697814941, "learning_rate": 1.6764482144639326e-06, "loss": 0.015041008591651917, "step": 5152 }, { "epoch": 1.3377466251298027, "grad_norm": 4.028415203094482, "learning_rate": 1.6756169273312606e-06, "loss": 0.018694579601287842, "step": 5153 }, { "epoch": 1.338006230529595, "grad_norm": 3.057328701019287, "learning_rate": 1.6747858301249148e-06, "loss": 0.03221721947193146, "step": 5154 }, { "epoch": 1.3382658359293873, "grad_norm": 2.3764381408691406, "learning_rate": 1.6739549229918154e-06, "loss": 0.06575323641300201, "step": 5155 }, { "epoch": 1.3385254413291796, "grad_norm": 2.1660666465759277, "learning_rate": 1.6731242060788489e-06, "loss": 0.013442840427160263, "step": 5156 }, { "epoch": 1.338785046728972, "grad_norm": 3.164903402328491, "learning_rate": 1.6722936795328637e-06, "loss": 0.032616548240184784, "step": 5157 }, { "epoch": 1.3390446521287642, "grad_norm": 2.6331424713134766, "learning_rate": 1.671463343500681e-06, "loss": 0.023242458701133728, "step": 5158 }, { "epoch": 1.3393042575285565, "grad_norm": 2.2003118991851807, "learning_rate": 1.670633198129083e-06, "loss": 0.09072147309780121, "step": 5159 }, { "epoch": 1.3395638629283488, "grad_norm": 2.0708038806915283, "learning_rate": 1.669803243564822e-06, "loss": 0.012780449353158474, "step": 5160 }, { "epoch": 1.3398234683281411, "grad_norm": 3.219597101211548, "learning_rate": 1.6689734799546145e-06, "loss": 0.018496904522180557, "step": 5161 }, { "epoch": 1.3400830737279334, "grad_norm": 3.48321270942688, "learning_rate": 1.6681439074451424e-06, "loss": 0.0342629998922348, "step": 5162 }, { "epoch": 1.3403426791277258, "grad_norm": 2.8782994747161865, "learning_rate": 1.6673145261830577e-06, "loss": 0.035702094435691833, "step": 5163 }, { "epoch": 1.340602284527518, "grad_norm": 2.433662176132202, "learning_rate": 1.6664853363149739e-06, "loss": 0.015583761036396027, "step": 5164 }, { "epoch": 1.3408618899273104, "grad_norm": 4.215040683746338, "learning_rate": 1.6656563379874724e-06, "loss": 0.0067232754081487656, "step": 5165 }, { "epoch": 1.3411214953271027, "grad_norm": 2.491732358932495, "learning_rate": 1.6648275313471039e-06, "loss": 0.005421106703579426, "step": 5166 }, { "epoch": 1.341381100726895, "grad_norm": 2.9797110557556152, "learning_rate": 1.6639989165403794e-06, "loss": 0.06430623680353165, "step": 5167 }, { "epoch": 1.3416407061266875, "grad_norm": 2.9293174743652344, "learning_rate": 1.6631704937137811e-06, "loss": 0.035406459122896194, "step": 5168 }, { "epoch": 1.3419003115264798, "grad_norm": 4.131493091583252, "learning_rate": 1.6623422630137542e-06, "loss": 0.012901929207146168, "step": 5169 }, { "epoch": 1.3421599169262721, "grad_norm": 2.0456788539886475, "learning_rate": 1.6615142245867107e-06, "loss": 0.07605002820491791, "step": 5170 }, { "epoch": 1.3424195223260644, "grad_norm": 2.5872089862823486, "learning_rate": 1.66068637857903e-06, "loss": 0.06320337951183319, "step": 5171 }, { "epoch": 1.3426791277258567, "grad_norm": 3.153604507446289, "learning_rate": 1.6598587251370557e-06, "loss": 0.05112612247467041, "step": 5172 }, { "epoch": 1.342938733125649, "grad_norm": 1.890228509902954, "learning_rate": 1.659031264407097e-06, "loss": 0.010861034505069256, "step": 5173 }, { "epoch": 1.3431983385254413, "grad_norm": 3.162839889526367, "learning_rate": 1.6582039965354313e-06, "loss": 0.012689793482422829, "step": 5174 }, { "epoch": 1.3434579439252337, "grad_norm": 6.2988691329956055, "learning_rate": 1.6573769216683e-06, "loss": 0.0458158440887928, "step": 5175 }, { "epoch": 1.343717549325026, "grad_norm": 3.6935434341430664, "learning_rate": 1.6565500399519116e-06, "loss": 0.028352104127407074, "step": 5176 }, { "epoch": 1.3439771547248183, "grad_norm": 3.8039886951446533, "learning_rate": 1.6557233515324398e-06, "loss": 0.029660530388355255, "step": 5177 }, { "epoch": 1.3442367601246106, "grad_norm": 2.915376901626587, "learning_rate": 1.6548968565560226e-06, "loss": 0.07270964980125427, "step": 5178 }, { "epoch": 1.3444963655244029, "grad_norm": 3.4468109607696533, "learning_rate": 1.6540705551687686e-06, "loss": 0.06390120089054108, "step": 5179 }, { "epoch": 1.3447559709241952, "grad_norm": 3.3545947074890137, "learning_rate": 1.653244447516745e-06, "loss": 0.036962565034627914, "step": 5180 }, { "epoch": 1.3450155763239875, "grad_norm": 3.7235116958618164, "learning_rate": 1.6524185337459909e-06, "loss": 0.015973815694451332, "step": 5181 }, { "epoch": 1.3452751817237798, "grad_norm": 2.386286973953247, "learning_rate": 1.6515928140025095e-06, "loss": 0.05940597131848335, "step": 5182 }, { "epoch": 1.345534787123572, "grad_norm": 3.3769774436950684, "learning_rate": 1.6507672884322685e-06, "loss": 0.05158163234591484, "step": 5183 }, { "epoch": 1.3457943925233644, "grad_norm": 2.554687023162842, "learning_rate": 1.649941957181202e-06, "loss": 0.04981694370508194, "step": 5184 }, { "epoch": 1.3460539979231567, "grad_norm": 3.419281244277954, "learning_rate": 1.6491168203952085e-06, "loss": 0.03939047455787659, "step": 5185 }, { "epoch": 1.3463136033229492, "grad_norm": 2.170781373977661, "learning_rate": 1.6482918782201537e-06, "loss": 0.060266971588134766, "step": 5186 }, { "epoch": 1.3465732087227416, "grad_norm": 2.039020299911499, "learning_rate": 1.6474671308018714e-06, "loss": 0.0671895295381546, "step": 5187 }, { "epoch": 1.3468328141225339, "grad_norm": 3.270986795425415, "learning_rate": 1.646642578286154e-06, "loss": 0.017443636432290077, "step": 5188 }, { "epoch": 1.3470924195223262, "grad_norm": 2.8618316650390625, "learning_rate": 1.645818220818766e-06, "loss": 0.014091571792960167, "step": 5189 }, { "epoch": 1.3473520249221185, "grad_norm": 3.1090247631073, "learning_rate": 1.644994058545433e-06, "loss": 0.05657120794057846, "step": 5190 }, { "epoch": 1.3476116303219108, "grad_norm": 2.839688539505005, "learning_rate": 1.6441700916118505e-06, "loss": 0.020478079095482826, "step": 5191 }, { "epoch": 1.347871235721703, "grad_norm": 1.7240523099899292, "learning_rate": 1.6433463201636755e-06, "loss": 0.009043015539646149, "step": 5192 }, { "epoch": 1.3481308411214954, "grad_norm": 3.036849021911621, "learning_rate": 1.642522744346531e-06, "loss": 0.023669593036174774, "step": 5193 }, { "epoch": 1.3483904465212877, "grad_norm": 3.474095106124878, "learning_rate": 1.6416993643060081e-06, "loss": 0.028547005727887154, "step": 5194 }, { "epoch": 1.34865005192108, "grad_norm": 2.658902645111084, "learning_rate": 1.6408761801876612e-06, "loss": 0.04362862557172775, "step": 5195 }, { "epoch": 1.3489096573208723, "grad_norm": 3.8867363929748535, "learning_rate": 1.6400531921370095e-06, "loss": 0.0192631296813488, "step": 5196 }, { "epoch": 1.3491692627206646, "grad_norm": 3.0280869007110596, "learning_rate": 1.6392304002995397e-06, "loss": 0.012442352250218391, "step": 5197 }, { "epoch": 1.349428868120457, "grad_norm": 2.9487485885620117, "learning_rate": 1.6384078048207008e-06, "loss": 0.07050319015979767, "step": 5198 }, { "epoch": 1.3496884735202492, "grad_norm": 4.096578121185303, "learning_rate": 1.6375854058459107e-06, "loss": 0.043896954506635666, "step": 5199 }, { "epoch": 1.3499480789200415, "grad_norm": 3.526982069015503, "learning_rate": 1.63676320352055e-06, "loss": 0.037886958569288254, "step": 5200 }, { "epoch": 1.3502076843198338, "grad_norm": 3.604409694671631, "learning_rate": 1.6359411979899648e-06, "loss": 0.04714202508330345, "step": 5201 }, { "epoch": 1.3504672897196262, "grad_norm": 3.3937225341796875, "learning_rate": 1.6351193893994675e-06, "loss": 0.024844810366630554, "step": 5202 }, { "epoch": 1.3507268951194185, "grad_norm": 4.036552906036377, "learning_rate": 1.6342977778943353e-06, "loss": 0.054023172706365585, "step": 5203 }, { "epoch": 1.3509865005192108, "grad_norm": 3.4213335514068604, "learning_rate": 1.6334763636198083e-06, "loss": 0.027469860389828682, "step": 5204 }, { "epoch": 1.351246105919003, "grad_norm": 3.2596166133880615, "learning_rate": 1.6326551467210965e-06, "loss": 0.04419202357530594, "step": 5205 }, { "epoch": 1.3515057113187954, "grad_norm": 3.476862907409668, "learning_rate": 1.6318341273433704e-06, "loss": 0.013914567418396473, "step": 5206 }, { "epoch": 1.3517653167185877, "grad_norm": 3.2111165523529053, "learning_rate": 1.6310133056317678e-06, "loss": 0.03786808252334595, "step": 5207 }, { "epoch": 1.35202492211838, "grad_norm": 1.923401951789856, "learning_rate": 1.6301926817313923e-06, "loss": 0.004068832844495773, "step": 5208 }, { "epoch": 1.3522845275181723, "grad_norm": 3.4811270236968994, "learning_rate": 1.629372255787309e-06, "loss": 0.033184703439474106, "step": 5209 }, { "epoch": 1.3525441329179646, "grad_norm": 2.7907016277313232, "learning_rate": 1.6285520279445539e-06, "loss": 0.044259846210479736, "step": 5210 }, { "epoch": 1.352803738317757, "grad_norm": 3.6586787700653076, "learning_rate": 1.627731998348121e-06, "loss": 0.031197287142276764, "step": 5211 }, { "epoch": 1.3530633437175492, "grad_norm": 4.071527481079102, "learning_rate": 1.626912167142974e-06, "loss": 0.03217703849077225, "step": 5212 }, { "epoch": 1.3533229491173415, "grad_norm": 2.535223960876465, "learning_rate": 1.6260925344740414e-06, "loss": 0.0065780808217823505, "step": 5213 }, { "epoch": 1.3535825545171338, "grad_norm": 3.306432008743286, "learning_rate": 1.6252731004862149e-06, "loss": 0.0403820276260376, "step": 5214 }, { "epoch": 1.3538421599169261, "grad_norm": 3.660262107849121, "learning_rate": 1.6244538653243508e-06, "loss": 0.013846756890416145, "step": 5215 }, { "epoch": 1.3541017653167187, "grad_norm": 2.786975383758545, "learning_rate": 1.6236348291332716e-06, "loss": 0.014058948494493961, "step": 5216 }, { "epoch": 1.354361370716511, "grad_norm": 2.339477300643921, "learning_rate": 1.6228159920577638e-06, "loss": 0.07245254516601562, "step": 5217 }, { "epoch": 1.3546209761163033, "grad_norm": 2.8303422927856445, "learning_rate": 1.6219973542425816e-06, "loss": 0.04032054543495178, "step": 5218 }, { "epoch": 1.3548805815160956, "grad_norm": 5.583521842956543, "learning_rate": 1.621178915832437e-06, "loss": 0.03345361351966858, "step": 5219 }, { "epoch": 1.355140186915888, "grad_norm": 2.6989200115203857, "learning_rate": 1.6203606769720148e-06, "loss": 0.07289420068264008, "step": 5220 }, { "epoch": 1.3553997923156802, "grad_norm": 3.4693145751953125, "learning_rate": 1.6195426378059587e-06, "loss": 0.07383227348327637, "step": 5221 }, { "epoch": 1.3556593977154725, "grad_norm": 2.7507245540618896, "learning_rate": 1.6187247984788811e-06, "loss": 0.06009713560342789, "step": 5222 }, { "epoch": 1.3559190031152648, "grad_norm": 2.8028249740600586, "learning_rate": 1.6179071591353563e-06, "loss": 0.02416723407804966, "step": 5223 }, { "epoch": 1.3561786085150571, "grad_norm": 3.033846139907837, "learning_rate": 1.6170897199199231e-06, "loss": 0.007054668385535479, "step": 5224 }, { "epoch": 1.3564382139148494, "grad_norm": 2.833050012588501, "learning_rate": 1.6162724809770885e-06, "loss": 0.04916681349277496, "step": 5225 }, { "epoch": 1.3566978193146417, "grad_norm": 2.826772451400757, "learning_rate": 1.6154554424513203e-06, "loss": 0.030374636873602867, "step": 5226 }, { "epoch": 1.356957424714434, "grad_norm": 2.6922223567962646, "learning_rate": 1.6146386044870514e-06, "loss": 0.014336109161376953, "step": 5227 }, { "epoch": 1.3572170301142263, "grad_norm": 2.789294719696045, "learning_rate": 1.6138219672286814e-06, "loss": 0.03035261109471321, "step": 5228 }, { "epoch": 1.3574766355140186, "grad_norm": 4.363864898681641, "learning_rate": 1.6130055308205723e-06, "loss": 0.017128922045230865, "step": 5229 }, { "epoch": 1.357736240913811, "grad_norm": 2.9774587154388428, "learning_rate": 1.612189295407052e-06, "loss": 0.051109347492456436, "step": 5230 }, { "epoch": 1.3579958463136033, "grad_norm": 2.9399309158325195, "learning_rate": 1.6113732611324118e-06, "loss": 0.06856729090213776, "step": 5231 }, { "epoch": 1.3582554517133956, "grad_norm": 1.8562043905258179, "learning_rate": 1.6105574281409073e-06, "loss": 0.0068979342468082905, "step": 5232 }, { "epoch": 1.3585150571131879, "grad_norm": 2.855806827545166, "learning_rate": 1.6097417965767608e-06, "loss": 0.04890657588839531, "step": 5233 }, { "epoch": 1.3587746625129804, "grad_norm": 2.216423511505127, "learning_rate": 1.608926366584156e-06, "loss": 0.06565973162651062, "step": 5234 }, { "epoch": 1.3590342679127727, "grad_norm": 3.210721015930176, "learning_rate": 1.6081111383072414e-06, "loss": 0.02121070772409439, "step": 5235 }, { "epoch": 1.359293873312565, "grad_norm": 2.6969258785247803, "learning_rate": 1.6072961118901328e-06, "loss": 0.06052573770284653, "step": 5236 }, { "epoch": 1.3595534787123573, "grad_norm": 2.4011528491973877, "learning_rate": 1.6064812874769073e-06, "loss": 0.07255013287067413, "step": 5237 }, { "epoch": 1.3598130841121496, "grad_norm": 2.7222306728363037, "learning_rate": 1.605666665211606e-06, "loss": 0.021517114713788033, "step": 5238 }, { "epoch": 1.360072689511942, "grad_norm": 3.1385128498077393, "learning_rate": 1.6048522452382375e-06, "loss": 0.04383926838636398, "step": 5239 }, { "epoch": 1.3603322949117342, "grad_norm": 3.07116961479187, "learning_rate": 1.604038027700771e-06, "loss": 0.04625442251563072, "step": 5240 }, { "epoch": 1.3605919003115265, "grad_norm": 3.0700483322143555, "learning_rate": 1.6032240127431433e-06, "loss": 0.05273239687085152, "step": 5241 }, { "epoch": 1.3608515057113189, "grad_norm": 3.633965492248535, "learning_rate": 1.6024102005092513e-06, "loss": 0.052040088921785355, "step": 5242 }, { "epoch": 1.3611111111111112, "grad_norm": 6.3263092041015625, "learning_rate": 1.6015965911429593e-06, "loss": 0.018451107665896416, "step": 5243 }, { "epoch": 1.3613707165109035, "grad_norm": 2.4477319717407227, "learning_rate": 1.600783184788096e-06, "loss": 0.08714009821414948, "step": 5244 }, { "epoch": 1.3616303219106958, "grad_norm": 2.7875587940216064, "learning_rate": 1.5999699815884518e-06, "loss": 0.06682421267032623, "step": 5245 }, { "epoch": 1.361889927310488, "grad_norm": 3.0435404777526855, "learning_rate": 1.5991569816877827e-06, "loss": 0.028354540467262268, "step": 5246 }, { "epoch": 1.3621495327102804, "grad_norm": 2.8947486877441406, "learning_rate": 1.5983441852298075e-06, "loss": 0.050023335963487625, "step": 5247 }, { "epoch": 1.3624091381100727, "grad_norm": 3.3342342376708984, "learning_rate": 1.5975315923582113e-06, "loss": 0.025317825376987457, "step": 5248 }, { "epoch": 1.362668743509865, "grad_norm": 2.8555521965026855, "learning_rate": 1.5967192032166429e-06, "loss": 0.05730310082435608, "step": 5249 }, { "epoch": 1.3629283489096573, "grad_norm": 2.19272780418396, "learning_rate": 1.5959070179487112e-06, "loss": 0.004408859182149172, "step": 5250 }, { "epoch": 1.3631879543094496, "grad_norm": 2.7816293239593506, "learning_rate": 1.5950950366979943e-06, "loss": 0.027027592062950134, "step": 5251 }, { "epoch": 1.363447559709242, "grad_norm": 1.7148629426956177, "learning_rate": 1.5942832596080304e-06, "loss": 0.011308502405881882, "step": 5252 }, { "epoch": 1.3637071651090342, "grad_norm": 2.6678569316864014, "learning_rate": 1.5934716868223244e-06, "loss": 0.034995704889297485, "step": 5253 }, { "epoch": 1.3639667705088265, "grad_norm": 2.884157180786133, "learning_rate": 1.592660318484343e-06, "loss": 0.016826216131448746, "step": 5254 }, { "epoch": 1.3642263759086188, "grad_norm": 3.907423257827759, "learning_rate": 1.5918491547375175e-06, "loss": 0.024981901049613953, "step": 5255 }, { "epoch": 1.3644859813084111, "grad_norm": 3.472278594970703, "learning_rate": 1.591038195725244e-06, "loss": 0.012040423229336739, "step": 5256 }, { "epoch": 1.3647455867082035, "grad_norm": 2.3713855743408203, "learning_rate": 1.5902274415908802e-06, "loss": 0.07785490900278091, "step": 5257 }, { "epoch": 1.3650051921079958, "grad_norm": 2.9986674785614014, "learning_rate": 1.5894168924777493e-06, "loss": 0.07427840679883957, "step": 5258 }, { "epoch": 1.365264797507788, "grad_norm": 2.6176161766052246, "learning_rate": 1.5886065485291385e-06, "loss": 0.07472435384988785, "step": 5259 }, { "epoch": 1.3655244029075804, "grad_norm": 3.2477807998657227, "learning_rate": 1.587796409888297e-06, "loss": 0.05703788250684738, "step": 5260 }, { "epoch": 1.3657840083073727, "grad_norm": 4.010383605957031, "learning_rate": 1.5869864766984397e-06, "loss": 0.05449774116277695, "step": 5261 }, { "epoch": 1.366043613707165, "grad_norm": 3.3385093212127686, "learning_rate": 1.586176749102744e-06, "loss": 0.0400555357336998, "step": 5262 }, { "epoch": 1.3663032191069573, "grad_norm": 3.696767807006836, "learning_rate": 1.5853672272443499e-06, "loss": 0.054015353322029114, "step": 5263 }, { "epoch": 1.3665628245067498, "grad_norm": 3.5727949142456055, "learning_rate": 1.5845579112663651e-06, "loss": 0.057375457137823105, "step": 5264 }, { "epoch": 1.3668224299065421, "grad_norm": 0.9229604601860046, "learning_rate": 1.5837488013118558e-06, "loss": 0.0021603330969810486, "step": 5265 }, { "epoch": 1.3670820353063344, "grad_norm": 3.5566024780273438, "learning_rate": 1.5829398975238542e-06, "loss": 0.04701432213187218, "step": 5266 }, { "epoch": 1.3673416407061267, "grad_norm": 2.9737391471862793, "learning_rate": 1.5821312000453575e-06, "loss": 0.03258872404694557, "step": 5267 }, { "epoch": 1.367601246105919, "grad_norm": 3.0639233589172363, "learning_rate": 1.581322709019324e-06, "loss": 0.024844210594892502, "step": 5268 }, { "epoch": 1.3678608515057114, "grad_norm": 2.6819851398468018, "learning_rate": 1.580514424588675e-06, "loss": 0.007257208693772554, "step": 5269 }, { "epoch": 1.3681204569055037, "grad_norm": 2.59183406829834, "learning_rate": 1.5797063468962998e-06, "loss": 0.011108295992016792, "step": 5270 }, { "epoch": 1.368380062305296, "grad_norm": 2.771688938140869, "learning_rate": 1.5788984760850445e-06, "loss": 0.007114533334970474, "step": 5271 }, { "epoch": 1.3686396677050883, "grad_norm": 2.8774030208587646, "learning_rate": 1.5780908122977265e-06, "loss": 0.07060928642749786, "step": 5272 }, { "epoch": 1.3688992731048806, "grad_norm": 2.8364994525909424, "learning_rate": 1.5772833556771177e-06, "loss": 0.00653162132948637, "step": 5273 }, { "epoch": 1.3691588785046729, "grad_norm": 1.9112673997879028, "learning_rate": 1.57647610636596e-06, "loss": 0.06485914438962936, "step": 5274 }, { "epoch": 1.3694184839044652, "grad_norm": 3.547715902328491, "learning_rate": 1.5756690645069573e-06, "loss": 0.007964758202433586, "step": 5275 }, { "epoch": 1.3696780893042575, "grad_norm": 3.6548216342926025, "learning_rate": 1.5748622302427755e-06, "loss": 0.039914608001708984, "step": 5276 }, { "epoch": 1.3699376947040498, "grad_norm": 4.636048793792725, "learning_rate": 1.5740556037160443e-06, "loss": 0.052653439342975616, "step": 5277 }, { "epoch": 1.370197300103842, "grad_norm": 3.6492648124694824, "learning_rate": 1.5732491850693557e-06, "loss": 0.016062408685684204, "step": 5278 }, { "epoch": 1.3704569055036344, "grad_norm": 3.4892358779907227, "learning_rate": 1.5724429744452674e-06, "loss": 0.05199816823005676, "step": 5279 }, { "epoch": 1.3707165109034267, "grad_norm": 3.7321908473968506, "learning_rate": 1.5716369719863003e-06, "loss": 0.0476544052362442, "step": 5280 }, { "epoch": 1.370976116303219, "grad_norm": 3.8495421409606934, "learning_rate": 1.5708311778349341e-06, "loss": 0.037423521280288696, "step": 5281 }, { "epoch": 1.3712357217030116, "grad_norm": 3.0640439987182617, "learning_rate": 1.5700255921336168e-06, "loss": 0.031237632036209106, "step": 5282 }, { "epoch": 1.3714953271028039, "grad_norm": 4.391622066497803, "learning_rate": 1.5692202150247566e-06, "loss": 0.022465217858552933, "step": 5283 }, { "epoch": 1.3717549325025962, "grad_norm": 2.9263367652893066, "learning_rate": 1.5684150466507263e-06, "loss": 0.0816192701458931, "step": 5284 }, { "epoch": 1.3720145379023885, "grad_norm": 3.6848275661468506, "learning_rate": 1.567610087153862e-06, "loss": 0.059549376368522644, "step": 5285 }, { "epoch": 1.3722741433021808, "grad_norm": 2.750619411468506, "learning_rate": 1.5668053366764596e-06, "loss": 0.04852554202079773, "step": 5286 }, { "epoch": 1.372533748701973, "grad_norm": 2.2862985134124756, "learning_rate": 1.566000795360783e-06, "loss": 0.016971275210380554, "step": 5287 }, { "epoch": 1.3727933541017654, "grad_norm": 2.140514373779297, "learning_rate": 1.565196463349056e-06, "loss": 0.05599784106016159, "step": 5288 }, { "epoch": 1.3730529595015577, "grad_norm": 4.393212795257568, "learning_rate": 1.564392340783465e-06, "loss": 0.008321471512317657, "step": 5289 }, { "epoch": 1.37331256490135, "grad_norm": 2.8467214107513428, "learning_rate": 1.563588427806162e-06, "loss": 0.024770528078079224, "step": 5290 }, { "epoch": 1.3735721703011423, "grad_norm": 2.4944674968719482, "learning_rate": 1.5627847245592598e-06, "loss": 0.06539513170719147, "step": 5291 }, { "epoch": 1.3738317757009346, "grad_norm": 2.8174729347229004, "learning_rate": 1.5619812311848336e-06, "loss": 0.06813803315162659, "step": 5292 }, { "epoch": 1.374091381100727, "grad_norm": 2.881749391555786, "learning_rate": 1.561177947824925e-06, "loss": 0.06630530953407288, "step": 5293 }, { "epoch": 1.3743509865005192, "grad_norm": 2.928558588027954, "learning_rate": 1.5603748746215338e-06, "loss": 0.011793941259384155, "step": 5294 }, { "epoch": 1.3746105919003115, "grad_norm": 2.377593517303467, "learning_rate": 1.5595720117166268e-06, "loss": 0.06675609201192856, "step": 5295 }, { "epoch": 1.3748701973001038, "grad_norm": 3.1143839359283447, "learning_rate": 1.5587693592521307e-06, "loss": 0.018235525116324425, "step": 5296 }, { "epoch": 1.3751298026998962, "grad_norm": 3.403409481048584, "learning_rate": 1.557966917369936e-06, "loss": 0.02003021538257599, "step": 5297 }, { "epoch": 1.3753894080996885, "grad_norm": 4.503032684326172, "learning_rate": 1.5571646862118968e-06, "loss": 0.01691364124417305, "step": 5298 }, { "epoch": 1.3756490134994808, "grad_norm": 4.386319637298584, "learning_rate": 1.5563626659198289e-06, "loss": 0.01147824339568615, "step": 5299 }, { "epoch": 1.375908618899273, "grad_norm": 3.3689844608306885, "learning_rate": 1.55556085663551e-06, "loss": 0.05453705042600632, "step": 5300 }, { "epoch": 1.3761682242990654, "grad_norm": 4.526474952697754, "learning_rate": 1.5547592585006839e-06, "loss": 0.03906508535146713, "step": 5301 }, { "epoch": 1.3764278296988577, "grad_norm": 3.0576300621032715, "learning_rate": 1.5539578716570524e-06, "loss": 0.05302969366312027, "step": 5302 }, { "epoch": 1.37668743509865, "grad_norm": 2.825286865234375, "learning_rate": 1.5531566962462853e-06, "loss": 0.060984350740909576, "step": 5303 }, { "epoch": 1.3769470404984423, "grad_norm": 5.335888385772705, "learning_rate": 1.552355732410008e-06, "loss": 0.01766115613281727, "step": 5304 }, { "epoch": 1.3772066458982346, "grad_norm": 4.13524866104126, "learning_rate": 1.551554980289815e-06, "loss": 0.024056021124124527, "step": 5305 }, { "epoch": 1.377466251298027, "grad_norm": 3.306325912475586, "learning_rate": 1.550754440027262e-06, "loss": 0.06816782802343369, "step": 5306 }, { "epoch": 1.3777258566978192, "grad_norm": 3.4558708667755127, "learning_rate": 1.5499541117638645e-06, "loss": 0.011968996375799179, "step": 5307 }, { "epoch": 1.3779854620976115, "grad_norm": 2.5341060161590576, "learning_rate": 1.5491539956411027e-06, "loss": 0.07594083994626999, "step": 5308 }, { "epoch": 1.3782450674974038, "grad_norm": 4.483608722686768, "learning_rate": 1.548354091800418e-06, "loss": 0.01858637109398842, "step": 5309 }, { "epoch": 1.3785046728971961, "grad_norm": 2.8211119174957275, "learning_rate": 1.5475544003832161e-06, "loss": 0.05278913304209709, "step": 5310 }, { "epoch": 1.3787642782969884, "grad_norm": 3.143230676651001, "learning_rate": 1.546754921530865e-06, "loss": 0.03708186000585556, "step": 5311 }, { "epoch": 1.379023883696781, "grad_norm": 3.4730567932128906, "learning_rate": 1.5459556553846915e-06, "loss": 0.022887738421559334, "step": 5312 }, { "epoch": 1.3792834890965733, "grad_norm": 3.785317897796631, "learning_rate": 1.5451566020859903e-06, "loss": 0.04742179438471794, "step": 5313 }, { "epoch": 1.3795430944963656, "grad_norm": 3.49407958984375, "learning_rate": 1.5443577617760137e-06, "loss": 0.02621706947684288, "step": 5314 }, { "epoch": 1.379802699896158, "grad_norm": 2.639049530029297, "learning_rate": 1.54355913459598e-06, "loss": 0.06059274077415466, "step": 5315 }, { "epoch": 1.3800623052959502, "grad_norm": 3.9571194648742676, "learning_rate": 1.5427607206870677e-06, "loss": 0.013797164894640446, "step": 5316 }, { "epoch": 1.3803219106957425, "grad_norm": 3.600458860397339, "learning_rate": 1.5419625201904164e-06, "loss": 0.02762152999639511, "step": 5317 }, { "epoch": 1.3805815160955348, "grad_norm": 2.9253993034362793, "learning_rate": 1.5411645332471326e-06, "loss": 0.049547746777534485, "step": 5318 }, { "epoch": 1.3808411214953271, "grad_norm": 4.616223335266113, "learning_rate": 1.5403667599982808e-06, "loss": 0.04520997032523155, "step": 5319 }, { "epoch": 1.3811007268951194, "grad_norm": 3.4902751445770264, "learning_rate": 1.5395692005848876e-06, "loss": 0.019877886399626732, "step": 5320 }, { "epoch": 1.3813603322949117, "grad_norm": 2.81955885887146, "learning_rate": 1.5387718551479452e-06, "loss": 0.044965945184230804, "step": 5321 }, { "epoch": 1.381619937694704, "grad_norm": 3.298353672027588, "learning_rate": 1.5379747238284059e-06, "loss": 0.017239034175872803, "step": 5322 }, { "epoch": 1.3818795430944963, "grad_norm": 2.757000207901001, "learning_rate": 1.5371778067671826e-06, "loss": 0.00857163779437542, "step": 5323 }, { "epoch": 1.3821391484942886, "grad_norm": 3.040670394897461, "learning_rate": 1.5363811041051541e-06, "loss": 0.009768092073500156, "step": 5324 }, { "epoch": 1.382398753894081, "grad_norm": 2.3142402172088623, "learning_rate": 1.5355846159831572e-06, "loss": 0.07944606989622116, "step": 5325 }, { "epoch": 1.3826583592938733, "grad_norm": 3.3616795539855957, "learning_rate": 1.5347883425419945e-06, "loss": 0.019179457798600197, "step": 5326 }, { "epoch": 1.3829179646936656, "grad_norm": 3.347592353820801, "learning_rate": 1.5339922839224285e-06, "loss": 0.01990669220685959, "step": 5327 }, { "epoch": 1.3831775700934579, "grad_norm": 3.0325026512145996, "learning_rate": 1.5331964402651827e-06, "loss": 0.04474595934152603, "step": 5328 }, { "epoch": 1.3834371754932502, "grad_norm": 2.6049787998199463, "learning_rate": 1.5324008117109463e-06, "loss": 0.018568741157650948, "step": 5329 }, { "epoch": 1.3836967808930427, "grad_norm": 3.205070734024048, "learning_rate": 1.531605398400367e-06, "loss": 0.0582435242831707, "step": 5330 }, { "epoch": 1.383956386292835, "grad_norm": 2.3195385932922363, "learning_rate": 1.5308102004740549e-06, "loss": 0.01723199523985386, "step": 5331 }, { "epoch": 1.3842159916926273, "grad_norm": 2.477810859680176, "learning_rate": 1.530015218072584e-06, "loss": 0.06585055589675903, "step": 5332 }, { "epoch": 1.3844755970924196, "grad_norm": 2.266148567199707, "learning_rate": 1.5292204513364884e-06, "loss": 0.018692249432206154, "step": 5333 }, { "epoch": 1.384735202492212, "grad_norm": 2.790992259979248, "learning_rate": 1.528425900406266e-06, "loss": 0.059543438255786896, "step": 5334 }, { "epoch": 1.3849948078920042, "grad_norm": 2.5562515258789062, "learning_rate": 1.5276315654223728e-06, "loss": 0.027047544717788696, "step": 5335 }, { "epoch": 1.3852544132917965, "grad_norm": 2.0255041122436523, "learning_rate": 1.5268374465252301e-06, "loss": 0.07311663031578064, "step": 5336 }, { "epoch": 1.3855140186915889, "grad_norm": 2.6574859619140625, "learning_rate": 1.5260435438552208e-06, "loss": 0.05583511292934418, "step": 5337 }, { "epoch": 1.3857736240913812, "grad_norm": 2.7795135974884033, "learning_rate": 1.5252498575526887e-06, "loss": 0.040243230760097504, "step": 5338 }, { "epoch": 1.3860332294911735, "grad_norm": 2.4934256076812744, "learning_rate": 1.524456387757938e-06, "loss": 0.05185706168413162, "step": 5339 }, { "epoch": 1.3862928348909658, "grad_norm": 0.5591345429420471, "learning_rate": 1.5236631346112364e-06, "loss": 0.0017010115552693605, "step": 5340 }, { "epoch": 1.386552440290758, "grad_norm": 2.92263126373291, "learning_rate": 1.5228700982528127e-06, "loss": 0.013405095785856247, "step": 5341 }, { "epoch": 1.3868120456905504, "grad_norm": 3.8451731204986572, "learning_rate": 1.5220772788228599e-06, "loss": 0.03161618858575821, "step": 5342 }, { "epoch": 1.3870716510903427, "grad_norm": 3.0400502681732178, "learning_rate": 1.5212846764615271e-06, "loss": 0.03594915568828583, "step": 5343 }, { "epoch": 1.387331256490135, "grad_norm": 3.9807043075561523, "learning_rate": 1.5204922913089305e-06, "loss": 0.04015627130866051, "step": 5344 }, { "epoch": 1.3875908618899273, "grad_norm": 1.8962018489837646, "learning_rate": 1.5197001235051446e-06, "loss": 0.004572484642267227, "step": 5345 }, { "epoch": 1.3878504672897196, "grad_norm": 3.1582183837890625, "learning_rate": 1.5189081731902062e-06, "loss": 0.050455838441848755, "step": 5346 }, { "epoch": 1.388110072689512, "grad_norm": 3.9283690452575684, "learning_rate": 1.5181164405041149e-06, "loss": 0.06426799297332764, "step": 5347 }, { "epoch": 1.3883696780893042, "grad_norm": 2.7156381607055664, "learning_rate": 1.5173249255868299e-06, "loss": 0.051872797310352325, "step": 5348 }, { "epoch": 1.3886292834890965, "grad_norm": 3.1763579845428467, "learning_rate": 1.5165336285782739e-06, "loss": 0.043035462498664856, "step": 5349 }, { "epoch": 1.3888888888888888, "grad_norm": 7.605897426605225, "learning_rate": 1.5157425496183303e-06, "loss": 0.03504219651222229, "step": 5350 }, { "epoch": 1.3891484942886811, "grad_norm": 1.9698936939239502, "learning_rate": 1.5149516888468423e-06, "loss": 0.014734704047441483, "step": 5351 }, { "epoch": 1.3894080996884735, "grad_norm": 3.6901919841766357, "learning_rate": 1.5141610464036177e-06, "loss": 0.018357716500759125, "step": 5352 }, { "epoch": 1.3896677050882658, "grad_norm": 3.090911626815796, "learning_rate": 1.513370622428423e-06, "loss": 0.03301706165075302, "step": 5353 }, { "epoch": 1.389927310488058, "grad_norm": 3.231222152709961, "learning_rate": 1.5125804170609865e-06, "loss": 0.016956407576799393, "step": 5354 }, { "epoch": 1.3901869158878504, "grad_norm": 2.9505832195281982, "learning_rate": 1.5117904304409997e-06, "loss": 0.03743545711040497, "step": 5355 }, { "epoch": 1.3904465212876427, "grad_norm": 3.452666759490967, "learning_rate": 1.511000662708113e-06, "loss": 0.017858345061540604, "step": 5356 }, { "epoch": 1.390706126687435, "grad_norm": 3.3499271869659424, "learning_rate": 1.5102111140019404e-06, "loss": 0.02654937095940113, "step": 5357 }, { "epoch": 1.3909657320872273, "grad_norm": 3.368187189102173, "learning_rate": 1.5094217844620558e-06, "loss": 0.02419961616396904, "step": 5358 }, { "epoch": 1.3912253374870196, "grad_norm": 3.2747678756713867, "learning_rate": 1.508632674227993e-06, "loss": 0.062242068350315094, "step": 5359 }, { "epoch": 1.3914849428868121, "grad_norm": 2.6237735748291016, "learning_rate": 1.5078437834392508e-06, "loss": 0.045051805675029755, "step": 5360 }, { "epoch": 1.3917445482866044, "grad_norm": 3.0257928371429443, "learning_rate": 1.507055112235286e-06, "loss": 0.01775451935827732, "step": 5361 }, { "epoch": 1.3920041536863967, "grad_norm": 2.506927728652954, "learning_rate": 1.5062666607555175e-06, "loss": 0.05220722407102585, "step": 5362 }, { "epoch": 1.392263759086189, "grad_norm": 2.6933867931365967, "learning_rate": 1.505478429139326e-06, "loss": 0.06323781609535217, "step": 5363 }, { "epoch": 1.3925233644859814, "grad_norm": 3.1164019107818604, "learning_rate": 1.5046904175260513e-06, "loss": 0.01537974365055561, "step": 5364 }, { "epoch": 1.3927829698857737, "grad_norm": 3.4627795219421387, "learning_rate": 1.5039026260549988e-06, "loss": 0.006446356885135174, "step": 5365 }, { "epoch": 1.393042575285566, "grad_norm": 2.1625709533691406, "learning_rate": 1.5031150548654288e-06, "loss": 0.06832472234964371, "step": 5366 }, { "epoch": 1.3933021806853583, "grad_norm": 2.3577070236206055, "learning_rate": 1.502327704096567e-06, "loss": 0.0807010680437088, "step": 5367 }, { "epoch": 1.3935617860851506, "grad_norm": 2.6658051013946533, "learning_rate": 1.5015405738876002e-06, "loss": 0.0038449964486062527, "step": 5368 }, { "epoch": 1.3938213914849429, "grad_norm": 3.1667346954345703, "learning_rate": 1.5007536643776737e-06, "loss": 0.057171016931533813, "step": 5369 }, { "epoch": 1.3940809968847352, "grad_norm": 0.792352020740509, "learning_rate": 1.4999669757058956e-06, "loss": 0.0019533629529178143, "step": 5370 }, { "epoch": 1.3943406022845275, "grad_norm": 2.3755409717559814, "learning_rate": 1.4991805080113337e-06, "loss": 0.05934114009141922, "step": 5371 }, { "epoch": 1.3946002076843198, "grad_norm": 4.37093448638916, "learning_rate": 1.4983942614330182e-06, "loss": 0.03481443226337433, "step": 5372 }, { "epoch": 1.394859813084112, "grad_norm": 2.997148036956787, "learning_rate": 1.4976082361099397e-06, "loss": 0.05416069179773331, "step": 5373 }, { "epoch": 1.3951194184839044, "grad_norm": 4.20932674407959, "learning_rate": 1.496822432181048e-06, "loss": 0.015049697831273079, "step": 5374 }, { "epoch": 1.3953790238836967, "grad_norm": 3.0957119464874268, "learning_rate": 1.4960368497852568e-06, "loss": 0.06118064373731613, "step": 5375 }, { "epoch": 1.395638629283489, "grad_norm": 3.0764167308807373, "learning_rate": 1.49525148906144e-06, "loss": 0.0512867271900177, "step": 5376 }, { "epoch": 1.3958982346832813, "grad_norm": 2.9195663928985596, "learning_rate": 1.4944663501484285e-06, "loss": 0.060207970440387726, "step": 5377 }, { "epoch": 1.3961578400830739, "grad_norm": 3.1208336353302, "learning_rate": 1.4936814331850188e-06, "loss": 0.027272991836071014, "step": 5378 }, { "epoch": 1.3964174454828662, "grad_norm": 3.4989888668060303, "learning_rate": 1.4928967383099655e-06, "loss": 0.05093466117978096, "step": 5379 }, { "epoch": 1.3966770508826585, "grad_norm": 3.018008232116699, "learning_rate": 1.4921122656619855e-06, "loss": 0.03427368402481079, "step": 5380 }, { "epoch": 1.3969366562824508, "grad_norm": 2.6325418949127197, "learning_rate": 1.4913280153797553e-06, "loss": 0.050762277096509933, "step": 5381 }, { "epoch": 1.397196261682243, "grad_norm": 2.6558210849761963, "learning_rate": 1.4905439876019113e-06, "loss": 0.06920285522937775, "step": 5382 }, { "epoch": 1.3974558670820354, "grad_norm": 2.5367090702056885, "learning_rate": 1.4897601824670534e-06, "loss": 0.02156192995607853, "step": 5383 }, { "epoch": 1.3977154724818277, "grad_norm": 3.2548933029174805, "learning_rate": 1.4889766001137395e-06, "loss": 0.038723934441804886, "step": 5384 }, { "epoch": 1.39797507788162, "grad_norm": 2.736265182495117, "learning_rate": 1.4881932406804883e-06, "loss": 0.05553359538316727, "step": 5385 }, { "epoch": 1.3982346832814123, "grad_norm": 2.5094339847564697, "learning_rate": 1.4874101043057814e-06, "loss": 0.05492932349443436, "step": 5386 }, { "epoch": 1.3984942886812046, "grad_norm": 3.9651544094085693, "learning_rate": 1.4866271911280578e-06, "loss": 0.024831503629684448, "step": 5387 }, { "epoch": 1.398753894080997, "grad_norm": 4.1602253913879395, "learning_rate": 1.4858445012857203e-06, "loss": 0.02962576225399971, "step": 5388 }, { "epoch": 1.3990134994807892, "grad_norm": 3.019474506378174, "learning_rate": 1.4850620349171297e-06, "loss": 0.06368156522512436, "step": 5389 }, { "epoch": 1.3992731048805815, "grad_norm": 2.9253652095794678, "learning_rate": 1.4842797921606073e-06, "loss": 0.05538095161318779, "step": 5390 }, { "epoch": 1.3995327102803738, "grad_norm": 2.2115461826324463, "learning_rate": 1.483497773154437e-06, "loss": 0.005787366069853306, "step": 5391 }, { "epoch": 1.3997923156801662, "grad_norm": 2.194420576095581, "learning_rate": 1.4827159780368616e-06, "loss": 0.01714272052049637, "step": 5392 }, { "epoch": 1.4000519210799585, "grad_norm": 3.4720754623413086, "learning_rate": 1.4819344069460833e-06, "loss": 0.02162069082260132, "step": 5393 }, { "epoch": 1.4003115264797508, "grad_norm": 2.7560153007507324, "learning_rate": 1.4811530600202683e-06, "loss": 0.01436261460185051, "step": 5394 }, { "epoch": 1.400571131879543, "grad_norm": 3.0758025646209717, "learning_rate": 1.4803719373975382e-06, "loss": 0.013661835342645645, "step": 5395 }, { "epoch": 1.4008307372793354, "grad_norm": 3.0753414630889893, "learning_rate": 1.4795910392159798e-06, "loss": 0.010886752977967262, "step": 5396 }, { "epoch": 1.4010903426791277, "grad_norm": 2.6700210571289062, "learning_rate": 1.4788103656136372e-06, "loss": 0.019323457032442093, "step": 5397 }, { "epoch": 1.40134994807892, "grad_norm": 2.119723081588745, "learning_rate": 1.478029916728515e-06, "loss": 0.05372871458530426, "step": 5398 }, { "epoch": 1.4016095534787123, "grad_norm": 3.9552133083343506, "learning_rate": 1.4772496926985808e-06, "loss": 0.010020659305155277, "step": 5399 }, { "epoch": 1.4018691588785046, "grad_norm": 1.2714637517929077, "learning_rate": 1.476469693661757e-06, "loss": 0.003016309579834342, "step": 5400 }, { "epoch": 1.402128764278297, "grad_norm": 2.6433348655700684, "learning_rate": 1.4756899197559317e-06, "loss": 0.0184391550719738, "step": 5401 }, { "epoch": 1.4023883696780892, "grad_norm": 3.567244529724121, "learning_rate": 1.4749103711189517e-06, "loss": 0.04515961557626724, "step": 5402 }, { "epoch": 1.4026479750778815, "grad_norm": 2.854715585708618, "learning_rate": 1.4741310478886217e-06, "loss": 0.0296135563403368, "step": 5403 }, { "epoch": 1.4029075804776738, "grad_norm": 2.4573686122894287, "learning_rate": 1.4733519502027093e-06, "loss": 0.06387132406234741, "step": 5404 }, { "epoch": 1.4031671858774661, "grad_norm": 2.518454074859619, "learning_rate": 1.47257307819894e-06, "loss": 0.011639833450317383, "step": 5405 }, { "epoch": 1.4034267912772584, "grad_norm": 3.5967087745666504, "learning_rate": 1.4717944320150013e-06, "loss": 0.03266852721571922, "step": 5406 }, { "epoch": 1.4036863966770508, "grad_norm": 3.0615906715393066, "learning_rate": 1.4710160117885414e-06, "loss": 0.05577445030212402, "step": 5407 }, { "epoch": 1.4039460020768433, "grad_norm": 3.858992099761963, "learning_rate": 1.4702378176571641e-06, "loss": 0.02038324624300003, "step": 5408 }, { "epoch": 1.4042056074766356, "grad_norm": 3.7875561714172363, "learning_rate": 1.4694598497584392e-06, "loss": 0.026490509510040283, "step": 5409 }, { "epoch": 1.404465212876428, "grad_norm": 2.7589800357818604, "learning_rate": 1.4686821082298914e-06, "loss": 0.05337507277727127, "step": 5410 }, { "epoch": 1.4047248182762202, "grad_norm": 3.496065616607666, "learning_rate": 1.4679045932090092e-06, "loss": 0.033706407994031906, "step": 5411 }, { "epoch": 1.4049844236760125, "grad_norm": 3.179184675216675, "learning_rate": 1.4671273048332389e-06, "loss": 0.00762689346447587, "step": 5412 }, { "epoch": 1.4052440290758048, "grad_norm": 2.1093766689300537, "learning_rate": 1.4663502432399861e-06, "loss": 0.07554690539836884, "step": 5413 }, { "epoch": 1.4055036344755971, "grad_norm": 3.3470163345336914, "learning_rate": 1.4655734085666198e-06, "loss": 0.06269665062427521, "step": 5414 }, { "epoch": 1.4057632398753894, "grad_norm": 2.3110411167144775, "learning_rate": 1.4647968009504654e-06, "loss": 0.006425561383366585, "step": 5415 }, { "epoch": 1.4060228452751817, "grad_norm": 3.0564043521881104, "learning_rate": 1.4640204205288084e-06, "loss": 0.027485061436891556, "step": 5416 }, { "epoch": 1.406282450674974, "grad_norm": 3.7948765754699707, "learning_rate": 1.4632442674388969e-06, "loss": 0.04942472651600838, "step": 5417 }, { "epoch": 1.4065420560747663, "grad_norm": 2.458103895187378, "learning_rate": 1.4624683418179353e-06, "loss": 0.02111843042075634, "step": 5418 }, { "epoch": 1.4068016614745587, "grad_norm": 4.057821273803711, "learning_rate": 1.461692643803091e-06, "loss": 0.015105810947716236, "step": 5419 }, { "epoch": 1.407061266874351, "grad_norm": 3.317621946334839, "learning_rate": 1.4609171735314892e-06, "loss": 0.029541097581386566, "step": 5420 }, { "epoch": 1.4073208722741433, "grad_norm": 3.2741751670837402, "learning_rate": 1.4601419311402142e-06, "loss": 0.02722989022731781, "step": 5421 }, { "epoch": 1.4075804776739356, "grad_norm": 3.3535001277923584, "learning_rate": 1.4593669167663126e-06, "loss": 0.028596902266144753, "step": 5422 }, { "epoch": 1.4078400830737279, "grad_norm": 3.103569984436035, "learning_rate": 1.4585921305467887e-06, "loss": 0.030134212225675583, "step": 5423 }, { "epoch": 1.4080996884735202, "grad_norm": 3.4396071434020996, "learning_rate": 1.4578175726186056e-06, "loss": 0.0070795342326164246, "step": 5424 }, { "epoch": 1.4083592938733125, "grad_norm": 3.3095667362213135, "learning_rate": 1.45704324311869e-06, "loss": 0.0667634829878807, "step": 5425 }, { "epoch": 1.408618899273105, "grad_norm": 3.6636929512023926, "learning_rate": 1.4562691421839237e-06, "loss": 0.020415296778082848, "step": 5426 }, { "epoch": 1.4088785046728973, "grad_norm": 3.8684473037719727, "learning_rate": 1.4554952699511498e-06, "loss": 0.03842253237962723, "step": 5427 }, { "epoch": 1.4091381100726896, "grad_norm": 3.0722594261169434, "learning_rate": 1.454721626557173e-06, "loss": 0.061334747821092606, "step": 5428 }, { "epoch": 1.409397715472482, "grad_norm": 4.049891471862793, "learning_rate": 1.4539482121387539e-06, "loss": 0.051763590425252914, "step": 5429 }, { "epoch": 1.4096573208722742, "grad_norm": 3.668318510055542, "learning_rate": 1.4531750268326166e-06, "loss": 0.0366942472755909, "step": 5430 }, { "epoch": 1.4099169262720666, "grad_norm": 3.1223809719085693, "learning_rate": 1.45240207077544e-06, "loss": 0.019870102405548096, "step": 5431 }, { "epoch": 1.4101765316718589, "grad_norm": 2.1878926753997803, "learning_rate": 1.4516293441038659e-06, "loss": 0.006283373571932316, "step": 5432 }, { "epoch": 1.4104361370716512, "grad_norm": 3.5335099697113037, "learning_rate": 1.450856846954496e-06, "loss": 0.009825298562645912, "step": 5433 }, { "epoch": 1.4106957424714435, "grad_norm": 3.235548973083496, "learning_rate": 1.4500845794638886e-06, "loss": 0.05118381977081299, "step": 5434 }, { "epoch": 1.4109553478712358, "grad_norm": 4.023785591125488, "learning_rate": 1.449312541768564e-06, "loss": 0.015591641888022423, "step": 5435 }, { "epoch": 1.411214953271028, "grad_norm": 4.634274005889893, "learning_rate": 1.4485407340049991e-06, "loss": 0.015698913484811783, "step": 5436 }, { "epoch": 1.4114745586708204, "grad_norm": 2.9307661056518555, "learning_rate": 1.447769156309633e-06, "loss": 0.05902448669075966, "step": 5437 }, { "epoch": 1.4117341640706127, "grad_norm": 3.5475656986236572, "learning_rate": 1.4469978088188646e-06, "loss": 0.019982829689979553, "step": 5438 }, { "epoch": 1.411993769470405, "grad_norm": 2.880810499191284, "learning_rate": 1.4462266916690474e-06, "loss": 0.01845821738243103, "step": 5439 }, { "epoch": 1.4122533748701973, "grad_norm": 3.56036114692688, "learning_rate": 1.4454558049964993e-06, "loss": 0.031970374286174774, "step": 5440 }, { "epoch": 1.4125129802699896, "grad_norm": 2.4174842834472656, "learning_rate": 1.4446851489374942e-06, "loss": 0.01668427512049675, "step": 5441 }, { "epoch": 1.412772585669782, "grad_norm": 2.96882963180542, "learning_rate": 1.4439147236282674e-06, "loss": 0.039055172353982925, "step": 5442 }, { "epoch": 1.4130321910695742, "grad_norm": 2.743403673171997, "learning_rate": 1.4431445292050123e-06, "loss": 0.04890858009457588, "step": 5443 }, { "epoch": 1.4132917964693665, "grad_norm": 3.00608229637146, "learning_rate": 1.4423745658038808e-06, "loss": 0.05608880892395973, "step": 5444 }, { "epoch": 1.4135514018691588, "grad_norm": 1.9741237163543701, "learning_rate": 1.4416048335609856e-06, "loss": 0.0052256835624575615, "step": 5445 }, { "epoch": 1.4138110072689511, "grad_norm": 2.12166166305542, "learning_rate": 1.4408353326123981e-06, "loss": 0.01542946882545948, "step": 5446 }, { "epoch": 1.4140706126687435, "grad_norm": 3.7755990028381348, "learning_rate": 1.4400660630941469e-06, "loss": 0.027860108762979507, "step": 5447 }, { "epoch": 1.4143302180685358, "grad_norm": 3.362858533859253, "learning_rate": 1.439297025142223e-06, "loss": 0.03534601628780365, "step": 5448 }, { "epoch": 1.414589823468328, "grad_norm": 6.779343128204346, "learning_rate": 1.4385282188925732e-06, "loss": 0.03303174674510956, "step": 5449 }, { "epoch": 1.4148494288681204, "grad_norm": 4.457540512084961, "learning_rate": 1.4377596444811064e-06, "loss": 0.03130035474896431, "step": 5450 }, { "epoch": 1.4151090342679127, "grad_norm": 4.049659252166748, "learning_rate": 1.4369913020436882e-06, "loss": 0.031078308820724487, "step": 5451 }, { "epoch": 1.415368639667705, "grad_norm": 4.123408794403076, "learning_rate": 1.436223191716143e-06, "loss": 0.025393227115273476, "step": 5452 }, { "epoch": 1.4156282450674973, "grad_norm": 4.365286827087402, "learning_rate": 1.4354553136342569e-06, "loss": 0.02088814228773117, "step": 5453 }, { "epoch": 1.4158878504672896, "grad_norm": 3.496019124984741, "learning_rate": 1.4346876679337721e-06, "loss": 0.05141846090555191, "step": 5454 }, { "epoch": 1.416147455867082, "grad_norm": 2.7733304500579834, "learning_rate": 1.43392025475039e-06, "loss": 0.0664464682340622, "step": 5455 }, { "epoch": 1.4164070612668744, "grad_norm": 2.9574086666107178, "learning_rate": 1.4331530742197738e-06, "loss": 0.04099483788013458, "step": 5456 }, { "epoch": 1.4166666666666667, "grad_norm": 3.311807870864868, "learning_rate": 1.432386126477542e-06, "loss": 0.03692995011806488, "step": 5457 }, { "epoch": 1.416926272066459, "grad_norm": 3.3313181400299072, "learning_rate": 1.431619411659273e-06, "loss": 0.012431449256837368, "step": 5458 }, { "epoch": 1.4171858774662514, "grad_norm": 2.8014698028564453, "learning_rate": 1.4308529299005059e-06, "loss": 0.017043909057974815, "step": 5459 }, { "epoch": 1.4174454828660437, "grad_norm": 2.8315935134887695, "learning_rate": 1.4300866813367356e-06, "loss": 0.019828252494335175, "step": 5460 }, { "epoch": 1.417705088265836, "grad_norm": 3.546445608139038, "learning_rate": 1.4293206661034198e-06, "loss": 0.009217898361384869, "step": 5461 }, { "epoch": 1.4179646936656283, "grad_norm": 3.33849835395813, "learning_rate": 1.4285548843359686e-06, "loss": 0.024624710902571678, "step": 5462 }, { "epoch": 1.4182242990654206, "grad_norm": 3.347395896911621, "learning_rate": 1.427789336169757e-06, "loss": 0.02398514747619629, "step": 5463 }, { "epoch": 1.4184839044652129, "grad_norm": 2.60357928276062, "learning_rate": 1.4270240217401171e-06, "loss": 0.014760426245629787, "step": 5464 }, { "epoch": 1.4187435098650052, "grad_norm": 4.007322788238525, "learning_rate": 1.426258941182338e-06, "loss": 0.01483954768627882, "step": 5465 }, { "epoch": 1.4190031152647975, "grad_norm": 2.741661548614502, "learning_rate": 1.4254940946316684e-06, "loss": 0.023295268416404724, "step": 5466 }, { "epoch": 1.4192627206645898, "grad_norm": 3.1567280292510986, "learning_rate": 1.4247294822233146e-06, "loss": 0.0392901748418808, "step": 5467 }, { "epoch": 1.4195223260643821, "grad_norm": 1.0066382884979248, "learning_rate": 1.423965104092444e-06, "loss": 0.002801422728225589, "step": 5468 }, { "epoch": 1.4197819314641744, "grad_norm": 3.0962891578674316, "learning_rate": 1.4232009603741822e-06, "loss": 0.046598441898822784, "step": 5469 }, { "epoch": 1.4200415368639667, "grad_norm": 2.2269177436828613, "learning_rate": 1.4224370512036095e-06, "loss": 0.008954273536801338, "step": 5470 }, { "epoch": 1.420301142263759, "grad_norm": 4.310489177703857, "learning_rate": 1.4216733767157698e-06, "loss": 0.017939060926437378, "step": 5471 }, { "epoch": 1.4205607476635513, "grad_norm": 2.774019718170166, "learning_rate": 1.420909937045662e-06, "loss": 0.025293197482824326, "step": 5472 }, { "epoch": 1.4208203530633436, "grad_norm": 2.6788814067840576, "learning_rate": 1.4201467323282453e-06, "loss": 0.0669211894273758, "step": 5473 }, { "epoch": 1.421079958463136, "grad_norm": 4.360323905944824, "learning_rate": 1.419383762698437e-06, "loss": 0.025227662175893784, "step": 5474 }, { "epoch": 1.4213395638629285, "grad_norm": 3.046386480331421, "learning_rate": 1.4186210282911118e-06, "loss": 0.03253013640642166, "step": 5475 }, { "epoch": 1.4215991692627208, "grad_norm": 3.3097355365753174, "learning_rate": 1.4178585292411046e-06, "loss": 0.045262306928634644, "step": 5476 }, { "epoch": 1.421858774662513, "grad_norm": 4.659126281738281, "learning_rate": 1.4170962656832077e-06, "loss": 0.024798313155770302, "step": 5477 }, { "epoch": 1.4221183800623054, "grad_norm": 3.3934969902038574, "learning_rate": 1.416334237752171e-06, "loss": 0.04484258592128754, "step": 5478 }, { "epoch": 1.4223779854620977, "grad_norm": 2.8307971954345703, "learning_rate": 1.4155724455827046e-06, "loss": 0.05788026750087738, "step": 5479 }, { "epoch": 1.42263759086189, "grad_norm": 3.6343419551849365, "learning_rate": 1.4148108893094756e-06, "loss": 0.0682704895734787, "step": 5480 }, { "epoch": 1.4228971962616823, "grad_norm": 2.7157649993896484, "learning_rate": 1.414049569067109e-06, "loss": 0.010240525007247925, "step": 5481 }, { "epoch": 1.4231568016614746, "grad_norm": 2.2403223514556885, "learning_rate": 1.41328848499019e-06, "loss": 0.07321378588676453, "step": 5482 }, { "epoch": 1.423416407061267, "grad_norm": 3.043458938598633, "learning_rate": 1.4125276372132594e-06, "loss": 0.0605616569519043, "step": 5483 }, { "epoch": 1.4236760124610592, "grad_norm": 3.211836338043213, "learning_rate": 1.4117670258708194e-06, "loss": 0.03384361416101456, "step": 5484 }, { "epoch": 1.4239356178608515, "grad_norm": 2.3954334259033203, "learning_rate": 1.4110066510973275e-06, "loss": 0.05633902922272682, "step": 5485 }, { "epoch": 1.4241952232606438, "grad_norm": 2.8844704627990723, "learning_rate": 1.4102465130272003e-06, "loss": 0.028413565829396248, "step": 5486 }, { "epoch": 1.4244548286604362, "grad_norm": 4.22982120513916, "learning_rate": 1.4094866117948136e-06, "loss": 0.03881216049194336, "step": 5487 }, { "epoch": 1.4247144340602285, "grad_norm": 3.170450210571289, "learning_rate": 1.4087269475345009e-06, "loss": 0.028737971559166908, "step": 5488 }, { "epoch": 1.4249740394600208, "grad_norm": 4.497649192810059, "learning_rate": 1.4079675203805516e-06, "loss": 0.014257300645112991, "step": 5489 }, { "epoch": 1.425233644859813, "grad_norm": 2.100099802017212, "learning_rate": 1.4072083304672174e-06, "loss": 0.009575434029102325, "step": 5490 }, { "epoch": 1.4254932502596054, "grad_norm": 1.923689365386963, "learning_rate": 1.4064493779287033e-06, "loss": 0.01795978844165802, "step": 5491 }, { "epoch": 1.4257528556593977, "grad_norm": 4.8506855964660645, "learning_rate": 1.4056906628991775e-06, "loss": 0.01761222630739212, "step": 5492 }, { "epoch": 1.42601246105919, "grad_norm": 3.2534215450286865, "learning_rate": 1.4049321855127608e-06, "loss": 0.02593657746911049, "step": 5493 }, { "epoch": 1.4262720664589823, "grad_norm": 2.182363271713257, "learning_rate": 1.4041739459035352e-06, "loss": 0.006518371403217316, "step": 5494 }, { "epoch": 1.4265316718587746, "grad_norm": 2.9192655086517334, "learning_rate": 1.4034159442055417e-06, "loss": 0.012710873037576675, "step": 5495 }, { "epoch": 1.426791277258567, "grad_norm": 4.196672439575195, "learning_rate": 1.4026581805527767e-06, "loss": 0.038774892687797546, "step": 5496 }, { "epoch": 1.4270508826583592, "grad_norm": 2.542278289794922, "learning_rate": 1.4019006550791953e-06, "loss": 0.0707901269197464, "step": 5497 }, { "epoch": 1.4273104880581515, "grad_norm": 3.0365076065063477, "learning_rate": 1.4011433679187098e-06, "loss": 0.028497770428657532, "step": 5498 }, { "epoch": 1.4275700934579438, "grad_norm": 0.2871536612510681, "learning_rate": 1.400386319205192e-06, "loss": 0.0013327656779438257, "step": 5499 }, { "epoch": 1.4278296988577361, "grad_norm": 2.4443719387054443, "learning_rate": 1.3996295090724723e-06, "loss": 0.07415620982646942, "step": 5500 }, { "epoch": 1.4280893042575284, "grad_norm": 3.4843122959136963, "learning_rate": 1.3988729376543347e-06, "loss": 0.04734112322330475, "step": 5501 }, { "epoch": 1.4283489096573208, "grad_norm": 2.435873031616211, "learning_rate": 1.3981166050845258e-06, "loss": 0.08182793855667114, "step": 5502 }, { "epoch": 1.428608515057113, "grad_norm": 3.4288718700408936, "learning_rate": 1.3973605114967465e-06, "loss": 0.0325295589864254, "step": 5503 }, { "epoch": 1.4288681204569054, "grad_norm": 2.9282772541046143, "learning_rate": 1.396604657024658e-06, "loss": 0.04887153580784798, "step": 5504 }, { "epoch": 1.429127725856698, "grad_norm": 2.4438908100128174, "learning_rate": 1.3958490418018778e-06, "loss": 0.05963820591568947, "step": 5505 }, { "epoch": 1.4293873312564902, "grad_norm": 3.1279048919677734, "learning_rate": 1.3950936659619795e-06, "loss": 0.027090951800346375, "step": 5506 }, { "epoch": 1.4296469366562825, "grad_norm": 1.9856539964675903, "learning_rate": 1.3943385296384996e-06, "loss": 0.00514470599591732, "step": 5507 }, { "epoch": 1.4299065420560748, "grad_norm": 3.327631711959839, "learning_rate": 1.3935836329649266e-06, "loss": 0.011512743309140205, "step": 5508 }, { "epoch": 1.4301661474558671, "grad_norm": 2.215794324874878, "learning_rate": 1.3928289760747088e-06, "loss": 0.015640906989574432, "step": 5509 }, { "epoch": 1.4304257528556594, "grad_norm": 2.909820795059204, "learning_rate": 1.3920745591012536e-06, "loss": 0.07704728096723557, "step": 5510 }, { "epoch": 1.4306853582554517, "grad_norm": 2.5055532455444336, "learning_rate": 1.3913203821779239e-06, "loss": 0.07950089871883392, "step": 5511 }, { "epoch": 1.430944963655244, "grad_norm": 2.8107686042785645, "learning_rate": 1.3905664454380401e-06, "loss": 0.04899918660521507, "step": 5512 }, { "epoch": 1.4312045690550363, "grad_norm": 3.4232568740844727, "learning_rate": 1.3898127490148832e-06, "loss": 0.01653757318854332, "step": 5513 }, { "epoch": 1.4314641744548287, "grad_norm": 1.8765257596969604, "learning_rate": 1.389059293041687e-06, "loss": 0.009892476722598076, "step": 5514 }, { "epoch": 1.431723779854621, "grad_norm": 3.3268914222717285, "learning_rate": 1.3883060776516473e-06, "loss": 0.025507662445306778, "step": 5515 }, { "epoch": 1.4319833852544133, "grad_norm": 2.4678800106048584, "learning_rate": 1.3875531029779137e-06, "loss": 0.08851361274719238, "step": 5516 }, { "epoch": 1.4322429906542056, "grad_norm": 1.940075159072876, "learning_rate": 1.3868003691535958e-06, "loss": 0.015221922658383846, "step": 5517 }, { "epoch": 1.4325025960539979, "grad_norm": 2.31122088432312, "learning_rate": 1.3860478763117596e-06, "loss": 0.0702415406703949, "step": 5518 }, { "epoch": 1.4327622014537902, "grad_norm": 4.078246116638184, "learning_rate": 1.3852956245854288e-06, "loss": 0.03159952536225319, "step": 5519 }, { "epoch": 1.4330218068535825, "grad_norm": 2.2325377464294434, "learning_rate": 1.3845436141075832e-06, "loss": 0.008700903505086899, "step": 5520 }, { "epoch": 1.4332814122533748, "grad_norm": 1.8349409103393555, "learning_rate": 1.3837918450111626e-06, "loss": 0.004950389266014099, "step": 5521 }, { "epoch": 1.433541017653167, "grad_norm": 4.3061347007751465, "learning_rate": 1.3830403174290611e-06, "loss": 0.03536340221762657, "step": 5522 }, { "epoch": 1.4338006230529596, "grad_norm": 2.7315502166748047, "learning_rate": 1.382289031494134e-06, "loss": 0.025850966572761536, "step": 5523 }, { "epoch": 1.434060228452752, "grad_norm": 3.088635206222534, "learning_rate": 1.3815379873391876e-06, "loss": 0.03222949430346489, "step": 5524 }, { "epoch": 1.4343198338525442, "grad_norm": 2.897919178009033, "learning_rate": 1.3807871850969918e-06, "loss": 0.008868289180099964, "step": 5525 }, { "epoch": 1.4345794392523366, "grad_norm": 2.741183280944824, "learning_rate": 1.3800366249002715e-06, "loss": 0.04607291519641876, "step": 5526 }, { "epoch": 1.4348390446521289, "grad_norm": 3.205880880355835, "learning_rate": 1.3792863068817079e-06, "loss": 0.026985064148902893, "step": 5527 }, { "epoch": 1.4350986500519212, "grad_norm": 2.909574031829834, "learning_rate": 1.3785362311739397e-06, "loss": 0.010308372788131237, "step": 5528 }, { "epoch": 1.4353582554517135, "grad_norm": 3.9148199558258057, "learning_rate": 1.377786397909563e-06, "loss": 0.02815936878323555, "step": 5529 }, { "epoch": 1.4356178608515058, "grad_norm": 3.4292914867401123, "learning_rate": 1.377036807221132e-06, "loss": 0.04326559975743294, "step": 5530 }, { "epoch": 1.435877466251298, "grad_norm": 3.160416841506958, "learning_rate": 1.376287459241158e-06, "loss": 0.04973938688635826, "step": 5531 }, { "epoch": 1.4361370716510904, "grad_norm": 2.906914710998535, "learning_rate": 1.3755383541021056e-06, "loss": 0.040706876665353775, "step": 5532 }, { "epoch": 1.4363966770508827, "grad_norm": 3.0945065021514893, "learning_rate": 1.3747894919364024e-06, "loss": 0.012017050758004189, "step": 5533 }, { "epoch": 1.436656282450675, "grad_norm": 2.8141562938690186, "learning_rate": 1.3740408728764277e-06, "loss": 0.051843371242284775, "step": 5534 }, { "epoch": 1.4369158878504673, "grad_norm": 2.620046615600586, "learning_rate": 1.3732924970545225e-06, "loss": 0.03125886991620064, "step": 5535 }, { "epoch": 1.4371754932502596, "grad_norm": 3.031437397003174, "learning_rate": 1.3725443646029811e-06, "loss": 0.03548790141940117, "step": 5536 }, { "epoch": 1.437435098650052, "grad_norm": 2.761317253112793, "learning_rate": 1.3717964756540557e-06, "loss": 0.050609420984983444, "step": 5537 }, { "epoch": 1.4376947040498442, "grad_norm": 3.044757604598999, "learning_rate": 1.3710488303399577e-06, "loss": 0.024136852473020554, "step": 5538 }, { "epoch": 1.4379543094496365, "grad_norm": 2.582613706588745, "learning_rate": 1.3703014287928528e-06, "loss": 0.05879734829068184, "step": 5539 }, { "epoch": 1.4382139148494288, "grad_norm": 2.9207265377044678, "learning_rate": 1.3695542711448636e-06, "loss": 0.02012554742395878, "step": 5540 }, { "epoch": 1.4384735202492211, "grad_norm": 2.551051616668701, "learning_rate": 1.368807357528072e-06, "loss": 0.008236581459641457, "step": 5541 }, { "epoch": 1.4387331256490135, "grad_norm": 3.1581294536590576, "learning_rate": 1.368060688074514e-06, "loss": 0.023723160848021507, "step": 5542 }, { "epoch": 1.4389927310488058, "grad_norm": 3.041185140609741, "learning_rate": 1.367314262916184e-06, "loss": 0.04740124195814133, "step": 5543 }, { "epoch": 1.439252336448598, "grad_norm": 3.758387565612793, "learning_rate": 1.3665680821850335e-06, "loss": 0.03917108476161957, "step": 5544 }, { "epoch": 1.4395119418483904, "grad_norm": 3.6824636459350586, "learning_rate": 1.365822146012969e-06, "loss": 0.03802955895662308, "step": 5545 }, { "epoch": 1.4397715472481827, "grad_norm": 3.0813024044036865, "learning_rate": 1.3650764545318562e-06, "loss": 0.05699845403432846, "step": 5546 }, { "epoch": 1.440031152647975, "grad_norm": 2.8184616565704346, "learning_rate": 1.3643310078735156e-06, "loss": 0.03174026682972908, "step": 5547 }, { "epoch": 1.4402907580477673, "grad_norm": 3.684149980545044, "learning_rate": 1.3635858061697246e-06, "loss": 0.011438229121267796, "step": 5548 }, { "epoch": 1.4405503634475596, "grad_norm": 2.7758054733276367, "learning_rate": 1.3628408495522195e-06, "loss": 0.01611577346920967, "step": 5549 }, { "epoch": 1.440809968847352, "grad_norm": 3.3726537227630615, "learning_rate": 1.36209613815269e-06, "loss": 0.015116479247808456, "step": 5550 }, { "epoch": 1.4410695742471442, "grad_norm": 3.326805591583252, "learning_rate": 1.3613516721027837e-06, "loss": 0.04004545509815216, "step": 5551 }, { "epoch": 1.4413291796469365, "grad_norm": 3.1406338214874268, "learning_rate": 1.360607451534107e-06, "loss": 0.06468753516674042, "step": 5552 }, { "epoch": 1.441588785046729, "grad_norm": 2.916735887527466, "learning_rate": 1.3598634765782192e-06, "loss": 0.02557801455259323, "step": 5553 }, { "epoch": 1.4418483904465214, "grad_norm": 2.343687057495117, "learning_rate": 1.3591197473666401e-06, "loss": 0.015706120058894157, "step": 5554 }, { "epoch": 1.4421079958463137, "grad_norm": 4.757694721221924, "learning_rate": 1.3583762640308412e-06, "loss": 0.029981523752212524, "step": 5555 }, { "epoch": 1.442367601246106, "grad_norm": 2.7918484210968018, "learning_rate": 1.3576330267022552e-06, "loss": 0.027983423322439194, "step": 5556 }, { "epoch": 1.4426272066458983, "grad_norm": 3.3286685943603516, "learning_rate": 1.3568900355122697e-06, "loss": 0.012528345920145512, "step": 5557 }, { "epoch": 1.4428868120456906, "grad_norm": 3.4435908794403076, "learning_rate": 1.356147290592228e-06, "loss": 0.027143297716975212, "step": 5558 }, { "epoch": 1.4431464174454829, "grad_norm": 2.7420706748962402, "learning_rate": 1.3554047920734301e-06, "loss": 0.020438970997929573, "step": 5559 }, { "epoch": 1.4434060228452752, "grad_norm": 2.0091402530670166, "learning_rate": 1.3546625400871321e-06, "loss": 0.0161232128739357, "step": 5560 }, { "epoch": 1.4436656282450675, "grad_norm": 2.8532185554504395, "learning_rate": 1.3539205347645485e-06, "loss": 0.05093225836753845, "step": 5561 }, { "epoch": 1.4439252336448598, "grad_norm": 3.997391939163208, "learning_rate": 1.3531787762368498e-06, "loss": 0.04313149303197861, "step": 5562 }, { "epoch": 1.4441848390446521, "grad_norm": 3.1189892292022705, "learning_rate": 1.3524372646351585e-06, "loss": 0.043036822229623795, "step": 5563 }, { "epoch": 1.4444444444444444, "grad_norm": 5.97576379776001, "learning_rate": 1.3516960000905601e-06, "loss": 0.023515576496720314, "step": 5564 }, { "epoch": 1.4447040498442367, "grad_norm": 2.458932638168335, "learning_rate": 1.3509549827340917e-06, "loss": 0.00589647376909852, "step": 5565 }, { "epoch": 1.444963655244029, "grad_norm": 2.9118852615356445, "learning_rate": 1.350214212696748e-06, "loss": 0.02433750405907631, "step": 5566 }, { "epoch": 1.4452232606438213, "grad_norm": 3.3269731998443604, "learning_rate": 1.3494736901094814e-06, "loss": 0.008548775687813759, "step": 5567 }, { "epoch": 1.4454828660436136, "grad_norm": 2.2572848796844482, "learning_rate": 1.3487334151031975e-06, "loss": 0.015024267137050629, "step": 5568 }, { "epoch": 1.445742471443406, "grad_norm": 3.277613401412964, "learning_rate": 1.347993387808762e-06, "loss": 0.022848526015877724, "step": 5569 }, { "epoch": 1.4460020768431983, "grad_norm": 2.4007785320281982, "learning_rate": 1.3472536083569937e-06, "loss": 0.017613261938095093, "step": 5570 }, { "epoch": 1.4462616822429908, "grad_norm": 3.512073278427124, "learning_rate": 1.3465140768786683e-06, "loss": 0.017155535519123077, "step": 5571 }, { "epoch": 1.446521287642783, "grad_norm": 2.452303171157837, "learning_rate": 1.3457747935045193e-06, "loss": 0.01986924186348915, "step": 5572 }, { "epoch": 1.4467808930425754, "grad_norm": 2.40484619140625, "learning_rate": 1.345035758365234e-06, "loss": 0.07038949429988861, "step": 5573 }, { "epoch": 1.4470404984423677, "grad_norm": 2.315112352371216, "learning_rate": 1.3442969715914568e-06, "loss": 0.058104924857616425, "step": 5574 }, { "epoch": 1.44730010384216, "grad_norm": 2.7638344764709473, "learning_rate": 1.343558433313789e-06, "loss": 0.01148636732250452, "step": 5575 }, { "epoch": 1.4475597092419523, "grad_norm": 2.9691109657287598, "learning_rate": 1.3428201436627863e-06, "loss": 0.036177169531583786, "step": 5576 }, { "epoch": 1.4478193146417446, "grad_norm": 2.8811819553375244, "learning_rate": 1.342082102768963e-06, "loss": 0.05667837709188461, "step": 5577 }, { "epoch": 1.448078920041537, "grad_norm": 2.5782341957092285, "learning_rate": 1.3413443107627872e-06, "loss": 0.0052407956682145596, "step": 5578 }, { "epoch": 1.4483385254413292, "grad_norm": 3.3786230087280273, "learning_rate": 1.340606767774682e-06, "loss": 0.03472353518009186, "step": 5579 }, { "epoch": 1.4485981308411215, "grad_norm": 2.449594497680664, "learning_rate": 1.3398694739350303e-06, "loss": 0.026222029700875282, "step": 5580 }, { "epoch": 1.4488577362409139, "grad_norm": 2.2920520305633545, "learning_rate": 1.3391324293741681e-06, "loss": 0.07133390009403229, "step": 5581 }, { "epoch": 1.4491173416407062, "grad_norm": 1.496405005455017, "learning_rate": 1.3383956342223872e-06, "loss": 0.009648941457271576, "step": 5582 }, { "epoch": 1.4493769470404985, "grad_norm": 3.368255615234375, "learning_rate": 1.3376590886099372e-06, "loss": 0.043732959777116776, "step": 5583 }, { "epoch": 1.4496365524402908, "grad_norm": 1.7641477584838867, "learning_rate": 1.3369227926670215e-06, "loss": 0.004149942193180323, "step": 5584 }, { "epoch": 1.449896157840083, "grad_norm": 2.014096736907959, "learning_rate": 1.336186746523802e-06, "loss": 0.005752372555434704, "step": 5585 }, { "epoch": 1.4501557632398754, "grad_norm": 3.5335755348205566, "learning_rate": 1.3354509503103922e-06, "loss": 0.04220882058143616, "step": 5586 }, { "epoch": 1.4504153686396677, "grad_norm": 4.3655524253845215, "learning_rate": 1.3347154041568655e-06, "loss": 0.040313854813575745, "step": 5587 }, { "epoch": 1.45067497403946, "grad_norm": 3.233182191848755, "learning_rate": 1.3339801081932507e-06, "loss": 0.03902493417263031, "step": 5588 }, { "epoch": 1.4509345794392523, "grad_norm": 2.8679051399230957, "learning_rate": 1.3332450625495298e-06, "loss": 0.01700419932603836, "step": 5589 }, { "epoch": 1.4511941848390446, "grad_norm": 3.1227710247039795, "learning_rate": 1.332510267355643e-06, "loss": 0.023269470781087875, "step": 5590 }, { "epoch": 1.451453790238837, "grad_norm": 2.766923189163208, "learning_rate": 1.3317757227414835e-06, "loss": 0.04842543601989746, "step": 5591 }, { "epoch": 1.4517133956386292, "grad_norm": 4.236832141876221, "learning_rate": 1.331041428836904e-06, "loss": 0.009630111977458, "step": 5592 }, { "epoch": 1.4519730010384215, "grad_norm": 2.853097677230835, "learning_rate": 1.3303073857717103e-06, "loss": 0.0192450862377882, "step": 5593 }, { "epoch": 1.4522326064382138, "grad_norm": 3.0084545612335205, "learning_rate": 1.3295735936756631e-06, "loss": 0.030420728027820587, "step": 5594 }, { "epoch": 1.4524922118380061, "grad_norm": 3.9443304538726807, "learning_rate": 1.328840052678482e-06, "loss": 0.03049728274345398, "step": 5595 }, { "epoch": 1.4527518172377984, "grad_norm": 3.1692631244659424, "learning_rate": 1.3281067629098394e-06, "loss": 0.015710262581706047, "step": 5596 }, { "epoch": 1.4530114226375908, "grad_norm": 2.9524426460266113, "learning_rate": 1.3273737244993632e-06, "loss": 0.07633224129676819, "step": 5597 }, { "epoch": 1.453271028037383, "grad_norm": 0.6455792188644409, "learning_rate": 1.3266409375766392e-06, "loss": 0.0015941886231303215, "step": 5598 }, { "epoch": 1.4535306334371754, "grad_norm": 2.8770391941070557, "learning_rate": 1.3259084022712065e-06, "loss": 0.01040082611143589, "step": 5599 }, { "epoch": 1.4537902388369677, "grad_norm": 4.051125526428223, "learning_rate": 1.3251761187125614e-06, "loss": 0.01802469603717327, "step": 5600 }, { "epoch": 1.4540498442367602, "grad_norm": 2.61966609954834, "learning_rate": 1.3244440870301545e-06, "loss": 0.02764837257564068, "step": 5601 }, { "epoch": 1.4543094496365525, "grad_norm": 3.5423405170440674, "learning_rate": 1.3237123073533915e-06, "loss": 0.012654447928071022, "step": 5602 }, { "epoch": 1.4545690550363448, "grad_norm": 3.362494707107544, "learning_rate": 1.3229807798116351e-06, "loss": 0.05093761160969734, "step": 5603 }, { "epoch": 1.4548286604361371, "grad_norm": 2.880491018295288, "learning_rate": 1.3222495045342032e-06, "loss": 0.016600335016846657, "step": 5604 }, { "epoch": 1.4550882658359294, "grad_norm": 2.515324592590332, "learning_rate": 1.3215184816503663e-06, "loss": 0.03999796509742737, "step": 5605 }, { "epoch": 1.4553478712357217, "grad_norm": 2.1613919734954834, "learning_rate": 1.3207877112893553e-06, "loss": 0.007305909879505634, "step": 5606 }, { "epoch": 1.455607476635514, "grad_norm": 3.185626268386841, "learning_rate": 1.3200571935803513e-06, "loss": 0.01512685976922512, "step": 5607 }, { "epoch": 1.4558670820353063, "grad_norm": 4.160577297210693, "learning_rate": 1.3193269286524945e-06, "loss": 0.01206127181649208, "step": 5608 }, { "epoch": 1.4561266874350987, "grad_norm": 3.68898868560791, "learning_rate": 1.318596916634879e-06, "loss": 0.05640795826911926, "step": 5609 }, { "epoch": 1.456386292834891, "grad_norm": 3.0971689224243164, "learning_rate": 1.3178671576565533e-06, "loss": 0.014005916193127632, "step": 5610 }, { "epoch": 1.4566458982346833, "grad_norm": 4.450399398803711, "learning_rate": 1.3171376518465228e-06, "loss": 0.014443768188357353, "step": 5611 }, { "epoch": 1.4569055036344756, "grad_norm": 2.939946174621582, "learning_rate": 1.3164083993337478e-06, "loss": 0.010825326666235924, "step": 5612 }, { "epoch": 1.4571651090342679, "grad_norm": 2.634770631790161, "learning_rate": 1.3156794002471417e-06, "loss": 0.08091435581445694, "step": 5613 }, { "epoch": 1.4574247144340602, "grad_norm": 2.7890102863311768, "learning_rate": 1.314950654715577e-06, "loss": 0.05863203853368759, "step": 5614 }, { "epoch": 1.4576843198338525, "grad_norm": 2.6687936782836914, "learning_rate": 1.3142221628678778e-06, "loss": 0.021814759820699692, "step": 5615 }, { "epoch": 1.4579439252336448, "grad_norm": 3.435612440109253, "learning_rate": 1.3134939248328265e-06, "loss": 0.05496367812156677, "step": 5616 }, { "epoch": 1.458203530633437, "grad_norm": 2.215452194213867, "learning_rate": 1.3127659407391564e-06, "loss": 0.022086307406425476, "step": 5617 }, { "epoch": 1.4584631360332294, "grad_norm": 2.432292938232422, "learning_rate": 1.3120382107155595e-06, "loss": 0.04733221232891083, "step": 5618 }, { "epoch": 1.458722741433022, "grad_norm": 3.238631248474121, "learning_rate": 1.3113107348906833e-06, "loss": 0.028318976983428, "step": 5619 }, { "epoch": 1.4589823468328142, "grad_norm": 3.987748622894287, "learning_rate": 1.3105835133931264e-06, "loss": 0.011463879607617855, "step": 5620 }, { "epoch": 1.4592419522326066, "grad_norm": 3.1838979721069336, "learning_rate": 1.3098565463514467e-06, "loss": 0.026805942878127098, "step": 5621 }, { "epoch": 1.4595015576323989, "grad_norm": 3.453988552093506, "learning_rate": 1.3091298338941538e-06, "loss": 0.033858004957437515, "step": 5622 }, { "epoch": 1.4597611630321912, "grad_norm": 1.7336219549179077, "learning_rate": 1.3084033761497156e-06, "loss": 0.0012969798408448696, "step": 5623 }, { "epoch": 1.4600207684319835, "grad_norm": 2.744288444519043, "learning_rate": 1.3076771732465521e-06, "loss": 0.015810295939445496, "step": 5624 }, { "epoch": 1.4602803738317758, "grad_norm": 2.886596441268921, "learning_rate": 1.306951225313039e-06, "loss": 0.062068141996860504, "step": 5625 }, { "epoch": 1.460539979231568, "grad_norm": 2.177112102508545, "learning_rate": 1.3062255324775086e-06, "loss": 0.004656776320189238, "step": 5626 }, { "epoch": 1.4607995846313604, "grad_norm": 2.625194549560547, "learning_rate": 1.3055000948682458e-06, "loss": 0.00848047062754631, "step": 5627 }, { "epoch": 1.4610591900311527, "grad_norm": 3.60833477973938, "learning_rate": 1.3047749126134912e-06, "loss": 0.02609230764210224, "step": 5628 }, { "epoch": 1.461318795430945, "grad_norm": 3.4027135372161865, "learning_rate": 1.304049985841441e-06, "loss": 0.011375264264643192, "step": 5629 }, { "epoch": 1.4615784008307373, "grad_norm": 2.4353837966918945, "learning_rate": 1.303325314680245e-06, "loss": 0.008944879285991192, "step": 5630 }, { "epoch": 1.4618380062305296, "grad_norm": 4.251057147979736, "learning_rate": 1.3026008992580096e-06, "loss": 0.040388599038124084, "step": 5631 }, { "epoch": 1.462097611630322, "grad_norm": 1.5010040998458862, "learning_rate": 1.3018767397027942e-06, "loss": 0.0021002613939344883, "step": 5632 }, { "epoch": 1.4623572170301142, "grad_norm": 3.7533209323883057, "learning_rate": 1.301152836142613e-06, "loss": 0.03337244689464569, "step": 5633 }, { "epoch": 1.4626168224299065, "grad_norm": 4.103625774383545, "learning_rate": 1.3004291887054366e-06, "loss": 0.03554448485374451, "step": 5634 }, { "epoch": 1.4628764278296988, "grad_norm": 2.5080997943878174, "learning_rate": 1.299705797519189e-06, "loss": 0.06516525894403458, "step": 5635 }, { "epoch": 1.4631360332294912, "grad_norm": 2.864478826522827, "learning_rate": 1.2989826627117482e-06, "loss": 0.043258607387542725, "step": 5636 }, { "epoch": 1.4633956386292835, "grad_norm": 3.4557244777679443, "learning_rate": 1.2982597844109498e-06, "loss": 0.021695619449019432, "step": 5637 }, { "epoch": 1.4636552440290758, "grad_norm": 3.062485456466675, "learning_rate": 1.29753716274458e-06, "loss": 0.007527291774749756, "step": 5638 }, { "epoch": 1.463914849428868, "grad_norm": 3.8290839195251465, "learning_rate": 1.2968147978403835e-06, "loss": 0.023224536329507828, "step": 5639 }, { "epoch": 1.4641744548286604, "grad_norm": 2.3263845443725586, "learning_rate": 1.2960926898260575e-06, "loss": 0.0692884624004364, "step": 5640 }, { "epoch": 1.4644340602284527, "grad_norm": 1.4705867767333984, "learning_rate": 1.2953708388292533e-06, "loss": 0.0022436524741351604, "step": 5641 }, { "epoch": 1.464693665628245, "grad_norm": 3.5374441146850586, "learning_rate": 1.294649244977579e-06, "loss": 0.0496765598654747, "step": 5642 }, { "epoch": 1.4649532710280373, "grad_norm": 3.256840229034424, "learning_rate": 1.2939279083985945e-06, "loss": 0.021979883313179016, "step": 5643 }, { "epoch": 1.4652128764278296, "grad_norm": 3.0772674083709717, "learning_rate": 1.2932068292198157e-06, "loss": 0.0673961341381073, "step": 5644 }, { "epoch": 1.465472481827622, "grad_norm": 3.9101998805999756, "learning_rate": 1.292486007568714e-06, "loss": 0.049717459827661514, "step": 5645 }, { "epoch": 1.4657320872274142, "grad_norm": 3.2477407455444336, "learning_rate": 1.2917654435727137e-06, "loss": 0.03882573917508125, "step": 5646 }, { "epoch": 1.4659916926272065, "grad_norm": 3.0016121864318848, "learning_rate": 1.2910451373591936e-06, "loss": 0.046581290662288666, "step": 5647 }, { "epoch": 1.4662512980269988, "grad_norm": 3.6435115337371826, "learning_rate": 1.290325089055487e-06, "loss": 0.03308318182826042, "step": 5648 }, { "epoch": 1.4665109034267914, "grad_norm": 4.296600818634033, "learning_rate": 1.2896052987888825e-06, "loss": 0.033354099839925766, "step": 5649 }, { "epoch": 1.4667705088265837, "grad_norm": 2.694214105606079, "learning_rate": 1.2888857666866243e-06, "loss": 0.031216267496347427, "step": 5650 }, { "epoch": 1.467030114226376, "grad_norm": 3.3262112140655518, "learning_rate": 1.2881664928759051e-06, "loss": 0.05192750319838524, "step": 5651 }, { "epoch": 1.4672897196261683, "grad_norm": 2.4228744506835938, "learning_rate": 1.2874474774838802e-06, "loss": 0.06431561708450317, "step": 5652 }, { "epoch": 1.4675493250259606, "grad_norm": 2.690624713897705, "learning_rate": 1.2867287206376522e-06, "loss": 0.03773769736289978, "step": 5653 }, { "epoch": 1.467808930425753, "grad_norm": 2.134608030319214, "learning_rate": 1.2860102224642823e-06, "loss": 0.05916380509734154, "step": 5654 }, { "epoch": 1.4680685358255452, "grad_norm": 2.6021060943603516, "learning_rate": 1.2852919830907846e-06, "loss": 0.04342065379023552, "step": 5655 }, { "epoch": 1.4683281412253375, "grad_norm": 5.666736125946045, "learning_rate": 1.284574002644126e-06, "loss": 0.012585963122546673, "step": 5656 }, { "epoch": 1.4685877466251298, "grad_norm": 2.567899465560913, "learning_rate": 1.2838562812512306e-06, "loss": 0.01220683567225933, "step": 5657 }, { "epoch": 1.4688473520249221, "grad_norm": 3.092604875564575, "learning_rate": 1.2831388190389747e-06, "loss": 0.033402249217033386, "step": 5658 }, { "epoch": 1.4691069574247144, "grad_norm": 2.5707128047943115, "learning_rate": 1.2824216161341879e-06, "loss": 0.066123366355896, "step": 5659 }, { "epoch": 1.4693665628245067, "grad_norm": 3.075641632080078, "learning_rate": 1.2817046726636573e-06, "loss": 0.02453552931547165, "step": 5660 }, { "epoch": 1.469626168224299, "grad_norm": 2.7928876876831055, "learning_rate": 1.28098798875412e-06, "loss": 0.06004033982753754, "step": 5661 }, { "epoch": 1.4698857736240913, "grad_norm": 2.269244432449341, "learning_rate": 1.280271564532271e-06, "loss": 0.0214004535228014, "step": 5662 }, { "epoch": 1.4701453790238836, "grad_norm": 2.3626651763916016, "learning_rate": 1.2795554001247574e-06, "loss": 0.011743780225515366, "step": 5663 }, { "epoch": 1.470404984423676, "grad_norm": 2.754566192626953, "learning_rate": 1.2788394956581795e-06, "loss": 0.07747219502925873, "step": 5664 }, { "epoch": 1.4706645898234683, "grad_norm": 2.502347707748413, "learning_rate": 1.2781238512590944e-06, "loss": 0.05757306516170502, "step": 5665 }, { "epoch": 1.4709241952232606, "grad_norm": 2.5764079093933105, "learning_rate": 1.277408467054011e-06, "loss": 0.01430437806993723, "step": 5666 }, { "epoch": 1.471183800623053, "grad_norm": 2.788682699203491, "learning_rate": 1.2766933431693919e-06, "loss": 0.017122259363532066, "step": 5667 }, { "epoch": 1.4714434060228454, "grad_norm": 1.7448253631591797, "learning_rate": 1.275978479731656e-06, "loss": 0.0030383560806512833, "step": 5668 }, { "epoch": 1.4717030114226377, "grad_norm": 3.2393157482147217, "learning_rate": 1.2752638768671738e-06, "loss": 0.03228742256760597, "step": 5669 }, { "epoch": 1.47196261682243, "grad_norm": 3.692042589187622, "learning_rate": 1.2745495347022722e-06, "loss": 0.034299179911613464, "step": 5670 }, { "epoch": 1.4722222222222223, "grad_norm": 3.099069595336914, "learning_rate": 1.273835453363229e-06, "loss": 0.0071732960641384125, "step": 5671 }, { "epoch": 1.4724818276220146, "grad_norm": 3.89491605758667, "learning_rate": 1.2731216329762775e-06, "loss": 0.03421349078416824, "step": 5672 }, { "epoch": 1.472741433021807, "grad_norm": 3.742518901824951, "learning_rate": 1.2724080736676064e-06, "loss": 0.023952975869178772, "step": 5673 }, { "epoch": 1.4730010384215992, "grad_norm": 7.046811580657959, "learning_rate": 1.2716947755633547e-06, "loss": 0.018884792923927307, "step": 5674 }, { "epoch": 1.4732606438213915, "grad_norm": 2.128220319747925, "learning_rate": 1.2709817387896173e-06, "loss": 0.0021973750554025173, "step": 5675 }, { "epoch": 1.4735202492211839, "grad_norm": 3.9738235473632812, "learning_rate": 1.2702689634724446e-06, "loss": 0.03337552398443222, "step": 5676 }, { "epoch": 1.4737798546209762, "grad_norm": 2.8980414867401123, "learning_rate": 1.2695564497378377e-06, "loss": 0.006616098806262016, "step": 5677 }, { "epoch": 1.4740394600207685, "grad_norm": 3.6451587677001953, "learning_rate": 1.2688441977117526e-06, "loss": 0.043202754110097885, "step": 5678 }, { "epoch": 1.4742990654205608, "grad_norm": 3.550499200820923, "learning_rate": 1.2681322075200988e-06, "loss": 0.08491185307502747, "step": 5679 }, { "epoch": 1.474558670820353, "grad_norm": 4.163633346557617, "learning_rate": 1.2674204792887406e-06, "loss": 0.057379432022571564, "step": 5680 }, { "epoch": 1.4748182762201454, "grad_norm": 3.9772839546203613, "learning_rate": 1.2667090131434961e-06, "loss": 0.013276180252432823, "step": 5681 }, { "epoch": 1.4750778816199377, "grad_norm": 1.6142669916152954, "learning_rate": 1.2659978092101343e-06, "loss": 0.0038394778966903687, "step": 5682 }, { "epoch": 1.47533748701973, "grad_norm": 1.1866074800491333, "learning_rate": 1.2652868676143806e-06, "loss": 0.007897595874965191, "step": 5683 }, { "epoch": 1.4755970924195223, "grad_norm": 3.599918842315674, "learning_rate": 1.2645761884819136e-06, "loss": 0.01659184694290161, "step": 5684 }, { "epoch": 1.4758566978193146, "grad_norm": 3.3085384368896484, "learning_rate": 1.263865771938365e-06, "loss": 0.07219898700714111, "step": 5685 }, { "epoch": 1.476116303219107, "grad_norm": 3.7680230140686035, "learning_rate": 1.2631556181093202e-06, "loss": 0.041830796748399734, "step": 5686 }, { "epoch": 1.4763759086188992, "grad_norm": 3.5754027366638184, "learning_rate": 1.2624457271203172e-06, "loss": 0.011071179062128067, "step": 5687 }, { "epoch": 1.4766355140186915, "grad_norm": 0.5186857581138611, "learning_rate": 1.2617360990968486e-06, "loss": 0.0013831551186740398, "step": 5688 }, { "epoch": 1.4768951194184838, "grad_norm": 4.061644554138184, "learning_rate": 1.2610267341643628e-06, "loss": 0.033389683812856674, "step": 5689 }, { "epoch": 1.4771547248182761, "grad_norm": 2.7417874336242676, "learning_rate": 1.260317632448256e-06, "loss": 0.06428968906402588, "step": 5690 }, { "epoch": 1.4774143302180685, "grad_norm": 3.109924793243408, "learning_rate": 1.2596087940738832e-06, "loss": 0.05709107220172882, "step": 5691 }, { "epoch": 1.4776739356178608, "grad_norm": 2.570093870162964, "learning_rate": 1.2589002191665495e-06, "loss": 0.05412986874580383, "step": 5692 }, { "epoch": 1.477933541017653, "grad_norm": 4.398414611816406, "learning_rate": 1.2581919078515161e-06, "loss": 0.01239063497632742, "step": 5693 }, { "epoch": 1.4781931464174454, "grad_norm": 2.9535398483276367, "learning_rate": 1.2574838602539955e-06, "loss": 0.04747781902551651, "step": 5694 }, { "epoch": 1.4784527518172377, "grad_norm": 3.4247353076934814, "learning_rate": 1.2567760764991536e-06, "loss": 0.06053014472126961, "step": 5695 }, { "epoch": 1.47871235721703, "grad_norm": 2.236872434616089, "learning_rate": 1.2560685567121116e-06, "loss": 0.011305796913802624, "step": 5696 }, { "epoch": 1.4789719626168225, "grad_norm": 2.3698322772979736, "learning_rate": 1.2553613010179427e-06, "loss": 0.006135367322713137, "step": 5697 }, { "epoch": 1.4792315680166148, "grad_norm": 3.8617732524871826, "learning_rate": 1.2546543095416718e-06, "loss": 0.02130528725683689, "step": 5698 }, { "epoch": 1.4794911734164071, "grad_norm": 4.491860866546631, "learning_rate": 1.2539475824082813e-06, "loss": 0.013006528839468956, "step": 5699 }, { "epoch": 1.4797507788161994, "grad_norm": 3.5353164672851562, "learning_rate": 1.2532411197427034e-06, "loss": 0.010170474648475647, "step": 5700 }, { "epoch": 1.4800103842159917, "grad_norm": 2.8874242305755615, "learning_rate": 1.2525349216698229e-06, "loss": 0.053281962871551514, "step": 5701 }, { "epoch": 1.480269989615784, "grad_norm": 2.6510791778564453, "learning_rate": 1.2518289883144819e-06, "loss": 0.06739163398742676, "step": 5702 }, { "epoch": 1.4805295950155763, "grad_norm": 1.6346694231033325, "learning_rate": 1.2511233198014716e-06, "loss": 0.006071333307772875, "step": 5703 }, { "epoch": 1.4807892004153687, "grad_norm": 3.524928092956543, "learning_rate": 1.2504179162555392e-06, "loss": 0.03137471526861191, "step": 5704 }, { "epoch": 1.481048805815161, "grad_norm": 3.477022171020508, "learning_rate": 1.2497127778013833e-06, "loss": 0.05090541020035744, "step": 5705 }, { "epoch": 1.4813084112149533, "grad_norm": 1.1649539470672607, "learning_rate": 1.2490079045636554e-06, "loss": 0.006531345192342997, "step": 5706 }, { "epoch": 1.4815680166147456, "grad_norm": 1.0950655937194824, "learning_rate": 1.248303296666963e-06, "loss": 0.001668934477493167, "step": 5707 }, { "epoch": 1.4818276220145379, "grad_norm": 3.5166900157928467, "learning_rate": 1.247598954235863e-06, "loss": 0.029239080846309662, "step": 5708 }, { "epoch": 1.4820872274143302, "grad_norm": 2.280034303665161, "learning_rate": 1.2468948773948669e-06, "loss": 0.013380623422563076, "step": 5709 }, { "epoch": 1.4823468328141225, "grad_norm": 2.458163022994995, "learning_rate": 1.2461910662684404e-06, "loss": 0.006340086460113525, "step": 5710 }, { "epoch": 1.4826064382139148, "grad_norm": 3.419656753540039, "learning_rate": 1.2454875209810005e-06, "loss": 0.06800481677055359, "step": 5711 }, { "epoch": 1.482866043613707, "grad_norm": 4.013350009918213, "learning_rate": 1.2447842416569192e-06, "loss": 0.042085498571395874, "step": 5712 }, { "epoch": 1.4831256490134994, "grad_norm": 3.9101905822753906, "learning_rate": 1.244081228420518e-06, "loss": 0.010996355675160885, "step": 5713 }, { "epoch": 1.4833852544132917, "grad_norm": 3.169821262359619, "learning_rate": 1.2433784813960743e-06, "loss": 0.019913293421268463, "step": 5714 }, { "epoch": 1.4836448598130842, "grad_norm": 3.101466178894043, "learning_rate": 1.2426760007078188e-06, "loss": 0.012070922181010246, "step": 5715 }, { "epoch": 1.4839044652128766, "grad_norm": 3.2470057010650635, "learning_rate": 1.2419737864799334e-06, "loss": 0.043886713683605194, "step": 5716 }, { "epoch": 1.4841640706126689, "grad_norm": 3.682007074356079, "learning_rate": 1.2412718388365532e-06, "loss": 0.036279670894145966, "step": 5717 }, { "epoch": 1.4844236760124612, "grad_norm": 3.3148248195648193, "learning_rate": 1.2405701579017658e-06, "loss": 0.021100103855133057, "step": 5718 }, { "epoch": 1.4846832814122535, "grad_norm": 2.7719130516052246, "learning_rate": 1.2398687437996133e-06, "loss": 0.007722376845777035, "step": 5719 }, { "epoch": 1.4849428868120458, "grad_norm": 3.3861286640167236, "learning_rate": 1.239167596654091e-06, "loss": 0.026570169255137444, "step": 5720 }, { "epoch": 1.485202492211838, "grad_norm": 2.3680920600891113, "learning_rate": 1.2384667165891424e-06, "loss": 0.050990667194128036, "step": 5721 }, { "epoch": 1.4854620976116304, "grad_norm": 4.453882694244385, "learning_rate": 1.2377661037286698e-06, "loss": 0.015359073877334595, "step": 5722 }, { "epoch": 1.4857217030114227, "grad_norm": 3.7834928035736084, "learning_rate": 1.2370657581965236e-06, "loss": 0.02334820106625557, "step": 5723 }, { "epoch": 1.485981308411215, "grad_norm": 3.821767568588257, "learning_rate": 1.2363656801165104e-06, "loss": 0.008950072340667248, "step": 5724 }, { "epoch": 1.4862409138110073, "grad_norm": 3.1343696117401123, "learning_rate": 1.2356658696123873e-06, "loss": 0.010574160143733025, "step": 5725 }, { "epoch": 1.4865005192107996, "grad_norm": 7.016843795776367, "learning_rate": 1.234966326807864e-06, "loss": 0.024500902742147446, "step": 5726 }, { "epoch": 1.486760124610592, "grad_norm": 4.686921119689941, "learning_rate": 1.2342670518266057e-06, "loss": 0.01531661581248045, "step": 5727 }, { "epoch": 1.4870197300103842, "grad_norm": 2.535703182220459, "learning_rate": 1.2335680447922266e-06, "loss": 0.004289806820452213, "step": 5728 }, { "epoch": 1.4872793354101765, "grad_norm": 3.1630115509033203, "learning_rate": 1.2328693058282945e-06, "loss": 0.05099719762802124, "step": 5729 }, { "epoch": 1.4875389408099688, "grad_norm": 3.298400402069092, "learning_rate": 1.2321708350583324e-06, "loss": 0.027887530624866486, "step": 5730 }, { "epoch": 1.4877985462097612, "grad_norm": 3.6208739280700684, "learning_rate": 1.231472632605813e-06, "loss": 0.026212412863969803, "step": 5731 }, { "epoch": 1.4880581516095535, "grad_norm": 3.7254831790924072, "learning_rate": 1.230774698594162e-06, "loss": 0.03808005899190903, "step": 5732 }, { "epoch": 1.4883177570093458, "grad_norm": 4.327530384063721, "learning_rate": 1.230077033146759e-06, "loss": 0.041089437901973724, "step": 5733 }, { "epoch": 1.488577362409138, "grad_norm": 2.690875291824341, "learning_rate": 1.2293796363869345e-06, "loss": 0.058862313628196716, "step": 5734 }, { "epoch": 1.4888369678089304, "grad_norm": 3.1817715167999268, "learning_rate": 1.2286825084379733e-06, "loss": 0.0572514683008194, "step": 5735 }, { "epoch": 1.4890965732087227, "grad_norm": 2.9952821731567383, "learning_rate": 1.227985649423111e-06, "loss": 0.04690461978316307, "step": 5736 }, { "epoch": 1.489356178608515, "grad_norm": 3.212470054626465, "learning_rate": 1.2272890594655356e-06, "loss": 0.0185970701277256, "step": 5737 }, { "epoch": 1.4896157840083073, "grad_norm": 4.2125020027160645, "learning_rate": 1.2265927386883898e-06, "loss": 0.024649163708090782, "step": 5738 }, { "epoch": 1.4898753894080996, "grad_norm": 2.7402985095977783, "learning_rate": 1.2258966872147664e-06, "loss": 0.04042331129312515, "step": 5739 }, { "epoch": 1.490134994807892, "grad_norm": 2.6411359310150146, "learning_rate": 1.2252009051677106e-06, "loss": 0.06101846694946289, "step": 5740 }, { "epoch": 1.4903946002076842, "grad_norm": 3.9329283237457275, "learning_rate": 1.2245053926702224e-06, "loss": 0.01050881203263998, "step": 5741 }, { "epoch": 1.4906542056074765, "grad_norm": 2.3567147254943848, "learning_rate": 1.2238101498452507e-06, "loss": 0.057314664125442505, "step": 5742 }, { "epoch": 1.4909138110072688, "grad_norm": 2.5778980255126953, "learning_rate": 1.2231151768157008e-06, "loss": 0.062469616532325745, "step": 5743 }, { "epoch": 1.4911734164070611, "grad_norm": 3.3325650691986084, "learning_rate": 1.2224204737044252e-06, "loss": 0.005799282342195511, "step": 5744 }, { "epoch": 1.4914330218068537, "grad_norm": 2.4969990253448486, "learning_rate": 1.221726040634233e-06, "loss": 0.019196882843971252, "step": 5745 }, { "epoch": 1.491692627206646, "grad_norm": 3.310720920562744, "learning_rate": 1.2210318777278844e-06, "loss": 0.012038907036185265, "step": 5746 }, { "epoch": 1.4919522326064383, "grad_norm": 2.970520257949829, "learning_rate": 1.2203379851080914e-06, "loss": 0.0573384091258049, "step": 5747 }, { "epoch": 1.4922118380062306, "grad_norm": 1.8393628597259521, "learning_rate": 1.219644362897518e-06, "loss": 0.010891802608966827, "step": 5748 }, { "epoch": 1.492471443406023, "grad_norm": 3.0697948932647705, "learning_rate": 1.21895101121878e-06, "loss": 0.036287572234869, "step": 5749 }, { "epoch": 1.4927310488058152, "grad_norm": 2.7908525466918945, "learning_rate": 1.218257930194447e-06, "loss": 0.018890127539634705, "step": 5750 }, { "epoch": 1.4929906542056075, "grad_norm": 3.7058019638061523, "learning_rate": 1.2175651199470408e-06, "loss": 0.02917269617319107, "step": 5751 }, { "epoch": 1.4932502596053998, "grad_norm": 3.730365514755249, "learning_rate": 1.2168725805990319e-06, "loss": 0.017030812799930573, "step": 5752 }, { "epoch": 1.4935098650051921, "grad_norm": 2.8731088638305664, "learning_rate": 1.2161803122728476e-06, "loss": 0.012867150828242302, "step": 5753 }, { "epoch": 1.4937694704049844, "grad_norm": 3.7790989875793457, "learning_rate": 1.2154883150908635e-06, "loss": 0.048714738339185715, "step": 5754 }, { "epoch": 1.4940290758047767, "grad_norm": 3.354414939880371, "learning_rate": 1.2147965891754104e-06, "loss": 0.00901571474969387, "step": 5755 }, { "epoch": 1.494288681204569, "grad_norm": 2.3252687454223633, "learning_rate": 1.2141051346487692e-06, "loss": 0.015591172501444817, "step": 5756 }, { "epoch": 1.4945482866043613, "grad_norm": 2.9819202423095703, "learning_rate": 1.2134139516331715e-06, "loss": 0.05305632948875427, "step": 5757 }, { "epoch": 1.4948078920041536, "grad_norm": 3.4609763622283936, "learning_rate": 1.212723040250805e-06, "loss": 0.04120009392499924, "step": 5758 }, { "epoch": 1.495067497403946, "grad_norm": 3.5947067737579346, "learning_rate": 1.212032400623806e-06, "loss": 0.015406759455800056, "step": 5759 }, { "epoch": 1.4953271028037383, "grad_norm": 3.0471863746643066, "learning_rate": 1.2113420328742629e-06, "loss": 0.032056137919425964, "step": 5760 }, { "epoch": 1.4955867082035306, "grad_norm": 2.2779457569122314, "learning_rate": 1.2106519371242188e-06, "loss": 0.004734097048640251, "step": 5761 }, { "epoch": 1.4958463136033229, "grad_norm": 2.7500197887420654, "learning_rate": 1.2099621134956656e-06, "loss": 0.02605261281132698, "step": 5762 }, { "epoch": 1.4961059190031154, "grad_norm": 3.0658929347991943, "learning_rate": 1.209272562110548e-06, "loss": 0.004039866384118795, "step": 5763 }, { "epoch": 1.4963655244029077, "grad_norm": 3.543619394302368, "learning_rate": 1.2085832830907642e-06, "loss": 0.012729662470519543, "step": 5764 }, { "epoch": 1.4966251298027, "grad_norm": 2.9636783599853516, "learning_rate": 1.2078942765581611e-06, "loss": 0.01638069562613964, "step": 5765 }, { "epoch": 1.4968847352024923, "grad_norm": 2.044745922088623, "learning_rate": 1.2072055426345414e-06, "loss": 0.006806391291320324, "step": 5766 }, { "epoch": 1.4971443406022846, "grad_norm": 2.4764158725738525, "learning_rate": 1.2065170814416564e-06, "loss": 0.016253609210252762, "step": 5767 }, { "epoch": 1.497403946002077, "grad_norm": 3.8448286056518555, "learning_rate": 1.2058288931012095e-06, "loss": 0.05014324188232422, "step": 5768 }, { "epoch": 1.4976635514018692, "grad_norm": 2.823460578918457, "learning_rate": 1.205140977734858e-06, "loss": 0.07938501238822937, "step": 5769 }, { "epoch": 1.4979231568016615, "grad_norm": 2.345582962036133, "learning_rate": 1.2044533354642095e-06, "loss": 0.012418979778885841, "step": 5770 }, { "epoch": 1.4981827622014539, "grad_norm": 3.45740008354187, "learning_rate": 1.2037659664108222e-06, "loss": 0.048607029020786285, "step": 5771 }, { "epoch": 1.4984423676012462, "grad_norm": 2.7457869052886963, "learning_rate": 1.2030788706962085e-06, "loss": 0.010167937725782394, "step": 5772 }, { "epoch": 1.4987019730010385, "grad_norm": 2.6137359142303467, "learning_rate": 1.20239204844183e-06, "loss": 0.06665525585412979, "step": 5773 }, { "epoch": 1.4989615784008308, "grad_norm": 5.887295246124268, "learning_rate": 1.2017054997691032e-06, "loss": 0.01926823891699314, "step": 5774 }, { "epoch": 1.499221183800623, "grad_norm": 3.8788089752197266, "learning_rate": 1.2010192247993915e-06, "loss": 0.03281896188855171, "step": 5775 }, { "epoch": 1.4994807892004154, "grad_norm": 2.4418280124664307, "learning_rate": 1.200333223654014e-06, "loss": 0.06122181937098503, "step": 5776 }, { "epoch": 1.4997403946002077, "grad_norm": 2.5627193450927734, "learning_rate": 1.19964749645424e-06, "loss": 0.006823606789112091, "step": 5777 }, { "epoch": 1.5, "grad_norm": 2.4211816787719727, "learning_rate": 1.1989620433212908e-06, "loss": 0.021990817040205002, "step": 5778 }, { "epoch": 1.5002596053997923, "grad_norm": 3.0048320293426514, "learning_rate": 1.1982768643763382e-06, "loss": 0.040059227496385574, "step": 5779 }, { "epoch": 1.5005192107995846, "grad_norm": 2.828242301940918, "learning_rate": 1.1975919597405054e-06, "loss": 0.043635763227939606, "step": 5780 }, { "epoch": 1.500778816199377, "grad_norm": 3.4721782207489014, "learning_rate": 1.1969073295348686e-06, "loss": 0.04909808933734894, "step": 5781 }, { "epoch": 1.5010384215991692, "grad_norm": 2.5976665019989014, "learning_rate": 1.1962229738804565e-06, "loss": 0.01563866063952446, "step": 5782 }, { "epoch": 1.5012980269989615, "grad_norm": 2.495990037918091, "learning_rate": 1.1955388928982444e-06, "loss": 0.024747837334871292, "step": 5783 }, { "epoch": 1.5015576323987538, "grad_norm": 4.30174446105957, "learning_rate": 1.1948550867091643e-06, "loss": 0.03034406527876854, "step": 5784 }, { "epoch": 1.5018172377985461, "grad_norm": 2.681401014328003, "learning_rate": 1.1941715554340965e-06, "loss": 0.026722021400928497, "step": 5785 }, { "epoch": 1.5020768431983385, "grad_norm": 2.8292019367218018, "learning_rate": 1.1934882991938737e-06, "loss": 0.03154021129012108, "step": 5786 }, { "epoch": 1.5023364485981308, "grad_norm": 2.9056146144866943, "learning_rate": 1.1928053181092804e-06, "loss": 0.07101008296012878, "step": 5787 }, { "epoch": 1.502596053997923, "grad_norm": 3.2877607345581055, "learning_rate": 1.1921226123010515e-06, "loss": 0.02178444340825081, "step": 5788 }, { "epoch": 1.5028556593977154, "grad_norm": 3.152373790740967, "learning_rate": 1.1914401818898745e-06, "loss": 0.03346139192581177, "step": 5789 }, { "epoch": 1.5031152647975077, "grad_norm": 1.5660367012023926, "learning_rate": 1.1907580269963874e-06, "loss": 0.008543318137526512, "step": 5790 }, { "epoch": 1.5033748701973, "grad_norm": 3.3422093391418457, "learning_rate": 1.1900761477411782e-06, "loss": 0.03013298101723194, "step": 5791 }, { "epoch": 1.5036344755970923, "grad_norm": 3.509394645690918, "learning_rate": 1.1893945442447892e-06, "loss": 0.03493134677410126, "step": 5792 }, { "epoch": 1.5038940809968846, "grad_norm": 3.1305036544799805, "learning_rate": 1.188713216627712e-06, "loss": 0.04808962345123291, "step": 5793 }, { "epoch": 1.504153686396677, "grad_norm": 4.027866840362549, "learning_rate": 1.1880321650103884e-06, "loss": 0.05197472497820854, "step": 5794 }, { "epoch": 1.5044132917964692, "grad_norm": 3.102457284927368, "learning_rate": 1.1873513895132145e-06, "loss": 0.04702337831258774, "step": 5795 }, { "epoch": 1.5046728971962615, "grad_norm": 2.7455861568450928, "learning_rate": 1.1866708902565346e-06, "loss": 0.02622465044260025, "step": 5796 }, { "epoch": 1.504932502596054, "grad_norm": 3.552788257598877, "learning_rate": 1.1859906673606462e-06, "loss": 0.030040714889764786, "step": 5797 }, { "epoch": 1.5051921079958464, "grad_norm": 2.810030698776245, "learning_rate": 1.1853107209457966e-06, "loss": 0.058097679167985916, "step": 5798 }, { "epoch": 1.5054517133956387, "grad_norm": 3.2256343364715576, "learning_rate": 1.1846310511321845e-06, "loss": 0.06532019376754761, "step": 5799 }, { "epoch": 1.505711318795431, "grad_norm": 1.8219887018203735, "learning_rate": 1.1839516580399607e-06, "loss": 0.00802621990442276, "step": 5800 }, { "epoch": 1.5059709241952233, "grad_norm": 2.7963616847991943, "learning_rate": 1.1832725417892263e-06, "loss": 0.010806080885231495, "step": 5801 }, { "epoch": 1.5062305295950156, "grad_norm": 3.011784076690674, "learning_rate": 1.1825937025000325e-06, "loss": 0.03464488685131073, "step": 5802 }, { "epoch": 1.5064901349948079, "grad_norm": 3.051129102706909, "learning_rate": 1.181915140292384e-06, "loss": 0.022261476144194603, "step": 5803 }, { "epoch": 1.5067497403946002, "grad_norm": 3.183440923690796, "learning_rate": 1.1812368552862336e-06, "loss": 0.02938353270292282, "step": 5804 }, { "epoch": 1.5070093457943925, "grad_norm": 3.289010763168335, "learning_rate": 1.1805588476014886e-06, "loss": 0.01705051027238369, "step": 5805 }, { "epoch": 1.5072689511941848, "grad_norm": 3.468224048614502, "learning_rate": 1.1798811173580024e-06, "loss": 0.04856455698609352, "step": 5806 }, { "epoch": 1.507528556593977, "grad_norm": 3.354290008544922, "learning_rate": 1.1792036646755838e-06, "loss": 0.011653386056423187, "step": 5807 }, { "epoch": 1.5077881619937694, "grad_norm": 3.4356706142425537, "learning_rate": 1.1785264896739914e-06, "loss": 0.04106544703245163, "step": 5808 }, { "epoch": 1.508047767393562, "grad_norm": 1.459548830986023, "learning_rate": 1.177849592472934e-06, "loss": 0.011926430277526379, "step": 5809 }, { "epoch": 1.5083073727933543, "grad_norm": 2.565600872039795, "learning_rate": 1.1771729731920707e-06, "loss": 0.05728050321340561, "step": 5810 }, { "epoch": 1.5085669781931466, "grad_norm": 2.4951517581939697, "learning_rate": 1.1764966319510124e-06, "loss": 0.06934474408626556, "step": 5811 }, { "epoch": 1.5088265835929389, "grad_norm": 3.132415294647217, "learning_rate": 1.1758205688693219e-06, "loss": 0.07012808322906494, "step": 5812 }, { "epoch": 1.5090861889927312, "grad_norm": 4.208166122436523, "learning_rate": 1.1751447840665105e-06, "loss": 0.033704109489917755, "step": 5813 }, { "epoch": 1.5093457943925235, "grad_norm": 3.745885133743286, "learning_rate": 1.1744692776620414e-06, "loss": 0.025851044803857803, "step": 5814 }, { "epoch": 1.5096053997923158, "grad_norm": 2.248831033706665, "learning_rate": 1.17379404977533e-06, "loss": 0.015854408964514732, "step": 5815 }, { "epoch": 1.509865005192108, "grad_norm": 1.6074650287628174, "learning_rate": 1.1731191005257399e-06, "loss": 0.008247744292020798, "step": 5816 }, { "epoch": 1.5101246105919004, "grad_norm": 0.5032666325569153, "learning_rate": 1.1724444300325864e-06, "loss": 0.0015974343987181783, "step": 5817 }, { "epoch": 1.5103842159916927, "grad_norm": 3.7386553287506104, "learning_rate": 1.1717700384151375e-06, "loss": 0.03127730265259743, "step": 5818 }, { "epoch": 1.510643821391485, "grad_norm": 2.8594396114349365, "learning_rate": 1.1710959257926084e-06, "loss": 0.06755969673395157, "step": 5819 }, { "epoch": 1.5109034267912773, "grad_norm": 2.4069395065307617, "learning_rate": 1.1704220922841678e-06, "loss": 0.007628558203577995, "step": 5820 }, { "epoch": 1.5111630321910696, "grad_norm": 3.550196886062622, "learning_rate": 1.169748538008934e-06, "loss": 0.044975847005844116, "step": 5821 }, { "epoch": 1.511422637590862, "grad_norm": 3.207016944885254, "learning_rate": 1.1690752630859748e-06, "loss": 0.022872360423207283, "step": 5822 }, { "epoch": 1.5116822429906542, "grad_norm": 3.068850517272949, "learning_rate": 1.1684022676343118e-06, "loss": 0.028935827314853668, "step": 5823 }, { "epoch": 1.5119418483904465, "grad_norm": 2.9543349742889404, "learning_rate": 1.1677295517729138e-06, "loss": 0.029891707003116608, "step": 5824 }, { "epoch": 1.5122014537902388, "grad_norm": 2.5608444213867188, "learning_rate": 1.1670571156207013e-06, "loss": 0.011808359064161777, "step": 5825 }, { "epoch": 1.5124610591900312, "grad_norm": 4.006289958953857, "learning_rate": 1.1663849592965466e-06, "loss": 0.028350843116641045, "step": 5826 }, { "epoch": 1.5127206645898235, "grad_norm": 3.189432382583618, "learning_rate": 1.1657130829192708e-06, "loss": 0.04801395907998085, "step": 5827 }, { "epoch": 1.5129802699896158, "grad_norm": 4.103120803833008, "learning_rate": 1.1650414866076472e-06, "loss": 0.0409429594874382, "step": 5828 }, { "epoch": 1.513239875389408, "grad_norm": 3.749119520187378, "learning_rate": 1.1643701704803977e-06, "loss": 0.03390670567750931, "step": 5829 }, { "epoch": 1.5134994807892004, "grad_norm": 3.545891046524048, "learning_rate": 1.1636991346561956e-06, "loss": 0.018796324729919434, "step": 5830 }, { "epoch": 1.5137590861889927, "grad_norm": 4.174508094787598, "learning_rate": 1.1630283792536657e-06, "loss": 0.02345024049282074, "step": 5831 }, { "epoch": 1.514018691588785, "grad_norm": 3.8405418395996094, "learning_rate": 1.1623579043913814e-06, "loss": 0.017189286649227142, "step": 5832 }, { "epoch": 1.5142782969885773, "grad_norm": 3.8506457805633545, "learning_rate": 1.1616877101878665e-06, "loss": 0.026645880192518234, "step": 5833 }, { "epoch": 1.5145379023883696, "grad_norm": 2.9262819290161133, "learning_rate": 1.161017796761598e-06, "loss": 0.04829345643520355, "step": 5834 }, { "epoch": 1.514797507788162, "grad_norm": 3.075695514678955, "learning_rate": 1.1603481642309992e-06, "loss": 0.006261915899813175, "step": 5835 }, { "epoch": 1.5150571131879542, "grad_norm": 3.7507665157318115, "learning_rate": 1.1596788127144485e-06, "loss": 0.041629351675510406, "step": 5836 }, { "epoch": 1.5153167185877465, "grad_norm": 3.1266841888427734, "learning_rate": 1.1590097423302683e-06, "loss": 0.006198530085384846, "step": 5837 }, { "epoch": 1.5155763239875388, "grad_norm": 2.790285587310791, "learning_rate": 1.1583409531967372e-06, "loss": 0.058342643082141876, "step": 5838 }, { "epoch": 1.5158359293873311, "grad_norm": 3.3141367435455322, "learning_rate": 1.1576724454320828e-06, "loss": 0.009098529815673828, "step": 5839 }, { "epoch": 1.5160955347871234, "grad_norm": 2.6615982055664062, "learning_rate": 1.1570042191544787e-06, "loss": 0.07016721367835999, "step": 5840 }, { "epoch": 1.5163551401869158, "grad_norm": 3.4891693592071533, "learning_rate": 1.156336274482055e-06, "loss": 0.010143887251615524, "step": 5841 }, { "epoch": 1.516614745586708, "grad_norm": 3.2767677307128906, "learning_rate": 1.1556686115328872e-06, "loss": 0.008378631435334682, "step": 5842 }, { "epoch": 1.5168743509865004, "grad_norm": 1.7013145685195923, "learning_rate": 1.1550012304250043e-06, "loss": 0.0033584069460630417, "step": 5843 }, { "epoch": 1.5171339563862927, "grad_norm": 2.7928454875946045, "learning_rate": 1.154334131276383e-06, "loss": 0.01216735690832138, "step": 5844 }, { "epoch": 1.5173935617860852, "grad_norm": 2.119633913040161, "learning_rate": 1.1536673142049507e-06, "loss": 0.020771795883774757, "step": 5845 }, { "epoch": 1.5176531671858775, "grad_norm": 3.1262214183807373, "learning_rate": 1.153000779328587e-06, "loss": 0.028731871396303177, "step": 5846 }, { "epoch": 1.5179127725856698, "grad_norm": 3.155864953994751, "learning_rate": 1.1523345267651188e-06, "loss": 0.014967849478125572, "step": 5847 }, { "epoch": 1.5181723779854621, "grad_norm": 3.396252393722534, "learning_rate": 1.1516685566323243e-06, "loss": 0.02346024662256241, "step": 5848 }, { "epoch": 1.5184319833852544, "grad_norm": 3.5473554134368896, "learning_rate": 1.1510028690479323e-06, "loss": 0.024132035672664642, "step": 5849 }, { "epoch": 1.5186915887850467, "grad_norm": 2.366640567779541, "learning_rate": 1.1503374641296204e-06, "loss": 0.03347735106945038, "step": 5850 }, { "epoch": 1.518951194184839, "grad_norm": 3.0165534019470215, "learning_rate": 1.1496723419950178e-06, "loss": 0.057550616562366486, "step": 5851 }, { "epoch": 1.5192107995846313, "grad_norm": 2.9102580547332764, "learning_rate": 1.149007502761703e-06, "loss": 0.03403250128030777, "step": 5852 }, { "epoch": 1.5194704049844237, "grad_norm": 3.523649215698242, "learning_rate": 1.1483429465472026e-06, "loss": 0.05083751305937767, "step": 5853 }, { "epoch": 1.519730010384216, "grad_norm": 4.775001525878906, "learning_rate": 1.1476786734689973e-06, "loss": 0.018808964639902115, "step": 5854 }, { "epoch": 1.5199896157840083, "grad_norm": 2.8505401611328125, "learning_rate": 1.1470146836445141e-06, "loss": 0.039344318211078644, "step": 5855 }, { "epoch": 1.5202492211838006, "grad_norm": 2.514307737350464, "learning_rate": 1.1463509771911305e-06, "loss": 0.08168670535087585, "step": 5856 }, { "epoch": 1.520508826583593, "grad_norm": 3.0099294185638428, "learning_rate": 1.1456875542261765e-06, "loss": 0.03906608372926712, "step": 5857 }, { "epoch": 1.5207684319833854, "grad_norm": 3.1663384437561035, "learning_rate": 1.145024414866928e-06, "loss": 0.04430078715085983, "step": 5858 }, { "epoch": 1.5210280373831777, "grad_norm": 3.497581958770752, "learning_rate": 1.1443615592306147e-06, "loss": 0.018035512417554855, "step": 5859 }, { "epoch": 1.52128764278297, "grad_norm": 1.8986517190933228, "learning_rate": 1.1436989874344135e-06, "loss": 0.01463153026998043, "step": 5860 }, { "epoch": 1.5215472481827623, "grad_norm": 4.579676628112793, "learning_rate": 1.143036699595451e-06, "loss": 0.01443540956825018, "step": 5861 }, { "epoch": 1.5218068535825546, "grad_norm": 2.4103033542633057, "learning_rate": 1.1423746958308066e-06, "loss": 0.044766202569007874, "step": 5862 }, { "epoch": 1.522066458982347, "grad_norm": 2.9009792804718018, "learning_rate": 1.1417129762575057e-06, "loss": 0.027354737743735313, "step": 5863 }, { "epoch": 1.5223260643821392, "grad_norm": 2.834103584289551, "learning_rate": 1.1410515409925252e-06, "loss": 0.05597903952002525, "step": 5864 }, { "epoch": 1.5225856697819315, "grad_norm": 2.994661808013916, "learning_rate": 1.1403903901527929e-06, "loss": 0.03704989328980446, "step": 5865 }, { "epoch": 1.5228452751817239, "grad_norm": 2.425139904022217, "learning_rate": 1.1397295238551847e-06, "loss": 0.0614374503493309, "step": 5866 }, { "epoch": 1.5231048805815162, "grad_norm": 1.995538592338562, "learning_rate": 1.1390689422165262e-06, "loss": 0.01276104710996151, "step": 5867 }, { "epoch": 1.5233644859813085, "grad_norm": 2.3710997104644775, "learning_rate": 1.1384086453535925e-06, "loss": 0.05553199350833893, "step": 5868 }, { "epoch": 1.5236240913811008, "grad_norm": 1.8698736429214478, "learning_rate": 1.1377486333831099e-06, "loss": 0.009433194994926453, "step": 5869 }, { "epoch": 1.523883696780893, "grad_norm": 2.6378262042999268, "learning_rate": 1.1370889064217545e-06, "loss": 0.01602780446410179, "step": 5870 }, { "epoch": 1.5241433021806854, "grad_norm": 2.8561370372772217, "learning_rate": 1.1364294645861483e-06, "loss": 0.020123250782489777, "step": 5871 }, { "epoch": 1.5244029075804777, "grad_norm": 1.6290838718414307, "learning_rate": 1.1357703079928676e-06, "loss": 0.007874570786952972, "step": 5872 }, { "epoch": 1.52466251298027, "grad_norm": 2.6181583404541016, "learning_rate": 1.1351114367584347e-06, "loss": 0.049482449889183044, "step": 5873 }, { "epoch": 1.5249221183800623, "grad_norm": 3.36913800239563, "learning_rate": 1.1344528509993246e-06, "loss": 0.022046983242034912, "step": 5874 }, { "epoch": 1.5251817237798546, "grad_norm": 4.775784969329834, "learning_rate": 1.1337945508319593e-06, "loss": 0.04133722931146622, "step": 5875 }, { "epoch": 1.525441329179647, "grad_norm": 1.4206997156143188, "learning_rate": 1.1331365363727106e-06, "loss": 0.011004609055817127, "step": 5876 }, { "epoch": 1.5257009345794392, "grad_norm": 3.336353302001953, "learning_rate": 1.1324788077379015e-06, "loss": 0.007985399104654789, "step": 5877 }, { "epoch": 1.5259605399792315, "grad_norm": 2.9996743202209473, "learning_rate": 1.1318213650438028e-06, "loss": 0.02710297890007496, "step": 5878 }, { "epoch": 1.5262201453790238, "grad_norm": 3.1303234100341797, "learning_rate": 1.131164208406635e-06, "loss": 0.026502477005124092, "step": 5879 }, { "epoch": 1.5264797507788161, "grad_norm": 4.976485729217529, "learning_rate": 1.130507337942569e-06, "loss": 0.02179352007806301, "step": 5880 }, { "epoch": 1.5267393561786085, "grad_norm": 3.2347896099090576, "learning_rate": 1.129850753767724e-06, "loss": 0.029787562787532806, "step": 5881 }, { "epoch": 1.5269989615784008, "grad_norm": 3.3829777240753174, "learning_rate": 1.1291944559981697e-06, "loss": 0.019488897174596786, "step": 5882 }, { "epoch": 1.527258566978193, "grad_norm": 3.658125877380371, "learning_rate": 1.128538444749924e-06, "loss": 0.01132073812186718, "step": 5883 }, { "epoch": 1.5275181723779854, "grad_norm": 3.555880308151245, "learning_rate": 1.1278827201389539e-06, "loss": 0.028456801548600197, "step": 5884 }, { "epoch": 1.5277777777777777, "grad_norm": 4.131979942321777, "learning_rate": 1.1272272822811784e-06, "loss": 0.022906839847564697, "step": 5885 }, { "epoch": 1.52803738317757, "grad_norm": 2.3374574184417725, "learning_rate": 1.1265721312924624e-06, "loss": 0.013410335406661034, "step": 5886 }, { "epoch": 1.5282969885773623, "grad_norm": 2.7692365646362305, "learning_rate": 1.1259172672886216e-06, "loss": 0.037529606372117996, "step": 5887 }, { "epoch": 1.5285565939771546, "grad_norm": 3.292290687561035, "learning_rate": 1.1252626903854218e-06, "loss": 0.012814875692129135, "step": 5888 }, { "epoch": 1.528816199376947, "grad_norm": 3.594191789627075, "learning_rate": 1.1246084006985764e-06, "loss": 0.04962912201881409, "step": 5889 }, { "epoch": 1.5290758047767392, "grad_norm": 4.514123916625977, "learning_rate": 1.1239543983437498e-06, "loss": 0.04294510930776596, "step": 5890 }, { "epoch": 1.5293354101765315, "grad_norm": 2.754479169845581, "learning_rate": 1.1233006834365542e-06, "loss": 0.0492207407951355, "step": 5891 }, { "epoch": 1.5295950155763238, "grad_norm": 2.711947441101074, "learning_rate": 1.1226472560925502e-06, "loss": 0.07849155366420746, "step": 5892 }, { "epoch": 1.5298546209761164, "grad_norm": 3.0993614196777344, "learning_rate": 1.1219941164272515e-06, "loss": 0.014363713562488556, "step": 5893 }, { "epoch": 1.5301142263759087, "grad_norm": 1.7754909992218018, "learning_rate": 1.1213412645561154e-06, "loss": 0.006305868737399578, "step": 5894 }, { "epoch": 1.530373831775701, "grad_norm": 2.7637438774108887, "learning_rate": 1.1206887005945521e-06, "loss": 0.01812119595706463, "step": 5895 }, { "epoch": 1.5306334371754933, "grad_norm": 3.3016374111175537, "learning_rate": 1.1200364246579213e-06, "loss": 0.02126535028219223, "step": 5896 }, { "epoch": 1.5308930425752856, "grad_norm": 2.052591562271118, "learning_rate": 1.119384436861529e-06, "loss": 0.021250933408737183, "step": 5897 }, { "epoch": 1.5311526479750779, "grad_norm": 2.9895212650299072, "learning_rate": 1.1187327373206323e-06, "loss": 0.007065082900226116, "step": 5898 }, { "epoch": 1.5314122533748702, "grad_norm": 8.455843925476074, "learning_rate": 1.118081326150436e-06, "loss": 0.037221506237983704, "step": 5899 }, { "epoch": 1.5316718587746625, "grad_norm": 2.796300172805786, "learning_rate": 1.117430203466095e-06, "loss": 0.011059356853365898, "step": 5900 }, { "epoch": 1.5319314641744548, "grad_norm": 3.084310531616211, "learning_rate": 1.1167793693827145e-06, "loss": 0.02183816023170948, "step": 5901 }, { "epoch": 1.532191069574247, "grad_norm": 2.6429240703582764, "learning_rate": 1.1161288240153446e-06, "loss": 0.007890337146818638, "step": 5902 }, { "epoch": 1.5324506749740394, "grad_norm": 2.9578566551208496, "learning_rate": 1.1154785674789883e-06, "loss": 0.026484716683626175, "step": 5903 }, { "epoch": 1.5327102803738317, "grad_norm": 0.8402738571166992, "learning_rate": 1.1148285998885956e-06, "loss": 0.001256778254173696, "step": 5904 }, { "epoch": 1.5329698857736243, "grad_norm": 3.26332950592041, "learning_rate": 1.114178921359066e-06, "loss": 0.023333091288805008, "step": 5905 }, { "epoch": 1.5332294911734166, "grad_norm": 3.7730109691619873, "learning_rate": 1.113529532005248e-06, "loss": 0.04137740284204483, "step": 5906 }, { "epoch": 1.5334890965732089, "grad_norm": 3.029895067214966, "learning_rate": 1.112880431941938e-06, "loss": 0.03700597584247589, "step": 5907 }, { "epoch": 1.5337487019730012, "grad_norm": 3.1106159687042236, "learning_rate": 1.112231621283883e-06, "loss": 0.03222542628645897, "step": 5908 }, { "epoch": 1.5340083073727935, "grad_norm": 3.1977121829986572, "learning_rate": 1.1115831001457775e-06, "loss": 0.04330768436193466, "step": 5909 }, { "epoch": 1.5342679127725858, "grad_norm": 3.5999271869659424, "learning_rate": 1.1109348686422643e-06, "loss": 0.026657454669475555, "step": 5910 }, { "epoch": 1.534527518172378, "grad_norm": 2.4589524269104004, "learning_rate": 1.1102869268879379e-06, "loss": 0.013249527662992477, "step": 5911 }, { "epoch": 1.5347871235721704, "grad_norm": 2.795060634613037, "learning_rate": 1.1096392749973373e-06, "loss": 0.05440690368413925, "step": 5912 }, { "epoch": 1.5350467289719627, "grad_norm": 2.463529586791992, "learning_rate": 1.1089919130849543e-06, "loss": 0.011209944263100624, "step": 5913 }, { "epoch": 1.535306334371755, "grad_norm": 2.538601875305176, "learning_rate": 1.108344841265227e-06, "loss": 0.006835168227553368, "step": 5914 }, { "epoch": 1.5355659397715473, "grad_norm": 4.179077625274658, "learning_rate": 1.1076980596525422e-06, "loss": 0.011930400505661964, "step": 5915 }, { "epoch": 1.5358255451713396, "grad_norm": 2.377591371536255, "learning_rate": 1.1070515683612373e-06, "loss": 0.006081688683480024, "step": 5916 }, { "epoch": 1.536085150571132, "grad_norm": 7.316437721252441, "learning_rate": 1.1064053675055967e-06, "loss": 0.03009680286049843, "step": 5917 }, { "epoch": 1.5363447559709242, "grad_norm": 4.604000568389893, "learning_rate": 1.1057594571998528e-06, "loss": 0.010308925062417984, "step": 5918 }, { "epoch": 1.5366043613707165, "grad_norm": 3.036271095275879, "learning_rate": 1.10511383755819e-06, "loss": 0.017564557492733, "step": 5919 }, { "epoch": 1.5368639667705088, "grad_norm": 2.4127721786499023, "learning_rate": 1.1044685086947376e-06, "loss": 0.05637914314866066, "step": 5920 }, { "epoch": 1.5371235721703012, "grad_norm": 3.4202463626861572, "learning_rate": 1.1038234707235742e-06, "loss": 0.017040932551026344, "step": 5921 }, { "epoch": 1.5373831775700935, "grad_norm": 3.411496877670288, "learning_rate": 1.1031787237587296e-06, "loss": 0.027829796075820923, "step": 5922 }, { "epoch": 1.5376427829698858, "grad_norm": 3.7535531520843506, "learning_rate": 1.102534267914179e-06, "loss": 0.01206792052835226, "step": 5923 }, { "epoch": 1.537902388369678, "grad_norm": 3.372633457183838, "learning_rate": 1.101890103303849e-06, "loss": 0.01460660807788372, "step": 5924 }, { "epoch": 1.5381619937694704, "grad_norm": 2.001741886138916, "learning_rate": 1.1012462300416104e-06, "loss": 0.012362674809992313, "step": 5925 }, { "epoch": 1.5384215991692627, "grad_norm": 2.524467706680298, "learning_rate": 1.100602648241287e-06, "loss": 0.010977743193507195, "step": 5926 }, { "epoch": 1.538681204569055, "grad_norm": 2.6488823890686035, "learning_rate": 1.0999593580166501e-06, "loss": 0.05004788190126419, "step": 5927 }, { "epoch": 1.5389408099688473, "grad_norm": 2.611088514328003, "learning_rate": 1.099316359481418e-06, "loss": 0.01723584532737732, "step": 5928 }, { "epoch": 1.5392004153686396, "grad_norm": 3.1085867881774902, "learning_rate": 1.0986736527492573e-06, "loss": 0.06258764117956161, "step": 5929 }, { "epoch": 1.539460020768432, "grad_norm": 3.662999391555786, "learning_rate": 1.0980312379337844e-06, "loss": 0.03667810559272766, "step": 5930 }, { "epoch": 1.5397196261682242, "grad_norm": 2.7448859214782715, "learning_rate": 1.097389115148563e-06, "loss": 0.031821850687265396, "step": 5931 }, { "epoch": 1.5399792315680165, "grad_norm": 0.3315354883670807, "learning_rate": 1.0967472845071078e-06, "loss": 0.0009421156719326973, "step": 5932 }, { "epoch": 1.5402388369678088, "grad_norm": 2.8016300201416016, "learning_rate": 1.0961057461228768e-06, "loss": 0.02451515942811966, "step": 5933 }, { "epoch": 1.5404984423676011, "grad_norm": 2.5996570587158203, "learning_rate": 1.0954645001092818e-06, "loss": 0.013837850652635098, "step": 5934 }, { "epoch": 1.5407580477673934, "grad_norm": 2.8171420097351074, "learning_rate": 1.0948235465796784e-06, "loss": 0.027617264539003372, "step": 5935 }, { "epoch": 1.5410176531671858, "grad_norm": 3.87735915184021, "learning_rate": 1.0941828856473738e-06, "loss": 0.022192714735865593, "step": 5936 }, { "epoch": 1.541277258566978, "grad_norm": 2.991854190826416, "learning_rate": 1.0935425174256224e-06, "loss": 0.015820685774087906, "step": 5937 }, { "epoch": 1.5415368639667704, "grad_norm": 2.5757644176483154, "learning_rate": 1.092902442027625e-06, "loss": 0.010922185145318508, "step": 5938 }, { "epoch": 1.5417964693665627, "grad_norm": 3.155240535736084, "learning_rate": 1.0922626595665343e-06, "loss": 0.008987260982394218, "step": 5939 }, { "epoch": 1.542056074766355, "grad_norm": 3.164280652999878, "learning_rate": 1.091623170155448e-06, "loss": 0.04527817666530609, "step": 5940 }, { "epoch": 1.5423156801661475, "grad_norm": 1.6450284719467163, "learning_rate": 1.0909839739074126e-06, "loss": 0.004179656505584717, "step": 5941 }, { "epoch": 1.5425752855659398, "grad_norm": 3.2368180751800537, "learning_rate": 1.0903450709354253e-06, "loss": 0.017238102853298187, "step": 5942 }, { "epoch": 1.5428348909657321, "grad_norm": 3.337493419647217, "learning_rate": 1.0897064613524273e-06, "loss": 0.02122383750975132, "step": 5943 }, { "epoch": 1.5430944963655244, "grad_norm": 1.6766753196716309, "learning_rate": 1.0890681452713123e-06, "loss": 0.004506271798163652, "step": 5944 }, { "epoch": 1.5433541017653167, "grad_norm": 3.020246744155884, "learning_rate": 1.0884301228049187e-06, "loss": 0.028921151533722878, "step": 5945 }, { "epoch": 1.543613707165109, "grad_norm": 4.2360687255859375, "learning_rate": 1.0877923940660338e-06, "loss": 0.020240239799022675, "step": 5946 }, { "epoch": 1.5438733125649013, "grad_norm": 3.635758399963379, "learning_rate": 1.087154959167395e-06, "loss": 0.030933810397982597, "step": 5947 }, { "epoch": 1.5441329179646937, "grad_norm": 2.7983295917510986, "learning_rate": 1.0865178182216852e-06, "loss": 0.010695334523916245, "step": 5948 }, { "epoch": 1.544392523364486, "grad_norm": 3.9720585346221924, "learning_rate": 1.0858809713415355e-06, "loss": 0.016079703345894814, "step": 5949 }, { "epoch": 1.5446521287642783, "grad_norm": 2.8361332416534424, "learning_rate": 1.0852444186395279e-06, "loss": 0.012436673045158386, "step": 5950 }, { "epoch": 1.5449117341640706, "grad_norm": 3.103612184524536, "learning_rate": 1.0846081602281889e-06, "loss": 0.03052602894604206, "step": 5951 }, { "epoch": 1.5451713395638629, "grad_norm": 2.8676254749298096, "learning_rate": 1.0839721962199942e-06, "loss": 0.015998665243387222, "step": 5952 }, { "epoch": 1.5454309449636554, "grad_norm": 2.629794120788574, "learning_rate": 1.083336526727369e-06, "loss": 0.04777194932103157, "step": 5953 }, { "epoch": 1.5456905503634477, "grad_norm": 2.7320492267608643, "learning_rate": 1.0827011518626836e-06, "loss": 0.018596835434436798, "step": 5954 }, { "epoch": 1.54595015576324, "grad_norm": 2.4632041454315186, "learning_rate": 1.0820660717382598e-06, "loss": 0.013032680377364159, "step": 5955 }, { "epoch": 1.5462097611630323, "grad_norm": 3.601128578186035, "learning_rate": 1.0814312864663625e-06, "loss": 0.028453655540943146, "step": 5956 }, { "epoch": 1.5464693665628246, "grad_norm": 3.366516351699829, "learning_rate": 1.0807967961592086e-06, "loss": 0.0043546357192099094, "step": 5957 }, { "epoch": 1.546728971962617, "grad_norm": 3.1094658374786377, "learning_rate": 1.0801626009289621e-06, "loss": 0.07725153863430023, "step": 5958 }, { "epoch": 1.5469885773624092, "grad_norm": 2.5645062923431396, "learning_rate": 1.0795287008877334e-06, "loss": 0.012869220227003098, "step": 5959 }, { "epoch": 1.5472481827622016, "grad_norm": 3.440135955810547, "learning_rate": 1.0788950961475819e-06, "loss": 0.04217222332954407, "step": 5960 }, { "epoch": 1.5475077881619939, "grad_norm": 2.6049022674560547, "learning_rate": 1.0782617868205138e-06, "loss": 0.004205100238323212, "step": 5961 }, { "epoch": 1.5477673935617862, "grad_norm": 3.1136844158172607, "learning_rate": 1.0776287730184837e-06, "loss": 0.06229017674922943, "step": 5962 }, { "epoch": 1.5480269989615785, "grad_norm": 3.753157377243042, "learning_rate": 1.0769960548533955e-06, "loss": 0.013720321469008923, "step": 5963 }, { "epoch": 1.5482866043613708, "grad_norm": 3.8420331478118896, "learning_rate": 1.076363632437097e-06, "loss": 0.046847812831401825, "step": 5964 }, { "epoch": 1.548546209761163, "grad_norm": 3.438944101333618, "learning_rate": 1.0757315058813877e-06, "loss": 0.02513648197054863, "step": 5965 }, { "epoch": 1.5488058151609554, "grad_norm": 3.453723907470703, "learning_rate": 1.0750996752980117e-06, "loss": 0.04370569437742233, "step": 5966 }, { "epoch": 1.5490654205607477, "grad_norm": 2.445791482925415, "learning_rate": 1.0744681407986636e-06, "loss": 0.08770877122879028, "step": 5967 }, { "epoch": 1.54932502596054, "grad_norm": 3.060591697692871, "learning_rate": 1.073836902494984e-06, "loss": 0.03683602809906006, "step": 5968 }, { "epoch": 1.5495846313603323, "grad_norm": 2.956697702407837, "learning_rate": 1.0732059604985602e-06, "loss": 0.05267028510570526, "step": 5969 }, { "epoch": 1.5498442367601246, "grad_norm": 4.267246246337891, "learning_rate": 1.0725753149209295e-06, "loss": 0.037859439849853516, "step": 5970 }, { "epoch": 1.550103842159917, "grad_norm": 2.6936538219451904, "learning_rate": 1.0719449658735754e-06, "loss": 0.006630256772041321, "step": 5971 }, { "epoch": 1.5503634475597092, "grad_norm": 6.04428243637085, "learning_rate": 1.0713149134679281e-06, "loss": 0.03080240823328495, "step": 5972 }, { "epoch": 1.5506230529595015, "grad_norm": 1.9412204027175903, "learning_rate": 1.0706851578153681e-06, "loss": 0.008763429708778858, "step": 5973 }, { "epoch": 1.5508826583592938, "grad_norm": 2.7169971466064453, "learning_rate": 1.0700556990272203e-06, "loss": 0.05808103829622269, "step": 5974 }, { "epoch": 1.5511422637590861, "grad_norm": 2.060997724533081, "learning_rate": 1.06942653721476e-06, "loss": 0.006253133062273264, "step": 5975 }, { "epoch": 1.5514018691588785, "grad_norm": 2.8811938762664795, "learning_rate": 1.0687976724892078e-06, "loss": 0.0465351939201355, "step": 5976 }, { "epoch": 1.5516614745586708, "grad_norm": 1.3896867036819458, "learning_rate": 1.0681691049617322e-06, "loss": 0.004280788823962212, "step": 5977 }, { "epoch": 1.551921079958463, "grad_norm": 3.0337748527526855, "learning_rate": 1.0675408347434504e-06, "loss": 0.03564197197556496, "step": 5978 }, { "epoch": 1.5521806853582554, "grad_norm": 1.8002755641937256, "learning_rate": 1.066912861945426e-06, "loss": 0.01017211377620697, "step": 5979 }, { "epoch": 1.5524402907580477, "grad_norm": 4.595710277557373, "learning_rate": 1.0662851866786694e-06, "loss": 0.027670785784721375, "step": 5980 }, { "epoch": 1.55269989615784, "grad_norm": 3.6261119842529297, "learning_rate": 1.0656578090541402e-06, "loss": 0.03224901109933853, "step": 5981 }, { "epoch": 1.5529595015576323, "grad_norm": 2.481628656387329, "learning_rate": 1.0650307291827443e-06, "loss": 0.06267958879470825, "step": 5982 }, { "epoch": 1.5532191069574246, "grad_norm": 3.380887746810913, "learning_rate": 1.0644039471753343e-06, "loss": 0.03854160010814667, "step": 5983 }, { "epoch": 1.553478712357217, "grad_norm": 2.4621968269348145, "learning_rate": 1.0637774631427117e-06, "loss": 0.011702205054461956, "step": 5984 }, { "epoch": 1.5537383177570092, "grad_norm": 2.226130723953247, "learning_rate": 1.0631512771956238e-06, "loss": 0.0031416364945471287, "step": 5985 }, { "epoch": 1.5539979231568015, "grad_norm": 3.5413131713867188, "learning_rate": 1.0625253894447676e-06, "loss": 0.04876924306154251, "step": 5986 }, { "epoch": 1.5542575285565938, "grad_norm": 2.7720773220062256, "learning_rate": 1.0618998000007832e-06, "loss": 0.007385364267975092, "step": 5987 }, { "epoch": 1.5545171339563861, "grad_norm": 3.0115268230438232, "learning_rate": 1.061274508974262e-06, "loss": 0.009553496725857258, "step": 5988 }, { "epoch": 1.5547767393561787, "grad_norm": 4.020471572875977, "learning_rate": 1.060649516475741e-06, "loss": 0.047843724489212036, "step": 5989 }, { "epoch": 1.555036344755971, "grad_norm": 2.220259666442871, "learning_rate": 1.0600248226157051e-06, "loss": 0.010022949427366257, "step": 5990 }, { "epoch": 1.5552959501557633, "grad_norm": 2.463595151901245, "learning_rate": 1.0594004275045849e-06, "loss": 0.008850405924022198, "step": 5991 }, { "epoch": 1.5555555555555556, "grad_norm": 1.3639888763427734, "learning_rate": 1.058776331252759e-06, "loss": 0.007533038966357708, "step": 5992 }, { "epoch": 1.5558151609553479, "grad_norm": 2.3090291023254395, "learning_rate": 1.058152533970554e-06, "loss": 0.013953842222690582, "step": 5993 }, { "epoch": 1.5560747663551402, "grad_norm": 3.239000082015991, "learning_rate": 1.0575290357682439e-06, "loss": 0.03794758394360542, "step": 5994 }, { "epoch": 1.5563343717549325, "grad_norm": 2.5675208568573, "learning_rate": 1.0569058367560468e-06, "loss": 0.0099995918571949, "step": 5995 }, { "epoch": 1.5565939771547248, "grad_norm": 2.961040496826172, "learning_rate": 1.0562829370441304e-06, "loss": 0.019205529242753983, "step": 5996 }, { "epoch": 1.5568535825545171, "grad_norm": 1.3577089309692383, "learning_rate": 1.055660336742611e-06, "loss": 0.008206043392419815, "step": 5997 }, { "epoch": 1.5571131879543094, "grad_norm": 2.5828208923339844, "learning_rate": 1.0550380359615487e-06, "loss": 0.05465840548276901, "step": 5998 }, { "epoch": 1.5573727933541017, "grad_norm": 2.6199018955230713, "learning_rate": 1.054416034810952e-06, "loss": 0.004468370694667101, "step": 5999 }, { "epoch": 1.557632398753894, "grad_norm": 3.215808629989624, "learning_rate": 1.0537943334007762e-06, "loss": 0.01633552648127079, "step": 6000 }, { "epoch": 1.5578920041536866, "grad_norm": 1.8288925886154175, "learning_rate": 1.0531729318409242e-06, "loss": 0.003055922221392393, "step": 6001 }, { "epoch": 1.5581516095534789, "grad_norm": 3.5175299644470215, "learning_rate": 1.052551830241247e-06, "loss": 0.04778058081865311, "step": 6002 }, { "epoch": 1.5584112149532712, "grad_norm": 2.6426141262054443, "learning_rate": 1.0519310287115383e-06, "loss": 0.008870366960763931, "step": 6003 }, { "epoch": 1.5586708203530635, "grad_norm": 5.750606060028076, "learning_rate": 1.051310527361544e-06, "loss": 0.030521035194396973, "step": 6004 }, { "epoch": 1.5589304257528558, "grad_norm": 2.556896686553955, "learning_rate": 1.0506903263009536e-06, "loss": 0.01867247372865677, "step": 6005 }, { "epoch": 1.559190031152648, "grad_norm": 4.304748058319092, "learning_rate": 1.0500704256394037e-06, "loss": 0.045260921120643616, "step": 6006 }, { "epoch": 1.5594496365524404, "grad_norm": 2.5528764724731445, "learning_rate": 1.04945082548648e-06, "loss": 0.014151622541248798, "step": 6007 }, { "epoch": 1.5597092419522327, "grad_norm": 2.6800103187561035, "learning_rate": 1.0488315259517124e-06, "loss": 0.053508490324020386, "step": 6008 }, { "epoch": 1.559968847352025, "grad_norm": 2.230407238006592, "learning_rate": 1.0482125271445798e-06, "loss": 0.013061224482953548, "step": 6009 }, { "epoch": 1.5602284527518173, "grad_norm": 3.309535026550293, "learning_rate": 1.0475938291745069e-06, "loss": 0.01066275779157877, "step": 6010 }, { "epoch": 1.5604880581516096, "grad_norm": 3.5520527362823486, "learning_rate": 1.0469754321508641e-06, "loss": 0.034037720412015915, "step": 6011 }, { "epoch": 1.560747663551402, "grad_norm": 3.1545779705047607, "learning_rate": 1.046357336182972e-06, "loss": 0.0816165953874588, "step": 6012 }, { "epoch": 1.5610072689511942, "grad_norm": 3.566540002822876, "learning_rate": 1.045739541380094e-06, "loss": 0.019017957150936127, "step": 6013 }, { "epoch": 1.5612668743509865, "grad_norm": 2.9281857013702393, "learning_rate": 1.0451220478514424e-06, "loss": 0.026902638375759125, "step": 6014 }, { "epoch": 1.5615264797507789, "grad_norm": 3.845435857772827, "learning_rate": 1.044504855706177e-06, "loss": 0.02646062709391117, "step": 6015 }, { "epoch": 1.5617860851505712, "grad_norm": 2.062934637069702, "learning_rate": 1.0438879650534016e-06, "loss": 0.0032737404108047485, "step": 6016 }, { "epoch": 1.5620456905503635, "grad_norm": 2.1087069511413574, "learning_rate": 1.0432713760021696e-06, "loss": 0.00741351256147027, "step": 6017 }, { "epoch": 1.5623052959501558, "grad_norm": 3.6050310134887695, "learning_rate": 1.0426550886614797e-06, "loss": 0.025592884048819542, "step": 6018 }, { "epoch": 1.562564901349948, "grad_norm": 3.8430216312408447, "learning_rate": 1.0420391031402763e-06, "loss": 0.014034494757652283, "step": 6019 }, { "epoch": 1.5628245067497404, "grad_norm": 3.217162847518921, "learning_rate": 1.041423419547453e-06, "loss": 0.04221802204847336, "step": 6020 }, { "epoch": 1.5630841121495327, "grad_norm": 2.9058213233947754, "learning_rate": 1.0408080379918481e-06, "loss": 0.01951325312256813, "step": 6021 }, { "epoch": 1.563343717549325, "grad_norm": 5.792534828186035, "learning_rate": 1.0401929585822459e-06, "loss": 0.04806225001811981, "step": 6022 }, { "epoch": 1.5636033229491173, "grad_norm": 3.315592050552368, "learning_rate": 1.0395781814273801e-06, "loss": 0.01541566289961338, "step": 6023 }, { "epoch": 1.5638629283489096, "grad_norm": 2.8245601654052734, "learning_rate": 1.0389637066359278e-06, "loss": 0.022315748035907745, "step": 6024 }, { "epoch": 1.564122533748702, "grad_norm": 3.2631349563598633, "learning_rate": 1.0383495343165154e-06, "loss": 0.012453874573111534, "step": 6025 }, { "epoch": 1.5643821391484942, "grad_norm": 4.988672256469727, "learning_rate": 1.037735664577713e-06, "loss": 0.020884178578853607, "step": 6026 }, { "epoch": 1.5646417445482865, "grad_norm": 4.18328857421875, "learning_rate": 1.0371220975280393e-06, "loss": 0.023733224719762802, "step": 6027 }, { "epoch": 1.5649013499480788, "grad_norm": 2.0866341590881348, "learning_rate": 1.0365088332759596e-06, "loss": 0.005020732060074806, "step": 6028 }, { "epoch": 1.5651609553478711, "grad_norm": 1.187843918800354, "learning_rate": 1.0358958719298845e-06, "loss": 0.004733959212899208, "step": 6029 }, { "epoch": 1.5654205607476634, "grad_norm": 1.1651556491851807, "learning_rate": 1.0352832135981711e-06, "loss": 0.0025369226932525635, "step": 6030 }, { "epoch": 1.5656801661474558, "grad_norm": 3.194705009460449, "learning_rate": 1.0346708583891237e-06, "loss": 0.037076301872730255, "step": 6031 }, { "epoch": 1.565939771547248, "grad_norm": 3.2901663780212402, "learning_rate": 1.034058806410993e-06, "loss": 0.0487261638045311, "step": 6032 }, { "epoch": 1.5661993769470404, "grad_norm": 3.3472490310668945, "learning_rate": 1.0334470577719755e-06, "loss": 0.01795840635895729, "step": 6033 }, { "epoch": 1.5664589823468327, "grad_norm": 2.7227914333343506, "learning_rate": 1.0328356125802136e-06, "loss": 0.05771200358867645, "step": 6034 }, { "epoch": 1.566718587746625, "grad_norm": 3.3750627040863037, "learning_rate": 1.0322244709437983e-06, "loss": 0.02964632771909237, "step": 6035 }, { "epoch": 1.5669781931464173, "grad_norm": 3.105283498764038, "learning_rate": 1.0316136329707649e-06, "loss": 0.017620190978050232, "step": 6036 }, { "epoch": 1.5672377985462098, "grad_norm": 3.271643877029419, "learning_rate": 1.0310030987690945e-06, "loss": 0.015950441360473633, "step": 6037 }, { "epoch": 1.5674974039460021, "grad_norm": 3.7980310916900635, "learning_rate": 1.030392868446717e-06, "loss": 0.02591252140700817, "step": 6038 }, { "epoch": 1.5677570093457944, "grad_norm": 2.9307236671447754, "learning_rate": 1.029782942111506e-06, "loss": 0.020546428859233856, "step": 6039 }, { "epoch": 1.5680166147455867, "grad_norm": 3.653198480606079, "learning_rate": 1.0291733198712838e-06, "loss": 0.02804916352033615, "step": 6040 }, { "epoch": 1.568276220145379, "grad_norm": 3.479595184326172, "learning_rate": 1.0285640018338166e-06, "loss": 0.03946448490023613, "step": 6041 }, { "epoch": 1.5685358255451713, "grad_norm": 2.896660566329956, "learning_rate": 1.027954988106818e-06, "loss": 0.006744787096977234, "step": 6042 }, { "epoch": 1.5687954309449637, "grad_norm": 3.280442714691162, "learning_rate": 1.0273462787979485e-06, "loss": 0.029451711103320122, "step": 6043 }, { "epoch": 1.569055036344756, "grad_norm": 3.379154920578003, "learning_rate": 1.0267378740148131e-06, "loss": 0.05830035358667374, "step": 6044 }, { "epoch": 1.5693146417445483, "grad_norm": 2.379673480987549, "learning_rate": 1.0261297738649637e-06, "loss": 0.011215245351195335, "step": 6045 }, { "epoch": 1.5695742471443406, "grad_norm": 3.769296407699585, "learning_rate": 1.0255219784558992e-06, "loss": 0.02834782376885414, "step": 6046 }, { "epoch": 1.5698338525441329, "grad_norm": 1.9880452156066895, "learning_rate": 1.0249144878950634e-06, "loss": 0.008802676573395729, "step": 6047 }, { "epoch": 1.5700934579439252, "grad_norm": 2.317026376724243, "learning_rate": 1.0243073022898472e-06, "loss": 0.009630342945456505, "step": 6048 }, { "epoch": 1.5703530633437177, "grad_norm": 3.4796464443206787, "learning_rate": 1.023700421747587e-06, "loss": 0.02168232575058937, "step": 6049 }, { "epoch": 1.57061266874351, "grad_norm": 2.623222827911377, "learning_rate": 1.0230938463755648e-06, "loss": 0.016206704080104828, "step": 6050 }, { "epoch": 1.5708722741433023, "grad_norm": 2.4216573238372803, "learning_rate": 1.0224875762810102e-06, "loss": 0.07093057781457901, "step": 6051 }, { "epoch": 1.5711318795430946, "grad_norm": 4.08283805847168, "learning_rate": 1.021881611571097e-06, "loss": 0.019354883581399918, "step": 6052 }, { "epoch": 1.571391484942887, "grad_norm": 3.6863906383514404, "learning_rate": 1.0212759523529456e-06, "loss": 0.033472225069999695, "step": 6053 }, { "epoch": 1.5716510903426792, "grad_norm": 3.449352741241455, "learning_rate": 1.020670598733624e-06, "loss": 0.056511420756578445, "step": 6054 }, { "epoch": 1.5719106957424716, "grad_norm": 2.113610029220581, "learning_rate": 1.0200655508201438e-06, "loss": 0.010506216436624527, "step": 6055 }, { "epoch": 1.5721703011422639, "grad_norm": 3.4443869590759277, "learning_rate": 1.019460808719465e-06, "loss": 0.05186551809310913, "step": 6056 }, { "epoch": 1.5724299065420562, "grad_norm": 3.22416353225708, "learning_rate": 1.0188563725384898e-06, "loss": 0.05768274515867233, "step": 6057 }, { "epoch": 1.5726895119418485, "grad_norm": 3.4916698932647705, "learning_rate": 1.01825224238407e-06, "loss": 0.03303146734833717, "step": 6058 }, { "epoch": 1.5729491173416408, "grad_norm": 2.516951560974121, "learning_rate": 1.0176484183630035e-06, "loss": 0.015669889748096466, "step": 6059 }, { "epoch": 1.573208722741433, "grad_norm": 2.544724941253662, "learning_rate": 1.0170449005820296e-06, "loss": 0.06131654232740402, "step": 6060 }, { "epoch": 1.5734683281412254, "grad_norm": 2.8742713928222656, "learning_rate": 1.0164416891478382e-06, "loss": 0.003119127359241247, "step": 6061 }, { "epoch": 1.5737279335410177, "grad_norm": 2.8629982471466064, "learning_rate": 1.0158387841670626e-06, "loss": 0.009582705795764923, "step": 6062 }, { "epoch": 1.57398753894081, "grad_norm": 1.8954085111618042, "learning_rate": 1.0152361857462836e-06, "loss": 0.010388179682195187, "step": 6063 }, { "epoch": 1.5742471443406023, "grad_norm": 3.4898860454559326, "learning_rate": 1.0146338939920262e-06, "loss": 0.026857294142246246, "step": 6064 }, { "epoch": 1.5745067497403946, "grad_norm": 4.275811195373535, "learning_rate": 1.014031909010761e-06, "loss": 0.02324884757399559, "step": 6065 }, { "epoch": 1.574766355140187, "grad_norm": 2.845806837081909, "learning_rate": 1.0134302309089065e-06, "loss": 0.03543885052204132, "step": 6066 }, { "epoch": 1.5750259605399792, "grad_norm": 3.0997753143310547, "learning_rate": 1.0128288597928252e-06, "loss": 0.024812277406454086, "step": 6067 }, { "epoch": 1.5752855659397715, "grad_norm": 3.6282761096954346, "learning_rate": 1.0122277957688244e-06, "loss": 0.016589421778917313, "step": 6068 }, { "epoch": 1.5755451713395638, "grad_norm": 2.621419668197632, "learning_rate": 1.0116270389431606e-06, "loss": 0.015888681635260582, "step": 6069 }, { "epoch": 1.5758047767393562, "grad_norm": 2.9376919269561768, "learning_rate": 1.0110265894220322e-06, "loss": 0.009184146299958229, "step": 6070 }, { "epoch": 1.5760643821391485, "grad_norm": 3.512225866317749, "learning_rate": 1.010426447311586e-06, "loss": 0.030322639271616936, "step": 6071 }, { "epoch": 1.5763239875389408, "grad_norm": 3.7092933654785156, "learning_rate": 1.0098266127179132e-06, "loss": 0.04086216166615486, "step": 6072 }, { "epoch": 1.576583592938733, "grad_norm": 2.623798131942749, "learning_rate": 1.00922708574705e-06, "loss": 0.01040917169302702, "step": 6073 }, { "epoch": 1.5768431983385254, "grad_norm": 3.223749876022339, "learning_rate": 1.0086278665049801e-06, "loss": 0.02709377184510231, "step": 6074 }, { "epoch": 1.5771028037383177, "grad_norm": 1.9332739114761353, "learning_rate": 1.0080289550976314e-06, "loss": 0.011402611620724201, "step": 6075 }, { "epoch": 1.57736240913811, "grad_norm": 2.950424909591675, "learning_rate": 1.0074303516308767e-06, "loss": 0.02946452796459198, "step": 6076 }, { "epoch": 1.5776220145379023, "grad_norm": 4.080645561218262, "learning_rate": 1.0068320562105369e-06, "loss": 0.04161008447408676, "step": 6077 }, { "epoch": 1.5778816199376946, "grad_norm": 3.0402414798736572, "learning_rate": 1.0062340689423762e-06, "loss": 0.00826173834502697, "step": 6078 }, { "epoch": 1.578141225337487, "grad_norm": 2.9152655601501465, "learning_rate": 1.0056363899321053e-06, "loss": 0.02694152668118477, "step": 6079 }, { "epoch": 1.5784008307372792, "grad_norm": 2.6714022159576416, "learning_rate": 1.0050390192853801e-06, "loss": 0.015560517087578773, "step": 6080 }, { "epoch": 1.5786604361370715, "grad_norm": 3.1887149810791016, "learning_rate": 1.0044419571078016e-06, "loss": 0.010191037319600582, "step": 6081 }, { "epoch": 1.5789200415368638, "grad_norm": 3.2625458240509033, "learning_rate": 1.0038452035049174e-06, "loss": 0.04398763179779053, "step": 6082 }, { "epoch": 1.5791796469366561, "grad_norm": 3.9084861278533936, "learning_rate": 1.0032487585822196e-06, "loss": 0.03183791786432266, "step": 6083 }, { "epoch": 1.5794392523364484, "grad_norm": 3.195626735687256, "learning_rate": 1.0026526224451455e-06, "loss": 0.011575297452509403, "step": 6084 }, { "epoch": 1.579698857736241, "grad_norm": 3.2028424739837646, "learning_rate": 1.0020567951990794e-06, "loss": 0.033326465636491776, "step": 6085 }, { "epoch": 1.5799584631360333, "grad_norm": 2.5487635135650635, "learning_rate": 1.0014612769493494e-06, "loss": 0.025050688534975052, "step": 6086 }, { "epoch": 1.5802180685358256, "grad_norm": 3.6733429431915283, "learning_rate": 1.000866067801229e-06, "loss": 0.016873840242624283, "step": 6087 }, { "epoch": 1.5804776739356179, "grad_norm": 2.394473075866699, "learning_rate": 1.0002711678599378e-06, "loss": 0.012501416727900505, "step": 6088 }, { "epoch": 1.5807372793354102, "grad_norm": 3.449936628341675, "learning_rate": 9.996765772306405e-07, "loss": 0.03254853934049606, "step": 6089 }, { "epoch": 1.5809968847352025, "grad_norm": 3.6401169300079346, "learning_rate": 9.990822960184488e-07, "loss": 0.011749513447284698, "step": 6090 }, { "epoch": 1.5812564901349948, "grad_norm": 2.5514729022979736, "learning_rate": 9.984883243284146e-07, "loss": 0.01423007994890213, "step": 6091 }, { "epoch": 1.5815160955347871, "grad_norm": 2.363478899002075, "learning_rate": 9.978946622655414e-07, "loss": 0.013342399150133133, "step": 6092 }, { "epoch": 1.5817757009345794, "grad_norm": 4.201878547668457, "learning_rate": 9.973013099347734e-07, "loss": 0.018962083384394646, "step": 6093 }, { "epoch": 1.5820353063343717, "grad_norm": 3.5073318481445312, "learning_rate": 9.967082674410028e-07, "loss": 0.031973592936992645, "step": 6094 }, { "epoch": 1.582294911734164, "grad_norm": 2.906618595123291, "learning_rate": 9.961155348890654e-07, "loss": 0.03085550107061863, "step": 6095 }, { "epoch": 1.5825545171339563, "grad_norm": 3.797339916229248, "learning_rate": 9.955231123837422e-07, "loss": 0.028221171349287033, "step": 6096 }, { "epoch": 1.5828141225337489, "grad_norm": 3.0179007053375244, "learning_rate": 9.94931000029761e-07, "loss": 0.03502753749489784, "step": 6097 }, { "epoch": 1.5830737279335412, "grad_norm": 2.9573159217834473, "learning_rate": 9.943391979317938e-07, "loss": 0.04319887235760689, "step": 6098 }, { "epoch": 1.5833333333333335, "grad_norm": 3.2662816047668457, "learning_rate": 9.93747706194456e-07, "loss": 0.010220288299024105, "step": 6099 }, { "epoch": 1.5835929387331258, "grad_norm": 2.8703689575195312, "learning_rate": 9.931565249223116e-07, "loss": 0.015798350796103477, "step": 6100 }, { "epoch": 1.583852544132918, "grad_norm": 3.4807207584381104, "learning_rate": 9.925656542198666e-07, "loss": 0.023348195478320122, "step": 6101 }, { "epoch": 1.5841121495327104, "grad_norm": 3.4962947368621826, "learning_rate": 9.919750941915748e-07, "loss": 0.03959228843450546, "step": 6102 }, { "epoch": 1.5843717549325027, "grad_norm": 3.4997990131378174, "learning_rate": 9.91384844941833e-07, "loss": 0.03350568562746048, "step": 6103 }, { "epoch": 1.584631360332295, "grad_norm": 1.7487963438034058, "learning_rate": 9.907949065749827e-07, "loss": 0.009753216058015823, "step": 6104 }, { "epoch": 1.5848909657320873, "grad_norm": 2.7095212936401367, "learning_rate": 9.902052791953131e-07, "loss": 0.031912922859191895, "step": 6105 }, { "epoch": 1.5851505711318796, "grad_norm": 3.0062754154205322, "learning_rate": 9.896159629070564e-07, "loss": 0.02004610374569893, "step": 6106 }, { "epoch": 1.585410176531672, "grad_norm": 2.920525312423706, "learning_rate": 9.890269578143894e-07, "loss": 0.05501376837491989, "step": 6107 }, { "epoch": 1.5856697819314642, "grad_norm": 2.9940052032470703, "learning_rate": 9.884382640214357e-07, "loss": 0.0314737930893898, "step": 6108 }, { "epoch": 1.5859293873312565, "grad_norm": 4.342431545257568, "learning_rate": 9.878498816322621e-07, "loss": 0.025952458381652832, "step": 6109 }, { "epoch": 1.5861889927310489, "grad_norm": 2.973430871963501, "learning_rate": 9.87261810750882e-07, "loss": 0.026779841631650925, "step": 6110 }, { "epoch": 1.5864485981308412, "grad_norm": 3.130018711090088, "learning_rate": 9.866740514812524e-07, "loss": 0.041291311383247375, "step": 6111 }, { "epoch": 1.5867082035306335, "grad_norm": 2.5331296920776367, "learning_rate": 9.86086603927275e-07, "loss": 0.006378462538123131, "step": 6112 }, { "epoch": 1.5869678089304258, "grad_norm": 3.067086935043335, "learning_rate": 9.854994681927992e-07, "loss": 0.007213269360363483, "step": 6113 }, { "epoch": 1.587227414330218, "grad_norm": 4.992260456085205, "learning_rate": 9.849126443816146e-07, "loss": 0.01241954043507576, "step": 6114 }, { "epoch": 1.5874870197300104, "grad_norm": 1.078452706336975, "learning_rate": 9.843261325974591e-07, "loss": 0.006132975220680237, "step": 6115 }, { "epoch": 1.5877466251298027, "grad_norm": 3.5178275108337402, "learning_rate": 9.837399329440155e-07, "loss": 0.0675501897931099, "step": 6116 }, { "epoch": 1.588006230529595, "grad_norm": 2.8373124599456787, "learning_rate": 9.831540455249097e-07, "loss": 0.056032367050647736, "step": 6117 }, { "epoch": 1.5882658359293873, "grad_norm": 2.1108086109161377, "learning_rate": 9.825684704437132e-07, "loss": 0.006789142265915871, "step": 6118 }, { "epoch": 1.5885254413291796, "grad_norm": 3.6741907596588135, "learning_rate": 9.819832078039419e-07, "loss": 0.06849038600921631, "step": 6119 }, { "epoch": 1.588785046728972, "grad_norm": 4.411643981933594, "learning_rate": 9.813982577090572e-07, "loss": 0.031538113951683044, "step": 6120 }, { "epoch": 1.5890446521287642, "grad_norm": 4.814372539520264, "learning_rate": 9.808136202624657e-07, "loss": 0.04158644378185272, "step": 6121 }, { "epoch": 1.5893042575285565, "grad_norm": 2.729505777359009, "learning_rate": 9.802292955675163e-07, "loss": 0.0094600236043334, "step": 6122 }, { "epoch": 1.5895638629283488, "grad_norm": 4.386035919189453, "learning_rate": 9.796452837275055e-07, "loss": 0.013875331729650497, "step": 6123 }, { "epoch": 1.5898234683281411, "grad_norm": 3.4915964603424072, "learning_rate": 9.790615848456725e-07, "loss": 0.045722462236881256, "step": 6124 }, { "epoch": 1.5900830737279334, "grad_norm": 3.5467965602874756, "learning_rate": 9.784781990252027e-07, "loss": 0.029604662209749222, "step": 6125 }, { "epoch": 1.5903426791277258, "grad_norm": 2.840479850769043, "learning_rate": 9.778951263692247e-07, "loss": 0.049474913626909256, "step": 6126 }, { "epoch": 1.590602284527518, "grad_norm": 2.598802089691162, "learning_rate": 9.773123669808118e-07, "loss": 0.01466023176908493, "step": 6127 }, { "epoch": 1.5908618899273104, "grad_norm": 3.21868634223938, "learning_rate": 9.767299209629841e-07, "loss": 0.05147789418697357, "step": 6128 }, { "epoch": 1.5911214953271027, "grad_norm": 1.7100703716278076, "learning_rate": 9.761477884187038e-07, "loss": 0.007700557820498943, "step": 6129 }, { "epoch": 1.591381100726895, "grad_norm": 4.2346014976501465, "learning_rate": 9.755659694508781e-07, "loss": 0.02938908524811268, "step": 6130 }, { "epoch": 1.5916407061266873, "grad_norm": 2.6797292232513428, "learning_rate": 9.749844641623605e-07, "loss": 0.01667741686105728, "step": 6131 }, { "epoch": 1.5919003115264796, "grad_norm": 2.711970090866089, "learning_rate": 9.744032726559465e-07, "loss": 0.047023653984069824, "step": 6132 }, { "epoch": 1.592159916926272, "grad_norm": 2.4399118423461914, "learning_rate": 9.738223950343789e-07, "loss": 0.006753067951649427, "step": 6133 }, { "epoch": 1.5924195223260644, "grad_norm": 3.2829954624176025, "learning_rate": 9.732418314003422e-07, "loss": 0.034496910870075226, "step": 6134 }, { "epoch": 1.5926791277258567, "grad_norm": 3.423786163330078, "learning_rate": 9.726615818564675e-07, "loss": 0.021308250725269318, "step": 6135 }, { "epoch": 1.592938733125649, "grad_norm": 0.33917051553726196, "learning_rate": 9.720816465053296e-07, "loss": 0.0009150725672952831, "step": 6136 }, { "epoch": 1.5931983385254413, "grad_norm": 3.277787685394287, "learning_rate": 9.715020254494475e-07, "loss": 0.020868228748440742, "step": 6137 }, { "epoch": 1.5934579439252337, "grad_norm": 3.5249621868133545, "learning_rate": 9.709227187912852e-07, "loss": 0.060721028596162796, "step": 6138 }, { "epoch": 1.593717549325026, "grad_norm": 3.3954949378967285, "learning_rate": 9.703437266332506e-07, "loss": 0.023404870182275772, "step": 6139 }, { "epoch": 1.5939771547248183, "grad_norm": 2.8597817420959473, "learning_rate": 9.697650490776968e-07, "loss": 0.05430147796869278, "step": 6140 }, { "epoch": 1.5942367601246106, "grad_norm": 2.513709306716919, "learning_rate": 9.691866862269197e-07, "loss": 0.03558141365647316, "step": 6141 }, { "epoch": 1.5944963655244029, "grad_norm": 1.8638453483581543, "learning_rate": 9.686086381831617e-07, "loss": 0.003033180721104145, "step": 6142 }, { "epoch": 1.5947559709241952, "grad_norm": 2.98964786529541, "learning_rate": 9.680309050486075e-07, "loss": 0.020236745476722717, "step": 6143 }, { "epoch": 1.5950155763239875, "grad_norm": 4.106471538543701, "learning_rate": 9.674534869253887e-07, "loss": 0.014724714681506157, "step": 6144 }, { "epoch": 1.5952751817237798, "grad_norm": 6.439882755279541, "learning_rate": 9.668763839155776e-07, "loss": 0.009732067584991455, "step": 6145 }, { "epoch": 1.5955347871235723, "grad_norm": 3.6945948600769043, "learning_rate": 9.662995961211935e-07, "loss": 0.023904025554656982, "step": 6146 }, { "epoch": 1.5957943925233646, "grad_norm": 1.9865097999572754, "learning_rate": 9.657231236442002e-07, "loss": 0.008635756559669971, "step": 6147 }, { "epoch": 1.596053997923157, "grad_norm": 3.6579415798187256, "learning_rate": 9.65146966586504e-07, "loss": 0.040000881999731064, "step": 6148 }, { "epoch": 1.5963136033229492, "grad_norm": 4.744409561157227, "learning_rate": 9.645711250499565e-07, "loss": 0.03286496549844742, "step": 6149 }, { "epoch": 1.5965732087227416, "grad_norm": 2.969107151031494, "learning_rate": 9.639955991363526e-07, "loss": 0.06531436741352081, "step": 6150 }, { "epoch": 1.5968328141225339, "grad_norm": 3.1150779724121094, "learning_rate": 9.634203889474324e-07, "loss": 0.043566010892391205, "step": 6151 }, { "epoch": 1.5970924195223262, "grad_norm": 2.7097208499908447, "learning_rate": 9.628454945848815e-07, "loss": 0.06402347981929779, "step": 6152 }, { "epoch": 1.5973520249221185, "grad_norm": 1.7122118473052979, "learning_rate": 9.622709161503257e-07, "loss": 0.01174967736005783, "step": 6153 }, { "epoch": 1.5976116303219108, "grad_norm": 3.4722299575805664, "learning_rate": 9.616966537453389e-07, "loss": 0.019518544897437096, "step": 6154 }, { "epoch": 1.597871235721703, "grad_norm": 3.3832573890686035, "learning_rate": 9.611227074714364e-07, "loss": 0.013232676312327385, "step": 6155 }, { "epoch": 1.5981308411214954, "grad_norm": 2.9614808559417725, "learning_rate": 9.6054907743008e-07, "loss": 0.047922633588314056, "step": 6156 }, { "epoch": 1.5983904465212877, "grad_norm": 1.4786707162857056, "learning_rate": 9.599757637226739e-07, "loss": 0.007860848680138588, "step": 6157 }, { "epoch": 1.59865005192108, "grad_norm": 3.102283239364624, "learning_rate": 9.594027664505656e-07, "loss": 0.04033895209431648, "step": 6158 }, { "epoch": 1.5989096573208723, "grad_norm": 0.3076881766319275, "learning_rate": 9.588300857150498e-07, "loss": 0.0011547085596248507, "step": 6159 }, { "epoch": 1.5991692627206646, "grad_norm": 3.2672650814056396, "learning_rate": 9.582577216173626e-07, "loss": 0.03365282341837883, "step": 6160 }, { "epoch": 1.599428868120457, "grad_norm": 2.580000877380371, "learning_rate": 9.576856742586837e-07, "loss": 0.010606668889522552, "step": 6161 }, { "epoch": 1.5996884735202492, "grad_norm": 0.16511359810829163, "learning_rate": 9.5711394374014e-07, "loss": 0.0006630689604207873, "step": 6162 }, { "epoch": 1.5999480789200415, "grad_norm": 2.951035737991333, "learning_rate": 9.565425301627987e-07, "loss": 0.022604361176490784, "step": 6163 }, { "epoch": 1.6002076843198338, "grad_norm": 3.064467430114746, "learning_rate": 9.55971433627674e-07, "loss": 0.0647471621632576, "step": 6164 }, { "epoch": 1.6004672897196262, "grad_norm": 3.4106855392456055, "learning_rate": 9.554006542357219e-07, "loss": 0.02752789296209812, "step": 6165 }, { "epoch": 1.6007268951194185, "grad_norm": 3.2884371280670166, "learning_rate": 9.548301920878428e-07, "loss": 0.01157587580382824, "step": 6166 }, { "epoch": 1.6009865005192108, "grad_norm": 4.096403121948242, "learning_rate": 9.542600472848825e-07, "loss": 0.012865867465734482, "step": 6167 }, { "epoch": 1.601246105919003, "grad_norm": 3.2804434299468994, "learning_rate": 9.536902199276284e-07, "loss": 0.04957238957285881, "step": 6168 }, { "epoch": 1.6015057113187954, "grad_norm": 3.2144761085510254, "learning_rate": 9.531207101168132e-07, "loss": 0.018266044557094574, "step": 6169 }, { "epoch": 1.6017653167185877, "grad_norm": 3.141803503036499, "learning_rate": 9.525515179531139e-07, "loss": 0.04258955270051956, "step": 6170 }, { "epoch": 1.60202492211838, "grad_norm": 3.4582624435424805, "learning_rate": 9.519826435371502e-07, "loss": 0.023374520242214203, "step": 6171 }, { "epoch": 1.6022845275181723, "grad_norm": 4.5663371086120605, "learning_rate": 9.514140869694852e-07, "loss": 0.020046064630150795, "step": 6172 }, { "epoch": 1.6025441329179646, "grad_norm": 1.554040551185608, "learning_rate": 9.508458483506281e-07, "loss": 0.003770199604332447, "step": 6173 }, { "epoch": 1.602803738317757, "grad_norm": 2.6044349670410156, "learning_rate": 9.502779277810292e-07, "loss": 0.013924801722168922, "step": 6174 }, { "epoch": 1.6030633437175492, "grad_norm": 1.8987237215042114, "learning_rate": 9.497103253610857e-07, "loss": 0.002564463298767805, "step": 6175 }, { "epoch": 1.6033229491173415, "grad_norm": 4.072655200958252, "learning_rate": 9.491430411911343e-07, "loss": 0.013372451066970825, "step": 6176 }, { "epoch": 1.6035825545171338, "grad_norm": 4.033962726593018, "learning_rate": 9.485760753714592e-07, "loss": 0.01780734956264496, "step": 6177 }, { "epoch": 1.6038421599169261, "grad_norm": 3.198970079421997, "learning_rate": 9.480094280022873e-07, "loss": 0.05272812396287918, "step": 6178 }, { "epoch": 1.6041017653167184, "grad_norm": 4.110971450805664, "learning_rate": 9.474430991837886e-07, "loss": 0.022348960861563683, "step": 6179 }, { "epoch": 1.6043613707165107, "grad_norm": 2.4352073669433594, "learning_rate": 9.468770890160767e-07, "loss": 0.008516788482666016, "step": 6180 }, { "epoch": 1.604620976116303, "grad_norm": 2.5047500133514404, "learning_rate": 9.463113975992086e-07, "loss": 0.03473004698753357, "step": 6181 }, { "epoch": 1.6048805815160956, "grad_norm": 3.511573076248169, "learning_rate": 9.457460250331868e-07, "loss": 0.008293426595628262, "step": 6182 }, { "epoch": 1.605140186915888, "grad_norm": 3.35414457321167, "learning_rate": 9.451809714179569e-07, "loss": 0.028327010571956635, "step": 6183 }, { "epoch": 1.6053997923156802, "grad_norm": 3.5216376781463623, "learning_rate": 9.446162368534051e-07, "loss": 0.030627399682998657, "step": 6184 }, { "epoch": 1.6056593977154725, "grad_norm": 3.352966070175171, "learning_rate": 9.440518214393657e-07, "loss": 0.04456070810556412, "step": 6185 }, { "epoch": 1.6059190031152648, "grad_norm": 4.235993385314941, "learning_rate": 9.434877252756125e-07, "loss": 0.02399051934480667, "step": 6186 }, { "epoch": 1.6061786085150571, "grad_norm": 3.689237356185913, "learning_rate": 9.429239484618669e-07, "loss": 0.020932668820023537, "step": 6187 }, { "epoch": 1.6064382139148494, "grad_norm": 3.747953414916992, "learning_rate": 9.423604910977904e-07, "loss": 0.028027527034282684, "step": 6188 }, { "epoch": 1.6066978193146417, "grad_norm": 3.52854585647583, "learning_rate": 9.41797353282989e-07, "loss": 0.036120131611824036, "step": 6189 }, { "epoch": 1.606957424714434, "grad_norm": 3.101658821105957, "learning_rate": 9.412345351170138e-07, "loss": 0.007800300605595112, "step": 6190 }, { "epoch": 1.6072170301142263, "grad_norm": 3.3265459537506104, "learning_rate": 9.406720366993575e-07, "loss": 0.005006606690585613, "step": 6191 }, { "epoch": 1.6074766355140186, "grad_norm": 3.1906399726867676, "learning_rate": 9.401098581294563e-07, "loss": 0.04364350810647011, "step": 6192 }, { "epoch": 1.607736240913811, "grad_norm": 3.5256781578063965, "learning_rate": 9.395479995066919e-07, "loss": 0.03330131247639656, "step": 6193 }, { "epoch": 1.6079958463136035, "grad_norm": 2.779489278793335, "learning_rate": 9.38986460930387e-07, "loss": 0.010108785703778267, "step": 6194 }, { "epoch": 1.6082554517133958, "grad_norm": 2.6837661266326904, "learning_rate": 9.384252424998094e-07, "loss": 0.048652976751327515, "step": 6195 }, { "epoch": 1.608515057113188, "grad_norm": 3.006563901901245, "learning_rate": 9.378643443141694e-07, "loss": 0.04660452902317047, "step": 6196 }, { "epoch": 1.6087746625129804, "grad_norm": 4.27303409576416, "learning_rate": 9.373037664726204e-07, "loss": 0.045731622725725174, "step": 6197 }, { "epoch": 1.6090342679127727, "grad_norm": 1.6190681457519531, "learning_rate": 9.367435090742608e-07, "loss": 0.0027612787671387196, "step": 6198 }, { "epoch": 1.609293873312565, "grad_norm": 2.7739996910095215, "learning_rate": 9.361835722181306e-07, "loss": 0.006829892750829458, "step": 6199 }, { "epoch": 1.6095534787123573, "grad_norm": 3.9914631843566895, "learning_rate": 9.356239560032135e-07, "loss": 0.023225314915180206, "step": 6200 }, { "epoch": 1.6098130841121496, "grad_norm": 2.2660019397735596, "learning_rate": 9.350646605284378e-07, "loss": 0.009583394974470139, "step": 6201 }, { "epoch": 1.610072689511942, "grad_norm": 2.9209604263305664, "learning_rate": 9.345056858926735e-07, "loss": 0.057963401079177856, "step": 6202 }, { "epoch": 1.6103322949117342, "grad_norm": 0.40014517307281494, "learning_rate": 9.339470321947341e-07, "loss": 0.00092344667064026, "step": 6203 }, { "epoch": 1.6105919003115265, "grad_norm": 2.4510629177093506, "learning_rate": 9.333886995333782e-07, "loss": 0.01651078835129738, "step": 6204 }, { "epoch": 1.6108515057113189, "grad_norm": 2.983813524246216, "learning_rate": 9.328306880073046e-07, "loss": 0.008586673066020012, "step": 6205 }, { "epoch": 1.6111111111111112, "grad_norm": 2.5668234825134277, "learning_rate": 9.322729977151587e-07, "loss": 0.023056602105498314, "step": 6206 }, { "epoch": 1.6113707165109035, "grad_norm": 2.3817849159240723, "learning_rate": 9.317156287555253e-07, "loss": 0.03580964356660843, "step": 6207 }, { "epoch": 1.6116303219106958, "grad_norm": 2.7694108486175537, "learning_rate": 9.311585812269356e-07, "loss": 0.020090855658054352, "step": 6208 }, { "epoch": 1.611889927310488, "grad_norm": 3.1682002544403076, "learning_rate": 9.306018552278631e-07, "loss": 0.010538499802350998, "step": 6209 }, { "epoch": 1.6121495327102804, "grad_norm": 3.121926784515381, "learning_rate": 9.300454508567242e-07, "loss": 0.04568599909543991, "step": 6210 }, { "epoch": 1.6124091381100727, "grad_norm": 2.6744377613067627, "learning_rate": 9.294893682118783e-07, "loss": 0.006390575785189867, "step": 6211 }, { "epoch": 1.612668743509865, "grad_norm": 0.9280980825424194, "learning_rate": 9.28933607391627e-07, "loss": 0.001186567242257297, "step": 6212 }, { "epoch": 1.6129283489096573, "grad_norm": 2.302028179168701, "learning_rate": 9.28378168494217e-07, "loss": 0.009248963557183743, "step": 6213 }, { "epoch": 1.6131879543094496, "grad_norm": 1.651628017425537, "learning_rate": 9.278230516178385e-07, "loss": 0.004169857129454613, "step": 6214 }, { "epoch": 1.613447559709242, "grad_norm": 2.9813406467437744, "learning_rate": 9.272682568606209e-07, "loss": 0.042229291051626205, "step": 6215 }, { "epoch": 1.6137071651090342, "grad_norm": 1.6270337104797363, "learning_rate": 9.267137843206411e-07, "loss": 0.0033367532305419445, "step": 6216 }, { "epoch": 1.6139667705088265, "grad_norm": 3.1471176147460938, "learning_rate": 9.261596340959161e-07, "loss": 0.06083158403635025, "step": 6217 }, { "epoch": 1.6142263759086188, "grad_norm": 3.4804041385650635, "learning_rate": 9.256058062844078e-07, "loss": 0.013958794996142387, "step": 6218 }, { "epoch": 1.6144859813084111, "grad_norm": 3.406001567840576, "learning_rate": 9.250523009840201e-07, "loss": 0.02247702330350876, "step": 6219 }, { "epoch": 1.6147455867082035, "grad_norm": 3.314631700515747, "learning_rate": 9.244991182925987e-07, "loss": 0.013049431145191193, "step": 6220 }, { "epoch": 1.6150051921079958, "grad_norm": 3.627424478530884, "learning_rate": 9.239462583079359e-07, "loss": 0.036884963512420654, "step": 6221 }, { "epoch": 1.615264797507788, "grad_norm": 3.185347318649292, "learning_rate": 9.233937211277633e-07, "loss": 0.011765316128730774, "step": 6222 }, { "epoch": 1.6155244029075804, "grad_norm": 2.836526870727539, "learning_rate": 9.228415068497565e-07, "loss": 0.06470315903425217, "step": 6223 }, { "epoch": 1.6157840083073727, "grad_norm": 3.1022279262542725, "learning_rate": 9.222896155715352e-07, "loss": 0.04444076120853424, "step": 6224 }, { "epoch": 1.616043613707165, "grad_norm": 2.0668444633483887, "learning_rate": 9.217380473906612e-07, "loss": 0.010489208623766899, "step": 6225 }, { "epoch": 1.6163032191069573, "grad_norm": 2.5111141204833984, "learning_rate": 9.211868024046381e-07, "loss": 0.006452539935708046, "step": 6226 }, { "epoch": 1.6165628245067496, "grad_norm": 3.999265193939209, "learning_rate": 9.206358807109147e-07, "loss": 0.03599698096513748, "step": 6227 }, { "epoch": 1.616822429906542, "grad_norm": 2.0947141647338867, "learning_rate": 9.200852824068801e-07, "loss": 0.006091102957725525, "step": 6228 }, { "epoch": 1.6170820353063342, "grad_norm": 3.477687358856201, "learning_rate": 9.195350075898684e-07, "loss": 0.058363959193229675, "step": 6229 }, { "epoch": 1.6173416407061267, "grad_norm": 3.6804697513580322, "learning_rate": 9.189850563571556e-07, "loss": 0.013045052997767925, "step": 6230 }, { "epoch": 1.617601246105919, "grad_norm": 4.258885860443115, "learning_rate": 9.184354288059596e-07, "loss": 0.029267016798257828, "step": 6231 }, { "epoch": 1.6178608515057114, "grad_norm": 1.9205913543701172, "learning_rate": 9.178861250334429e-07, "loss": 0.005340700037777424, "step": 6232 }, { "epoch": 1.6181204569055037, "grad_norm": 2.736069679260254, "learning_rate": 9.173371451367097e-07, "loss": 0.04615345969796181, "step": 6233 }, { "epoch": 1.618380062305296, "grad_norm": 3.4274587631225586, "learning_rate": 9.167884892128059e-07, "loss": 0.03272530436515808, "step": 6234 }, { "epoch": 1.6186396677050883, "grad_norm": 4.098651885986328, "learning_rate": 9.162401573587229e-07, "loss": 0.031225530430674553, "step": 6235 }, { "epoch": 1.6188992731048806, "grad_norm": 1.7923115491867065, "learning_rate": 9.156921496713922e-07, "loss": 0.004138548858463764, "step": 6236 }, { "epoch": 1.6191588785046729, "grad_norm": 4.109671115875244, "learning_rate": 9.151444662476902e-07, "loss": 0.030615121126174927, "step": 6237 }, { "epoch": 1.6194184839044652, "grad_norm": 2.738294839859009, "learning_rate": 9.145971071844332e-07, "loss": 0.006910540163516998, "step": 6238 }, { "epoch": 1.6196780893042575, "grad_norm": 2.0200741291046143, "learning_rate": 9.140500725783818e-07, "loss": 0.017971470952033997, "step": 6239 }, { "epoch": 1.6199376947040498, "grad_norm": 1.8651198148727417, "learning_rate": 9.135033625262412e-07, "loss": 0.00799874309450388, "step": 6240 }, { "epoch": 1.620197300103842, "grad_norm": 3.3249027729034424, "learning_rate": 9.129569771246555e-07, "loss": 0.016350315883755684, "step": 6241 }, { "epoch": 1.6204569055036346, "grad_norm": 3.3234634399414062, "learning_rate": 9.124109164702135e-07, "loss": 0.05235758051276207, "step": 6242 }, { "epoch": 1.620716510903427, "grad_norm": 3.0311832427978516, "learning_rate": 9.118651806594459e-07, "loss": 0.05780566856265068, "step": 6243 }, { "epoch": 1.6209761163032192, "grad_norm": 3.3323862552642822, "learning_rate": 9.113197697888265e-07, "loss": 0.05563471093773842, "step": 6244 }, { "epoch": 1.6212357217030116, "grad_norm": 3.9782495498657227, "learning_rate": 9.107746839547729e-07, "loss": 0.051855579018592834, "step": 6245 }, { "epoch": 1.6214953271028039, "grad_norm": 3.7407596111297607, "learning_rate": 9.102299232536413e-07, "loss": 0.04263303801417351, "step": 6246 }, { "epoch": 1.6217549325025962, "grad_norm": 3.0922303199768066, "learning_rate": 9.096854877817346e-07, "loss": 0.03180374205112457, "step": 6247 }, { "epoch": 1.6220145379023885, "grad_norm": 2.854586362838745, "learning_rate": 9.091413776352954e-07, "loss": 0.01840135082602501, "step": 6248 }, { "epoch": 1.6222741433021808, "grad_norm": 1.8726528882980347, "learning_rate": 9.085975929105113e-07, "loss": 0.008592139929533005, "step": 6249 }, { "epoch": 1.622533748701973, "grad_norm": 3.342017889022827, "learning_rate": 9.080541337035102e-07, "loss": 0.0281352736055851, "step": 6250 }, { "epoch": 1.6227933541017654, "grad_norm": 3.509253978729248, "learning_rate": 9.075110001103627e-07, "loss": 0.02354069985449314, "step": 6251 }, { "epoch": 1.6230529595015577, "grad_norm": 1.4827650785446167, "learning_rate": 9.069681922270837e-07, "loss": 0.002726782113313675, "step": 6252 }, { "epoch": 1.62331256490135, "grad_norm": 3.0934789180755615, "learning_rate": 9.064257101496285e-07, "loss": 0.0263258945196867, "step": 6253 }, { "epoch": 1.6235721703011423, "grad_norm": 3.0609211921691895, "learning_rate": 9.058835539738947e-07, "loss": 0.025172844529151917, "step": 6254 }, { "epoch": 1.6238317757009346, "grad_norm": 2.948484182357788, "learning_rate": 9.053417237957243e-07, "loss": 0.016120420768857002, "step": 6255 }, { "epoch": 1.624091381100727, "grad_norm": 4.078478813171387, "learning_rate": 9.048002197109004e-07, "loss": 0.014840452931821346, "step": 6256 }, { "epoch": 1.6243509865005192, "grad_norm": 3.7958827018737793, "learning_rate": 9.042590418151474e-07, "loss": 0.024302681908011436, "step": 6257 }, { "epoch": 1.6246105919003115, "grad_norm": 1.3955037593841553, "learning_rate": 9.037181902041348e-07, "loss": 0.0025991343427449465, "step": 6258 }, { "epoch": 1.6248701973001038, "grad_norm": 2.266611099243164, "learning_rate": 9.031776649734708e-07, "loss": 0.004517972469329834, "step": 6259 }, { "epoch": 1.6251298026998962, "grad_norm": 1.8087559938430786, "learning_rate": 9.026374662187102e-07, "loss": 0.0040753623470664024, "step": 6260 }, { "epoch": 1.6253894080996885, "grad_norm": 2.6795966625213623, "learning_rate": 9.020975940353463e-07, "loss": 0.00861053355038166, "step": 6261 }, { "epoch": 1.6256490134994808, "grad_norm": 2.5125415325164795, "learning_rate": 9.015580485188158e-07, "loss": 0.005007945001125336, "step": 6262 }, { "epoch": 1.625908618899273, "grad_norm": 3.1064627170562744, "learning_rate": 9.010188297644994e-07, "loss": 0.0059574199840426445, "step": 6263 }, { "epoch": 1.6261682242990654, "grad_norm": 3.4887313842773438, "learning_rate": 9.004799378677176e-07, "loss": 0.014192299917340279, "step": 6264 }, { "epoch": 1.6264278296988577, "grad_norm": 3.5170044898986816, "learning_rate": 8.999413729237339e-07, "loss": 0.02039887197315693, "step": 6265 }, { "epoch": 1.62668743509865, "grad_norm": 3.179920196533203, "learning_rate": 8.994031350277552e-07, "loss": 0.04664822667837143, "step": 6266 }, { "epoch": 1.6269470404984423, "grad_norm": 4.359062194824219, "learning_rate": 8.988652242749288e-07, "loss": 0.033048808574676514, "step": 6267 }, { "epoch": 1.6272066458982346, "grad_norm": 1.14756178855896, "learning_rate": 8.983276407603464e-07, "loss": 0.0011077288072556257, "step": 6268 }, { "epoch": 1.627466251298027, "grad_norm": 3.837064027786255, "learning_rate": 8.977903845790384e-07, "loss": 0.033143606036901474, "step": 6269 }, { "epoch": 1.6277258566978192, "grad_norm": 3.415752410888672, "learning_rate": 8.972534558259803e-07, "loss": 0.02869417890906334, "step": 6270 }, { "epoch": 1.6279854620976115, "grad_norm": 2.7601659297943115, "learning_rate": 8.967168545960897e-07, "loss": 0.019203707575798035, "step": 6271 }, { "epoch": 1.6282450674974038, "grad_norm": 2.9687585830688477, "learning_rate": 8.961805809842244e-07, "loss": 0.0067484090104699135, "step": 6272 }, { "epoch": 1.6285046728971961, "grad_norm": 1.8302379846572876, "learning_rate": 8.956446350851855e-07, "loss": 0.002812274731695652, "step": 6273 }, { "epoch": 1.6287642782969884, "grad_norm": 4.065577507019043, "learning_rate": 8.951090169937158e-07, "loss": 0.016119638457894325, "step": 6274 }, { "epoch": 1.6290238836967808, "grad_norm": 2.96661114692688, "learning_rate": 8.945737268045004e-07, "loss": 0.008681626990437508, "step": 6275 }, { "epoch": 1.629283489096573, "grad_norm": 2.8156352043151855, "learning_rate": 8.940387646121673e-07, "loss": 0.012669159099459648, "step": 6276 }, { "epoch": 1.6295430944963654, "grad_norm": 3.5892229080200195, "learning_rate": 8.935041305112838e-07, "loss": 0.031024504452943802, "step": 6277 }, { "epoch": 1.629802699896158, "grad_norm": 3.914933681488037, "learning_rate": 8.929698245963625e-07, "loss": 0.02088421955704689, "step": 6278 }, { "epoch": 1.6300623052959502, "grad_norm": 3.2086057662963867, "learning_rate": 8.92435846961856e-07, "loss": 0.00781452376395464, "step": 6279 }, { "epoch": 1.6303219106957425, "grad_norm": 4.1268792152404785, "learning_rate": 8.919021977021584e-07, "loss": 0.015597132965922356, "step": 6280 }, { "epoch": 1.6305815160955348, "grad_norm": 3.540346384048462, "learning_rate": 8.91368876911608e-07, "loss": 0.03341173008084297, "step": 6281 }, { "epoch": 1.6308411214953271, "grad_norm": 4.50174617767334, "learning_rate": 8.908358846844828e-07, "loss": 0.007051398046314716, "step": 6282 }, { "epoch": 1.6311007268951194, "grad_norm": 4.15086555480957, "learning_rate": 8.903032211150042e-07, "loss": 0.033158764243125916, "step": 6283 }, { "epoch": 1.6313603322949117, "grad_norm": 2.7589011192321777, "learning_rate": 8.89770886297335e-07, "loss": 0.007499732077121735, "step": 6284 }, { "epoch": 1.631619937694704, "grad_norm": 0.3975200653076172, "learning_rate": 8.892388803255788e-07, "loss": 0.0008273522253148258, "step": 6285 }, { "epoch": 1.6318795430944963, "grad_norm": 3.5449492931365967, "learning_rate": 8.887072032937833e-07, "loss": 0.02133350633084774, "step": 6286 }, { "epoch": 1.6321391484942886, "grad_norm": 3.368863582611084, "learning_rate": 8.881758552959366e-07, "loss": 0.03520522639155388, "step": 6287 }, { "epoch": 1.632398753894081, "grad_norm": 1.974590539932251, "learning_rate": 8.876448364259677e-07, "loss": 0.00710722291842103, "step": 6288 }, { "epoch": 1.6326583592938733, "grad_norm": 3.521822929382324, "learning_rate": 8.871141467777501e-07, "loss": 0.013557312078773975, "step": 6289 }, { "epoch": 1.6329179646936658, "grad_norm": 3.494239330291748, "learning_rate": 8.865837864450962e-07, "loss": 0.04008118063211441, "step": 6290 }, { "epoch": 1.633177570093458, "grad_norm": 4.035830497741699, "learning_rate": 8.86053755521763e-07, "loss": 0.03896914795041084, "step": 6291 }, { "epoch": 1.6334371754932504, "grad_norm": 4.232991695404053, "learning_rate": 8.855240541014471e-07, "loss": 0.021505244076251984, "step": 6292 }, { "epoch": 1.6336967808930427, "grad_norm": 2.7246384620666504, "learning_rate": 8.849946822777871e-07, "loss": 0.009394727647304535, "step": 6293 }, { "epoch": 1.633956386292835, "grad_norm": 4.612369537353516, "learning_rate": 8.844656401443646e-07, "loss": 0.018270274624228477, "step": 6294 }, { "epoch": 1.6342159916926273, "grad_norm": 3.5301713943481445, "learning_rate": 8.839369277947021e-07, "loss": 0.019335264340043068, "step": 6295 }, { "epoch": 1.6344755970924196, "grad_norm": 3.4679064750671387, "learning_rate": 8.83408545322263e-07, "loss": 0.021579794585704803, "step": 6296 }, { "epoch": 1.634735202492212, "grad_norm": 3.689269542694092, "learning_rate": 8.828804928204543e-07, "loss": 0.021173428744077682, "step": 6297 }, { "epoch": 1.6349948078920042, "grad_norm": 3.2913925647735596, "learning_rate": 8.823527703826227e-07, "loss": 0.02490595355629921, "step": 6298 }, { "epoch": 1.6352544132917965, "grad_norm": 2.099931001663208, "learning_rate": 8.818253781020586e-07, "loss": 0.0061635905876755714, "step": 6299 }, { "epoch": 1.6355140186915889, "grad_norm": 3.017674207687378, "learning_rate": 8.812983160719912e-07, "loss": 0.05624103546142578, "step": 6300 }, { "epoch": 1.6357736240913812, "grad_norm": 3.262799024581909, "learning_rate": 8.807715843855941e-07, "loss": 0.015537171624600887, "step": 6301 }, { "epoch": 1.6360332294911735, "grad_norm": 3.728830099105835, "learning_rate": 8.802451831359815e-07, "loss": 0.03410881757736206, "step": 6302 }, { "epoch": 1.6362928348909658, "grad_norm": 3.342712879180908, "learning_rate": 8.797191124162091e-07, "loss": 0.041989535093307495, "step": 6303 }, { "epoch": 1.636552440290758, "grad_norm": 3.0705630779266357, "learning_rate": 8.791933723192731e-07, "loss": 0.01963530294597149, "step": 6304 }, { "epoch": 1.6368120456905504, "grad_norm": 0.8469759225845337, "learning_rate": 8.786679629381136e-07, "loss": 0.0014018892543390393, "step": 6305 }, { "epoch": 1.6370716510903427, "grad_norm": 3.2306458950042725, "learning_rate": 8.781428843656106e-07, "loss": 0.04595763236284256, "step": 6306 }, { "epoch": 1.637331256490135, "grad_norm": 0.802387535572052, "learning_rate": 8.776181366945858e-07, "loss": 0.001759340986609459, "step": 6307 }, { "epoch": 1.6375908618899273, "grad_norm": 0.9528518319129944, "learning_rate": 8.77093720017802e-07, "loss": 0.00207728473469615, "step": 6308 }, { "epoch": 1.6378504672897196, "grad_norm": 0.39413923025131226, "learning_rate": 8.765696344279645e-07, "loss": 0.000654019764624536, "step": 6309 }, { "epoch": 1.638110072689512, "grad_norm": 3.7092182636260986, "learning_rate": 8.76045880017721e-07, "loss": 0.030885450541973114, "step": 6310 }, { "epoch": 1.6383696780893042, "grad_norm": 3.9228837490081787, "learning_rate": 8.755224568796569e-07, "loss": 0.011118771508336067, "step": 6311 }, { "epoch": 1.6386292834890965, "grad_norm": 4.214446544647217, "learning_rate": 8.749993651063033e-07, "loss": 0.024022050201892853, "step": 6312 }, { "epoch": 1.6388888888888888, "grad_norm": 4.357154846191406, "learning_rate": 8.744766047901291e-07, "loss": 0.018209030851721764, "step": 6313 }, { "epoch": 1.6391484942886811, "grad_norm": 0.6214494109153748, "learning_rate": 8.739541760235484e-07, "loss": 0.0017717878799885511, "step": 6314 }, { "epoch": 1.6394080996884735, "grad_norm": 3.5976858139038086, "learning_rate": 8.734320788989133e-07, "loss": 0.028707237914204597, "step": 6315 }, { "epoch": 1.6396677050882658, "grad_norm": 2.50875186920166, "learning_rate": 8.729103135085184e-07, "loss": 0.008361726067960262, "step": 6316 }, { "epoch": 1.639927310488058, "grad_norm": 1.4033913612365723, "learning_rate": 8.723888799446012e-07, "loss": 0.005497168283909559, "step": 6317 }, { "epoch": 1.6401869158878504, "grad_norm": 3.00372314453125, "learning_rate": 8.718677782993383e-07, "loss": 0.026855342090129852, "step": 6318 }, { "epoch": 1.6404465212876427, "grad_norm": 3.8788540363311768, "learning_rate": 8.713470086648482e-07, "loss": 0.05520676448941231, "step": 6319 }, { "epoch": 1.640706126687435, "grad_norm": 1.7539432048797607, "learning_rate": 8.70826571133192e-07, "loss": 0.014298020862042904, "step": 6320 }, { "epoch": 1.6409657320872273, "grad_norm": 3.000450849533081, "learning_rate": 8.703064657963703e-07, "loss": 0.016876021400094032, "step": 6321 }, { "epoch": 1.6412253374870196, "grad_norm": 1.8309894800186157, "learning_rate": 8.697866927463264e-07, "loss": 0.004653786309063435, "step": 6322 }, { "epoch": 1.641484942886812, "grad_norm": 1.2884624004364014, "learning_rate": 8.692672520749443e-07, "loss": 0.0019092869479209185, "step": 6323 }, { "epoch": 1.6417445482866042, "grad_norm": 2.841343879699707, "learning_rate": 8.687481438740481e-07, "loss": 0.010245898738503456, "step": 6324 }, { "epoch": 1.6420041536863965, "grad_norm": 1.8629539012908936, "learning_rate": 8.682293682354059e-07, "loss": 0.009092053398489952, "step": 6325 }, { "epoch": 1.642263759086189, "grad_norm": 2.2615981101989746, "learning_rate": 8.677109252507246e-07, "loss": 0.007680367678403854, "step": 6326 }, { "epoch": 1.6425233644859814, "grad_norm": 3.429805040359497, "learning_rate": 8.671928150116523e-07, "loss": 0.04156149551272392, "step": 6327 }, { "epoch": 1.6427829698857737, "grad_norm": 4.222134590148926, "learning_rate": 8.666750376097805e-07, "loss": 0.046783093363046646, "step": 6328 }, { "epoch": 1.643042575285566, "grad_norm": 3.514639377593994, "learning_rate": 8.661575931366387e-07, "loss": 0.0365617498755455, "step": 6329 }, { "epoch": 1.6433021806853583, "grad_norm": 3.502140998840332, "learning_rate": 8.656404816837011e-07, "loss": 0.007361392490565777, "step": 6330 }, { "epoch": 1.6435617860851506, "grad_norm": 4.123145580291748, "learning_rate": 8.651237033423801e-07, "loss": 0.011138255707919598, "step": 6331 }, { "epoch": 1.6438213914849429, "grad_norm": 3.624683380126953, "learning_rate": 8.646072582040296e-07, "loss": 0.049295395612716675, "step": 6332 }, { "epoch": 1.6440809968847352, "grad_norm": 4.709841728210449, "learning_rate": 8.640911463599475e-07, "loss": 0.01821630820631981, "step": 6333 }, { "epoch": 1.6443406022845275, "grad_norm": 4.071410179138184, "learning_rate": 8.635753679013681e-07, "loss": 0.0353439599275589, "step": 6334 }, { "epoch": 1.6446002076843198, "grad_norm": 3.4382565021514893, "learning_rate": 8.6305992291947e-07, "loss": 0.005636458285152912, "step": 6335 }, { "epoch": 1.644859813084112, "grad_norm": 3.1016793251037598, "learning_rate": 8.62544811505373e-07, "loss": 0.013086723163723946, "step": 6336 }, { "epoch": 1.6451194184839044, "grad_norm": 3.0383589267730713, "learning_rate": 8.620300337501361e-07, "loss": 0.035729795694351196, "step": 6337 }, { "epoch": 1.645379023883697, "grad_norm": 0.32543614506721497, "learning_rate": 8.615155897447606e-07, "loss": 0.0009263016399927437, "step": 6338 }, { "epoch": 1.6456386292834893, "grad_norm": 2.4671847820281982, "learning_rate": 8.610014795801879e-07, "loss": 0.0039337133057415485, "step": 6339 }, { "epoch": 1.6458982346832816, "grad_norm": 2.9297943115234375, "learning_rate": 8.604877033473011e-07, "loss": 0.00961710512638092, "step": 6340 }, { "epoch": 1.6461578400830739, "grad_norm": 3.437617540359497, "learning_rate": 8.599742611369255e-07, "loss": 0.03187389671802521, "step": 6341 }, { "epoch": 1.6464174454828662, "grad_norm": 3.9885623455047607, "learning_rate": 8.594611530398235e-07, "loss": 0.010496541857719421, "step": 6342 }, { "epoch": 1.6466770508826585, "grad_norm": 3.130986452102661, "learning_rate": 8.589483791467024e-07, "loss": 0.02625282295048237, "step": 6343 }, { "epoch": 1.6469366562824508, "grad_norm": 3.171977996826172, "learning_rate": 8.584359395482081e-07, "loss": 0.053170640021562576, "step": 6344 }, { "epoch": 1.647196261682243, "grad_norm": 2.419464111328125, "learning_rate": 8.579238343349295e-07, "loss": 0.016293536871671677, "step": 6345 }, { "epoch": 1.6474558670820354, "grad_norm": 2.7332701683044434, "learning_rate": 8.574120635973937e-07, "loss": 0.04688971862196922, "step": 6346 }, { "epoch": 1.6477154724818277, "grad_norm": 2.791783571243286, "learning_rate": 8.569006274260702e-07, "loss": 0.02004963532090187, "step": 6347 }, { "epoch": 1.64797507788162, "grad_norm": 3.1464483737945557, "learning_rate": 8.563895259113703e-07, "loss": 0.03490329161286354, "step": 6348 }, { "epoch": 1.6482346832814123, "grad_norm": 1.9200564622879028, "learning_rate": 8.558787591436437e-07, "loss": 0.004944679327309132, "step": 6349 }, { "epoch": 1.6484942886812046, "grad_norm": 0.6366289258003235, "learning_rate": 8.553683272131827e-07, "loss": 0.0014133594231680036, "step": 6350 }, { "epoch": 1.648753894080997, "grad_norm": 3.7457633018493652, "learning_rate": 8.548582302102207e-07, "loss": 0.04576563090085983, "step": 6351 }, { "epoch": 1.6490134994807892, "grad_norm": 3.831439971923828, "learning_rate": 8.543484682249301e-07, "loss": 0.03156231716275215, "step": 6352 }, { "epoch": 1.6492731048805815, "grad_norm": 3.7519631385803223, "learning_rate": 8.53839041347426e-07, "loss": 0.030068064108490944, "step": 6353 }, { "epoch": 1.6495327102803738, "grad_norm": 3.1648285388946533, "learning_rate": 8.53329949667763e-07, "loss": 0.013948727399110794, "step": 6354 }, { "epoch": 1.6497923156801662, "grad_norm": 3.180616617202759, "learning_rate": 8.528211932759367e-07, "loss": 0.009796250611543655, "step": 6355 }, { "epoch": 1.6500519210799585, "grad_norm": 3.1238644123077393, "learning_rate": 8.52312772261884e-07, "loss": 0.013414863497018814, "step": 6356 }, { "epoch": 1.6503115264797508, "grad_norm": 3.462353467941284, "learning_rate": 8.518046867154823e-07, "loss": 0.01077299565076828, "step": 6357 }, { "epoch": 1.650571131879543, "grad_norm": 3.04500412940979, "learning_rate": 8.512969367265483e-07, "loss": 0.05838344618678093, "step": 6358 }, { "epoch": 1.6508307372793354, "grad_norm": 2.5683019161224365, "learning_rate": 8.507895223848417e-07, "loss": 0.010360484011471272, "step": 6359 }, { "epoch": 1.6510903426791277, "grad_norm": 3.34124493598938, "learning_rate": 8.502824437800615e-07, "loss": 0.029284697026014328, "step": 6360 }, { "epoch": 1.65134994807892, "grad_norm": 4.344941139221191, "learning_rate": 8.49775701001847e-07, "loss": 0.03357694298028946, "step": 6361 }, { "epoch": 1.6516095534787123, "grad_norm": 3.515305519104004, "learning_rate": 8.492692941397798e-07, "loss": 0.042077530175447464, "step": 6362 }, { "epoch": 1.6518691588785046, "grad_norm": 3.8822269439697266, "learning_rate": 8.487632232833798e-07, "loss": 0.03472302481532097, "step": 6363 }, { "epoch": 1.652128764278297, "grad_norm": 3.975931406021118, "learning_rate": 8.482574885221106e-07, "loss": 0.03344598039984703, "step": 6364 }, { "epoch": 1.6523883696780892, "grad_norm": 3.7961037158966064, "learning_rate": 8.477520899453719e-07, "loss": 0.011423686519265175, "step": 6365 }, { "epoch": 1.6526479750778815, "grad_norm": 2.0839247703552246, "learning_rate": 8.472470276425082e-07, "loss": 0.002109101042151451, "step": 6366 }, { "epoch": 1.6529075804776738, "grad_norm": 2.9539146423339844, "learning_rate": 8.467423017028035e-07, "loss": 0.024418216198682785, "step": 6367 }, { "epoch": 1.6531671858774661, "grad_norm": 3.8978538513183594, "learning_rate": 8.462379122154807e-07, "loss": 0.045634523034095764, "step": 6368 }, { "epoch": 1.6534267912772584, "grad_norm": 1.9697946310043335, "learning_rate": 8.457338592697046e-07, "loss": 0.006588227581232786, "step": 6369 }, { "epoch": 1.6536863966770508, "grad_norm": 1.3172307014465332, "learning_rate": 8.452301429545798e-07, "loss": 0.005939727649092674, "step": 6370 }, { "epoch": 1.653946002076843, "grad_norm": 1.9861348867416382, "learning_rate": 8.447267633591525e-07, "loss": 0.015351837500929832, "step": 6371 }, { "epoch": 1.6542056074766354, "grad_norm": 0.9102775454521179, "learning_rate": 8.442237205724094e-07, "loss": 0.006438607349991798, "step": 6372 }, { "epoch": 1.6544652128764277, "grad_norm": 3.513108253479004, "learning_rate": 8.437210146832749e-07, "loss": 0.016156122088432312, "step": 6373 }, { "epoch": 1.6547248182762202, "grad_norm": 3.7381644248962402, "learning_rate": 8.432186457806175e-07, "loss": 0.025896910578012466, "step": 6374 }, { "epoch": 1.6549844236760125, "grad_norm": 3.282007932662964, "learning_rate": 8.427166139532436e-07, "loss": 0.028163187205791473, "step": 6375 }, { "epoch": 1.6552440290758048, "grad_norm": 4.3254313468933105, "learning_rate": 8.422149192899018e-07, "loss": 0.025999292731285095, "step": 6376 }, { "epoch": 1.6555036344755971, "grad_norm": 4.941933631896973, "learning_rate": 8.417135618792799e-07, "loss": 0.009279453195631504, "step": 6377 }, { "epoch": 1.6557632398753894, "grad_norm": 2.400312662124634, "learning_rate": 8.412125418100055e-07, "loss": 0.011716829612851143, "step": 6378 }, { "epoch": 1.6560228452751817, "grad_norm": 2.7463231086730957, "learning_rate": 8.407118591706493e-07, "loss": 0.010266785509884357, "step": 6379 }, { "epoch": 1.656282450674974, "grad_norm": 3.7379939556121826, "learning_rate": 8.402115140497189e-07, "loss": 0.02372916042804718, "step": 6380 }, { "epoch": 1.6565420560747663, "grad_norm": 2.9849185943603516, "learning_rate": 8.397115065356641e-07, "loss": 0.06215682625770569, "step": 6381 }, { "epoch": 1.6568016614745587, "grad_norm": 3.5847058296203613, "learning_rate": 8.392118367168757e-07, "loss": 0.028187669813632965, "step": 6382 }, { "epoch": 1.657061266874351, "grad_norm": 3.5332610607147217, "learning_rate": 8.38712504681683e-07, "loss": 0.013135828077793121, "step": 6383 }, { "epoch": 1.6573208722741433, "grad_norm": 3.2431626319885254, "learning_rate": 8.38213510518357e-07, "loss": 0.026581162586808205, "step": 6384 }, { "epoch": 1.6575804776739356, "grad_norm": 3.248507022857666, "learning_rate": 8.377148543151084e-07, "loss": 0.020539481192827225, "step": 6385 }, { "epoch": 1.657840083073728, "grad_norm": 3.2645254135131836, "learning_rate": 8.372165361600876e-07, "loss": 0.014064807444810867, "step": 6386 }, { "epoch": 1.6580996884735204, "grad_norm": 7.261911392211914, "learning_rate": 8.367185561413866e-07, "loss": 0.035302821546792984, "step": 6387 }, { "epoch": 1.6583592938733127, "grad_norm": 1.3807730674743652, "learning_rate": 8.362209143470368e-07, "loss": 0.005199024453759193, "step": 6388 }, { "epoch": 1.658618899273105, "grad_norm": 2.0933837890625, "learning_rate": 8.357236108650091e-07, "loss": 0.008152617141604424, "step": 6389 }, { "epoch": 1.6588785046728973, "grad_norm": 3.1377220153808594, "learning_rate": 8.352266457832158e-07, "loss": 0.011254603043198586, "step": 6390 }, { "epoch": 1.6591381100726896, "grad_norm": 3.4704413414001465, "learning_rate": 8.347300191895097e-07, "loss": 0.011438967660069466, "step": 6391 }, { "epoch": 1.659397715472482, "grad_norm": 3.21793532371521, "learning_rate": 8.342337311716817e-07, "loss": 0.026771191507577896, "step": 6392 }, { "epoch": 1.6596573208722742, "grad_norm": 3.185232162475586, "learning_rate": 8.337377818174651e-07, "loss": 0.05331999436020851, "step": 6393 }, { "epoch": 1.6599169262720666, "grad_norm": 2.7640106678009033, "learning_rate": 8.332421712145316e-07, "loss": 0.004700541961938143, "step": 6394 }, { "epoch": 1.6601765316718589, "grad_norm": 3.2928643226623535, "learning_rate": 8.327468994504953e-07, "loss": 0.023846011608839035, "step": 6395 }, { "epoch": 1.6604361370716512, "grad_norm": 3.1906216144561768, "learning_rate": 8.32251966612907e-07, "loss": 0.022157978266477585, "step": 6396 }, { "epoch": 1.6606957424714435, "grad_norm": 3.0635364055633545, "learning_rate": 8.317573727892603e-07, "loss": 0.062458284199237823, "step": 6397 }, { "epoch": 1.6609553478712358, "grad_norm": 3.117058038711548, "learning_rate": 8.312631180669889e-07, "loss": 0.03994814306497574, "step": 6398 }, { "epoch": 1.661214953271028, "grad_norm": 3.7577571868896484, "learning_rate": 8.307692025334648e-07, "loss": 0.015923133119940758, "step": 6399 }, { "epoch": 1.6614745586708204, "grad_norm": 2.9621219635009766, "learning_rate": 8.302756262760009e-07, "loss": 0.07337778806686401, "step": 6400 }, { "epoch": 1.6617341640706127, "grad_norm": 2.0298948287963867, "learning_rate": 8.297823893818502e-07, "loss": 0.017979411408305168, "step": 6401 }, { "epoch": 1.661993769470405, "grad_norm": 2.431849956512451, "learning_rate": 8.292894919382058e-07, "loss": 0.00869200099259615, "step": 6402 }, { "epoch": 1.6622533748701973, "grad_norm": 3.409864902496338, "learning_rate": 8.287969340322017e-07, "loss": 0.05323103070259094, "step": 6403 }, { "epoch": 1.6625129802699896, "grad_norm": 5.192052841186523, "learning_rate": 8.283047157509091e-07, "loss": 0.009694178588688374, "step": 6404 }, { "epoch": 1.662772585669782, "grad_norm": 3.977591037750244, "learning_rate": 8.278128371813423e-07, "loss": 0.02652200497686863, "step": 6405 }, { "epoch": 1.6630321910695742, "grad_norm": 3.5325093269348145, "learning_rate": 8.273212984104531e-07, "loss": 0.04271761327981949, "step": 6406 }, { "epoch": 1.6632917964693665, "grad_norm": 2.473989963531494, "learning_rate": 8.268300995251353e-07, "loss": 0.0045914603397250175, "step": 6407 }, { "epoch": 1.6635514018691588, "grad_norm": 2.999769926071167, "learning_rate": 8.263392406122212e-07, "loss": 0.03471498191356659, "step": 6408 }, { "epoch": 1.6638110072689511, "grad_norm": 3.8201801776885986, "learning_rate": 8.258487217584827e-07, "loss": 0.025014884769916534, "step": 6409 }, { "epoch": 1.6640706126687435, "grad_norm": 1.484875202178955, "learning_rate": 8.253585430506333e-07, "loss": 0.009508023038506508, "step": 6410 }, { "epoch": 1.6643302180685358, "grad_norm": 1.8526355028152466, "learning_rate": 8.248687045753256e-07, "loss": 0.007081165909767151, "step": 6411 }, { "epoch": 1.664589823468328, "grad_norm": 2.9204273223876953, "learning_rate": 8.243792064191503e-07, "loss": 0.051019467413425446, "step": 6412 }, { "epoch": 1.6648494288681204, "grad_norm": 3.2106804847717285, "learning_rate": 8.238900486686413e-07, "loss": 0.04260481148958206, "step": 6413 }, { "epoch": 1.6651090342679127, "grad_norm": 3.2391631603240967, "learning_rate": 8.234012314102695e-07, "loss": 0.05309060215950012, "step": 6414 }, { "epoch": 1.665368639667705, "grad_norm": 3.6318347454071045, "learning_rate": 8.22912754730446e-07, "loss": 0.024876806885004044, "step": 6415 }, { "epoch": 1.6656282450674973, "grad_norm": 3.1102609634399414, "learning_rate": 8.224246187155238e-07, "loss": 0.008879496715962887, "step": 6416 }, { "epoch": 1.6658878504672896, "grad_norm": 2.000725507736206, "learning_rate": 8.219368234517927e-07, "loss": 0.01605064421892166, "step": 6417 }, { "epoch": 1.666147455867082, "grad_norm": 4.604510307312012, "learning_rate": 8.21449369025485e-07, "loss": 0.02541188709437847, "step": 6418 }, { "epoch": 1.6664070612668742, "grad_norm": 2.2127373218536377, "learning_rate": 8.209622555227709e-07, "loss": 0.008107330650091171, "step": 6419 }, { "epoch": 1.6666666666666665, "grad_norm": 3.1255528926849365, "learning_rate": 8.204754830297602e-07, "loss": 0.04003912955522537, "step": 6420 }, { "epoch": 1.6669262720664588, "grad_norm": 1.8027892112731934, "learning_rate": 8.199890516325046e-07, "loss": 0.0038906875997781754, "step": 6421 }, { "epoch": 1.6671858774662511, "grad_norm": 3.6439099311828613, "learning_rate": 8.195029614169934e-07, "loss": 0.028468817472457886, "step": 6422 }, { "epoch": 1.6674454828660437, "grad_norm": 3.2232837677001953, "learning_rate": 8.190172124691553e-07, "loss": 0.05172080174088478, "step": 6423 }, { "epoch": 1.667705088265836, "grad_norm": 3.95742130279541, "learning_rate": 8.185318048748611e-07, "loss": 0.008704915642738342, "step": 6424 }, { "epoch": 1.6679646936656283, "grad_norm": 3.259974718093872, "learning_rate": 8.180467387199184e-07, "loss": 0.05797459930181503, "step": 6425 }, { "epoch": 1.6682242990654206, "grad_norm": 3.4400947093963623, "learning_rate": 8.175620140900774e-07, "loss": 0.027158517390489578, "step": 6426 }, { "epoch": 1.6684839044652129, "grad_norm": 5.184450626373291, "learning_rate": 8.170776310710246e-07, "loss": 0.00956491194665432, "step": 6427 }, { "epoch": 1.6687435098650052, "grad_norm": 3.503016710281372, "learning_rate": 8.165935897483881e-07, "loss": 0.010947895236313343, "step": 6428 }, { "epoch": 1.6690031152647975, "grad_norm": 3.921929121017456, "learning_rate": 8.161098902077366e-07, "loss": 0.026501862332224846, "step": 6429 }, { "epoch": 1.6692627206645898, "grad_norm": 4.839776039123535, "learning_rate": 8.156265325345764e-07, "loss": 0.023077718913555145, "step": 6430 }, { "epoch": 1.6695223260643821, "grad_norm": 4.239302158355713, "learning_rate": 8.151435168143537e-07, "loss": 0.045218575745821, "step": 6431 }, { "epoch": 1.6697819314641744, "grad_norm": 3.3897922039031982, "learning_rate": 8.146608431324548e-07, "loss": 0.013075046241283417, "step": 6432 }, { "epoch": 1.6700415368639667, "grad_norm": 3.0841240882873535, "learning_rate": 8.141785115742049e-07, "loss": 0.06026573106646538, "step": 6433 }, { "epoch": 1.670301142263759, "grad_norm": 4.563730239868164, "learning_rate": 8.136965222248713e-07, "loss": 0.015615588054060936, "step": 6434 }, { "epoch": 1.6705607476635516, "grad_norm": 3.3248767852783203, "learning_rate": 8.132148751696556e-07, "loss": 0.03405916690826416, "step": 6435 }, { "epoch": 1.6708203530633439, "grad_norm": 3.256875991821289, "learning_rate": 8.127335704937042e-07, "loss": 0.035996925085783005, "step": 6436 }, { "epoch": 1.6710799584631362, "grad_norm": 3.441119909286499, "learning_rate": 8.122526082820994e-07, "loss": 0.0496491864323616, "step": 6437 }, { "epoch": 1.6713395638629285, "grad_norm": 3.978097438812256, "learning_rate": 8.117719886198655e-07, "loss": 0.024197574704885483, "step": 6438 }, { "epoch": 1.6715991692627208, "grad_norm": 3.818082571029663, "learning_rate": 8.112917115919644e-07, "loss": 0.012410547584295273, "step": 6439 }, { "epoch": 1.671858774662513, "grad_norm": 3.24015212059021, "learning_rate": 8.108117772832977e-07, "loss": 0.018194761127233505, "step": 6440 }, { "epoch": 1.6721183800623054, "grad_norm": 4.218336582183838, "learning_rate": 8.103321857787077e-07, "loss": 0.03967740386724472, "step": 6441 }, { "epoch": 1.6723779854620977, "grad_norm": 3.01521897315979, "learning_rate": 8.09852937162975e-07, "loss": 0.0229684729129076, "step": 6442 }, { "epoch": 1.67263759086189, "grad_norm": 3.766181230545044, "learning_rate": 8.093740315208188e-07, "loss": 0.037218108773231506, "step": 6443 }, { "epoch": 1.6728971962616823, "grad_norm": 3.5935475826263428, "learning_rate": 8.088954689368999e-07, "loss": 0.026434144005179405, "step": 6444 }, { "epoch": 1.6731568016614746, "grad_norm": 4.3064446449279785, "learning_rate": 8.084172494958168e-07, "loss": 0.01483951322734356, "step": 6445 }, { "epoch": 1.673416407061267, "grad_norm": 3.4636902809143066, "learning_rate": 8.07939373282107e-07, "loss": 0.025772299617528915, "step": 6446 }, { "epoch": 1.6736760124610592, "grad_norm": 3.1917760372161865, "learning_rate": 8.074618403802494e-07, "loss": 0.006489669904112816, "step": 6447 }, { "epoch": 1.6739356178608515, "grad_norm": 2.5216224193573, "learning_rate": 8.069846508746594e-07, "loss": 0.027469849213957787, "step": 6448 }, { "epoch": 1.6741952232606438, "grad_norm": 2.729254722595215, "learning_rate": 8.065078048496946e-07, "loss": 0.006752275861799717, "step": 6449 }, { "epoch": 1.6744548286604362, "grad_norm": 3.0287740230560303, "learning_rate": 8.060313023896499e-07, "loss": 0.011807207949459553, "step": 6450 }, { "epoch": 1.6747144340602285, "grad_norm": 3.198945999145508, "learning_rate": 8.055551435787595e-07, "loss": 0.020756887272000313, "step": 6451 }, { "epoch": 1.6749740394600208, "grad_norm": 3.619731903076172, "learning_rate": 8.050793285011983e-07, "loss": 0.00949259102344513, "step": 6452 }, { "epoch": 1.675233644859813, "grad_norm": 2.4125235080718994, "learning_rate": 8.04603857241079e-07, "loss": 0.005848838482052088, "step": 6453 }, { "epoch": 1.6754932502596054, "grad_norm": 3.184860944747925, "learning_rate": 8.041287298824535e-07, "loss": 0.00889834389090538, "step": 6454 }, { "epoch": 1.6757528556593977, "grad_norm": 4.177186489105225, "learning_rate": 8.036539465093146e-07, "loss": 0.015264016576111317, "step": 6455 }, { "epoch": 1.67601246105919, "grad_norm": 3.0442473888397217, "learning_rate": 8.031795072055921e-07, "loss": 0.014340493828058243, "step": 6456 }, { "epoch": 1.6762720664589823, "grad_norm": 2.4232611656188965, "learning_rate": 8.027054120551572e-07, "loss": 0.010968885384500027, "step": 6457 }, { "epoch": 1.6765316718587746, "grad_norm": 3.7109827995300293, "learning_rate": 8.022316611418176e-07, "loss": 0.024571722373366356, "step": 6458 }, { "epoch": 1.676791277258567, "grad_norm": 2.233961343765259, "learning_rate": 8.017582545493221e-07, "loss": 0.008666323497891426, "step": 6459 }, { "epoch": 1.6770508826583592, "grad_norm": 3.5355875492095947, "learning_rate": 8.01285192361359e-07, "loss": 0.033879853785037994, "step": 6460 }, { "epoch": 1.6773104880581515, "grad_norm": 2.23047137260437, "learning_rate": 8.008124746615542e-07, "loss": 0.0036722877994179726, "step": 6461 }, { "epoch": 1.6775700934579438, "grad_norm": 3.70931077003479, "learning_rate": 8.003401015334734e-07, "loss": 0.009451786056160927, "step": 6462 }, { "epoch": 1.6778296988577361, "grad_norm": 4.004913330078125, "learning_rate": 7.99868073060621e-07, "loss": 0.009199281223118305, "step": 6463 }, { "epoch": 1.6780893042575284, "grad_norm": 3.7745964527130127, "learning_rate": 7.99396389326441e-07, "loss": 0.016115952283143997, "step": 6464 }, { "epoch": 1.6783489096573208, "grad_norm": 3.4442901611328125, "learning_rate": 7.989250504143174e-07, "loss": 0.04937400668859482, "step": 6465 }, { "epoch": 1.678608515057113, "grad_norm": 3.505033493041992, "learning_rate": 7.984540564075702e-07, "loss": 0.01330711878836155, "step": 6466 }, { "epoch": 1.6788681204569054, "grad_norm": 4.038443565368652, "learning_rate": 7.979834073894621e-07, "loss": 0.022707972675561905, "step": 6467 }, { "epoch": 1.6791277258566977, "grad_norm": 3.886838674545288, "learning_rate": 7.975131034431917e-07, "loss": 0.024106372147798538, "step": 6468 }, { "epoch": 1.67938733125649, "grad_norm": 3.690767526626587, "learning_rate": 7.970431446518991e-07, "loss": 0.03981778770685196, "step": 6469 }, { "epoch": 1.6796469366562823, "grad_norm": 4.452856063842773, "learning_rate": 7.965735310986615e-07, "loss": 0.004755803849548101, "step": 6470 }, { "epoch": 1.6799065420560748, "grad_norm": 3.775479555130005, "learning_rate": 7.961042628664958e-07, "loss": 0.024917641654610634, "step": 6471 }, { "epoch": 1.6801661474558671, "grad_norm": 5.834166049957275, "learning_rate": 7.956353400383586e-07, "loss": 0.03541434928774834, "step": 6472 }, { "epoch": 1.6804257528556594, "grad_norm": 4.646366596221924, "learning_rate": 7.951667626971443e-07, "loss": 0.020701013505458832, "step": 6473 }, { "epoch": 1.6806853582554517, "grad_norm": 3.2134363651275635, "learning_rate": 7.94698530925686e-07, "loss": 0.020239219069480896, "step": 6474 }, { "epoch": 1.680944963655244, "grad_norm": 3.3501760959625244, "learning_rate": 7.942306448067574e-07, "loss": 0.031161336228251457, "step": 6475 }, { "epoch": 1.6812045690550363, "grad_norm": 2.4588284492492676, "learning_rate": 7.937631044230699e-07, "loss": 0.008303170092403889, "step": 6476 }, { "epoch": 1.6814641744548287, "grad_norm": 2.8293254375457764, "learning_rate": 7.932959098572731e-07, "loss": 0.01873927377164364, "step": 6477 }, { "epoch": 1.681723779854621, "grad_norm": 2.7846689224243164, "learning_rate": 7.928290611919573e-07, "loss": 0.009944518096745014, "step": 6478 }, { "epoch": 1.6819833852544133, "grad_norm": 3.3280386924743652, "learning_rate": 7.923625585096496e-07, "loss": 0.025667252019047737, "step": 6479 }, { "epoch": 1.6822429906542056, "grad_norm": 3.2167162895202637, "learning_rate": 7.91896401892818e-07, "loss": 0.010152647271752357, "step": 6480 }, { "epoch": 1.6825025960539979, "grad_norm": 2.2323784828186035, "learning_rate": 7.914305914238682e-07, "loss": 0.011616010218858719, "step": 6481 }, { "epoch": 1.6827622014537902, "grad_norm": 2.691784381866455, "learning_rate": 7.909651271851438e-07, "loss": 0.01210303045809269, "step": 6482 }, { "epoch": 1.6830218068535827, "grad_norm": 3.7834041118621826, "learning_rate": 7.905000092589296e-07, "loss": 0.021686777472496033, "step": 6483 }, { "epoch": 1.683281412253375, "grad_norm": 3.250490665435791, "learning_rate": 7.900352377274471e-07, "loss": 0.01900169439613819, "step": 6484 }, { "epoch": 1.6835410176531673, "grad_norm": 0.46719685196876526, "learning_rate": 7.895708126728569e-07, "loss": 0.0011608973145484924, "step": 6485 }, { "epoch": 1.6838006230529596, "grad_norm": 2.764799118041992, "learning_rate": 7.891067341772596e-07, "loss": 0.015270952135324478, "step": 6486 }, { "epoch": 1.684060228452752, "grad_norm": 1.1453287601470947, "learning_rate": 7.886430023226927e-07, "loss": 0.0016340838046744466, "step": 6487 }, { "epoch": 1.6843198338525442, "grad_norm": 1.9352349042892456, "learning_rate": 7.881796171911349e-07, "loss": 0.012249326333403587, "step": 6488 }, { "epoch": 1.6845794392523366, "grad_norm": 0.37584200501441956, "learning_rate": 7.877165788644998e-07, "loss": 0.0012216975446790457, "step": 6489 }, { "epoch": 1.6848390446521289, "grad_norm": 3.181793689727783, "learning_rate": 7.872538874246436e-07, "loss": 0.01700512506067753, "step": 6490 }, { "epoch": 1.6850986500519212, "grad_norm": 2.9417643547058105, "learning_rate": 7.867915429533596e-07, "loss": 0.008935804478824139, "step": 6491 }, { "epoch": 1.6853582554517135, "grad_norm": 2.901386022567749, "learning_rate": 7.863295455323791e-07, "loss": 0.07698969542980194, "step": 6492 }, { "epoch": 1.6856178608515058, "grad_norm": 3.454136610031128, "learning_rate": 7.85867895243373e-07, "loss": 0.02887050062417984, "step": 6493 }, { "epoch": 1.685877466251298, "grad_norm": 1.9751789569854736, "learning_rate": 7.854065921679499e-07, "loss": 0.009357589296996593, "step": 6494 }, { "epoch": 1.6861370716510904, "grad_norm": 2.7051584720611572, "learning_rate": 7.849456363876579e-07, "loss": 0.03356238082051277, "step": 6495 }, { "epoch": 1.6863966770508827, "grad_norm": 2.816725492477417, "learning_rate": 7.844850279839847e-07, "loss": 0.05660402029752731, "step": 6496 }, { "epoch": 1.686656282450675, "grad_norm": 3.3552117347717285, "learning_rate": 7.84024767038353e-07, "loss": 0.02507217600941658, "step": 6497 }, { "epoch": 1.6869158878504673, "grad_norm": 3.646453380584717, "learning_rate": 7.835648536321285e-07, "loss": 0.016895785927772522, "step": 6498 }, { "epoch": 1.6871754932502596, "grad_norm": 4.845649719238281, "learning_rate": 7.831052878466124e-07, "loss": 0.0373159721493721, "step": 6499 }, { "epoch": 1.687435098650052, "grad_norm": 4.166831016540527, "learning_rate": 7.826460697630449e-07, "loss": 0.005663788877427578, "step": 6500 }, { "epoch": 1.6876947040498442, "grad_norm": 1.903027892112732, "learning_rate": 7.821871994626062e-07, "loss": 0.00741304038092494, "step": 6501 }, { "epoch": 1.6879543094496365, "grad_norm": 1.959965467453003, "learning_rate": 7.817286770264134e-07, "loss": 0.004682219587266445, "step": 6502 }, { "epoch": 1.6882139148494288, "grad_norm": 2.819579601287842, "learning_rate": 7.812705025355234e-07, "loss": 0.047773633152246475, "step": 6503 }, { "epoch": 1.6884735202492211, "grad_norm": 1.4733517169952393, "learning_rate": 7.808126760709309e-07, "loss": 0.003702795132994652, "step": 6504 }, { "epoch": 1.6887331256490135, "grad_norm": 1.713907241821289, "learning_rate": 7.803551977135681e-07, "loss": 0.00669999560341239, "step": 6505 }, { "epoch": 1.6889927310488058, "grad_norm": 3.337724447250366, "learning_rate": 7.798980675443079e-07, "loss": 0.016020603477954865, "step": 6506 }, { "epoch": 1.689252336448598, "grad_norm": 2.8657288551330566, "learning_rate": 7.794412856439602e-07, "loss": 0.014313379302620888, "step": 6507 }, { "epoch": 1.6895119418483904, "grad_norm": 3.9104363918304443, "learning_rate": 7.789848520932729e-07, "loss": 0.010765399783849716, "step": 6508 }, { "epoch": 1.6897715472481827, "grad_norm": 2.253044605255127, "learning_rate": 7.785287669729338e-07, "loss": 0.008060355670750141, "step": 6509 }, { "epoch": 1.690031152647975, "grad_norm": 3.608293294906616, "learning_rate": 7.780730303635675e-07, "loss": 0.013906141743063927, "step": 6510 }, { "epoch": 1.6902907580477673, "grad_norm": 2.911606788635254, "learning_rate": 7.776176423457391e-07, "loss": 0.003390318714082241, "step": 6511 }, { "epoch": 1.6905503634475596, "grad_norm": 3.9165003299713135, "learning_rate": 7.771626029999496e-07, "loss": 0.019283883273601532, "step": 6512 }, { "epoch": 1.690809968847352, "grad_norm": 1.8600307703018188, "learning_rate": 7.767079124066397e-07, "loss": 0.0017474581254646182, "step": 6513 }, { "epoch": 1.6910695742471442, "grad_norm": 3.480700731277466, "learning_rate": 7.762535706461887e-07, "loss": 0.017559191212058067, "step": 6514 }, { "epoch": 1.6913291796469365, "grad_norm": 3.0916476249694824, "learning_rate": 7.757995777989139e-07, "loss": 0.01046350784599781, "step": 6515 }, { "epoch": 1.6915887850467288, "grad_norm": 3.3264412879943848, "learning_rate": 7.753459339450699e-07, "loss": 0.037174928933382034, "step": 6516 }, { "epoch": 1.6918483904465211, "grad_norm": 2.8978967666625977, "learning_rate": 7.748926391648516e-07, "loss": 0.015291769988834858, "step": 6517 }, { "epoch": 1.6921079958463134, "grad_norm": 1.852832555770874, "learning_rate": 7.744396935383904e-07, "loss": 0.00801929272711277, "step": 6518 }, { "epoch": 1.692367601246106, "grad_norm": 0.200540229678154, "learning_rate": 7.739870971457581e-07, "loss": 0.0006090738461352885, "step": 6519 }, { "epoch": 1.6926272066458983, "grad_norm": 1.6933934688568115, "learning_rate": 7.735348500669611e-07, "loss": 0.006622548680752516, "step": 6520 }, { "epoch": 1.6928868120456906, "grad_norm": 1.9299124479293823, "learning_rate": 7.730829523819477e-07, "loss": 0.012229658663272858, "step": 6521 }, { "epoch": 1.6931464174454829, "grad_norm": 3.3779876232147217, "learning_rate": 7.726314041706031e-07, "loss": 0.022638805210590363, "step": 6522 }, { "epoch": 1.6934060228452752, "grad_norm": 3.1913537979125977, "learning_rate": 7.721802055127509e-07, "loss": 0.009862751699984074, "step": 6523 }, { "epoch": 1.6936656282450675, "grad_norm": 1.0021547079086304, "learning_rate": 7.717293564881521e-07, "loss": 0.006610387936234474, "step": 6524 }, { "epoch": 1.6939252336448598, "grad_norm": 3.048490047454834, "learning_rate": 7.712788571765065e-07, "loss": 0.04230085760354996, "step": 6525 }, { "epoch": 1.6941848390446521, "grad_norm": 3.5705807209014893, "learning_rate": 7.708287076574523e-07, "loss": 0.012846507132053375, "step": 6526 }, { "epoch": 1.6944444444444444, "grad_norm": 2.4673256874084473, "learning_rate": 7.703789080105657e-07, "loss": 0.012727444991469383, "step": 6527 }, { "epoch": 1.6947040498442367, "grad_norm": 2.6922216415405273, "learning_rate": 7.699294583153605e-07, "loss": 0.02995895966887474, "step": 6528 }, { "epoch": 1.694963655244029, "grad_norm": 3.10754132270813, "learning_rate": 7.694803586512899e-07, "loss": 0.02723102644085884, "step": 6529 }, { "epoch": 1.6952232606438213, "grad_norm": 2.263672351837158, "learning_rate": 7.690316090977437e-07, "loss": 0.010910633020102978, "step": 6530 }, { "epoch": 1.6954828660436139, "grad_norm": 3.823530673980713, "learning_rate": 7.685832097340508e-07, "loss": 0.019407376646995544, "step": 6531 }, { "epoch": 1.6957424714434062, "grad_norm": 4.804922103881836, "learning_rate": 7.681351606394784e-07, "loss": 0.03408906236290932, "step": 6532 }, { "epoch": 1.6960020768431985, "grad_norm": 3.7214455604553223, "learning_rate": 7.676874618932302e-07, "loss": 0.033528633415699005, "step": 6533 }, { "epoch": 1.6962616822429908, "grad_norm": 4.043628692626953, "learning_rate": 7.672401135744503e-07, "loss": 0.029993902891874313, "step": 6534 }, { "epoch": 1.696521287642783, "grad_norm": 3.2379655838012695, "learning_rate": 7.667931157622192e-07, "loss": 0.010746853426098824, "step": 6535 }, { "epoch": 1.6967808930425754, "grad_norm": 3.1931397914886475, "learning_rate": 7.663464685355555e-07, "loss": 0.01798638142645359, "step": 6536 }, { "epoch": 1.6970404984423677, "grad_norm": 3.682373523712158, "learning_rate": 7.65900171973417e-07, "loss": 0.013868847861886024, "step": 6537 }, { "epoch": 1.69730010384216, "grad_norm": 3.1112983226776123, "learning_rate": 7.65454226154698e-07, "loss": 0.024776708334684372, "step": 6538 }, { "epoch": 1.6975597092419523, "grad_norm": 1.1418631076812744, "learning_rate": 7.650086311582314e-07, "loss": 0.002217551926150918, "step": 6539 }, { "epoch": 1.6978193146417446, "grad_norm": 3.2932186126708984, "learning_rate": 7.645633870627889e-07, "loss": 0.04008318483829498, "step": 6540 }, { "epoch": 1.698078920041537, "grad_norm": 1.9734019041061401, "learning_rate": 7.641184939470788e-07, "loss": 0.008177625946700573, "step": 6541 }, { "epoch": 1.6983385254413292, "grad_norm": 1.7189891338348389, "learning_rate": 7.636739518897488e-07, "loss": 0.004881901200860739, "step": 6542 }, { "epoch": 1.6985981308411215, "grad_norm": 2.92519474029541, "learning_rate": 7.632297609693828e-07, "loss": 0.016546033322811127, "step": 6543 }, { "epoch": 1.6988577362409139, "grad_norm": 2.9367363452911377, "learning_rate": 7.627859212645041e-07, "loss": 0.007555441465228796, "step": 6544 }, { "epoch": 1.6991173416407062, "grad_norm": 2.8347811698913574, "learning_rate": 7.623424328535738e-07, "loss": 0.008092363364994526, "step": 6545 }, { "epoch": 1.6993769470404985, "grad_norm": 3.4461450576782227, "learning_rate": 7.618992958149897e-07, "loss": 0.005291393026709557, "step": 6546 }, { "epoch": 1.6996365524402908, "grad_norm": 3.03998064994812, "learning_rate": 7.614565102270885e-07, "loss": 0.025440076366066933, "step": 6547 }, { "epoch": 1.699896157840083, "grad_norm": 2.9763870239257812, "learning_rate": 7.610140761681449e-07, "loss": 0.0045150816440582275, "step": 6548 }, { "epoch": 1.7001557632398754, "grad_norm": 3.135955810546875, "learning_rate": 7.605719937163705e-07, "loss": 0.011625543236732483, "step": 6549 }, { "epoch": 1.7004153686396677, "grad_norm": 3.7157771587371826, "learning_rate": 7.601302629499168e-07, "loss": 0.022042911499738693, "step": 6550 }, { "epoch": 1.70067497403946, "grad_norm": 2.18599009513855, "learning_rate": 7.596888839468696e-07, "loss": 0.005219435319304466, "step": 6551 }, { "epoch": 1.7009345794392523, "grad_norm": 4.508427619934082, "learning_rate": 7.592478567852557e-07, "loss": 0.021418005228042603, "step": 6552 }, { "epoch": 1.7011941848390446, "grad_norm": 3.7571189403533936, "learning_rate": 7.588071815430393e-07, "loss": 0.01378002017736435, "step": 6553 }, { "epoch": 1.701453790238837, "grad_norm": 3.795490264892578, "learning_rate": 7.583668582981202e-07, "loss": 0.034062378108501434, "step": 6554 }, { "epoch": 1.7017133956386292, "grad_norm": 3.705200433731079, "learning_rate": 7.579268871283385e-07, "loss": 0.04837518185377121, "step": 6555 }, { "epoch": 1.7019730010384215, "grad_norm": 1.6264441013336182, "learning_rate": 7.574872681114705e-07, "loss": 0.002166888676583767, "step": 6556 }, { "epoch": 1.7022326064382138, "grad_norm": 2.0188910961151123, "learning_rate": 7.570480013252313e-07, "loss": 0.012497052550315857, "step": 6557 }, { "epoch": 1.7024922118380061, "grad_norm": 3.8252177238464355, "learning_rate": 7.566090868472728e-07, "loss": 0.013839149847626686, "step": 6558 }, { "epoch": 1.7027518172377984, "grad_norm": 1.8449355363845825, "learning_rate": 7.56170524755185e-07, "loss": 0.00547043327242136, "step": 6559 }, { "epoch": 1.7030114226375908, "grad_norm": 3.089223861694336, "learning_rate": 7.557323151264958e-07, "loss": 0.009506243281066418, "step": 6560 }, { "epoch": 1.703271028037383, "grad_norm": 3.3353357315063477, "learning_rate": 7.552944580386709e-07, "loss": 0.021657731384038925, "step": 6561 }, { "epoch": 1.7035306334371754, "grad_norm": 0.1835448145866394, "learning_rate": 7.548569535691127e-07, "loss": 0.00037428553332574666, "step": 6562 }, { "epoch": 1.7037902388369677, "grad_norm": 0.7953515648841858, "learning_rate": 7.544198017951626e-07, "loss": 0.0017260992899537086, "step": 6563 }, { "epoch": 1.70404984423676, "grad_norm": 3.5113279819488525, "learning_rate": 7.539830027940986e-07, "loss": 0.008001181297004223, "step": 6564 }, { "epoch": 1.7043094496365523, "grad_norm": 1.8832628726959229, "learning_rate": 7.535465566431374e-07, "loss": 0.0029644975438714027, "step": 6565 }, { "epoch": 1.7045690550363446, "grad_norm": 3.498697280883789, "learning_rate": 7.531104634194323e-07, "loss": 0.04998214542865753, "step": 6566 }, { "epoch": 1.7048286604361371, "grad_norm": 2.6253437995910645, "learning_rate": 7.526747232000737e-07, "loss": 0.06006632745265961, "step": 6567 }, { "epoch": 1.7050882658359294, "grad_norm": 3.220494270324707, "learning_rate": 7.522393360620922e-07, "loss": 0.012476436793804169, "step": 6568 }, { "epoch": 1.7053478712357217, "grad_norm": 2.9501123428344727, "learning_rate": 7.518043020824535e-07, "loss": 0.017778100445866585, "step": 6569 }, { "epoch": 1.705607476635514, "grad_norm": 3.007998466491699, "learning_rate": 7.513696213380608e-07, "loss": 0.02035800740122795, "step": 6570 }, { "epoch": 1.7058670820353063, "grad_norm": 2.762369155883789, "learning_rate": 7.509352939057573e-07, "loss": 0.03056504763662815, "step": 6571 }, { "epoch": 1.7061266874350987, "grad_norm": 3.8262107372283936, "learning_rate": 7.505013198623207e-07, "loss": 0.02151675336062908, "step": 6572 }, { "epoch": 1.706386292834891, "grad_norm": 2.377413511276245, "learning_rate": 7.500676992844686e-07, "loss": 0.005626301746815443, "step": 6573 }, { "epoch": 1.7066458982346833, "grad_norm": 1.8737565279006958, "learning_rate": 7.49634432248855e-07, "loss": 0.0030990485101938248, "step": 6574 }, { "epoch": 1.7069055036344756, "grad_norm": 3.4892914295196533, "learning_rate": 7.492015188320711e-07, "loss": 0.030729614198207855, "step": 6575 }, { "epoch": 1.7071651090342679, "grad_norm": 1.9855594635009766, "learning_rate": 7.487689591106467e-07, "loss": 0.013360189273953438, "step": 6576 }, { "epoch": 1.7074247144340602, "grad_norm": 3.8169362545013428, "learning_rate": 7.483367531610483e-07, "loss": 0.025804320350289345, "step": 6577 }, { "epoch": 1.7076843198338525, "grad_norm": 3.2440788745880127, "learning_rate": 7.479049010596795e-07, "loss": 0.04757821559906006, "step": 6578 }, { "epoch": 1.707943925233645, "grad_norm": 3.6907854080200195, "learning_rate": 7.474734028828828e-07, "loss": 0.06458847969770432, "step": 6579 }, { "epoch": 1.7082035306334373, "grad_norm": 3.6698343753814697, "learning_rate": 7.470422587069368e-07, "loss": 0.05195442959666252, "step": 6580 }, { "epoch": 1.7084631360332296, "grad_norm": 2.332422971725464, "learning_rate": 7.466114686080576e-07, "loss": 0.02038254216313362, "step": 6581 }, { "epoch": 1.708722741433022, "grad_norm": 3.762707233428955, "learning_rate": 7.461810326623988e-07, "loss": 0.012397633865475655, "step": 6582 }, { "epoch": 1.7089823468328142, "grad_norm": 1.7522586584091187, "learning_rate": 7.457509509460522e-07, "loss": 0.0065737501718103886, "step": 6583 }, { "epoch": 1.7092419522326066, "grad_norm": 3.186176061630249, "learning_rate": 7.453212235350471e-07, "loss": 0.012157928198575974, "step": 6584 }, { "epoch": 1.7095015576323989, "grad_norm": 2.238546371459961, "learning_rate": 7.448918505053476e-07, "loss": 0.006195914000272751, "step": 6585 }, { "epoch": 1.7097611630321912, "grad_norm": 3.9307198524475098, "learning_rate": 7.444628319328586e-07, "loss": 0.010296422988176346, "step": 6586 }, { "epoch": 1.7100207684319835, "grad_norm": 4.357461929321289, "learning_rate": 7.440341678934197e-07, "loss": 0.01602371782064438, "step": 6587 }, { "epoch": 1.7102803738317758, "grad_norm": 3.172919273376465, "learning_rate": 7.436058584628098e-07, "loss": 0.02040778286755085, "step": 6588 }, { "epoch": 1.710539979231568, "grad_norm": 1.3713523149490356, "learning_rate": 7.431779037167437e-07, "loss": 0.0028329435735940933, "step": 6589 }, { "epoch": 1.7107995846313604, "grad_norm": 2.365938663482666, "learning_rate": 7.42750303730874e-07, "loss": 0.015546602196991444, "step": 6590 }, { "epoch": 1.7110591900311527, "grad_norm": 2.837078332901001, "learning_rate": 7.423230585807907e-07, "loss": 0.026363270357251167, "step": 6591 }, { "epoch": 1.711318795430945, "grad_norm": 3.3757004737854004, "learning_rate": 7.418961683420211e-07, "loss": 0.009520218707621098, "step": 6592 }, { "epoch": 1.7115784008307373, "grad_norm": 1.9403102397918701, "learning_rate": 7.414696330900291e-07, "loss": 0.007690707221627235, "step": 6593 }, { "epoch": 1.7118380062305296, "grad_norm": 2.935065507888794, "learning_rate": 7.410434529002172e-07, "loss": 0.05043267831206322, "step": 6594 }, { "epoch": 1.712097611630322, "grad_norm": 2.947601318359375, "learning_rate": 7.406176278479232e-07, "loss": 0.012836006470024586, "step": 6595 }, { "epoch": 1.7123572170301142, "grad_norm": 2.5192408561706543, "learning_rate": 7.401921580084244e-07, "loss": 0.020999547094106674, "step": 6596 }, { "epoch": 1.7126168224299065, "grad_norm": 3.0099642276763916, "learning_rate": 7.397670434569335e-07, "loss": 0.01486119069159031, "step": 6597 }, { "epoch": 1.7128764278296988, "grad_norm": 2.8167195320129395, "learning_rate": 7.393422842686008e-07, "loss": 0.010005250573158264, "step": 6598 }, { "epoch": 1.7131360332294912, "grad_norm": 2.8514769077301025, "learning_rate": 7.389178805185145e-07, "loss": 0.014783293940126896, "step": 6599 }, { "epoch": 1.7133956386292835, "grad_norm": 2.8276102542877197, "learning_rate": 7.384938322816995e-07, "loss": 0.05235888808965683, "step": 6600 }, { "epoch": 1.7136552440290758, "grad_norm": 3.068269968032837, "learning_rate": 7.380701396331168e-07, "loss": 0.0664195790886879, "step": 6601 }, { "epoch": 1.713914849428868, "grad_norm": 2.6156890392303467, "learning_rate": 7.376468026476669e-07, "loss": 0.017859751358628273, "step": 6602 }, { "epoch": 1.7141744548286604, "grad_norm": 4.402809143066406, "learning_rate": 7.372238214001852e-07, "loss": 0.034651875495910645, "step": 6603 }, { "epoch": 1.7144340602284527, "grad_norm": 2.1184778213500977, "learning_rate": 7.368011959654456e-07, "loss": 0.016988079994916916, "step": 6604 }, { "epoch": 1.714693665628245, "grad_norm": 0.2073780745267868, "learning_rate": 7.363789264181586e-07, "loss": 0.0008425647974945605, "step": 6605 }, { "epoch": 1.7149532710280373, "grad_norm": 4.033547878265381, "learning_rate": 7.359570128329711e-07, "loss": 0.004261649213731289, "step": 6606 }, { "epoch": 1.7152128764278296, "grad_norm": 2.253596305847168, "learning_rate": 7.355354552844693e-07, "loss": 0.010966073721647263, "step": 6607 }, { "epoch": 1.715472481827622, "grad_norm": 2.8706648349761963, "learning_rate": 7.351142538471728e-07, "loss": 0.04433171823620796, "step": 6608 }, { "epoch": 1.7157320872274142, "grad_norm": 1.168245792388916, "learning_rate": 7.346934085955416e-07, "loss": 0.009667986072599888, "step": 6609 }, { "epoch": 1.7159916926272065, "grad_norm": 3.1256256103515625, "learning_rate": 7.342729196039716e-07, "loss": 0.012911785393953323, "step": 6610 }, { "epoch": 1.7162512980269988, "grad_norm": 2.7646665573120117, "learning_rate": 7.338527869467954e-07, "loss": 0.05750744789838791, "step": 6611 }, { "epoch": 1.7165109034267911, "grad_norm": 3.083343505859375, "learning_rate": 7.33433010698283e-07, "loss": 0.02394505962729454, "step": 6612 }, { "epoch": 1.7167705088265834, "grad_norm": 3.1628270149230957, "learning_rate": 7.330135909326405e-07, "loss": 0.04290040582418442, "step": 6613 }, { "epoch": 1.7170301142263757, "grad_norm": 2.891624927520752, "learning_rate": 7.325945277240124e-07, "loss": 0.01273668184876442, "step": 6614 }, { "epoch": 1.7172897196261683, "grad_norm": 3.2984111309051514, "learning_rate": 7.3217582114648e-07, "loss": 0.027673548087477684, "step": 6615 }, { "epoch": 1.7175493250259606, "grad_norm": 3.085846424102783, "learning_rate": 7.317574712740595e-07, "loss": 0.028594806790351868, "step": 6616 }, { "epoch": 1.717808930425753, "grad_norm": 2.823049783706665, "learning_rate": 7.313394781807066e-07, "loss": 0.06785829365253448, "step": 6617 }, { "epoch": 1.7180685358255452, "grad_norm": 2.820676803588867, "learning_rate": 7.309218419403132e-07, "loss": 0.007762151770293713, "step": 6618 }, { "epoch": 1.7183281412253375, "grad_norm": 2.2301719188690186, "learning_rate": 7.305045626267074e-07, "loss": 0.005967581644654274, "step": 6619 }, { "epoch": 1.7185877466251298, "grad_norm": 2.6756086349487305, "learning_rate": 7.300876403136544e-07, "loss": 0.010314048267900944, "step": 6620 }, { "epoch": 1.7188473520249221, "grad_norm": 3.4244682788848877, "learning_rate": 7.296710750748563e-07, "loss": 0.05672452226281166, "step": 6621 }, { "epoch": 1.7191069574247144, "grad_norm": 3.0579748153686523, "learning_rate": 7.292548669839531e-07, "loss": 0.006689727772027254, "step": 6622 }, { "epoch": 1.7193665628245067, "grad_norm": 3.004546642303467, "learning_rate": 7.288390161145209e-07, "loss": 0.0071349721401929855, "step": 6623 }, { "epoch": 1.719626168224299, "grad_norm": 1.8343489170074463, "learning_rate": 7.284235225400716e-07, "loss": 0.00925525277853012, "step": 6624 }, { "epoch": 1.7198857736240913, "grad_norm": 1.7950550317764282, "learning_rate": 7.280083863340559e-07, "loss": 0.00545080192387104, "step": 6625 }, { "epoch": 1.7201453790238836, "grad_norm": 3.8107941150665283, "learning_rate": 7.275936075698597e-07, "loss": 0.01740892231464386, "step": 6626 }, { "epoch": 1.7204049844236762, "grad_norm": 0.9868530631065369, "learning_rate": 7.271791863208067e-07, "loss": 0.0066360775381326675, "step": 6627 }, { "epoch": 1.7206645898234685, "grad_norm": 3.7897393703460693, "learning_rate": 7.267651226601577e-07, "loss": 0.055849507451057434, "step": 6628 }, { "epoch": 1.7209241952232608, "grad_norm": 1.2760803699493408, "learning_rate": 7.263514166611083e-07, "loss": 0.006538380868732929, "step": 6629 }, { "epoch": 1.721183800623053, "grad_norm": 3.5678186416625977, "learning_rate": 7.259380683967936e-07, "loss": 0.060901544988155365, "step": 6630 }, { "epoch": 1.7214434060228454, "grad_norm": 0.9815926551818848, "learning_rate": 7.255250779402835e-07, "loss": 0.0020378432236611843, "step": 6631 }, { "epoch": 1.7217030114226377, "grad_norm": 2.4183261394500732, "learning_rate": 7.251124453645849e-07, "loss": 0.006964255589991808, "step": 6632 }, { "epoch": 1.72196261682243, "grad_norm": 3.0078976154327393, "learning_rate": 7.247001707426424e-07, "loss": 0.029397744685411453, "step": 6633 }, { "epoch": 1.7222222222222223, "grad_norm": 3.4990451335906982, "learning_rate": 7.242882541473367e-07, "loss": 0.013356601819396019, "step": 6634 }, { "epoch": 1.7224818276220146, "grad_norm": 1.1878929138183594, "learning_rate": 7.238766956514844e-07, "loss": 0.0028451597318053246, "step": 6635 }, { "epoch": 1.722741433021807, "grad_norm": 5.048838138580322, "learning_rate": 7.23465495327841e-07, "loss": 0.006069720257073641, "step": 6636 }, { "epoch": 1.7230010384215992, "grad_norm": 2.850024938583374, "learning_rate": 7.230546532490958e-07, "loss": 0.012046624906361103, "step": 6637 }, { "epoch": 1.7232606438213915, "grad_norm": 2.7402048110961914, "learning_rate": 7.226441694878774e-07, "loss": 0.047145046293735504, "step": 6638 }, { "epoch": 1.7235202492211839, "grad_norm": 3.41383957862854, "learning_rate": 7.222340441167498e-07, "loss": 0.026373714208602905, "step": 6639 }, { "epoch": 1.7237798546209762, "grad_norm": 4.139690399169922, "learning_rate": 7.218242772082131e-07, "loss": 0.009440688416361809, "step": 6640 }, { "epoch": 1.7240394600207685, "grad_norm": 1.4290059804916382, "learning_rate": 7.214148688347055e-07, "loss": 0.008016031235456467, "step": 6641 }, { "epoch": 1.7242990654205608, "grad_norm": 3.703932523727417, "learning_rate": 7.210058190686003e-07, "loss": 0.023080900311470032, "step": 6642 }, { "epoch": 1.724558670820353, "grad_norm": 4.242494583129883, "learning_rate": 7.205971279822084e-07, "loss": 0.020109646022319794, "step": 6643 }, { "epoch": 1.7248182762201454, "grad_norm": 3.1225714683532715, "learning_rate": 7.201887956477774e-07, "loss": 0.03289896994829178, "step": 6644 }, { "epoch": 1.7250778816199377, "grad_norm": 2.709897518157959, "learning_rate": 7.197808221374904e-07, "loss": 0.012979939579963684, "step": 6645 }, { "epoch": 1.72533748701973, "grad_norm": 0.4980267584323883, "learning_rate": 7.193732075234693e-07, "loss": 0.001501723425462842, "step": 6646 }, { "epoch": 1.7255970924195223, "grad_norm": 2.9223859310150146, "learning_rate": 7.189659518777687e-07, "loss": 0.012917038053274155, "step": 6647 }, { "epoch": 1.7258566978193146, "grad_norm": 0.7873444557189941, "learning_rate": 7.185590552723834e-07, "loss": 0.0041123428381979465, "step": 6648 }, { "epoch": 1.726116303219107, "grad_norm": 1.3436779975891113, "learning_rate": 7.181525177792436e-07, "loss": 0.003121880814433098, "step": 6649 }, { "epoch": 1.7263759086188992, "grad_norm": 2.89882493019104, "learning_rate": 7.177463394702157e-07, "loss": 0.010377001017332077, "step": 6650 }, { "epoch": 1.7266355140186915, "grad_norm": 1.9910063743591309, "learning_rate": 7.173405204171028e-07, "loss": 0.008170676417648792, "step": 6651 }, { "epoch": 1.7268951194184838, "grad_norm": 1.1514396667480469, "learning_rate": 7.169350606916434e-07, "loss": 0.005223106127232313, "step": 6652 }, { "epoch": 1.7271547248182761, "grad_norm": 1.9446513652801514, "learning_rate": 7.165299603655141e-07, "loss": 0.01588950678706169, "step": 6653 }, { "epoch": 1.7274143302180685, "grad_norm": 3.663466691970825, "learning_rate": 7.161252195103286e-07, "loss": 0.015783511102199554, "step": 6654 }, { "epoch": 1.7276739356178608, "grad_norm": 2.6324336528778076, "learning_rate": 7.157208381976338e-07, "loss": 0.010842731222510338, "step": 6655 }, { "epoch": 1.727933541017653, "grad_norm": 3.246001958847046, "learning_rate": 7.153168164989166e-07, "loss": 0.07157844305038452, "step": 6656 }, { "epoch": 1.7281931464174454, "grad_norm": 2.7978692054748535, "learning_rate": 7.149131544855976e-07, "loss": 0.006991298869252205, "step": 6657 }, { "epoch": 1.7284527518172377, "grad_norm": 4.3332695960998535, "learning_rate": 7.145098522290362e-07, "loss": 0.012540281750261784, "step": 6658 }, { "epoch": 1.72871235721703, "grad_norm": 3.4891955852508545, "learning_rate": 7.141069098005259e-07, "loss": 0.022404352203011513, "step": 6659 }, { "epoch": 1.7289719626168223, "grad_norm": 2.632378101348877, "learning_rate": 7.137043272712981e-07, "loss": 0.0080064432695508, "step": 6660 }, { "epoch": 1.7292315680166146, "grad_norm": 3.808530330657959, "learning_rate": 7.133021047125206e-07, "loss": 0.019568881019949913, "step": 6661 }, { "epoch": 1.729491173416407, "grad_norm": 2.698237180709839, "learning_rate": 7.129002421952968e-07, "loss": 0.00613414216786623, "step": 6662 }, { "epoch": 1.7297507788161994, "grad_norm": 1.7279105186462402, "learning_rate": 7.124987397906661e-07, "loss": 0.0035234796814620495, "step": 6663 }, { "epoch": 1.7300103842159917, "grad_norm": 3.083264112472534, "learning_rate": 7.120975975696064e-07, "loss": 0.006871948484331369, "step": 6664 }, { "epoch": 1.730269989615784, "grad_norm": 4.228390693664551, "learning_rate": 7.116968156030297e-07, "loss": 0.032303377985954285, "step": 6665 }, { "epoch": 1.7305295950155763, "grad_norm": 2.1593825817108154, "learning_rate": 7.112963939617845e-07, "loss": 0.007485940121114254, "step": 6666 }, { "epoch": 1.7307892004153687, "grad_norm": 1.9973362684249878, "learning_rate": 7.108963327166571e-07, "loss": 0.004317970015108585, "step": 6667 }, { "epoch": 1.731048805815161, "grad_norm": 1.8187450170516968, "learning_rate": 7.104966319383684e-07, "loss": 0.012403477914631367, "step": 6668 }, { "epoch": 1.7313084112149533, "grad_norm": 3.951836585998535, "learning_rate": 7.100972916975771e-07, "loss": 0.016332661733031273, "step": 6669 }, { "epoch": 1.7315680166147456, "grad_norm": 3.582843542098999, "learning_rate": 7.096983120648771e-07, "loss": 0.012747138738632202, "step": 6670 }, { "epoch": 1.7318276220145379, "grad_norm": 2.849862813949585, "learning_rate": 7.092996931107983e-07, "loss": 0.025616541504859924, "step": 6671 }, { "epoch": 1.7320872274143302, "grad_norm": 0.22848843038082123, "learning_rate": 7.089014349058085e-07, "loss": 0.0006748751038685441, "step": 6672 }, { "epoch": 1.7323468328141225, "grad_norm": 2.6518678665161133, "learning_rate": 7.085035375203098e-07, "loss": 0.011755448766052723, "step": 6673 }, { "epoch": 1.7326064382139148, "grad_norm": 2.943934202194214, "learning_rate": 7.081060010246415e-07, "loss": 0.06989133358001709, "step": 6674 }, { "epoch": 1.7328660436137073, "grad_norm": 2.7858283519744873, "learning_rate": 7.077088254890792e-07, "loss": 0.05080224201083183, "step": 6675 }, { "epoch": 1.7331256490134996, "grad_norm": 4.042202949523926, "learning_rate": 7.07312010983834e-07, "loss": 0.024070819839835167, "step": 6676 }, { "epoch": 1.733385254413292, "grad_norm": 1.479217290878296, "learning_rate": 7.069155575790547e-07, "loss": 0.007702617906033993, "step": 6677 }, { "epoch": 1.7336448598130842, "grad_norm": 2.3025591373443604, "learning_rate": 7.065194653448234e-07, "loss": 0.004485404584556818, "step": 6678 }, { "epoch": 1.7339044652128766, "grad_norm": 0.20069104433059692, "learning_rate": 7.061237343511613e-07, "loss": 0.0006803389405831695, "step": 6679 }, { "epoch": 1.7341640706126689, "grad_norm": 3.648226022720337, "learning_rate": 7.057283646680249e-07, "loss": 0.07115595042705536, "step": 6680 }, { "epoch": 1.7344236760124612, "grad_norm": 3.8987791538238525, "learning_rate": 7.05333356365306e-07, "loss": 0.021808845922350883, "step": 6681 }, { "epoch": 1.7346832814122535, "grad_norm": 3.1996445655822754, "learning_rate": 7.04938709512833e-07, "loss": 0.0600089505314827, "step": 6682 }, { "epoch": 1.7349428868120458, "grad_norm": 3.809582233428955, "learning_rate": 7.045444241803699e-07, "loss": 0.03536929935216904, "step": 6683 }, { "epoch": 1.735202492211838, "grad_norm": 3.0830318927764893, "learning_rate": 7.04150500437618e-07, "loss": 0.031897615641355515, "step": 6684 }, { "epoch": 1.7354620976116304, "grad_norm": 4.281317710876465, "learning_rate": 7.037569383542147e-07, "loss": 0.04047711193561554, "step": 6685 }, { "epoch": 1.7357217030114227, "grad_norm": 3.1510207653045654, "learning_rate": 7.033637379997308e-07, "loss": 0.013242918998003006, "step": 6686 }, { "epoch": 1.735981308411215, "grad_norm": 3.6655352115631104, "learning_rate": 7.029708994436771e-07, "loss": 0.01605059951543808, "step": 6687 }, { "epoch": 1.7362409138110073, "grad_norm": 1.6060118675231934, "learning_rate": 7.02578422755497e-07, "loss": 0.005458214785903692, "step": 6688 }, { "epoch": 1.7365005192107996, "grad_norm": 4.0309953689575195, "learning_rate": 7.021863080045726e-07, "loss": 0.00710046011954546, "step": 6689 }, { "epoch": 1.736760124610592, "grad_norm": 2.5108795166015625, "learning_rate": 7.017945552602201e-07, "loss": 0.0023537587840110064, "step": 6690 }, { "epoch": 1.7370197300103842, "grad_norm": 3.7861196994781494, "learning_rate": 7.014031645916923e-07, "loss": 0.02507307380437851, "step": 6691 }, { "epoch": 1.7372793354101765, "grad_norm": 1.9272336959838867, "learning_rate": 7.010121360681787e-07, "loss": 0.0029343829955905676, "step": 6692 }, { "epoch": 1.7375389408099688, "grad_norm": 3.1610028743743896, "learning_rate": 7.006214697588036e-07, "loss": 0.01517052948474884, "step": 6693 }, { "epoch": 1.7377985462097612, "grad_norm": 2.6449337005615234, "learning_rate": 7.002311657326279e-07, "loss": 0.003327274927869439, "step": 6694 }, { "epoch": 1.7380581516095535, "grad_norm": 3.3916032314300537, "learning_rate": 6.99841224058649e-07, "loss": 0.03282388299703598, "step": 6695 }, { "epoch": 1.7383177570093458, "grad_norm": 2.8945271968841553, "learning_rate": 6.994516448057993e-07, "loss": 0.012856882065534592, "step": 6696 }, { "epoch": 1.738577362409138, "grad_norm": 3.5227901935577393, "learning_rate": 6.99062428042947e-07, "loss": 0.03399036079645157, "step": 6697 }, { "epoch": 1.7388369678089304, "grad_norm": 3.516752004623413, "learning_rate": 6.986735738388976e-07, "loss": 0.010832247324287891, "step": 6698 }, { "epoch": 1.7390965732087227, "grad_norm": 3.01906418800354, "learning_rate": 6.982850822623907e-07, "loss": 0.019694257527589798, "step": 6699 }, { "epoch": 1.739356178608515, "grad_norm": 3.6812689304351807, "learning_rate": 6.978969533821036e-07, "loss": 0.007711370941251516, "step": 6700 }, { "epoch": 1.7396157840083073, "grad_norm": 4.343751430511475, "learning_rate": 6.975091872666484e-07, "loss": 0.016630183905363083, "step": 6701 }, { "epoch": 1.7398753894080996, "grad_norm": 3.820472002029419, "learning_rate": 6.971217839845726e-07, "loss": 0.06302672624588013, "step": 6702 }, { "epoch": 1.740134994807892, "grad_norm": 3.0890328884124756, "learning_rate": 6.96734743604361e-07, "loss": 0.014193685725331306, "step": 6703 }, { "epoch": 1.7403946002076842, "grad_norm": 0.28570854663848877, "learning_rate": 6.963480661944335e-07, "loss": 0.0006409839261323214, "step": 6704 }, { "epoch": 1.7406542056074765, "grad_norm": 4.037911415100098, "learning_rate": 6.959617518231448e-07, "loss": 0.021594934165477753, "step": 6705 }, { "epoch": 1.7409138110072688, "grad_norm": 3.7465784549713135, "learning_rate": 6.955758005587876e-07, "loss": 0.03185145929455757, "step": 6706 }, { "epoch": 1.7411734164070611, "grad_norm": 3.2385621070861816, "learning_rate": 6.951902124695885e-07, "loss": 0.004296362400054932, "step": 6707 }, { "epoch": 1.7414330218068534, "grad_norm": 3.2632739543914795, "learning_rate": 6.948049876237116e-07, "loss": 0.02071473002433777, "step": 6708 }, { "epoch": 1.7416926272066457, "grad_norm": 1.8980467319488525, "learning_rate": 6.944201260892544e-07, "loss": 0.005171028431504965, "step": 6709 }, { "epoch": 1.741952232606438, "grad_norm": 3.1955087184906006, "learning_rate": 6.940356279342524e-07, "loss": 0.05338333547115326, "step": 6710 }, { "epoch": 1.7422118380062306, "grad_norm": 3.6195435523986816, "learning_rate": 6.936514932266763e-07, "loss": 0.05620258301496506, "step": 6711 }, { "epoch": 1.742471443406023, "grad_norm": 2.7295804023742676, "learning_rate": 6.932677220344321e-07, "loss": 0.00943269208073616, "step": 6712 }, { "epoch": 1.7427310488058152, "grad_norm": 2.495143413543701, "learning_rate": 6.928843144253619e-07, "loss": 0.006940901279449463, "step": 6713 }, { "epoch": 1.7429906542056075, "grad_norm": 0.42308226227760315, "learning_rate": 6.925012704672426e-07, "loss": 0.0008499260293319821, "step": 6714 }, { "epoch": 1.7432502596053998, "grad_norm": 3.6880979537963867, "learning_rate": 6.921185902277882e-07, "loss": 0.01955091394484043, "step": 6715 }, { "epoch": 1.7435098650051921, "grad_norm": 4.202269554138184, "learning_rate": 6.917362737746487e-07, "loss": 0.008313152939081192, "step": 6716 }, { "epoch": 1.7437694704049844, "grad_norm": 4.366441249847412, "learning_rate": 6.913543211754075e-07, "loss": 0.01611809805035591, "step": 6717 }, { "epoch": 1.7440290758047767, "grad_norm": 3.4993772506713867, "learning_rate": 6.909727324975854e-07, "loss": 0.012711777351796627, "step": 6718 }, { "epoch": 1.744288681204569, "grad_norm": 3.4731433391571045, "learning_rate": 6.905915078086392e-07, "loss": 0.03756825625896454, "step": 6719 }, { "epoch": 1.7445482866043613, "grad_norm": 3.5231101512908936, "learning_rate": 6.902106471759595e-07, "loss": 0.02534506283700466, "step": 6720 }, { "epoch": 1.7448078920041536, "grad_norm": 0.6310194134712219, "learning_rate": 6.898301506668749e-07, "loss": 0.0016158393118530512, "step": 6721 }, { "epoch": 1.745067497403946, "grad_norm": 4.379998207092285, "learning_rate": 6.894500183486478e-07, "loss": 0.002767335856333375, "step": 6722 }, { "epoch": 1.7453271028037385, "grad_norm": 1.9568932056427002, "learning_rate": 6.890702502884775e-07, "loss": 0.004655622411519289, "step": 6723 }, { "epoch": 1.7455867082035308, "grad_norm": 1.7693287134170532, "learning_rate": 6.886908465534975e-07, "loss": 0.004552572499960661, "step": 6724 }, { "epoch": 1.745846313603323, "grad_norm": 5.323751926422119, "learning_rate": 6.883118072107783e-07, "loss": 0.006946527864784002, "step": 6725 }, { "epoch": 1.7461059190031154, "grad_norm": 1.9705495834350586, "learning_rate": 6.879331323273253e-07, "loss": 0.0015906556509435177, "step": 6726 }, { "epoch": 1.7463655244029077, "grad_norm": 0.6972590684890747, "learning_rate": 6.875548219700792e-07, "loss": 0.005150377284735441, "step": 6727 }, { "epoch": 1.7466251298027, "grad_norm": 0.8895103335380554, "learning_rate": 6.871768762059167e-07, "loss": 0.0012771316105499864, "step": 6728 }, { "epoch": 1.7468847352024923, "grad_norm": 2.7333948612213135, "learning_rate": 6.867992951016506e-07, "loss": 0.021767495200037956, "step": 6729 }, { "epoch": 1.7471443406022846, "grad_norm": 1.8312541246414185, "learning_rate": 6.864220787240275e-07, "loss": 0.01571844145655632, "step": 6730 }, { "epoch": 1.747403946002077, "grad_norm": 2.7196362018585205, "learning_rate": 6.860452271397317e-07, "loss": 0.041942231357097626, "step": 6731 }, { "epoch": 1.7476635514018692, "grad_norm": 3.697737455368042, "learning_rate": 6.856687404153814e-07, "loss": 0.03426438570022583, "step": 6732 }, { "epoch": 1.7479231568016615, "grad_norm": 3.168400287628174, "learning_rate": 6.852926186175305e-07, "loss": 0.049306049942970276, "step": 6733 }, { "epoch": 1.7481827622014539, "grad_norm": 3.5022430419921875, "learning_rate": 6.849168618126695e-07, "loss": 0.01560159120708704, "step": 6734 }, { "epoch": 1.7484423676012462, "grad_norm": 2.724135637283325, "learning_rate": 6.845414700672231e-07, "loss": 0.010084942914545536, "step": 6735 }, { "epoch": 1.7487019730010385, "grad_norm": 0.30058765411376953, "learning_rate": 6.841664434475518e-07, "loss": 0.0006061226013116539, "step": 6736 }, { "epoch": 1.7489615784008308, "grad_norm": 3.473623275756836, "learning_rate": 6.837917820199525e-07, "loss": 0.007856374606490135, "step": 6737 }, { "epoch": 1.749221183800623, "grad_norm": 3.6463375091552734, "learning_rate": 6.834174858506555e-07, "loss": 0.007113378960639238, "step": 6738 }, { "epoch": 1.7494807892004154, "grad_norm": 3.4209110736846924, "learning_rate": 6.830435550058297e-07, "loss": 0.04645133018493652, "step": 6739 }, { "epoch": 1.7497403946002077, "grad_norm": 4.533265113830566, "learning_rate": 6.826699895515751e-07, "loss": 0.009557580575346947, "step": 6740 }, { "epoch": 1.75, "grad_norm": 3.339768886566162, "learning_rate": 6.822967895539311e-07, "loss": 0.028034649789333344, "step": 6741 }, { "epoch": 1.7502596053997923, "grad_norm": 2.233752965927124, "learning_rate": 6.81923955078871e-07, "loss": 0.003939586225897074, "step": 6742 }, { "epoch": 1.7505192107995846, "grad_norm": 2.9432528018951416, "learning_rate": 6.815514861923028e-07, "loss": 0.0023201650474220514, "step": 6743 }, { "epoch": 1.750778816199377, "grad_norm": 2.986764669418335, "learning_rate": 6.811793829600707e-07, "loss": 0.019790902733802795, "step": 6744 }, { "epoch": 1.7510384215991692, "grad_norm": 2.938213586807251, "learning_rate": 6.808076454479534e-07, "loss": 0.019539792090654373, "step": 6745 }, { "epoch": 1.7512980269989615, "grad_norm": 2.7746734619140625, "learning_rate": 6.80436273721667e-07, "loss": 0.040044765919446945, "step": 6746 }, { "epoch": 1.7515576323987538, "grad_norm": 3.3882904052734375, "learning_rate": 6.800652678468604e-07, "loss": 0.025070955976843834, "step": 6747 }, { "epoch": 1.7518172377985461, "grad_norm": 2.0236904621124268, "learning_rate": 6.79694627889119e-07, "loss": 0.005746589973568916, "step": 6748 }, { "epoch": 1.7520768431983385, "grad_norm": 2.9200503826141357, "learning_rate": 6.793243539139636e-07, "loss": 0.06163045018911362, "step": 6749 }, { "epoch": 1.7523364485981308, "grad_norm": 1.9388186931610107, "learning_rate": 6.789544459868508e-07, "loss": 0.011255376040935516, "step": 6750 }, { "epoch": 1.752596053997923, "grad_norm": 2.527268171310425, "learning_rate": 6.785849041731705e-07, "loss": 0.027286987751722336, "step": 6751 }, { "epoch": 1.7528556593977154, "grad_norm": 2.548001289367676, "learning_rate": 6.782157285382508e-07, "loss": 0.01591774821281433, "step": 6752 }, { "epoch": 1.7531152647975077, "grad_norm": 3.538219928741455, "learning_rate": 6.778469191473519e-07, "loss": 0.02420109324157238, "step": 6753 }, { "epoch": 1.7533748701973, "grad_norm": 3.1503400802612305, "learning_rate": 6.774784760656725e-07, "loss": 0.06582622230052948, "step": 6754 }, { "epoch": 1.7536344755970923, "grad_norm": 5.106628894805908, "learning_rate": 6.771103993583437e-07, "loss": 0.013836177065968513, "step": 6755 }, { "epoch": 1.7538940809968846, "grad_norm": 1.8538472652435303, "learning_rate": 6.767426890904332e-07, "loss": 0.006780738942325115, "step": 6756 }, { "epoch": 1.754153686396677, "grad_norm": 3.563234329223633, "learning_rate": 6.763753453269445e-07, "loss": 0.012356158345937729, "step": 6757 }, { "epoch": 1.7544132917964692, "grad_norm": 3.288883686065674, "learning_rate": 6.760083681328147e-07, "loss": 0.04507237672805786, "step": 6758 }, { "epoch": 1.7546728971962615, "grad_norm": 3.469593048095703, "learning_rate": 6.756417575729171e-07, "loss": 0.022608526051044464, "step": 6759 }, { "epoch": 1.754932502596054, "grad_norm": 3.017664670944214, "learning_rate": 6.752755137120606e-07, "loss": 0.015882890671491623, "step": 6760 }, { "epoch": 1.7551921079958464, "grad_norm": 4.334394931793213, "learning_rate": 6.749096366149881e-07, "loss": 0.016031693667173386, "step": 6761 }, { "epoch": 1.7554517133956387, "grad_norm": 2.4617652893066406, "learning_rate": 6.745441263463791e-07, "loss": 0.009032823145389557, "step": 6762 }, { "epoch": 1.755711318795431, "grad_norm": 1.9846141338348389, "learning_rate": 6.741789829708467e-07, "loss": 0.007500392850488424, "step": 6763 }, { "epoch": 1.7559709241952233, "grad_norm": 1.332529902458191, "learning_rate": 6.738142065529396e-07, "loss": 0.008433151990175247, "step": 6764 }, { "epoch": 1.7562305295950156, "grad_norm": 2.9491255283355713, "learning_rate": 6.73449797157143e-07, "loss": 0.02632039226591587, "step": 6765 }, { "epoch": 1.7564901349948079, "grad_norm": 4.9016242027282715, "learning_rate": 6.730857548478753e-07, "loss": 0.013203748501837254, "step": 6766 }, { "epoch": 1.7567497403946002, "grad_norm": 3.9513986110687256, "learning_rate": 6.727220796894913e-07, "loss": 0.031347326934337616, "step": 6767 }, { "epoch": 1.7570093457943925, "grad_norm": 3.4698598384857178, "learning_rate": 6.723587717462801e-07, "loss": 0.013443847186863422, "step": 6768 }, { "epoch": 1.7572689511941848, "grad_norm": 0.5600963830947876, "learning_rate": 6.719958310824662e-07, "loss": 0.0016395627753809094, "step": 6769 }, { "epoch": 1.757528556593977, "grad_norm": 2.253600835800171, "learning_rate": 6.7163325776221e-07, "loss": 0.015538196079432964, "step": 6770 }, { "epoch": 1.7577881619937694, "grad_norm": 2.5643861293792725, "learning_rate": 6.712710518496049e-07, "loss": 0.007119936402887106, "step": 6771 }, { "epoch": 1.758047767393562, "grad_norm": 3.9596595764160156, "learning_rate": 6.709092134086813e-07, "loss": 0.010771846398711205, "step": 6772 }, { "epoch": 1.7583073727933543, "grad_norm": 3.7533490657806396, "learning_rate": 6.705477425034046e-07, "loss": 0.046891115605831146, "step": 6773 }, { "epoch": 1.7585669781931466, "grad_norm": 2.6345183849334717, "learning_rate": 6.701866391976733e-07, "loss": 0.04839062690734863, "step": 6774 }, { "epoch": 1.7588265835929389, "grad_norm": 0.9939109086990356, "learning_rate": 6.698259035553228e-07, "loss": 0.002884508343413472, "step": 6775 }, { "epoch": 1.7590861889927312, "grad_norm": 4.001338958740234, "learning_rate": 6.694655356401226e-07, "loss": 0.0071336934342980385, "step": 6776 }, { "epoch": 1.7593457943925235, "grad_norm": 3.3931281566619873, "learning_rate": 6.691055355157783e-07, "loss": 0.036548178642988205, "step": 6777 }, { "epoch": 1.7596053997923158, "grad_norm": 2.4372150897979736, "learning_rate": 6.687459032459289e-07, "loss": 0.01058248896151781, "step": 6778 }, { "epoch": 1.759865005192108, "grad_norm": 3.7418463230133057, "learning_rate": 6.683866388941495e-07, "loss": 0.011854610405862331, "step": 6779 }, { "epoch": 1.7601246105919004, "grad_norm": 1.724069595336914, "learning_rate": 6.680277425239496e-07, "loss": 0.006678570993244648, "step": 6780 }, { "epoch": 1.7603842159916927, "grad_norm": 2.6168508529663086, "learning_rate": 6.676692141987744e-07, "loss": 0.0070467982441186905, "step": 6781 }, { "epoch": 1.760643821391485, "grad_norm": 3.4557511806488037, "learning_rate": 6.673110539820023e-07, "loss": 0.02511872723698616, "step": 6782 }, { "epoch": 1.7609034267912773, "grad_norm": 5.168121337890625, "learning_rate": 6.669532619369493e-07, "loss": 0.019368598237633705, "step": 6783 }, { "epoch": 1.7611630321910696, "grad_norm": 3.336196184158325, "learning_rate": 6.665958381268634e-07, "loss": 0.04718714952468872, "step": 6784 }, { "epoch": 1.761422637590862, "grad_norm": 3.5389528274536133, "learning_rate": 6.662387826149303e-07, "loss": 0.014479748904705048, "step": 6785 }, { "epoch": 1.7616822429906542, "grad_norm": 3.4235668182373047, "learning_rate": 6.658820954642687e-07, "loss": 0.02130476012825966, "step": 6786 }, { "epoch": 1.7619418483904465, "grad_norm": 2.965144157409668, "learning_rate": 6.655257767379321e-07, "loss": 0.04178417846560478, "step": 6787 }, { "epoch": 1.7622014537902388, "grad_norm": 3.6413681507110596, "learning_rate": 6.651698264989105e-07, "loss": 0.018269900232553482, "step": 6788 }, { "epoch": 1.7624610591900312, "grad_norm": 3.143186569213867, "learning_rate": 6.648142448101274e-07, "loss": 0.0353337787091732, "step": 6789 }, { "epoch": 1.7627206645898235, "grad_norm": 2.011137008666992, "learning_rate": 6.644590317344409e-07, "loss": 0.005368477664887905, "step": 6790 }, { "epoch": 1.7629802699896158, "grad_norm": 3.0601933002471924, "learning_rate": 6.641041873346453e-07, "loss": 0.01896212249994278, "step": 6791 }, { "epoch": 1.763239875389408, "grad_norm": 1.7265394926071167, "learning_rate": 6.637497116734686e-07, "loss": 0.005221987143158913, "step": 6792 }, { "epoch": 1.7634994807892004, "grad_norm": 3.3566877841949463, "learning_rate": 6.633956048135741e-07, "loss": 0.026387225836515427, "step": 6793 }, { "epoch": 1.7637590861889927, "grad_norm": 3.028756618499756, "learning_rate": 6.630418668175598e-07, "loss": 0.016718151047825813, "step": 6794 }, { "epoch": 1.764018691588785, "grad_norm": 3.2851858139038086, "learning_rate": 6.626884977479582e-07, "loss": 0.009526827372610569, "step": 6795 }, { "epoch": 1.7642782969885773, "grad_norm": 1.5602900981903076, "learning_rate": 6.623354976672376e-07, "loss": 0.002374837873503566, "step": 6796 }, { "epoch": 1.7645379023883696, "grad_norm": 1.8206210136413574, "learning_rate": 6.619828666377998e-07, "loss": 0.006247790530323982, "step": 6797 }, { "epoch": 1.764797507788162, "grad_norm": 3.5871548652648926, "learning_rate": 6.616306047219813e-07, "loss": 0.05282904580235481, "step": 6798 }, { "epoch": 1.7650571131879542, "grad_norm": 4.4891462326049805, "learning_rate": 6.612787119820552e-07, "loss": 0.021915223449468613, "step": 6799 }, { "epoch": 1.7653167185877465, "grad_norm": 1.7495943307876587, "learning_rate": 6.609271884802272e-07, "loss": 0.003150533651933074, "step": 6800 }, { "epoch": 1.7655763239875388, "grad_norm": 2.184279441833496, "learning_rate": 6.605760342786391e-07, "loss": 0.012974824756383896, "step": 6801 }, { "epoch": 1.7658359293873311, "grad_norm": 3.0743956565856934, "learning_rate": 6.602252494393663e-07, "loss": 0.010689778253436089, "step": 6802 }, { "epoch": 1.7660955347871234, "grad_norm": 0.10291801393032074, "learning_rate": 6.598748340244198e-07, "loss": 0.0004155303759034723, "step": 6803 }, { "epoch": 1.7663551401869158, "grad_norm": 0.7934589385986328, "learning_rate": 6.595247880957459e-07, "loss": 0.0012173751601949334, "step": 6804 }, { "epoch": 1.766614745586708, "grad_norm": 4.138736248016357, "learning_rate": 6.59175111715223e-07, "loss": 0.04869571700692177, "step": 6805 }, { "epoch": 1.7668743509865004, "grad_norm": 3.6069021224975586, "learning_rate": 6.588258049446673e-07, "loss": 0.01823591999709606, "step": 6806 }, { "epoch": 1.7671339563862927, "grad_norm": 2.8307745456695557, "learning_rate": 6.584768678458274e-07, "loss": 0.009510008618235588, "step": 6807 }, { "epoch": 1.7673935617860852, "grad_norm": 3.6334586143493652, "learning_rate": 6.58128300480388e-07, "loss": 0.052397988736629486, "step": 6808 }, { "epoch": 1.7676531671858775, "grad_norm": 4.0190019607543945, "learning_rate": 6.577801029099676e-07, "loss": 0.02964167669415474, "step": 6809 }, { "epoch": 1.7679127725856698, "grad_norm": 3.224714756011963, "learning_rate": 6.574322751961193e-07, "loss": 0.006470905616879463, "step": 6810 }, { "epoch": 1.7681723779854621, "grad_norm": 2.4165775775909424, "learning_rate": 6.570848174003316e-07, "loss": 0.013051712885499, "step": 6811 }, { "epoch": 1.7684319833852544, "grad_norm": 4.167196750640869, "learning_rate": 6.567377295840267e-07, "loss": 0.02665158174932003, "step": 6812 }, { "epoch": 1.7686915887850467, "grad_norm": 3.234539270401001, "learning_rate": 6.563910118085616e-07, "loss": 0.013163525611162186, "step": 6813 }, { "epoch": 1.768951194184839, "grad_norm": 3.506413459777832, "learning_rate": 6.560446641352288e-07, "loss": 0.011179955676198006, "step": 6814 }, { "epoch": 1.7692107995846313, "grad_norm": 3.7634449005126953, "learning_rate": 6.556986866252535e-07, "loss": 0.01837964542210102, "step": 6815 }, { "epoch": 1.7694704049844237, "grad_norm": 3.6442415714263916, "learning_rate": 6.553530793397975e-07, "loss": 0.02522469311952591, "step": 6816 }, { "epoch": 1.769730010384216, "grad_norm": 1.5495799779891968, "learning_rate": 6.550078423399565e-07, "loss": 0.0022788161877542734, "step": 6817 }, { "epoch": 1.7699896157840083, "grad_norm": 3.258178949356079, "learning_rate": 6.546629756867594e-07, "loss": 0.05054329335689545, "step": 6818 }, { "epoch": 1.7702492211838006, "grad_norm": 0.9993057250976562, "learning_rate": 6.543184794411715e-07, "loss": 0.0020415224134922028, "step": 6819 }, { "epoch": 1.770508826583593, "grad_norm": 2.930988073348999, "learning_rate": 6.539743536640921e-07, "loss": 0.010376280173659325, "step": 6820 }, { "epoch": 1.7707684319833854, "grad_norm": 1.2567139863967896, "learning_rate": 6.536305984163538e-07, "loss": 0.0031372574158012867, "step": 6821 }, { "epoch": 1.7710280373831777, "grad_norm": 2.7951533794403076, "learning_rate": 6.532872137587254e-07, "loss": 0.007096076849848032, "step": 6822 }, { "epoch": 1.77128764278297, "grad_norm": 3.768291473388672, "learning_rate": 6.529441997519091e-07, "loss": 0.014612225815653801, "step": 6823 }, { "epoch": 1.7715472481827623, "grad_norm": 1.755523443222046, "learning_rate": 6.526015564565423e-07, "loss": 0.004861809778958559, "step": 6824 }, { "epoch": 1.7718068535825546, "grad_norm": 3.3448429107666016, "learning_rate": 6.522592839331962e-07, "loss": 0.06292746216058731, "step": 6825 }, { "epoch": 1.772066458982347, "grad_norm": 3.3006510734558105, "learning_rate": 6.519173822423768e-07, "loss": 0.018029993399977684, "step": 6826 }, { "epoch": 1.7723260643821392, "grad_norm": 3.8891677856445312, "learning_rate": 6.51575851444525e-07, "loss": 0.024074634537100792, "step": 6827 }, { "epoch": 1.7725856697819315, "grad_norm": 0.7435476183891296, "learning_rate": 6.512346916000144e-07, "loss": 0.0020633076783269644, "step": 6828 }, { "epoch": 1.7728452751817239, "grad_norm": 2.744201183319092, "learning_rate": 6.508939027691548e-07, "loss": 0.010158403776586056, "step": 6829 }, { "epoch": 1.7731048805815162, "grad_norm": 3.0876808166503906, "learning_rate": 6.505534850121904e-07, "loss": 0.003589861560612917, "step": 6830 }, { "epoch": 1.7733644859813085, "grad_norm": 3.2831411361694336, "learning_rate": 6.502134383892993e-07, "loss": 0.06129506230354309, "step": 6831 }, { "epoch": 1.7736240913811008, "grad_norm": 3.0952107906341553, "learning_rate": 6.498737629605933e-07, "loss": 0.013307017274200916, "step": 6832 }, { "epoch": 1.773883696780893, "grad_norm": 3.5891129970550537, "learning_rate": 6.49534458786119e-07, "loss": 0.019885020330548286, "step": 6833 }, { "epoch": 1.7741433021806854, "grad_norm": 4.709648609161377, "learning_rate": 6.491955259258584e-07, "loss": 0.005767159163951874, "step": 6834 }, { "epoch": 1.7744029075804777, "grad_norm": 2.373448133468628, "learning_rate": 6.488569644397272e-07, "loss": 0.009485282003879547, "step": 6835 }, { "epoch": 1.77466251298027, "grad_norm": 3.266495943069458, "learning_rate": 6.485187743875743e-07, "loss": 0.005505390465259552, "step": 6836 }, { "epoch": 1.7749221183800623, "grad_norm": 1.777173399925232, "learning_rate": 6.481809558291848e-07, "loss": 0.011437567882239819, "step": 6837 }, { "epoch": 1.7751817237798546, "grad_norm": 3.052675485610962, "learning_rate": 6.478435088242768e-07, "loss": 0.037862472236156464, "step": 6838 }, { "epoch": 1.775441329179647, "grad_norm": 3.2488815784454346, "learning_rate": 6.475064334325039e-07, "loss": 0.055387143045663834, "step": 6839 }, { "epoch": 1.7757009345794392, "grad_norm": 0.8274136781692505, "learning_rate": 6.471697297134526e-07, "loss": 0.00418046535924077, "step": 6840 }, { "epoch": 1.7759605399792315, "grad_norm": 4.028328895568848, "learning_rate": 6.468333977266446e-07, "loss": 0.03475591540336609, "step": 6841 }, { "epoch": 1.7762201453790238, "grad_norm": 2.736717939376831, "learning_rate": 6.464974375315357e-07, "loss": 0.01913820207118988, "step": 6842 }, { "epoch": 1.7764797507788161, "grad_norm": 3.643744468688965, "learning_rate": 6.461618491875163e-07, "loss": 0.009635990485548973, "step": 6843 }, { "epoch": 1.7767393561786085, "grad_norm": 2.7199409008026123, "learning_rate": 6.4582663275391e-07, "loss": 0.010041560046374798, "step": 6844 }, { "epoch": 1.7769989615784008, "grad_norm": 3.633131504058838, "learning_rate": 6.454917882899764e-07, "loss": 0.03681700676679611, "step": 6845 }, { "epoch": 1.777258566978193, "grad_norm": 3.031078815460205, "learning_rate": 6.451573158549071e-07, "loss": 0.006433183792978525, "step": 6846 }, { "epoch": 1.7775181723779854, "grad_norm": 4.107684135437012, "learning_rate": 6.448232155078303e-07, "loss": 0.012976864352822304, "step": 6847 }, { "epoch": 1.7777777777777777, "grad_norm": 2.7427492141723633, "learning_rate": 6.44489487307807e-07, "loss": 0.017543237656354904, "step": 6848 }, { "epoch": 1.77803738317757, "grad_norm": 4.119384288787842, "learning_rate": 6.441561313138321e-07, "loss": 0.027259450405836105, "step": 6849 }, { "epoch": 1.7782969885773623, "grad_norm": 1.890599012374878, "learning_rate": 6.438231475848361e-07, "loss": 0.002081215148791671, "step": 6850 }, { "epoch": 1.7785565939771546, "grad_norm": 3.22890567779541, "learning_rate": 6.434905361796825e-07, "loss": 0.03421521186828613, "step": 6851 }, { "epoch": 1.778816199376947, "grad_norm": 4.335741996765137, "learning_rate": 6.431582971571693e-07, "loss": 0.008258726447820663, "step": 6852 }, { "epoch": 1.7790758047767392, "grad_norm": 3.021148443222046, "learning_rate": 6.428264305760288e-07, "loss": 0.018118560314178467, "step": 6853 }, { "epoch": 1.7793354101765315, "grad_norm": 2.1852471828460693, "learning_rate": 6.424949364949279e-07, "loss": 0.014073822647333145, "step": 6854 }, { "epoch": 1.7795950155763238, "grad_norm": 0.9970529079437256, "learning_rate": 6.421638149724661e-07, "loss": 0.006199127063155174, "step": 6855 }, { "epoch": 1.7798546209761164, "grad_norm": 3.098447561264038, "learning_rate": 6.418330660671794e-07, "loss": 0.05799482390284538, "step": 6856 }, { "epoch": 1.7801142263759087, "grad_norm": 1.5086991786956787, "learning_rate": 6.415026898375355e-07, "loss": 0.0062859393656253815, "step": 6857 }, { "epoch": 1.780373831775701, "grad_norm": 2.107557773590088, "learning_rate": 6.411726863419386e-07, "loss": 0.007470071315765381, "step": 6858 }, { "epoch": 1.7806334371754933, "grad_norm": 2.5345447063446045, "learning_rate": 6.408430556387243e-07, "loss": 0.004332543816417456, "step": 6859 }, { "epoch": 1.7808930425752856, "grad_norm": 1.9312664270401, "learning_rate": 6.405137977861643e-07, "loss": 0.007901996374130249, "step": 6860 }, { "epoch": 1.7811526479750779, "grad_norm": 0.7749754786491394, "learning_rate": 6.401849128424645e-07, "loss": 0.0009869299829006195, "step": 6861 }, { "epoch": 1.7814122533748702, "grad_norm": 1.7840900421142578, "learning_rate": 6.398564008657638e-07, "loss": 0.008745675906538963, "step": 6862 }, { "epoch": 1.7816718587746625, "grad_norm": 3.0711777210235596, "learning_rate": 6.395282619141354e-07, "loss": 0.03846690431237221, "step": 6863 }, { "epoch": 1.7819314641744548, "grad_norm": 3.307011365890503, "learning_rate": 6.392004960455867e-07, "loss": 0.04416513815522194, "step": 6864 }, { "epoch": 1.782191069574247, "grad_norm": 1.8042155504226685, "learning_rate": 6.388731033180594e-07, "loss": 0.015284794382750988, "step": 6865 }, { "epoch": 1.7824506749740394, "grad_norm": 1.5033276081085205, "learning_rate": 6.385460837894296e-07, "loss": 0.0027903977315872908, "step": 6866 }, { "epoch": 1.7827102803738317, "grad_norm": 3.3726606369018555, "learning_rate": 6.382194375175058e-07, "loss": 0.024531256407499313, "step": 6867 }, { "epoch": 1.7829698857736243, "grad_norm": 2.7844908237457275, "learning_rate": 6.378931645600321e-07, "loss": 0.03851975128054619, "step": 6868 }, { "epoch": 1.7832294911734166, "grad_norm": 3.15934681892395, "learning_rate": 6.375672649746861e-07, "loss": 0.005620317533612251, "step": 6869 }, { "epoch": 1.7834890965732089, "grad_norm": 2.1903164386749268, "learning_rate": 6.372417388190796e-07, "loss": 0.0046920147724449635, "step": 6870 }, { "epoch": 1.7837487019730012, "grad_norm": 2.70976185798645, "learning_rate": 6.369165861507577e-07, "loss": 0.02464272268116474, "step": 6871 }, { "epoch": 1.7840083073727935, "grad_norm": 0.9211921691894531, "learning_rate": 6.365918070271999e-07, "loss": 0.003133049700409174, "step": 6872 }, { "epoch": 1.7842679127725858, "grad_norm": 2.956355094909668, "learning_rate": 6.362674015058207e-07, "loss": 0.0030799750238656998, "step": 6873 }, { "epoch": 1.784527518172378, "grad_norm": 0.8580063581466675, "learning_rate": 6.359433696439665e-07, "loss": 0.006333754397928715, "step": 6874 }, { "epoch": 1.7847871235721704, "grad_norm": 3.793696880340576, "learning_rate": 6.356197114989187e-07, "loss": 0.014122933149337769, "step": 6875 }, { "epoch": 1.7850467289719627, "grad_norm": 3.7127676010131836, "learning_rate": 6.352964271278934e-07, "loss": 0.025180798023939133, "step": 6876 }, { "epoch": 1.785306334371755, "grad_norm": 3.902630090713501, "learning_rate": 6.349735165880393e-07, "loss": 0.028569061309099197, "step": 6877 }, { "epoch": 1.7855659397715473, "grad_norm": 3.775418519973755, "learning_rate": 6.346509799364402e-07, "loss": 0.028935082256793976, "step": 6878 }, { "epoch": 1.7858255451713396, "grad_norm": 3.4240245819091797, "learning_rate": 6.343288172301125e-07, "loss": 0.018918970599770546, "step": 6879 }, { "epoch": 1.786085150571132, "grad_norm": 2.83894681930542, "learning_rate": 6.340070285260071e-07, "loss": 0.0343601293861866, "step": 6880 }, { "epoch": 1.7863447559709242, "grad_norm": 0.5303726196289062, "learning_rate": 6.3368561388101e-07, "loss": 0.0035155811347067356, "step": 6881 }, { "epoch": 1.7866043613707165, "grad_norm": 4.293386936187744, "learning_rate": 6.333645733519387e-07, "loss": 0.011040191166102886, "step": 6882 }, { "epoch": 1.7868639667705088, "grad_norm": 3.0574090480804443, "learning_rate": 6.330439069955461e-07, "loss": 0.00853293389081955, "step": 6883 }, { "epoch": 1.7871235721703012, "grad_norm": 2.573887586593628, "learning_rate": 6.327236148685192e-07, "loss": 0.008413946256041527, "step": 6884 }, { "epoch": 1.7873831775700935, "grad_norm": 1.7327661514282227, "learning_rate": 6.32403697027478e-07, "loss": 0.005026938393712044, "step": 6885 }, { "epoch": 1.7876427829698858, "grad_norm": 3.4009759426116943, "learning_rate": 6.320841535289765e-07, "loss": 0.01184148620814085, "step": 6886 }, { "epoch": 1.787902388369678, "grad_norm": 0.45807790756225586, "learning_rate": 6.317649844295029e-07, "loss": 0.0010863055940717459, "step": 6887 }, { "epoch": 1.7881619937694704, "grad_norm": 4.289025783538818, "learning_rate": 6.314461897854787e-07, "loss": 0.008567826822400093, "step": 6888 }, { "epoch": 1.7884215991692627, "grad_norm": 3.2967817783355713, "learning_rate": 6.311277696532603e-07, "loss": 0.012888813391327858, "step": 6889 }, { "epoch": 1.788681204569055, "grad_norm": 1.6683943271636963, "learning_rate": 6.308097240891359e-07, "loss": 0.007292124442756176, "step": 6890 }, { "epoch": 1.7889408099688473, "grad_norm": 3.9165313243865967, "learning_rate": 6.304920531493292e-07, "loss": 0.014228618703782558, "step": 6891 }, { "epoch": 1.7892004153686396, "grad_norm": 3.519594192504883, "learning_rate": 6.301747568899975e-07, "loss": 0.024870648980140686, "step": 6892 }, { "epoch": 1.789460020768432, "grad_norm": 1.7355681657791138, "learning_rate": 6.298578353672312e-07, "loss": 0.011632425710558891, "step": 6893 }, { "epoch": 1.7897196261682242, "grad_norm": 1.855838656425476, "learning_rate": 6.29541288637055e-07, "loss": 0.010687712579965591, "step": 6894 }, { "epoch": 1.7899792315680165, "grad_norm": 3.5343098640441895, "learning_rate": 6.292251167554264e-07, "loss": 0.009731059893965721, "step": 6895 }, { "epoch": 1.7902388369678088, "grad_norm": 3.6866612434387207, "learning_rate": 6.289093197782378e-07, "loss": 0.04715351015329361, "step": 6896 }, { "epoch": 1.7904984423676011, "grad_norm": 4.159236431121826, "learning_rate": 6.285938977613156e-07, "loss": 0.02011169120669365, "step": 6897 }, { "epoch": 1.7907580477673934, "grad_norm": 2.3832035064697266, "learning_rate": 6.282788507604179e-07, "loss": 0.015655752271413803, "step": 6898 }, { "epoch": 1.7910176531671858, "grad_norm": 3.819085121154785, "learning_rate": 6.279641788312388e-07, "loss": 0.0140121690928936, "step": 6899 }, { "epoch": 1.791277258566978, "grad_norm": 1.9381555318832397, "learning_rate": 6.276498820294043e-07, "loss": 0.009739688597619534, "step": 6900 }, { "epoch": 1.7915368639667704, "grad_norm": 3.113513946533203, "learning_rate": 6.273359604104756e-07, "loss": 0.006382535677403212, "step": 6901 }, { "epoch": 1.7917964693665627, "grad_norm": 3.929111957550049, "learning_rate": 6.270224140299466e-07, "loss": 0.012211046181619167, "step": 6902 }, { "epoch": 1.792056074766355, "grad_norm": 1.5011059045791626, "learning_rate": 6.267092429432446e-07, "loss": 0.004798330366611481, "step": 6903 }, { "epoch": 1.7923156801661475, "grad_norm": 3.893578052520752, "learning_rate": 6.26396447205732e-07, "loss": 0.009314008057117462, "step": 6904 }, { "epoch": 1.7925752855659398, "grad_norm": 1.3026649951934814, "learning_rate": 6.260840268727032e-07, "loss": 0.0032716162968426943, "step": 6905 }, { "epoch": 1.7928348909657321, "grad_norm": 2.931820869445801, "learning_rate": 6.257719819993873e-07, "loss": 0.005503038410097361, "step": 6906 }, { "epoch": 1.7930944963655244, "grad_norm": 4.199589729309082, "learning_rate": 6.254603126409466e-07, "loss": 0.022122835740447044, "step": 6907 }, { "epoch": 1.7933541017653167, "grad_norm": 3.530630588531494, "learning_rate": 6.251490188524768e-07, "loss": 0.0527593232691288, "step": 6908 }, { "epoch": 1.793613707165109, "grad_norm": 3.909437417984009, "learning_rate": 6.248381006890083e-07, "loss": 0.007340181153267622, "step": 6909 }, { "epoch": 1.7938733125649013, "grad_norm": 3.136969804763794, "learning_rate": 6.24527558205504e-07, "loss": 0.007048319093883038, "step": 6910 }, { "epoch": 1.7941329179646937, "grad_norm": 5.555108547210693, "learning_rate": 6.2421739145686e-07, "loss": 0.007462511770427227, "step": 6911 }, { "epoch": 1.794392523364486, "grad_norm": 2.6133177280426025, "learning_rate": 6.239076004979075e-07, "loss": 0.01030048355460167, "step": 6912 }, { "epoch": 1.7946521287642783, "grad_norm": 2.8464183807373047, "learning_rate": 6.235981853834102e-07, "loss": 0.00820869579911232, "step": 6913 }, { "epoch": 1.7949117341640706, "grad_norm": 3.359769821166992, "learning_rate": 6.232891461680653e-07, "loss": 0.014823254197835922, "step": 6914 }, { "epoch": 1.7951713395638629, "grad_norm": 4.568560600280762, "learning_rate": 6.229804829065044e-07, "loss": 0.010357799008488655, "step": 6915 }, { "epoch": 1.7954309449636554, "grad_norm": 3.328420877456665, "learning_rate": 6.226721956532921e-07, "loss": 0.02037295699119568, "step": 6916 }, { "epoch": 1.7956905503634477, "grad_norm": 2.075956106185913, "learning_rate": 6.223642844629255e-07, "loss": 0.005476000718772411, "step": 6917 }, { "epoch": 1.79595015576324, "grad_norm": 3.4852938652038574, "learning_rate": 6.220567493898376e-07, "loss": 0.008186204358935356, "step": 6918 }, { "epoch": 1.7962097611630323, "grad_norm": 2.597346544265747, "learning_rate": 6.217495904883928e-07, "loss": 0.0046342480927705765, "step": 6919 }, { "epoch": 1.7964693665628246, "grad_norm": 2.5576462745666504, "learning_rate": 6.214428078128902e-07, "loss": 0.0042450204491615295, "step": 6920 }, { "epoch": 1.796728971962617, "grad_norm": 3.384932041168213, "learning_rate": 6.211364014175613e-07, "loss": 0.03986620903015137, "step": 6921 }, { "epoch": 1.7969885773624092, "grad_norm": 4.1736674308776855, "learning_rate": 6.208303713565721e-07, "loss": 0.0028645561542361975, "step": 6922 }, { "epoch": 1.7972481827622016, "grad_norm": 3.1435022354125977, "learning_rate": 6.205247176840217e-07, "loss": 0.011217517778277397, "step": 6923 }, { "epoch": 1.7975077881619939, "grad_norm": 0.3541259765625, "learning_rate": 6.202194404539431e-07, "loss": 0.0008552783983759582, "step": 6924 }, { "epoch": 1.7977673935617862, "grad_norm": 3.4912161827087402, "learning_rate": 6.199145397203017e-07, "loss": 0.01971941813826561, "step": 6925 }, { "epoch": 1.7980269989615785, "grad_norm": 3.0768496990203857, "learning_rate": 6.19610015536997e-07, "loss": 0.008721331134438515, "step": 6926 }, { "epoch": 1.7982866043613708, "grad_norm": 3.2046515941619873, "learning_rate": 6.193058679578619e-07, "loss": 0.010095011442899704, "step": 6927 }, { "epoch": 1.798546209761163, "grad_norm": 3.200411319732666, "learning_rate": 6.190020970366636e-07, "loss": 0.0378638319671154, "step": 6928 }, { "epoch": 1.7988058151609554, "grad_norm": 3.377743721008301, "learning_rate": 6.186987028271007e-07, "loss": 0.03129266947507858, "step": 6929 }, { "epoch": 1.7990654205607477, "grad_norm": 3.1808555126190186, "learning_rate": 6.183956853828069e-07, "loss": 0.013623185455799103, "step": 6930 }, { "epoch": 1.79932502596054, "grad_norm": 1.947422742843628, "learning_rate": 6.180930447573486e-07, "loss": 0.006222531199455261, "step": 6931 }, { "epoch": 1.7995846313603323, "grad_norm": 0.25543099641799927, "learning_rate": 6.177907810042259e-07, "loss": 0.0008611818193458021, "step": 6932 }, { "epoch": 1.7998442367601246, "grad_norm": 4.251255989074707, "learning_rate": 6.174888941768721e-07, "loss": 0.01124393567442894, "step": 6933 }, { "epoch": 1.800103842159917, "grad_norm": 3.6829872131347656, "learning_rate": 6.171873843286534e-07, "loss": 0.03657840937376022, "step": 6934 }, { "epoch": 1.8003634475597092, "grad_norm": 4.158417701721191, "learning_rate": 6.168862515128703e-07, "loss": 0.012569420039653778, "step": 6935 }, { "epoch": 1.8006230529595015, "grad_norm": 1.3193581104278564, "learning_rate": 6.165854957827565e-07, "loss": 0.00345940631814301, "step": 6936 }, { "epoch": 1.8008826583592938, "grad_norm": 1.0771920680999756, "learning_rate": 6.16285117191478e-07, "loss": 0.0011226062197238207, "step": 6937 }, { "epoch": 1.8011422637590861, "grad_norm": 3.504666805267334, "learning_rate": 6.159851157921354e-07, "loss": 0.012825222685933113, "step": 6938 }, { "epoch": 1.8014018691588785, "grad_norm": 4.7401838302612305, "learning_rate": 6.156854916377618e-07, "loss": 0.00962941162288189, "step": 6939 }, { "epoch": 1.8016614745586708, "grad_norm": 3.456876754760742, "learning_rate": 6.153862447813236e-07, "loss": 0.05361020937561989, "step": 6940 }, { "epoch": 1.801921079958463, "grad_norm": 2.840851068496704, "learning_rate": 6.150873752757217e-07, "loss": 0.003316809656098485, "step": 6941 }, { "epoch": 1.8021806853582554, "grad_norm": 2.740072727203369, "learning_rate": 6.147888831737882e-07, "loss": 0.011059267446398735, "step": 6942 }, { "epoch": 1.8024402907580477, "grad_norm": 2.1258368492126465, "learning_rate": 6.144907685282909e-07, "loss": 0.010606501251459122, "step": 6943 }, { "epoch": 1.80269989615784, "grad_norm": 0.3821415305137634, "learning_rate": 6.14193031391929e-07, "loss": 0.000620124745182693, "step": 6944 }, { "epoch": 1.8029595015576323, "grad_norm": 3.6188974380493164, "learning_rate": 6.13895671817335e-07, "loss": 0.03564505651593208, "step": 6945 }, { "epoch": 1.8032191069574246, "grad_norm": 4.716024875640869, "learning_rate": 6.135986898570765e-07, "loss": 0.023232024163007736, "step": 6946 }, { "epoch": 1.803478712357217, "grad_norm": 3.588139772415161, "learning_rate": 6.133020855636526e-07, "loss": 0.019323866814374924, "step": 6947 }, { "epoch": 1.8037383177570092, "grad_norm": 3.6251025199890137, "learning_rate": 6.130058589894958e-07, "loss": 0.012199854478240013, "step": 6948 }, { "epoch": 1.8039979231568015, "grad_norm": 2.6251678466796875, "learning_rate": 6.127100101869725e-07, "loss": 0.004952020011842251, "step": 6949 }, { "epoch": 1.8042575285565938, "grad_norm": 4.1827826499938965, "learning_rate": 6.124145392083817e-07, "loss": 0.012883604504168034, "step": 6950 }, { "epoch": 1.8045171339563861, "grad_norm": 3.6693406105041504, "learning_rate": 6.121194461059566e-07, "loss": 0.005462259985506535, "step": 6951 }, { "epoch": 1.8047767393561787, "grad_norm": 4.021783351898193, "learning_rate": 6.118247309318625e-07, "loss": 0.03258778154850006, "step": 6952 }, { "epoch": 1.805036344755971, "grad_norm": 3.671996831893921, "learning_rate": 6.115303937381976e-07, "loss": 0.02516121044754982, "step": 6953 }, { "epoch": 1.8052959501557633, "grad_norm": 1.3565993309020996, "learning_rate": 6.112364345769953e-07, "loss": 0.006696558091789484, "step": 6954 }, { "epoch": 1.8055555555555556, "grad_norm": 3.8793036937713623, "learning_rate": 6.1094285350022e-07, "loss": 0.030811317265033722, "step": 6955 }, { "epoch": 1.8058151609553479, "grad_norm": 1.4917216300964355, "learning_rate": 6.1064965055977e-07, "loss": 0.005079471506178379, "step": 6956 }, { "epoch": 1.8060747663551402, "grad_norm": 2.916713237762451, "learning_rate": 6.103568258074773e-07, "loss": 0.00683183316141367, "step": 6957 }, { "epoch": 1.8063343717549325, "grad_norm": 1.5576506853103638, "learning_rate": 6.100643792951065e-07, "loss": 0.0021658907644450665, "step": 6958 }, { "epoch": 1.8065939771547248, "grad_norm": 2.525228500366211, "learning_rate": 6.097723110743556e-07, "loss": 0.0079415924847126, "step": 6959 }, { "epoch": 1.8068535825545171, "grad_norm": 3.5618088245391846, "learning_rate": 6.094806211968551e-07, "loss": 0.009609947912395, "step": 6960 }, { "epoch": 1.8071131879543094, "grad_norm": 2.039515972137451, "learning_rate": 6.091893097141695e-07, "loss": 0.008354172110557556, "step": 6961 }, { "epoch": 1.8073727933541017, "grad_norm": 0.6710286140441895, "learning_rate": 6.088983766777959e-07, "loss": 0.0010775683913379908, "step": 6962 }, { "epoch": 1.807632398753894, "grad_norm": 3.2464635372161865, "learning_rate": 6.086078221391644e-07, "loss": 0.02497411146759987, "step": 6963 }, { "epoch": 1.8078920041536866, "grad_norm": 2.28790545463562, "learning_rate": 6.083176461496389e-07, "loss": 0.004848433658480644, "step": 6964 }, { "epoch": 1.8081516095534789, "grad_norm": 3.209519386291504, "learning_rate": 6.08027848760515e-07, "loss": 0.015671204775571823, "step": 6965 }, { "epoch": 1.8084112149532712, "grad_norm": 3.2852163314819336, "learning_rate": 6.077384300230233e-07, "loss": 0.0493098720908165, "step": 6966 }, { "epoch": 1.8086708203530635, "grad_norm": 1.7368196249008179, "learning_rate": 6.074493899883259e-07, "loss": 0.00860656425356865, "step": 6967 }, { "epoch": 1.8089304257528558, "grad_norm": 3.647980213165283, "learning_rate": 6.07160728707518e-07, "loss": 0.030119579285383224, "step": 6968 }, { "epoch": 1.809190031152648, "grad_norm": 1.6186102628707886, "learning_rate": 6.068724462316291e-07, "loss": 0.002917930483818054, "step": 6969 }, { "epoch": 1.8094496365524404, "grad_norm": 1.4053648710250854, "learning_rate": 6.065845426116207e-07, "loss": 0.0022591850720345974, "step": 6970 }, { "epoch": 1.8097092419522327, "grad_norm": 1.3529627323150635, "learning_rate": 6.062970178983869e-07, "loss": 0.0021610092371702194, "step": 6971 }, { "epoch": 1.809968847352025, "grad_norm": 2.246551513671875, "learning_rate": 6.060098721427566e-07, "loss": 0.0048848786391317844, "step": 6972 }, { "epoch": 1.8102284527518173, "grad_norm": 0.13441790640354156, "learning_rate": 6.057231053954895e-07, "loss": 0.00035599188413470984, "step": 6973 }, { "epoch": 1.8104880581516096, "grad_norm": 2.980151891708374, "learning_rate": 6.054367177072803e-07, "loss": 0.04300699755549431, "step": 6974 }, { "epoch": 1.810747663551402, "grad_norm": 2.9939584732055664, "learning_rate": 6.051507091287554e-07, "loss": 0.009501419961452484, "step": 6975 }, { "epoch": 1.8110072689511942, "grad_norm": 3.2414660453796387, "learning_rate": 6.048650797104746e-07, "loss": 0.042747847735881805, "step": 6976 }, { "epoch": 1.8112668743509865, "grad_norm": 1.5345797538757324, "learning_rate": 6.045798295029305e-07, "loss": 0.0060408334247767925, "step": 6977 }, { "epoch": 1.8115264797507789, "grad_norm": 1.4969427585601807, "learning_rate": 6.042949585565491e-07, "loss": 0.010610582306981087, "step": 6978 }, { "epoch": 1.8117860851505712, "grad_norm": 0.7517175078392029, "learning_rate": 6.040104669216889e-07, "loss": 0.0006913719698786736, "step": 6979 }, { "epoch": 1.8120456905503635, "grad_norm": 2.8612189292907715, "learning_rate": 6.037263546486413e-07, "loss": 0.010163962841033936, "step": 6980 }, { "epoch": 1.8123052959501558, "grad_norm": 3.4519615173339844, "learning_rate": 6.03442621787631e-07, "loss": 0.014956715516746044, "step": 6981 }, { "epoch": 1.812564901349948, "grad_norm": 1.2531012296676636, "learning_rate": 6.031592683888158e-07, "loss": 0.0038722024764865637, "step": 6982 }, { "epoch": 1.8128245067497404, "grad_norm": 4.3471174240112305, "learning_rate": 6.028762945022857e-07, "loss": 0.010016867890954018, "step": 6983 }, { "epoch": 1.8130841121495327, "grad_norm": 3.236496925354004, "learning_rate": 6.02593700178064e-07, "loss": 0.009639021009206772, "step": 6984 }, { "epoch": 1.813343717549325, "grad_norm": 3.471963882446289, "learning_rate": 6.023114854661073e-07, "loss": 0.05438290908932686, "step": 6985 }, { "epoch": 1.8136033229491173, "grad_norm": 3.550933361053467, "learning_rate": 6.020296504163043e-07, "loss": 0.03837460279464722, "step": 6986 }, { "epoch": 1.8138629283489096, "grad_norm": 4.321341037750244, "learning_rate": 6.017481950784767e-07, "loss": 0.0023815520107746124, "step": 6987 }, { "epoch": 1.814122533748702, "grad_norm": 3.8060388565063477, "learning_rate": 6.014671195023805e-07, "loss": 0.009259969927370548, "step": 6988 }, { "epoch": 1.8143821391484942, "grad_norm": 3.619994878768921, "learning_rate": 6.01186423737702e-07, "loss": 0.016569115221500397, "step": 6989 }, { "epoch": 1.8146417445482865, "grad_norm": 3.4112160205841064, "learning_rate": 6.009061078340631e-07, "loss": 0.0168294720351696, "step": 6990 }, { "epoch": 1.8149013499480788, "grad_norm": 3.146388530731201, "learning_rate": 6.00626171841016e-07, "loss": 0.011191140860319138, "step": 6991 }, { "epoch": 1.8151609553478711, "grad_norm": 3.449345350265503, "learning_rate": 6.003466158080478e-07, "loss": 0.012065384536981583, "step": 6992 }, { "epoch": 1.8154205607476634, "grad_norm": 0.9558647274971008, "learning_rate": 6.000674397845777e-07, "loss": 0.005764275789260864, "step": 6993 }, { "epoch": 1.8156801661474558, "grad_norm": 1.2803564071655273, "learning_rate": 5.997886438199571e-07, "loss": 0.0008782373042777181, "step": 6994 }, { "epoch": 1.815939771547248, "grad_norm": 3.526914119720459, "learning_rate": 5.995102279634708e-07, "loss": 0.0475056953728199, "step": 6995 }, { "epoch": 1.8161993769470404, "grad_norm": 2.306823253631592, "learning_rate": 5.992321922643364e-07, "loss": 0.002191924722865224, "step": 6996 }, { "epoch": 1.8164589823468327, "grad_norm": 3.628221273422241, "learning_rate": 5.98954536771705e-07, "loss": 0.006174693815410137, "step": 6997 }, { "epoch": 1.816718587746625, "grad_norm": 4.444052696228027, "learning_rate": 5.986772615346586e-07, "loss": 0.013852547854185104, "step": 6998 }, { "epoch": 1.8169781931464173, "grad_norm": 3.3113865852355957, "learning_rate": 5.984003666022135e-07, "loss": 0.009275845251977444, "step": 6999 }, { "epoch": 1.8172377985462098, "grad_norm": 2.3739986419677734, "learning_rate": 5.981238520233188e-07, "loss": 0.014856441877782345, "step": 7000 }, { "epoch": 1.8174974039460021, "grad_norm": 0.7455753087997437, "learning_rate": 5.978477178468556e-07, "loss": 0.0015881425933912396, "step": 7001 }, { "epoch": 1.8177570093457944, "grad_norm": 0.40028420090675354, "learning_rate": 5.975719641216379e-07, "loss": 0.0007370141102001071, "step": 7002 }, { "epoch": 1.8180166147455867, "grad_norm": 3.0383310317993164, "learning_rate": 5.972965908964129e-07, "loss": 0.02165853977203369, "step": 7003 }, { "epoch": 1.818276220145379, "grad_norm": 4.055642604827881, "learning_rate": 5.970215982198602e-07, "loss": 0.022263310849666595, "step": 7004 }, { "epoch": 1.8185358255451713, "grad_norm": 3.372014045715332, "learning_rate": 5.967469861405927e-07, "loss": 0.04636305570602417, "step": 7005 }, { "epoch": 1.8187954309449637, "grad_norm": 3.4019722938537598, "learning_rate": 5.964727547071548e-07, "loss": 0.02443978562951088, "step": 7006 }, { "epoch": 1.819055036344756, "grad_norm": 0.7548400163650513, "learning_rate": 5.961989039680243e-07, "loss": 0.00219791941344738, "step": 7007 }, { "epoch": 1.8193146417445483, "grad_norm": 1.401965856552124, "learning_rate": 5.959254339716127e-07, "loss": 0.008125463500618935, "step": 7008 }, { "epoch": 1.8195742471443406, "grad_norm": 2.116973400115967, "learning_rate": 5.956523447662624e-07, "loss": 0.00390164484269917, "step": 7009 }, { "epoch": 1.8198338525441329, "grad_norm": 0.520858108997345, "learning_rate": 5.953796364002491e-07, "loss": 0.0014028237201273441, "step": 7010 }, { "epoch": 1.8200934579439252, "grad_norm": 1.174877643585205, "learning_rate": 5.951073089217822e-07, "loss": 0.0020381310023367405, "step": 7011 }, { "epoch": 1.8203530633437177, "grad_norm": 3.6080310344696045, "learning_rate": 5.948353623790024e-07, "loss": 0.01895083114504814, "step": 7012 }, { "epoch": 1.82061266874351, "grad_norm": 2.598787307739258, "learning_rate": 5.94563796819984e-07, "loss": 0.013466758653521538, "step": 7013 }, { "epoch": 1.8208722741433023, "grad_norm": 0.20043371617794037, "learning_rate": 5.942926122927334e-07, "loss": 0.0006486686179414392, "step": 7014 }, { "epoch": 1.8211318795430946, "grad_norm": 1.8768329620361328, "learning_rate": 5.940218088451894e-07, "loss": 0.007868119515478611, "step": 7015 }, { "epoch": 1.821391484942887, "grad_norm": 3.599135398864746, "learning_rate": 5.937513865252247e-07, "loss": 0.014768057502806187, "step": 7016 }, { "epoch": 1.8216510903426792, "grad_norm": 0.5672491192817688, "learning_rate": 5.934813453806434e-07, "loss": 0.0006830699858255684, "step": 7017 }, { "epoch": 1.8219106957424716, "grad_norm": 3.1814303398132324, "learning_rate": 5.932116854591819e-07, "loss": 0.03348381072282791, "step": 7018 }, { "epoch": 1.8221703011422639, "grad_norm": 3.5040595531463623, "learning_rate": 5.929424068085109e-07, "loss": 0.01632661744952202, "step": 7019 }, { "epoch": 1.8224299065420562, "grad_norm": 4.57963752746582, "learning_rate": 5.926735094762325e-07, "loss": 0.004622966982424259, "step": 7020 }, { "epoch": 1.8226895119418485, "grad_norm": 3.5711560249328613, "learning_rate": 5.924049935098812e-07, "loss": 0.03102152794599533, "step": 7021 }, { "epoch": 1.8229491173416408, "grad_norm": 3.5935418605804443, "learning_rate": 5.921368589569246e-07, "loss": 0.016689550131559372, "step": 7022 }, { "epoch": 1.823208722741433, "grad_norm": 1.5331605672836304, "learning_rate": 5.918691058647626e-07, "loss": 0.010564660653471947, "step": 7023 }, { "epoch": 1.8234683281412254, "grad_norm": 5.929174900054932, "learning_rate": 5.916017342807287e-07, "loss": 0.01334880106151104, "step": 7024 }, { "epoch": 1.8237279335410177, "grad_norm": 2.4626896381378174, "learning_rate": 5.913347442520865e-07, "loss": 0.004478354938328266, "step": 7025 }, { "epoch": 1.82398753894081, "grad_norm": 1.5689023733139038, "learning_rate": 5.910681358260355e-07, "loss": 0.009111052379012108, "step": 7026 }, { "epoch": 1.8242471443406023, "grad_norm": 3.5884296894073486, "learning_rate": 5.908019090497045e-07, "loss": 0.027170028537511826, "step": 7027 }, { "epoch": 1.8245067497403946, "grad_norm": 3.95920467376709, "learning_rate": 5.905360639701571e-07, "loss": 0.012853268533945084, "step": 7028 }, { "epoch": 1.824766355140187, "grad_norm": 3.29030442237854, "learning_rate": 5.902706006343887e-07, "loss": 0.024886325001716614, "step": 7029 }, { "epoch": 1.8250259605399792, "grad_norm": 3.8361921310424805, "learning_rate": 5.900055190893262e-07, "loss": 0.013335099443793297, "step": 7030 }, { "epoch": 1.8252855659397715, "grad_norm": 2.079817056655884, "learning_rate": 5.897408193818311e-07, "loss": 0.009867381304502487, "step": 7031 }, { "epoch": 1.8255451713395638, "grad_norm": 1.9813200235366821, "learning_rate": 5.894765015586959e-07, "loss": 0.01490000355988741, "step": 7032 }, { "epoch": 1.8258047767393562, "grad_norm": 1.8575278520584106, "learning_rate": 5.892125656666454e-07, "loss": 0.0068956660106778145, "step": 7033 }, { "epoch": 1.8260643821391485, "grad_norm": 3.448775053024292, "learning_rate": 5.889490117523381e-07, "loss": 0.02950458973646164, "step": 7034 }, { "epoch": 1.8263239875389408, "grad_norm": 3.4978489875793457, "learning_rate": 5.886858398623635e-07, "loss": 0.03038087859749794, "step": 7035 }, { "epoch": 1.826583592938733, "grad_norm": 3.5886032581329346, "learning_rate": 5.884230500432455e-07, "loss": 0.012799002230167389, "step": 7036 }, { "epoch": 1.8268431983385254, "grad_norm": 3.3359766006469727, "learning_rate": 5.881606423414383e-07, "loss": 0.02334151789546013, "step": 7037 }, { "epoch": 1.8271028037383177, "grad_norm": 3.6775033473968506, "learning_rate": 5.878986168033301e-07, "loss": 0.011555593460798264, "step": 7038 }, { "epoch": 1.82736240913811, "grad_norm": 3.233130693435669, "learning_rate": 5.876369734752411e-07, "loss": 0.01073676347732544, "step": 7039 }, { "epoch": 1.8276220145379023, "grad_norm": 3.651139736175537, "learning_rate": 5.873757124034234e-07, "loss": 0.03751562535762787, "step": 7040 }, { "epoch": 1.8278816199376946, "grad_norm": 3.679121494293213, "learning_rate": 5.871148336340623e-07, "loss": 0.01725878193974495, "step": 7041 }, { "epoch": 1.828141225337487, "grad_norm": 3.2067627906799316, "learning_rate": 5.868543372132751e-07, "loss": 0.01965573988854885, "step": 7042 }, { "epoch": 1.8284008307372792, "grad_norm": 3.8926138877868652, "learning_rate": 5.865942231871113e-07, "loss": 0.042241305112838745, "step": 7043 }, { "epoch": 1.8286604361370715, "grad_norm": 1.1591095924377441, "learning_rate": 5.863344916015538e-07, "loss": 0.00839969888329506, "step": 7044 }, { "epoch": 1.8289200415368638, "grad_norm": 3.0023715496063232, "learning_rate": 5.86075142502517e-07, "loss": 0.004199528135359287, "step": 7045 }, { "epoch": 1.8291796469366561, "grad_norm": 3.101323366165161, "learning_rate": 5.858161759358472e-07, "loss": 0.02901623398065567, "step": 7046 }, { "epoch": 1.8294392523364484, "grad_norm": 2.0477123260498047, "learning_rate": 5.855575919473248e-07, "loss": 0.009711059741675854, "step": 7047 }, { "epoch": 1.829698857736241, "grad_norm": 3.7085459232330322, "learning_rate": 5.852993905826606e-07, "loss": 0.0268210768699646, "step": 7048 }, { "epoch": 1.8299584631360333, "grad_norm": 3.3020238876342773, "learning_rate": 5.850415718874988e-07, "loss": 0.011284196749329567, "step": 7049 }, { "epoch": 1.8302180685358256, "grad_norm": 3.6583423614501953, "learning_rate": 5.847841359074165e-07, "loss": 0.02949637547135353, "step": 7050 }, { "epoch": 1.8304776739356179, "grad_norm": 4.301614761352539, "learning_rate": 5.845270826879221e-07, "loss": 0.03691096603870392, "step": 7051 }, { "epoch": 1.8307372793354102, "grad_norm": 4.586429595947266, "learning_rate": 5.842704122744569e-07, "loss": 0.013885335996747017, "step": 7052 }, { "epoch": 1.8309968847352025, "grad_norm": 4.063645839691162, "learning_rate": 5.840141247123938e-07, "loss": 0.0362071767449379, "step": 7053 }, { "epoch": 1.8312564901349948, "grad_norm": 2.907838821411133, "learning_rate": 5.837582200470388e-07, "loss": 0.02315063588321209, "step": 7054 }, { "epoch": 1.8315160955347871, "grad_norm": 3.590538501739502, "learning_rate": 5.835026983236308e-07, "loss": 0.008855603635311127, "step": 7055 }, { "epoch": 1.8317757009345794, "grad_norm": 3.5721724033355713, "learning_rate": 5.832475595873389e-07, "loss": 0.050543684512376785, "step": 7056 }, { "epoch": 1.8320353063343717, "grad_norm": 3.3843212127685547, "learning_rate": 5.829928038832668e-07, "loss": 0.025886505842208862, "step": 7057 }, { "epoch": 1.832294911734164, "grad_norm": 1.5745261907577515, "learning_rate": 5.827384312564488e-07, "loss": 0.0077426075004041195, "step": 7058 }, { "epoch": 1.8325545171339563, "grad_norm": 3.085902452468872, "learning_rate": 5.824844417518527e-07, "loss": 0.011346879415214062, "step": 7059 }, { "epoch": 1.8328141225337489, "grad_norm": 1.3684861660003662, "learning_rate": 5.822308354143776e-07, "loss": 0.0077264090068638325, "step": 7060 }, { "epoch": 1.8330737279335412, "grad_norm": 4.623274326324463, "learning_rate": 5.819776122888553e-07, "loss": 0.012361578643321991, "step": 7061 }, { "epoch": 1.8333333333333335, "grad_norm": 3.7633562088012695, "learning_rate": 5.817247724200503e-07, "loss": 0.021721381694078445, "step": 7062 }, { "epoch": 1.8335929387331258, "grad_norm": 3.0100598335266113, "learning_rate": 5.814723158526586e-07, "loss": 0.012411229312419891, "step": 7063 }, { "epoch": 1.833852544132918, "grad_norm": 3.4207093715667725, "learning_rate": 5.812202426313087e-07, "loss": 0.022995907813310623, "step": 7064 }, { "epoch": 1.8341121495327104, "grad_norm": 1.444404125213623, "learning_rate": 5.809685528005614e-07, "loss": 0.0011777363251894712, "step": 7065 }, { "epoch": 1.8343717549325027, "grad_norm": 2.7653048038482666, "learning_rate": 5.807172464049096e-07, "loss": 0.0062105851247906685, "step": 7066 }, { "epoch": 1.834631360332295, "grad_norm": 3.342733144760132, "learning_rate": 5.80466323488779e-07, "loss": 0.03547011688351631, "step": 7067 }, { "epoch": 1.8348909657320873, "grad_norm": 1.049144983291626, "learning_rate": 5.802157840965264e-07, "loss": 0.004635167308151722, "step": 7068 }, { "epoch": 1.8351505711318796, "grad_norm": 4.003378391265869, "learning_rate": 5.799656282724417e-07, "loss": 0.008895551785826683, "step": 7069 }, { "epoch": 1.835410176531672, "grad_norm": 2.1500985622406006, "learning_rate": 5.797158560607467e-07, "loss": 0.0050185080617666245, "step": 7070 }, { "epoch": 1.8356697819314642, "grad_norm": 1.6077436208724976, "learning_rate": 5.794664675055956e-07, "loss": 0.0034540374763309956, "step": 7071 }, { "epoch": 1.8359293873312565, "grad_norm": 0.04448244720697403, "learning_rate": 5.792174626510742e-07, "loss": 0.0002595281112007797, "step": 7072 }, { "epoch": 1.8361889927310489, "grad_norm": 1.2363572120666504, "learning_rate": 5.789688415412015e-07, "loss": 0.0017347835237160325, "step": 7073 }, { "epoch": 1.8364485981308412, "grad_norm": 2.842181921005249, "learning_rate": 5.787206042199273e-07, "loss": 0.019145676866173744, "step": 7074 }, { "epoch": 1.8367082035306335, "grad_norm": 1.7919797897338867, "learning_rate": 5.784727507311344e-07, "loss": 0.0017974607180804014, "step": 7075 }, { "epoch": 1.8369678089304258, "grad_norm": 1.0012985467910767, "learning_rate": 5.78225281118638e-07, "loss": 0.0022034388966858387, "step": 7076 }, { "epoch": 1.837227414330218, "grad_norm": 7.750362873077393, "learning_rate": 5.779781954261845e-07, "loss": 0.015982676297426224, "step": 7077 }, { "epoch": 1.8374870197300104, "grad_norm": 3.0946130752563477, "learning_rate": 5.777314936974541e-07, "loss": 0.022563399747014046, "step": 7078 }, { "epoch": 1.8377466251298027, "grad_norm": 3.2990081310272217, "learning_rate": 5.774851759760566e-07, "loss": 0.008501628413796425, "step": 7079 }, { "epoch": 1.838006230529595, "grad_norm": 1.787011981010437, "learning_rate": 5.77239242305536e-07, "loss": 0.0016326650511473417, "step": 7080 }, { "epoch": 1.8382658359293873, "grad_norm": 1.6351354122161865, "learning_rate": 5.769936927293679e-07, "loss": 0.0026905802078545094, "step": 7081 }, { "epoch": 1.8385254413291796, "grad_norm": 2.027900457382202, "learning_rate": 5.767485272909598e-07, "loss": 0.009324157610535622, "step": 7082 }, { "epoch": 1.838785046728972, "grad_norm": 4.393670558929443, "learning_rate": 5.765037460336512e-07, "loss": 0.010523463599383831, "step": 7083 }, { "epoch": 1.8390446521287642, "grad_norm": 2.1743927001953125, "learning_rate": 5.762593490007135e-07, "loss": 0.008360138162970543, "step": 7084 }, { "epoch": 1.8393042575285565, "grad_norm": 3.322706937789917, "learning_rate": 5.76015336235351e-07, "loss": 0.04958253726363182, "step": 7085 }, { "epoch": 1.8395638629283488, "grad_norm": 3.466615915298462, "learning_rate": 5.757717077807e-07, "loss": 0.02961065247654915, "step": 7086 }, { "epoch": 1.8398234683281411, "grad_norm": 3.5038559436798096, "learning_rate": 5.75528463679827e-07, "loss": 0.019983287900686264, "step": 7087 }, { "epoch": 1.8400830737279334, "grad_norm": 3.7427308559417725, "learning_rate": 5.752856039757334e-07, "loss": 0.020308803766965866, "step": 7088 }, { "epoch": 1.8403426791277258, "grad_norm": 1.658260464668274, "learning_rate": 5.750431287113501e-07, "loss": 0.003639827948063612, "step": 7089 }, { "epoch": 1.840602284527518, "grad_norm": 1.5748709440231323, "learning_rate": 5.74801037929542e-07, "loss": 0.0037407386116683483, "step": 7090 }, { "epoch": 1.8408618899273104, "grad_norm": 1.3648571968078613, "learning_rate": 5.745593316731051e-07, "loss": 0.0023956201039254665, "step": 7091 }, { "epoch": 1.8411214953271027, "grad_norm": 4.175821304321289, "learning_rate": 5.74318009984767e-07, "loss": 0.019672539085149765, "step": 7092 }, { "epoch": 1.841381100726895, "grad_norm": 5.043835639953613, "learning_rate": 5.740770729071887e-07, "loss": 0.022978678345680237, "step": 7093 }, { "epoch": 1.8416407061266873, "grad_norm": 3.6334564685821533, "learning_rate": 5.738365204829617e-07, "loss": 0.022434528917074203, "step": 7094 }, { "epoch": 1.8419003115264796, "grad_norm": 3.7542288303375244, "learning_rate": 5.735963527546102e-07, "loss": 0.015052085742354393, "step": 7095 }, { "epoch": 1.842159916926272, "grad_norm": 2.2192728519439697, "learning_rate": 5.733565697645906e-07, "loss": 0.013689680956304073, "step": 7096 }, { "epoch": 1.8424195223260644, "grad_norm": 2.399747371673584, "learning_rate": 5.731171715552908e-07, "loss": 0.007351499050855637, "step": 7097 }, { "epoch": 1.8426791277258567, "grad_norm": 0.2887953221797943, "learning_rate": 5.728781581690312e-07, "loss": 0.0007884094375185668, "step": 7098 }, { "epoch": 1.842938733125649, "grad_norm": 2.3234448432922363, "learning_rate": 5.726395296480637e-07, "loss": 0.005710428114980459, "step": 7099 }, { "epoch": 1.8431983385254413, "grad_norm": 2.2738232612609863, "learning_rate": 5.724012860345723e-07, "loss": 0.0057991743087768555, "step": 7100 }, { "epoch": 1.8434579439252337, "grad_norm": 0.4602138102054596, "learning_rate": 5.721634273706735e-07, "loss": 0.0008758516050875187, "step": 7101 }, { "epoch": 1.843717549325026, "grad_norm": 0.9877591729164124, "learning_rate": 5.719259536984147e-07, "loss": 0.005209319293498993, "step": 7102 }, { "epoch": 1.8439771547248183, "grad_norm": 4.2949042320251465, "learning_rate": 5.716888650597758e-07, "loss": 0.022962044924497604, "step": 7103 }, { "epoch": 1.8442367601246106, "grad_norm": 2.5399343967437744, "learning_rate": 5.714521614966689e-07, "loss": 0.004080238752067089, "step": 7104 }, { "epoch": 1.8444963655244029, "grad_norm": 1.7642534971237183, "learning_rate": 5.71215843050938e-07, "loss": 0.0027413556817919016, "step": 7105 }, { "epoch": 1.8447559709241952, "grad_norm": 0.8971071243286133, "learning_rate": 5.709799097643583e-07, "loss": 0.0037831070367246866, "step": 7106 }, { "epoch": 1.8450155763239875, "grad_norm": 3.182342290878296, "learning_rate": 5.707443616786376e-07, "loss": 0.043732769787311554, "step": 7107 }, { "epoch": 1.8452751817237798, "grad_norm": 5.833613395690918, "learning_rate": 5.705091988354152e-07, "loss": 0.018668219447135925, "step": 7108 }, { "epoch": 1.8455347871235723, "grad_norm": 2.6696887016296387, "learning_rate": 5.702744212762634e-07, "loss": 0.008118730038404465, "step": 7109 }, { "epoch": 1.8457943925233646, "grad_norm": 1.7512924671173096, "learning_rate": 5.700400290426841e-07, "loss": 0.0021751532331109047, "step": 7110 }, { "epoch": 1.846053997923157, "grad_norm": 3.5882346630096436, "learning_rate": 5.698060221761133e-07, "loss": 0.007855894975364208, "step": 7111 }, { "epoch": 1.8463136033229492, "grad_norm": 3.630547285079956, "learning_rate": 5.695724007179182e-07, "loss": 0.005717086605727673, "step": 7112 }, { "epoch": 1.8465732087227416, "grad_norm": 4.5339460372924805, "learning_rate": 5.693391647093975e-07, "loss": 0.012584453448653221, "step": 7113 }, { "epoch": 1.8468328141225339, "grad_norm": 1.2533798217773438, "learning_rate": 5.69106314191782e-07, "loss": 0.003536083735525608, "step": 7114 }, { "epoch": 1.8470924195223262, "grad_norm": 3.8368823528289795, "learning_rate": 5.688738492062342e-07, "loss": 0.012136375531554222, "step": 7115 }, { "epoch": 1.8473520249221185, "grad_norm": 1.9401341676712036, "learning_rate": 5.686417697938483e-07, "loss": 0.005061618983745575, "step": 7116 }, { "epoch": 1.8476116303219108, "grad_norm": 0.7772173285484314, "learning_rate": 5.684100759956516e-07, "loss": 0.0024106325581669807, "step": 7117 }, { "epoch": 1.847871235721703, "grad_norm": 3.761279582977295, "learning_rate": 5.681787678526015e-07, "loss": 0.006759183946996927, "step": 7118 }, { "epoch": 1.8481308411214954, "grad_norm": 3.537787914276123, "learning_rate": 5.679478454055882e-07, "loss": 0.025211181491613388, "step": 7119 }, { "epoch": 1.8483904465212877, "grad_norm": 3.2937166690826416, "learning_rate": 5.677173086954331e-07, "loss": 0.012541592121124268, "step": 7120 }, { "epoch": 1.84865005192108, "grad_norm": 1.561453104019165, "learning_rate": 5.674871577628907e-07, "loss": 0.004180294461548328, "step": 7121 }, { "epoch": 1.8489096573208723, "grad_norm": 3.280066967010498, "learning_rate": 5.672573926486457e-07, "loss": 0.04233796149492264, "step": 7122 }, { "epoch": 1.8491692627206646, "grad_norm": 3.3115274906158447, "learning_rate": 5.670280133933152e-07, "loss": 0.008944882079958916, "step": 7123 }, { "epoch": 1.849428868120457, "grad_norm": 2.946366310119629, "learning_rate": 5.66799020037449e-07, "loss": 0.005780375096946955, "step": 7124 }, { "epoch": 1.8496884735202492, "grad_norm": 2.155876874923706, "learning_rate": 5.66570412621527e-07, "loss": 0.012810872867703438, "step": 7125 }, { "epoch": 1.8499480789200415, "grad_norm": 4.4317169189453125, "learning_rate": 5.663421911859622e-07, "loss": 0.008655600249767303, "step": 7126 }, { "epoch": 1.8502076843198338, "grad_norm": 4.323980808258057, "learning_rate": 5.661143557710988e-07, "loss": 0.01847391575574875, "step": 7127 }, { "epoch": 1.8504672897196262, "grad_norm": 3.2636611461639404, "learning_rate": 5.658869064172132e-07, "loss": 0.03874236345291138, "step": 7128 }, { "epoch": 1.8507268951194185, "grad_norm": 3.6104323863983154, "learning_rate": 5.656598431645127e-07, "loss": 0.011435156688094139, "step": 7129 }, { "epoch": 1.8509865005192108, "grad_norm": 3.7538981437683105, "learning_rate": 5.654331660531374e-07, "loss": 0.008328190073370934, "step": 7130 }, { "epoch": 1.851246105919003, "grad_norm": 3.463080406188965, "learning_rate": 5.652068751231583e-07, "loss": 0.012163519859313965, "step": 7131 }, { "epoch": 1.8515057113187954, "grad_norm": 2.334101915359497, "learning_rate": 5.649809704145786e-07, "loss": 0.005374250002205372, "step": 7132 }, { "epoch": 1.8517653167185877, "grad_norm": 5.338383197784424, "learning_rate": 5.64755451967333e-07, "loss": 0.007122978568077087, "step": 7133 }, { "epoch": 1.85202492211838, "grad_norm": 3.727155923843384, "learning_rate": 5.645303198212878e-07, "loss": 0.003394525730982423, "step": 7134 }, { "epoch": 1.8522845275181723, "grad_norm": 1.6449477672576904, "learning_rate": 5.643055740162418e-07, "loss": 0.007911695167422295, "step": 7135 }, { "epoch": 1.8525441329179646, "grad_norm": 4.718925952911377, "learning_rate": 5.640812145919245e-07, "loss": 0.02020779438316822, "step": 7136 }, { "epoch": 1.852803738317757, "grad_norm": 3.369560718536377, "learning_rate": 5.638572415879972e-07, "loss": 0.008175553753972054, "step": 7137 }, { "epoch": 1.8530633437175492, "grad_norm": 0.793965756893158, "learning_rate": 5.63633655044054e-07, "loss": 0.0014391692820936441, "step": 7138 }, { "epoch": 1.8533229491173415, "grad_norm": 4.304140567779541, "learning_rate": 5.63410454999619e-07, "loss": 0.01962214522063732, "step": 7139 }, { "epoch": 1.8535825545171338, "grad_norm": 3.2093827724456787, "learning_rate": 5.6318764149415e-07, "loss": 0.04231564700603485, "step": 7140 }, { "epoch": 1.8538421599169261, "grad_norm": 1.6093882322311401, "learning_rate": 5.629652145670338e-07, "loss": 0.0034819089341908693, "step": 7141 }, { "epoch": 1.8541017653167184, "grad_norm": 2.5263829231262207, "learning_rate": 5.627431742575914e-07, "loss": 0.007821520790457726, "step": 7142 }, { "epoch": 1.8543613707165107, "grad_norm": 2.815302610397339, "learning_rate": 5.625215206050745e-07, "loss": 0.004946630448102951, "step": 7143 }, { "epoch": 1.854620976116303, "grad_norm": 3.9211678504943848, "learning_rate": 5.623002536486661e-07, "loss": 0.020301908254623413, "step": 7144 }, { "epoch": 1.8548805815160956, "grad_norm": 3.9504342079162598, "learning_rate": 5.620793734274812e-07, "loss": 0.02839423343539238, "step": 7145 }, { "epoch": 1.855140186915888, "grad_norm": 3.2007758617401123, "learning_rate": 5.618588799805662e-07, "loss": 0.03455976024270058, "step": 7146 }, { "epoch": 1.8553997923156802, "grad_norm": 3.570683479309082, "learning_rate": 5.616387733468992e-07, "loss": 0.0406106561422348, "step": 7147 }, { "epoch": 1.8556593977154725, "grad_norm": 4.390569686889648, "learning_rate": 5.614190535653909e-07, "loss": 0.007733501959592104, "step": 7148 }, { "epoch": 1.8559190031152648, "grad_norm": 3.117478847503662, "learning_rate": 5.611997206748815e-07, "loss": 0.004188652150332928, "step": 7149 }, { "epoch": 1.8561786085150571, "grad_norm": 5.984469413757324, "learning_rate": 5.60980774714145e-07, "loss": 0.00792638398706913, "step": 7150 }, { "epoch": 1.8564382139148494, "grad_norm": 0.4005078673362732, "learning_rate": 5.607622157218851e-07, "loss": 0.000796155072748661, "step": 7151 }, { "epoch": 1.8566978193146417, "grad_norm": 2.738675355911255, "learning_rate": 5.60544043736739e-07, "loss": 0.02384260855615139, "step": 7152 }, { "epoch": 1.856957424714434, "grad_norm": 1.093251347541809, "learning_rate": 5.60326258797274e-07, "loss": 0.0022700419649481773, "step": 7153 }, { "epoch": 1.8572170301142263, "grad_norm": 2.1349518299102783, "learning_rate": 5.601088609419893e-07, "loss": 0.004811184946447611, "step": 7154 }, { "epoch": 1.8574766355140186, "grad_norm": 0.776496410369873, "learning_rate": 5.598918502093165e-07, "loss": 0.0014503768179565668, "step": 7155 }, { "epoch": 1.857736240913811, "grad_norm": 0.13045115768909454, "learning_rate": 5.596752266376175e-07, "loss": 0.0005589823704212904, "step": 7156 }, { "epoch": 1.8579958463136035, "grad_norm": 3.4161815643310547, "learning_rate": 5.594589902651866e-07, "loss": 0.04931575059890747, "step": 7157 }, { "epoch": 1.8582554517133958, "grad_norm": 1.846421480178833, "learning_rate": 5.592431411302498e-07, "loss": 0.004526838660240173, "step": 7158 }, { "epoch": 1.858515057113188, "grad_norm": 0.5407582521438599, "learning_rate": 5.590276792709639e-07, "loss": 0.0010127852438017726, "step": 7159 }, { "epoch": 1.8587746625129804, "grad_norm": 3.51926326751709, "learning_rate": 5.588126047254173e-07, "loss": 0.014258893206715584, "step": 7160 }, { "epoch": 1.8590342679127727, "grad_norm": 4.105363845825195, "learning_rate": 5.58597917531631e-07, "loss": 0.013165263459086418, "step": 7161 }, { "epoch": 1.859293873312565, "grad_norm": 1.670058012008667, "learning_rate": 5.583836177275561e-07, "loss": 0.005371667444705963, "step": 7162 }, { "epoch": 1.8595534787123573, "grad_norm": 5.300353050231934, "learning_rate": 5.581697053510768e-07, "loss": 0.011267207562923431, "step": 7163 }, { "epoch": 1.8598130841121496, "grad_norm": 6.524885654449463, "learning_rate": 5.579561804400068e-07, "loss": 0.009709961712360382, "step": 7164 }, { "epoch": 1.860072689511942, "grad_norm": 2.4842875003814697, "learning_rate": 5.577430430320932e-07, "loss": 0.00626692920923233, "step": 7165 }, { "epoch": 1.8603322949117342, "grad_norm": 2.483290910720825, "learning_rate": 5.575302931650135e-07, "loss": 0.0068099768832325935, "step": 7166 }, { "epoch": 1.8605919003115265, "grad_norm": 3.0638113021850586, "learning_rate": 5.573179308763772e-07, "loss": 0.0346846804022789, "step": 7167 }, { "epoch": 1.8608515057113189, "grad_norm": 0.8278377056121826, "learning_rate": 5.571059562037244e-07, "loss": 0.0034276144579052925, "step": 7168 }, { "epoch": 1.8611111111111112, "grad_norm": 2.5782344341278076, "learning_rate": 5.568943691845285e-07, "loss": 0.005408242344856262, "step": 7169 }, { "epoch": 1.8613707165109035, "grad_norm": 0.1761815845966339, "learning_rate": 5.566831698561925e-07, "loss": 0.00043746549636125565, "step": 7170 }, { "epoch": 1.8616303219106958, "grad_norm": 3.1515188217163086, "learning_rate": 5.56472358256052e-07, "loss": 0.03613464534282684, "step": 7171 }, { "epoch": 1.861889927310488, "grad_norm": 2.8545632362365723, "learning_rate": 5.56261934421373e-07, "loss": 0.013117045164108276, "step": 7172 }, { "epoch": 1.8621495327102804, "grad_norm": 3.016235113143921, "learning_rate": 5.560518983893542e-07, "loss": 0.022322459146380424, "step": 7173 }, { "epoch": 1.8624091381100727, "grad_norm": 3.7423996925354004, "learning_rate": 5.558422501971251e-07, "loss": 0.027379300445318222, "step": 7174 }, { "epoch": 1.862668743509865, "grad_norm": 1.5080194473266602, "learning_rate": 5.556329898817467e-07, "loss": 0.004652729257941246, "step": 7175 }, { "epoch": 1.8629283489096573, "grad_norm": 0.44829031825065613, "learning_rate": 5.554241174802112e-07, "loss": 0.0006378047401085496, "step": 7176 }, { "epoch": 1.8631879543094496, "grad_norm": 1.1782101392745972, "learning_rate": 5.552156330294427e-07, "loss": 0.007852421142160892, "step": 7177 }, { "epoch": 1.863447559709242, "grad_norm": 2.5438218116760254, "learning_rate": 5.55007536566296e-07, "loss": 0.0055859629064798355, "step": 7178 }, { "epoch": 1.8637071651090342, "grad_norm": 2.3218510150909424, "learning_rate": 5.547998281275589e-07, "loss": 0.005704367533326149, "step": 7179 }, { "epoch": 1.8639667705088265, "grad_norm": 0.9625756740570068, "learning_rate": 5.545925077499481e-07, "loss": 0.0016445027431473136, "step": 7180 }, { "epoch": 1.8642263759086188, "grad_norm": 3.1386353969573975, "learning_rate": 5.543855754701142e-07, "loss": 0.0072293877601623535, "step": 7181 }, { "epoch": 1.8644859813084111, "grad_norm": 2.591862440109253, "learning_rate": 5.541790313246371e-07, "loss": 0.008219035342335701, "step": 7182 }, { "epoch": 1.8647455867082035, "grad_norm": 2.610032320022583, "learning_rate": 5.539728753500302e-07, "loss": 0.005704126320779324, "step": 7183 }, { "epoch": 1.8650051921079958, "grad_norm": 4.411314010620117, "learning_rate": 5.537671075827364e-07, "loss": 0.005945143289864063, "step": 7184 }, { "epoch": 1.865264797507788, "grad_norm": 1.9354870319366455, "learning_rate": 5.535617280591308e-07, "loss": 0.001599668525159359, "step": 7185 }, { "epoch": 1.8655244029075804, "grad_norm": 2.517610788345337, "learning_rate": 5.5335673681552e-07, "loss": 0.0033024405129253864, "step": 7186 }, { "epoch": 1.8657840083073727, "grad_norm": 3.364359140396118, "learning_rate": 5.531521338881417e-07, "loss": 0.02345675230026245, "step": 7187 }, { "epoch": 1.866043613707165, "grad_norm": 3.4418294429779053, "learning_rate": 5.529479193131649e-07, "loss": 0.019726581871509552, "step": 7188 }, { "epoch": 1.8663032191069573, "grad_norm": 2.008610248565674, "learning_rate": 5.527440931266902e-07, "loss": 0.00878022238612175, "step": 7189 }, { "epoch": 1.8665628245067496, "grad_norm": 4.429602146148682, "learning_rate": 5.525406553647495e-07, "loss": 0.004426583647727966, "step": 7190 }, { "epoch": 1.866822429906542, "grad_norm": 2.2417447566986084, "learning_rate": 5.523376060633054e-07, "loss": 0.0022897268645465374, "step": 7191 }, { "epoch": 1.8670820353063342, "grad_norm": 0.37852466106414795, "learning_rate": 5.52134945258253e-07, "loss": 0.0008110635681077838, "step": 7192 }, { "epoch": 1.8673416407061267, "grad_norm": 2.7983505725860596, "learning_rate": 5.519326729854176e-07, "loss": 0.008265362121164799, "step": 7193 }, { "epoch": 1.867601246105919, "grad_norm": 2.4673924446105957, "learning_rate": 5.517307892805569e-07, "loss": 0.013634277507662773, "step": 7194 }, { "epoch": 1.8678608515057114, "grad_norm": 2.312252998352051, "learning_rate": 5.515292941793587e-07, "loss": 0.002969796769320965, "step": 7195 }, { "epoch": 1.8681204569055037, "grad_norm": 0.2513655126094818, "learning_rate": 5.513281877174431e-07, "loss": 0.0004394722345750779, "step": 7196 }, { "epoch": 1.868380062305296, "grad_norm": 3.2886953353881836, "learning_rate": 5.51127469930361e-07, "loss": 0.0028169448487460613, "step": 7197 }, { "epoch": 1.8686396677050883, "grad_norm": 3.3936071395874023, "learning_rate": 5.509271408535949e-07, "loss": 0.009902828373014927, "step": 7198 }, { "epoch": 1.8688992731048806, "grad_norm": 4.184501647949219, "learning_rate": 5.507272005225578e-07, "loss": 0.0061059873551130295, "step": 7199 }, { "epoch": 1.8691588785046729, "grad_norm": 0.8170734643936157, "learning_rate": 5.505276489725951e-07, "loss": 0.001241199905052781, "step": 7200 }, { "epoch": 1.8694184839044652, "grad_norm": 2.287517786026001, "learning_rate": 5.503284862389828e-07, "loss": 0.019588161259889603, "step": 7201 }, { "epoch": 1.8696780893042575, "grad_norm": 6.660347938537598, "learning_rate": 5.501297123569287e-07, "loss": 0.015693582594394684, "step": 7202 }, { "epoch": 1.8699376947040498, "grad_norm": 3.461780548095703, "learning_rate": 5.499313273615706e-07, "loss": 0.011009378358721733, "step": 7203 }, { "epoch": 1.870197300103842, "grad_norm": 0.9545326828956604, "learning_rate": 5.49733331287979e-07, "loss": 0.0028477865271270275, "step": 7204 }, { "epoch": 1.8704569055036346, "grad_norm": 2.580543279647827, "learning_rate": 5.495357241711554e-07, "loss": 0.0064238011837005615, "step": 7205 }, { "epoch": 1.870716510903427, "grad_norm": 4.0080695152282715, "learning_rate": 5.493385060460316e-07, "loss": 0.030093207955360413, "step": 7206 }, { "epoch": 1.8709761163032192, "grad_norm": 1.9151995182037354, "learning_rate": 5.491416769474714e-07, "loss": 0.004491385538130999, "step": 7207 }, { "epoch": 1.8712357217030116, "grad_norm": 4.215667247772217, "learning_rate": 5.489452369102696e-07, "loss": 0.020673386752605438, "step": 7208 }, { "epoch": 1.8714953271028039, "grad_norm": 3.7569406032562256, "learning_rate": 5.487491859691525e-07, "loss": 0.04476607218384743, "step": 7209 }, { "epoch": 1.8717549325025962, "grad_norm": 3.1414988040924072, "learning_rate": 5.485535241587777e-07, "loss": 0.01688188873231411, "step": 7210 }, { "epoch": 1.8720145379023885, "grad_norm": 2.8639626502990723, "learning_rate": 5.483582515137331e-07, "loss": 0.002575002843514085, "step": 7211 }, { "epoch": 1.8722741433021808, "grad_norm": 3.8543989658355713, "learning_rate": 5.481633680685388e-07, "loss": 0.013828856870532036, "step": 7212 }, { "epoch": 1.872533748701973, "grad_norm": 2.746713876724243, "learning_rate": 5.479688738576457e-07, "loss": 0.022401494905352592, "step": 7213 }, { "epoch": 1.8727933541017654, "grad_norm": 3.194331407546997, "learning_rate": 5.477747689154355e-07, "loss": 0.018883036449551582, "step": 7214 }, { "epoch": 1.8730529595015577, "grad_norm": 0.17832297086715698, "learning_rate": 5.475810532762226e-07, "loss": 0.0005130681674927473, "step": 7215 }, { "epoch": 1.87331256490135, "grad_norm": 2.9029738903045654, "learning_rate": 5.473877269742502e-07, "loss": 0.007590333931148052, "step": 7216 }, { "epoch": 1.8735721703011423, "grad_norm": 4.180056095123291, "learning_rate": 5.471947900436947e-07, "loss": 0.003917967900633812, "step": 7217 }, { "epoch": 1.8738317757009346, "grad_norm": 2.52348256111145, "learning_rate": 5.470022425186627e-07, "loss": 0.0068127806298434734, "step": 7218 }, { "epoch": 1.874091381100727, "grad_norm": 3.182650566101074, "learning_rate": 5.468100844331922e-07, "loss": 0.011525554582476616, "step": 7219 }, { "epoch": 1.8743509865005192, "grad_norm": 3.2641055583953857, "learning_rate": 5.466183158212525e-07, "loss": 0.010724728927016258, "step": 7220 }, { "epoch": 1.8746105919003115, "grad_norm": 3.4809577465057373, "learning_rate": 5.464269367167439e-07, "loss": 0.014837164431810379, "step": 7221 }, { "epoch": 1.8748701973001038, "grad_norm": 2.8623108863830566, "learning_rate": 5.462359471534976e-07, "loss": 0.026659101247787476, "step": 7222 }, { "epoch": 1.8751298026998962, "grad_norm": 4.17201566696167, "learning_rate": 5.460453471652765e-07, "loss": 0.008377711288630962, "step": 7223 }, { "epoch": 1.8753894080996885, "grad_norm": 3.603309392929077, "learning_rate": 5.45855136785774e-07, "loss": 0.026816479861736298, "step": 7224 }, { "epoch": 1.8756490134994808, "grad_norm": 3.070174217224121, "learning_rate": 5.45665316048615e-07, "loss": 0.016912182793021202, "step": 7225 }, { "epoch": 1.875908618899273, "grad_norm": 1.52097487449646, "learning_rate": 5.454758849873559e-07, "loss": 0.003343141172081232, "step": 7226 }, { "epoch": 1.8761682242990654, "grad_norm": 3.086895227432251, "learning_rate": 5.45286843635483e-07, "loss": 0.029720058664679527, "step": 7227 }, { "epoch": 1.8764278296988577, "grad_norm": 2.81097149848938, "learning_rate": 5.450981920264152e-07, "loss": 0.04662569984793663, "step": 7228 }, { "epoch": 1.87668743509865, "grad_norm": 0.8906770944595337, "learning_rate": 5.449099301935015e-07, "loss": 0.001204008120112121, "step": 7229 }, { "epoch": 1.8769470404984423, "grad_norm": 1.9909980297088623, "learning_rate": 5.447220581700222e-07, "loss": 0.0037109069526195526, "step": 7230 }, { "epoch": 1.8772066458982346, "grad_norm": 3.490443468093872, "learning_rate": 5.445345759891889e-07, "loss": 0.013842765241861343, "step": 7231 }, { "epoch": 1.877466251298027, "grad_norm": 2.2851061820983887, "learning_rate": 5.44347483684144e-07, "loss": 0.006698129698634148, "step": 7232 }, { "epoch": 1.8777258566978192, "grad_norm": 5.417111396789551, "learning_rate": 5.441607812879614e-07, "loss": 0.018455851823091507, "step": 7233 }, { "epoch": 1.8779854620976115, "grad_norm": 1.066818356513977, "learning_rate": 5.439744688336453e-07, "loss": 0.005530647002160549, "step": 7234 }, { "epoch": 1.8782450674974038, "grad_norm": 3.686638116836548, "learning_rate": 5.43788546354132e-07, "loss": 0.018717268481850624, "step": 7235 }, { "epoch": 1.8785046728971961, "grad_norm": 2.6859853267669678, "learning_rate": 5.436030138822881e-07, "loss": 0.005903633777052164, "step": 7236 }, { "epoch": 1.8787642782969884, "grad_norm": 2.107656717300415, "learning_rate": 5.434178714509117e-07, "loss": 0.003259404795244336, "step": 7237 }, { "epoch": 1.8790238836967808, "grad_norm": 2.2127602100372314, "learning_rate": 5.432331190927316e-07, "loss": 0.007179875858128071, "step": 7238 }, { "epoch": 1.879283489096573, "grad_norm": 3.0169830322265625, "learning_rate": 5.430487568404078e-07, "loss": 0.011541426181793213, "step": 7239 }, { "epoch": 1.8795430944963654, "grad_norm": 2.9448814392089844, "learning_rate": 5.428647847265313e-07, "loss": 0.012855229899287224, "step": 7240 }, { "epoch": 1.879802699896158, "grad_norm": 3.493669271469116, "learning_rate": 5.426812027836241e-07, "loss": 0.010572058148682117, "step": 7241 }, { "epoch": 1.8800623052959502, "grad_norm": 0.743689239025116, "learning_rate": 5.424980110441392e-07, "loss": 0.0013743180315941572, "step": 7242 }, { "epoch": 1.8803219106957425, "grad_norm": 3.935844659805298, "learning_rate": 5.423152095404608e-07, "loss": 0.05681348964571953, "step": 7243 }, { "epoch": 1.8805815160955348, "grad_norm": 3.448451042175293, "learning_rate": 5.421327983049046e-07, "loss": 0.011920860037207603, "step": 7244 }, { "epoch": 1.8808411214953271, "grad_norm": 2.353099822998047, "learning_rate": 5.419507773697157e-07, "loss": 0.00569689366966486, "step": 7245 }, { "epoch": 1.8811007268951194, "grad_norm": 3.0266711711883545, "learning_rate": 5.417691467670719e-07, "loss": 0.015487240627408028, "step": 7246 }, { "epoch": 1.8813603322949117, "grad_norm": 3.9751827716827393, "learning_rate": 5.415879065290811e-07, "loss": 0.021895935758948326, "step": 7247 }, { "epoch": 1.881619937694704, "grad_norm": 3.0422070026397705, "learning_rate": 5.414070566877826e-07, "loss": 0.0459723137319088, "step": 7248 }, { "epoch": 1.8818795430944963, "grad_norm": 3.948197364807129, "learning_rate": 5.412265972751465e-07, "loss": 0.03145270794630051, "step": 7249 }, { "epoch": 1.8821391484942886, "grad_norm": 2.3137829303741455, "learning_rate": 5.410465283230735e-07, "loss": 0.00966072827577591, "step": 7250 }, { "epoch": 1.882398753894081, "grad_norm": 3.4626858234405518, "learning_rate": 5.408668498633959e-07, "loss": 0.016112498939037323, "step": 7251 }, { "epoch": 1.8826583592938733, "grad_norm": 3.534564971923828, "learning_rate": 5.406875619278772e-07, "loss": 0.01894550397992134, "step": 7252 }, { "epoch": 1.8829179646936658, "grad_norm": 2.925053358078003, "learning_rate": 5.405086645482108e-07, "loss": 0.009492949582636356, "step": 7253 }, { "epoch": 1.883177570093458, "grad_norm": 4.154564380645752, "learning_rate": 5.40330157756022e-07, "loss": 0.017029855400323868, "step": 7254 }, { "epoch": 1.8834371754932504, "grad_norm": 4.246998310089111, "learning_rate": 5.401520415828664e-07, "loss": 0.009703803807497025, "step": 7255 }, { "epoch": 1.8836967808930427, "grad_norm": 3.454150915145874, "learning_rate": 5.399743160602311e-07, "loss": 0.015948303043842316, "step": 7256 }, { "epoch": 1.883956386292835, "grad_norm": 3.598790168762207, "learning_rate": 5.397969812195336e-07, "loss": 0.006467207334935665, "step": 7257 }, { "epoch": 1.8842159916926273, "grad_norm": 3.3288800716400146, "learning_rate": 5.396200370921232e-07, "loss": 0.059959813952445984, "step": 7258 }, { "epoch": 1.8844755970924196, "grad_norm": 1.4638714790344238, "learning_rate": 5.394434837092792e-07, "loss": 0.0019132217857986689, "step": 7259 }, { "epoch": 1.884735202492212, "grad_norm": 1.3412830829620361, "learning_rate": 5.392673211022122e-07, "loss": 0.00938809011131525, "step": 7260 }, { "epoch": 1.8849948078920042, "grad_norm": 3.606576442718506, "learning_rate": 5.390915493020638e-07, "loss": 0.0047776540741324425, "step": 7261 }, { "epoch": 1.8852544132917965, "grad_norm": 3.3555006980895996, "learning_rate": 5.389161683399063e-07, "loss": 0.004362087696790695, "step": 7262 }, { "epoch": 1.8855140186915889, "grad_norm": 3.1855862140655518, "learning_rate": 5.387411782467429e-07, "loss": 0.012738591060042381, "step": 7263 }, { "epoch": 1.8857736240913812, "grad_norm": 3.7816483974456787, "learning_rate": 5.385665790535087e-07, "loss": 0.01085919700562954, "step": 7264 }, { "epoch": 1.8860332294911735, "grad_norm": 3.9882829189300537, "learning_rate": 5.383923707910677e-07, "loss": 0.03162121772766113, "step": 7265 }, { "epoch": 1.8862928348909658, "grad_norm": 3.6622562408447266, "learning_rate": 5.382185534902165e-07, "loss": 0.016385158523917198, "step": 7266 }, { "epoch": 1.886552440290758, "grad_norm": 3.80850887298584, "learning_rate": 5.380451271816824e-07, "loss": 0.03729844093322754, "step": 7267 }, { "epoch": 1.8868120456905504, "grad_norm": 2.7870898246765137, "learning_rate": 5.378720918961223e-07, "loss": 0.046886831521987915, "step": 7268 }, { "epoch": 1.8870716510903427, "grad_norm": 1.8105816841125488, "learning_rate": 5.376994476641252e-07, "loss": 0.0026136694941669703, "step": 7269 }, { "epoch": 1.887331256490135, "grad_norm": 0.45910704135894775, "learning_rate": 5.375271945162114e-07, "loss": 0.0014578080736100674, "step": 7270 }, { "epoch": 1.8875908618899273, "grad_norm": 0.9575899243354797, "learning_rate": 5.373553324828303e-07, "loss": 0.002250192454084754, "step": 7271 }, { "epoch": 1.8878504672897196, "grad_norm": 6.580790042877197, "learning_rate": 5.371838615943637e-07, "loss": 0.004559775348752737, "step": 7272 }, { "epoch": 1.888110072689512, "grad_norm": 1.7367874383926392, "learning_rate": 5.370127818811234e-07, "loss": 0.0030949432402849197, "step": 7273 }, { "epoch": 1.8883696780893042, "grad_norm": 1.4002525806427002, "learning_rate": 5.368420933733525e-07, "loss": 0.0016581735108047724, "step": 7274 }, { "epoch": 1.8886292834890965, "grad_norm": 2.6662404537200928, "learning_rate": 5.366717961012253e-07, "loss": 0.006574234459549189, "step": 7275 }, { "epoch": 1.8888888888888888, "grad_norm": 1.2118979692459106, "learning_rate": 5.365018900948458e-07, "loss": 0.00251657678745687, "step": 7276 }, { "epoch": 1.8891484942886811, "grad_norm": 4.364222526550293, "learning_rate": 5.363323753842498e-07, "loss": 0.043774671852588654, "step": 7277 }, { "epoch": 1.8894080996884735, "grad_norm": 0.5665795207023621, "learning_rate": 5.361632519994034e-07, "loss": 0.0007564707193523645, "step": 7278 }, { "epoch": 1.8896677050882658, "grad_norm": 0.9731966257095337, "learning_rate": 5.359945199702042e-07, "loss": 0.00578647293150425, "step": 7279 }, { "epoch": 1.889927310488058, "grad_norm": 3.5008347034454346, "learning_rate": 5.358261793264796e-07, "loss": 0.048304103314876556, "step": 7280 }, { "epoch": 1.8901869158878504, "grad_norm": 4.560802459716797, "learning_rate": 5.356582300979887e-07, "loss": 0.0058380248956382275, "step": 7281 }, { "epoch": 1.8904465212876427, "grad_norm": 2.9150936603546143, "learning_rate": 5.354906723144211e-07, "loss": 0.031040124595165253, "step": 7282 }, { "epoch": 1.890706126687435, "grad_norm": 0.70963454246521, "learning_rate": 5.353235060053971e-07, "loss": 0.0010192464105784893, "step": 7283 }, { "epoch": 1.8909657320872273, "grad_norm": 3.933180570602417, "learning_rate": 5.351567312004675e-07, "loss": 0.015720581635832787, "step": 7284 }, { "epoch": 1.8912253374870196, "grad_norm": 1.288228154182434, "learning_rate": 5.349903479291153e-07, "loss": 0.0013334538089111447, "step": 7285 }, { "epoch": 1.891484942886812, "grad_norm": 2.3968162536621094, "learning_rate": 5.348243562207519e-07, "loss": 0.005296013318002224, "step": 7286 }, { "epoch": 1.8917445482866042, "grad_norm": 3.219050645828247, "learning_rate": 5.34658756104722e-07, "loss": 0.012814042158424854, "step": 7287 }, { "epoch": 1.8920041536863965, "grad_norm": 3.9721415042877197, "learning_rate": 5.344935476102992e-07, "loss": 0.028216734528541565, "step": 7288 }, { "epoch": 1.892263759086189, "grad_norm": 4.03662109375, "learning_rate": 5.34328730766689e-07, "loss": 0.018811242654919624, "step": 7289 }, { "epoch": 1.8925233644859814, "grad_norm": 4.790661334991455, "learning_rate": 5.341643056030269e-07, "loss": 0.004457185976207256, "step": 7290 }, { "epoch": 1.8927829698857737, "grad_norm": 2.514679431915283, "learning_rate": 5.340002721483798e-07, "loss": 0.003978288266807795, "step": 7291 }, { "epoch": 1.893042575285566, "grad_norm": 1.2271389961242676, "learning_rate": 5.338366304317445e-07, "loss": 0.0041125742718577385, "step": 7292 }, { "epoch": 1.8933021806853583, "grad_norm": 2.4340782165527344, "learning_rate": 5.336733804820501e-07, "loss": 0.004425420425832272, "step": 7293 }, { "epoch": 1.8935617860851506, "grad_norm": 2.348083734512329, "learning_rate": 5.335105223281544e-07, "loss": 0.01597687602043152, "step": 7294 }, { "epoch": 1.8938213914849429, "grad_norm": 3.5190837383270264, "learning_rate": 5.333480559988477e-07, "loss": 0.035007357597351074, "step": 7295 }, { "epoch": 1.8940809968847352, "grad_norm": 1.222572684288025, "learning_rate": 5.331859815228501e-07, "loss": 0.00440414110198617, "step": 7296 }, { "epoch": 1.8943406022845275, "grad_norm": 3.69303035736084, "learning_rate": 5.330242989288127e-07, "loss": 0.018776241689920425, "step": 7297 }, { "epoch": 1.8946002076843198, "grad_norm": 4.252735614776611, "learning_rate": 5.328630082453173e-07, "loss": 0.011939945630729198, "step": 7298 }, { "epoch": 1.894859813084112, "grad_norm": 2.197715997695923, "learning_rate": 5.327021095008762e-07, "loss": 0.0068251630291342735, "step": 7299 }, { "epoch": 1.8951194184839044, "grad_norm": 1.1852282285690308, "learning_rate": 5.325416027239327e-07, "loss": 0.0040390812791883945, "step": 7300 }, { "epoch": 1.895379023883697, "grad_norm": 2.392341136932373, "learning_rate": 5.323814879428608e-07, "loss": 0.0028111040592193604, "step": 7301 }, { "epoch": 1.8956386292834893, "grad_norm": 2.7576687335968018, "learning_rate": 5.322217651859652e-07, "loss": 0.0105920834466815, "step": 7302 }, { "epoch": 1.8958982346832816, "grad_norm": 0.995998203754425, "learning_rate": 5.320624344814812e-07, "loss": 0.0019453081768006086, "step": 7303 }, { "epoch": 1.8961578400830739, "grad_norm": 3.782513380050659, "learning_rate": 5.319034958575744e-07, "loss": 0.020331889390945435, "step": 7304 }, { "epoch": 1.8964174454828662, "grad_norm": 1.357105016708374, "learning_rate": 5.317449493423419e-07, "loss": 0.007803682703524828, "step": 7305 }, { "epoch": 1.8966770508826585, "grad_norm": 2.554581880569458, "learning_rate": 5.315867949638114e-07, "loss": 0.007596006616950035, "step": 7306 }, { "epoch": 1.8969366562824508, "grad_norm": 3.815256118774414, "learning_rate": 5.314290327499404e-07, "loss": 0.0129060884937644, "step": 7307 }, { "epoch": 1.897196261682243, "grad_norm": 3.342313528060913, "learning_rate": 5.312716627286179e-07, "loss": 0.044858433306217194, "step": 7308 }, { "epoch": 1.8974558670820354, "grad_norm": 3.452237606048584, "learning_rate": 5.311146849276632e-07, "loss": 0.018825359642505646, "step": 7309 }, { "epoch": 1.8977154724818277, "grad_norm": 0.763543963432312, "learning_rate": 5.309580993748266e-07, "loss": 0.00120442733168602, "step": 7310 }, { "epoch": 1.89797507788162, "grad_norm": 3.495640277862549, "learning_rate": 5.308019060977887e-07, "loss": 0.0309741348028183, "step": 7311 }, { "epoch": 1.8982346832814123, "grad_norm": 2.063239336013794, "learning_rate": 5.306461051241607e-07, "loss": 0.005873039364814758, "step": 7312 }, { "epoch": 1.8984942886812046, "grad_norm": 2.9647064208984375, "learning_rate": 5.304906964814849e-07, "loss": 0.053515445441007614, "step": 7313 }, { "epoch": 1.898753894080997, "grad_norm": 3.4339711666107178, "learning_rate": 5.303356801972342e-07, "loss": 0.03909539431333542, "step": 7314 }, { "epoch": 1.8990134994807892, "grad_norm": 1.8369718790054321, "learning_rate": 5.301810562988113e-07, "loss": 0.005993004888296127, "step": 7315 }, { "epoch": 1.8992731048805815, "grad_norm": 3.2289018630981445, "learning_rate": 5.300268248135509e-07, "loss": 0.032065413892269135, "step": 7316 }, { "epoch": 1.8995327102803738, "grad_norm": 2.2413487434387207, "learning_rate": 5.29872985768717e-07, "loss": 0.0026971311308443546, "step": 7317 }, { "epoch": 1.8997923156801662, "grad_norm": 2.516416549682617, "learning_rate": 5.297195391915051e-07, "loss": 0.0021684293169528246, "step": 7318 }, { "epoch": 1.9000519210799585, "grad_norm": 0.3273851275444031, "learning_rate": 5.295664851090409e-07, "loss": 0.00038279881118796766, "step": 7319 }, { "epoch": 1.9003115264797508, "grad_norm": 2.890582323074341, "learning_rate": 5.29413823548381e-07, "loss": 0.010318968445062637, "step": 7320 }, { "epoch": 1.900571131879543, "grad_norm": 3.694385528564453, "learning_rate": 5.292615545365124e-07, "loss": 0.017938531935214996, "step": 7321 }, { "epoch": 1.9008307372793354, "grad_norm": 2.8751845359802246, "learning_rate": 5.29109678100353e-07, "loss": 0.00680511724203825, "step": 7322 }, { "epoch": 1.9010903426791277, "grad_norm": 3.02504301071167, "learning_rate": 5.289581942667505e-07, "loss": 0.024816162884235382, "step": 7323 }, { "epoch": 1.90134994807892, "grad_norm": 3.1884517669677734, "learning_rate": 5.288071030624846e-07, "loss": 0.00552557036280632, "step": 7324 }, { "epoch": 1.9016095534787123, "grad_norm": 3.2413339614868164, "learning_rate": 5.286564045142638e-07, "loss": 0.05562442168593407, "step": 7325 }, { "epoch": 1.9018691588785046, "grad_norm": 4.909611225128174, "learning_rate": 5.28506098648729e-07, "loss": 0.014870261773467064, "step": 7326 }, { "epoch": 1.902128764278297, "grad_norm": 3.882551670074463, "learning_rate": 5.283561854924504e-07, "loss": 0.03616086021065712, "step": 7327 }, { "epoch": 1.9023883696780892, "grad_norm": 1.5932416915893555, "learning_rate": 5.282066650719291e-07, "loss": 0.003779376856982708, "step": 7328 }, { "epoch": 1.9026479750778815, "grad_norm": 2.372866630554199, "learning_rate": 5.280575374135975e-07, "loss": 0.0068401312455534935, "step": 7329 }, { "epoch": 1.9029075804776738, "grad_norm": 3.2511723041534424, "learning_rate": 5.279088025438173e-07, "loss": 0.017934761941432953, "step": 7330 }, { "epoch": 1.9031671858774661, "grad_norm": 1.4790539741516113, "learning_rate": 5.277604604888816e-07, "loss": 0.0029670055955648422, "step": 7331 }, { "epoch": 1.9034267912772584, "grad_norm": 4.877974987030029, "learning_rate": 5.27612511275014e-07, "loss": 0.013830922544002533, "step": 7332 }, { "epoch": 1.9036863966770508, "grad_norm": 3.9312546253204346, "learning_rate": 5.274649549283685e-07, "loss": 0.014198722317814827, "step": 7333 }, { "epoch": 1.903946002076843, "grad_norm": 0.530758798122406, "learning_rate": 5.273177914750298e-07, "loss": 0.005976269952952862, "step": 7334 }, { "epoch": 1.9042056074766354, "grad_norm": 1.8012012243270874, "learning_rate": 5.271710209410127e-07, "loss": 0.005977187305688858, "step": 7335 }, { "epoch": 1.9044652128764277, "grad_norm": 3.6038882732391357, "learning_rate": 5.270246433522629e-07, "loss": 0.028635257855057716, "step": 7336 }, { "epoch": 1.9047248182762202, "grad_norm": 2.4913783073425293, "learning_rate": 5.268786587346573e-07, "loss": 0.007311176508665085, "step": 7337 }, { "epoch": 1.9049844236760125, "grad_norm": 2.6666409969329834, "learning_rate": 5.267330671140016e-07, "loss": 0.008848625235259533, "step": 7338 }, { "epoch": 1.9052440290758048, "grad_norm": 2.6562962532043457, "learning_rate": 5.265878685160339e-07, "loss": 0.008085154928267002, "step": 7339 }, { "epoch": 1.9055036344755971, "grad_norm": 2.434439182281494, "learning_rate": 5.264430629664214e-07, "loss": 0.005936004687100649, "step": 7340 }, { "epoch": 1.9057632398753894, "grad_norm": 1.2130906581878662, "learning_rate": 5.262986504907629e-07, "loss": 0.003856787458062172, "step": 7341 }, { "epoch": 1.9060228452751817, "grad_norm": 3.7958977222442627, "learning_rate": 5.261546311145868e-07, "loss": 0.03601822257041931, "step": 7342 }, { "epoch": 1.906282450674974, "grad_norm": 3.5391156673431396, "learning_rate": 5.260110048633527e-07, "loss": 0.012443960644304752, "step": 7343 }, { "epoch": 1.9065420560747663, "grad_norm": 0.6935815215110779, "learning_rate": 5.258677717624502e-07, "loss": 0.000883410859387368, "step": 7344 }, { "epoch": 1.9068016614745587, "grad_norm": 1.513117790222168, "learning_rate": 5.257249318372001e-07, "loss": 0.0028248419985175133, "step": 7345 }, { "epoch": 1.907061266874351, "grad_norm": 2.3631629943847656, "learning_rate": 5.255824851128525e-07, "loss": 0.004496270790696144, "step": 7346 }, { "epoch": 1.9073208722741433, "grad_norm": 2.9416041374206543, "learning_rate": 5.254404316145894e-07, "loss": 0.009164223447442055, "step": 7347 }, { "epoch": 1.9075804776739356, "grad_norm": 3.9482288360595703, "learning_rate": 5.252987713675222e-07, "loss": 0.03226816654205322, "step": 7348 }, { "epoch": 1.907840083073728, "grad_norm": 3.344122886657715, "learning_rate": 5.251575043966932e-07, "loss": 0.034996818751096725, "step": 7349 }, { "epoch": 1.9080996884735204, "grad_norm": 4.836042881011963, "learning_rate": 5.250166307270754e-07, "loss": 0.019465966150164604, "step": 7350 }, { "epoch": 1.9083592938733127, "grad_norm": 3.515657901763916, "learning_rate": 5.24876150383572e-07, "loss": 0.038871265947818756, "step": 7351 }, { "epoch": 1.908618899273105, "grad_norm": 3.0425338745117188, "learning_rate": 5.247360633910168e-07, "loss": 0.040029626339673996, "step": 7352 }, { "epoch": 1.9088785046728973, "grad_norm": 3.539876699447632, "learning_rate": 5.245963697741734e-07, "loss": 0.010048026219010353, "step": 7353 }, { "epoch": 1.9091381100726896, "grad_norm": 1.392513632774353, "learning_rate": 5.244570695577371e-07, "loss": 0.004029575735330582, "step": 7354 }, { "epoch": 1.909397715472482, "grad_norm": 4.892919063568115, "learning_rate": 5.243181627663325e-07, "loss": 0.007318521849811077, "step": 7355 }, { "epoch": 1.9096573208722742, "grad_norm": 1.8759351968765259, "learning_rate": 5.241796494245157e-07, "loss": 0.0074917469173669815, "step": 7356 }, { "epoch": 1.9099169262720666, "grad_norm": 1.6370978355407715, "learning_rate": 5.240415295567719e-07, "loss": 0.0015551972901448607, "step": 7357 }, { "epoch": 1.9101765316718589, "grad_norm": 3.383622407913208, "learning_rate": 5.239038031875184e-07, "loss": 0.035518985241651535, "step": 7358 }, { "epoch": 1.9104361370716512, "grad_norm": 0.8061134815216064, "learning_rate": 5.237664703411014e-07, "loss": 0.005321032367646694, "step": 7359 }, { "epoch": 1.9106957424714435, "grad_norm": 1.152718186378479, "learning_rate": 5.236295310417988e-07, "loss": 0.0030961271841078997, "step": 7360 }, { "epoch": 1.9109553478712358, "grad_norm": 3.5079574584960938, "learning_rate": 5.234929853138177e-07, "loss": 0.005065418779850006, "step": 7361 }, { "epoch": 1.911214953271028, "grad_norm": 1.7293915748596191, "learning_rate": 5.233568331812964e-07, "loss": 0.004795414861291647, "step": 7362 }, { "epoch": 1.9114745586708204, "grad_norm": 3.2312021255493164, "learning_rate": 5.232210746683039e-07, "loss": 0.01054077036678791, "step": 7363 }, { "epoch": 1.9117341640706127, "grad_norm": 3.2103822231292725, "learning_rate": 5.230857097988388e-07, "loss": 0.006395399570465088, "step": 7364 }, { "epoch": 1.911993769470405, "grad_norm": 3.8517510890960693, "learning_rate": 5.229507385968308e-07, "loss": 0.00899852067232132, "step": 7365 }, { "epoch": 1.9122533748701973, "grad_norm": 3.0247135162353516, "learning_rate": 5.228161610861392e-07, "loss": 0.005335697904229164, "step": 7366 }, { "epoch": 1.9125129802699896, "grad_norm": 3.037320137023926, "learning_rate": 5.226819772905546e-07, "loss": 0.032640691846609116, "step": 7367 }, { "epoch": 1.912772585669782, "grad_norm": 5.431070804595947, "learning_rate": 5.22548187233798e-07, "loss": 0.019723663106560707, "step": 7368 }, { "epoch": 1.9130321910695742, "grad_norm": 4.226266384124756, "learning_rate": 5.224147909395197e-07, "loss": 0.0057436153292655945, "step": 7369 }, { "epoch": 1.9132917964693665, "grad_norm": 2.0372371673583984, "learning_rate": 5.222817884313015e-07, "loss": 0.004501032643020153, "step": 7370 }, { "epoch": 1.9135514018691588, "grad_norm": 1.1985567808151245, "learning_rate": 5.22149179732655e-07, "loss": 0.002699565142393112, "step": 7371 }, { "epoch": 1.9138110072689511, "grad_norm": 3.9642748832702637, "learning_rate": 5.220169648670227e-07, "loss": 0.04214935749769211, "step": 7372 }, { "epoch": 1.9140706126687435, "grad_norm": 0.45934659242630005, "learning_rate": 5.21885143857777e-07, "loss": 0.0008055178332142532, "step": 7373 }, { "epoch": 1.9143302180685358, "grad_norm": 3.4429280757904053, "learning_rate": 5.217537167282206e-07, "loss": 0.026696044951677322, "step": 7374 }, { "epoch": 1.914589823468328, "grad_norm": 2.3368706703186035, "learning_rate": 5.21622683501587e-07, "loss": 0.007990151643753052, "step": 7375 }, { "epoch": 1.9148494288681204, "grad_norm": 1.5238317251205444, "learning_rate": 5.214920442010399e-07, "loss": 0.004705457948148251, "step": 7376 }, { "epoch": 1.9151090342679127, "grad_norm": 3.026582956314087, "learning_rate": 5.213617988496733e-07, "loss": 0.030547762289643288, "step": 7377 }, { "epoch": 1.915368639667705, "grad_norm": 3.8285772800445557, "learning_rate": 5.212319474705119e-07, "loss": 0.016008328646421432, "step": 7378 }, { "epoch": 1.9156282450674973, "grad_norm": 1.8787899017333984, "learning_rate": 5.211024900865099e-07, "loss": 0.006647368893027306, "step": 7379 }, { "epoch": 1.9158878504672896, "grad_norm": 0.7164727449417114, "learning_rate": 5.209734267205526e-07, "loss": 0.0025840094313025475, "step": 7380 }, { "epoch": 1.916147455867082, "grad_norm": 3.1667559146881104, "learning_rate": 5.208447573954557e-07, "loss": 0.00850214995443821, "step": 7381 }, { "epoch": 1.9164070612668742, "grad_norm": 3.8324592113494873, "learning_rate": 5.207164821339647e-07, "loss": 0.00424692127853632, "step": 7382 }, { "epoch": 1.9166666666666665, "grad_norm": 3.462336301803589, "learning_rate": 5.205886009587559e-07, "loss": 0.03922060877084732, "step": 7383 }, { "epoch": 1.9169262720664588, "grad_norm": 0.703113317489624, "learning_rate": 5.204611138924358e-07, "loss": 0.0008918328676372766, "step": 7384 }, { "epoch": 1.9171858774662511, "grad_norm": 5.013307094573975, "learning_rate": 5.203340209575409e-07, "loss": 0.010534149594604969, "step": 7385 }, { "epoch": 1.9174454828660437, "grad_norm": 3.627676486968994, "learning_rate": 5.202073221765384e-07, "loss": 0.03274713456630707, "step": 7386 }, { "epoch": 1.917705088265836, "grad_norm": 4.401220321655273, "learning_rate": 5.20081017571826e-07, "loss": 0.011151568964123726, "step": 7387 }, { "epoch": 1.9179646936656283, "grad_norm": 2.1927788257598877, "learning_rate": 5.199551071657315e-07, "loss": 0.010118339210748672, "step": 7388 }, { "epoch": 1.9182242990654206, "grad_norm": 2.455526828765869, "learning_rate": 5.198295909805126e-07, "loss": 0.005674346350133419, "step": 7389 }, { "epoch": 1.9184839044652129, "grad_norm": 2.898406505584717, "learning_rate": 5.197044690383577e-07, "loss": 0.004931179340928793, "step": 7390 }, { "epoch": 1.9187435098650052, "grad_norm": 2.755795955657959, "learning_rate": 5.195797413613861e-07, "loss": 0.006056215148419142, "step": 7391 }, { "epoch": 1.9190031152647975, "grad_norm": 1.2113862037658691, "learning_rate": 5.194554079716461e-07, "loss": 0.0014076285297051072, "step": 7392 }, { "epoch": 1.9192627206645898, "grad_norm": 0.9744245409965515, "learning_rate": 5.19331468891117e-07, "loss": 0.003911563195288181, "step": 7393 }, { "epoch": 1.9195223260643821, "grad_norm": 3.5524888038635254, "learning_rate": 5.192079241417092e-07, "loss": 0.03795580938458443, "step": 7394 }, { "epoch": 1.9197819314641744, "grad_norm": 3.4128639698028564, "learning_rate": 5.190847737452617e-07, "loss": 0.038230523467063904, "step": 7395 }, { "epoch": 1.9200415368639667, "grad_norm": 3.5739452838897705, "learning_rate": 5.189620177235454e-07, "loss": 0.03375254198908806, "step": 7396 }, { "epoch": 1.920301142263759, "grad_norm": 3.612445592880249, "learning_rate": 5.1883965609826e-07, "loss": 0.011233151890337467, "step": 7397 }, { "epoch": 1.9205607476635516, "grad_norm": 3.5029959678649902, "learning_rate": 5.187176888910366e-07, "loss": 0.007068208418786526, "step": 7398 }, { "epoch": 1.9208203530633439, "grad_norm": 1.9621347188949585, "learning_rate": 5.185961161234364e-07, "loss": 0.002132358495146036, "step": 7399 }, { "epoch": 1.9210799584631362, "grad_norm": 3.4085943698883057, "learning_rate": 5.184749378169505e-07, "loss": 0.019220300018787384, "step": 7400 }, { "epoch": 1.9213395638629285, "grad_norm": 1.3124860525131226, "learning_rate": 5.183541539930002e-07, "loss": 0.005154673475772142, "step": 7401 }, { "epoch": 1.9215991692627208, "grad_norm": 1.1050400733947754, "learning_rate": 5.182337646729378e-07, "loss": 0.002155675319954753, "step": 7402 }, { "epoch": 1.921858774662513, "grad_norm": 1.0719507932662964, "learning_rate": 5.18113769878045e-07, "loss": 0.003554792609065771, "step": 7403 }, { "epoch": 1.9221183800623054, "grad_norm": 2.736698865890503, "learning_rate": 5.179941696295345e-07, "loss": 0.011583346873521805, "step": 7404 }, { "epoch": 1.9223779854620977, "grad_norm": 3.816786050796509, "learning_rate": 5.178749639485482e-07, "loss": 0.021633464843034744, "step": 7405 }, { "epoch": 1.92263759086189, "grad_norm": 1.0041849613189697, "learning_rate": 5.177561528561597e-07, "loss": 0.0012919409200549126, "step": 7406 }, { "epoch": 1.9228971962616823, "grad_norm": 3.276735305786133, "learning_rate": 5.176377363733718e-07, "loss": 0.012422065250575542, "step": 7407 }, { "epoch": 1.9231568016614746, "grad_norm": 4.562203884124756, "learning_rate": 5.175197145211176e-07, "loss": 0.016067229211330414, "step": 7408 }, { "epoch": 1.923416407061267, "grad_norm": 0.7352254986763, "learning_rate": 5.174020873202611e-07, "loss": 0.0009393100626766682, "step": 7409 }, { "epoch": 1.9236760124610592, "grad_norm": 2.3213441371917725, "learning_rate": 5.172848547915956e-07, "loss": 0.010902194306254387, "step": 7410 }, { "epoch": 1.9239356178608515, "grad_norm": 2.7896230220794678, "learning_rate": 5.171680169558456e-07, "loss": 0.0067147924564778805, "step": 7411 }, { "epoch": 1.9241952232606438, "grad_norm": 1.9270116090774536, "learning_rate": 5.170515738336647e-07, "loss": 0.00751909613609314, "step": 7412 }, { "epoch": 1.9244548286604362, "grad_norm": 0.83894282579422, "learning_rate": 5.169355254456381e-07, "loss": 0.0013897283934056759, "step": 7413 }, { "epoch": 1.9247144340602285, "grad_norm": 1.7230794429779053, "learning_rate": 5.168198718122804e-07, "loss": 0.002248867880553007, "step": 7414 }, { "epoch": 1.9249740394600208, "grad_norm": 3.3004918098449707, "learning_rate": 5.167046129540361e-07, "loss": 0.02841327339410782, "step": 7415 }, { "epoch": 1.925233644859813, "grad_norm": 3.535629987716675, "learning_rate": 5.165897488912805e-07, "loss": 0.008566059172153473, "step": 7416 }, { "epoch": 1.9254932502596054, "grad_norm": 1.708513617515564, "learning_rate": 5.164752796443191e-07, "loss": 0.00748826190829277, "step": 7417 }, { "epoch": 1.9257528556593977, "grad_norm": 3.6380200386047363, "learning_rate": 5.163612052333871e-07, "loss": 0.019357619807124138, "step": 7418 }, { "epoch": 1.92601246105919, "grad_norm": 1.0699188709259033, "learning_rate": 5.162475256786508e-07, "loss": 0.0012880319263786077, "step": 7419 }, { "epoch": 1.9262720664589823, "grad_norm": 3.5032198429107666, "learning_rate": 5.161342410002055e-07, "loss": 0.047088898718357086, "step": 7420 }, { "epoch": 1.9265316718587746, "grad_norm": 2.945965528488159, "learning_rate": 5.160213512180778e-07, "loss": 0.010212360881268978, "step": 7421 }, { "epoch": 1.926791277258567, "grad_norm": 4.419790744781494, "learning_rate": 5.159088563522239e-07, "loss": 0.03333428129553795, "step": 7422 }, { "epoch": 1.9270508826583592, "grad_norm": 1.0953372716903687, "learning_rate": 5.157967564225303e-07, "loss": 0.002548329997807741, "step": 7423 }, { "epoch": 1.9273104880581515, "grad_norm": 2.3052515983581543, "learning_rate": 5.156850514488136e-07, "loss": 0.012372257187962532, "step": 7424 }, { "epoch": 1.9275700934579438, "grad_norm": 2.9239494800567627, "learning_rate": 5.155737414508208e-07, "loss": 0.00505107780918479, "step": 7425 }, { "epoch": 1.9278296988577361, "grad_norm": 3.2507975101470947, "learning_rate": 5.154628264482289e-07, "loss": 0.010514398105442524, "step": 7426 }, { "epoch": 1.9280893042575284, "grad_norm": 3.1115729808807373, "learning_rate": 5.153523064606452e-07, "loss": 0.030960705131292343, "step": 7427 }, { "epoch": 1.9283489096573208, "grad_norm": 4.438064098358154, "learning_rate": 5.15242181507607e-07, "loss": 0.02418280392885208, "step": 7428 }, { "epoch": 1.928608515057113, "grad_norm": 3.0034148693084717, "learning_rate": 5.151324516085819e-07, "loss": 0.011442085728049278, "step": 7429 }, { "epoch": 1.9288681204569054, "grad_norm": 4.330939292907715, "learning_rate": 5.15023116782968e-07, "loss": 0.03301768749952316, "step": 7430 }, { "epoch": 1.9291277258566977, "grad_norm": 2.597620725631714, "learning_rate": 5.149141770500926e-07, "loss": 0.009593287482857704, "step": 7431 }, { "epoch": 1.92938733125649, "grad_norm": 5.595078945159912, "learning_rate": 5.148056324292142e-07, "loss": 0.014011834748089314, "step": 7432 }, { "epoch": 1.9296469366562823, "grad_norm": 2.9650135040283203, "learning_rate": 5.146974829395207e-07, "loss": 0.006618788931518793, "step": 7433 }, { "epoch": 1.9299065420560748, "grad_norm": 3.6165928840637207, "learning_rate": 5.145897286001304e-07, "loss": 0.010207789950072765, "step": 7434 }, { "epoch": 1.9301661474558671, "grad_norm": 3.4890637397766113, "learning_rate": 5.144823694300921e-07, "loss": 0.032481953501701355, "step": 7435 }, { "epoch": 1.9304257528556594, "grad_norm": 0.4647129476070404, "learning_rate": 5.143754054483844e-07, "loss": 0.007458180654793978, "step": 7436 }, { "epoch": 1.9306853582554517, "grad_norm": 4.59808349609375, "learning_rate": 5.14268836673916e-07, "loss": 0.010500750504434109, "step": 7437 }, { "epoch": 1.930944963655244, "grad_norm": 0.4271654784679413, "learning_rate": 5.141626631255257e-07, "loss": 0.0007476229802705348, "step": 7438 }, { "epoch": 1.9312045690550363, "grad_norm": 2.557283878326416, "learning_rate": 5.140568848219826e-07, "loss": 0.005484872497618198, "step": 7439 }, { "epoch": 1.9314641744548287, "grad_norm": 1.5878856182098389, "learning_rate": 5.139515017819862e-07, "loss": 0.004363042768090963, "step": 7440 }, { "epoch": 1.931723779854621, "grad_norm": 2.7438716888427734, "learning_rate": 5.138465140241657e-07, "loss": 0.00536190252751112, "step": 7441 }, { "epoch": 1.9319833852544133, "grad_norm": 1.6202818155288696, "learning_rate": 5.137419215670798e-07, "loss": 0.003169568022713065, "step": 7442 }, { "epoch": 1.9322429906542056, "grad_norm": 3.766322612762451, "learning_rate": 5.136377244292192e-07, "loss": 0.02965123951435089, "step": 7443 }, { "epoch": 1.9325025960539979, "grad_norm": 3.7943456172943115, "learning_rate": 5.135339226290027e-07, "loss": 0.008554988540709019, "step": 7444 }, { "epoch": 1.9327622014537902, "grad_norm": 3.5003767013549805, "learning_rate": 5.134305161847805e-07, "loss": 0.00970907136797905, "step": 7445 }, { "epoch": 1.9330218068535827, "grad_norm": 1.8602070808410645, "learning_rate": 5.133275051148322e-07, "loss": 0.008375140838325024, "step": 7446 }, { "epoch": 1.933281412253375, "grad_norm": 4.071442604064941, "learning_rate": 5.132248894373682e-07, "loss": 0.014138109982013702, "step": 7447 }, { "epoch": 1.9335410176531673, "grad_norm": 3.6398959159851074, "learning_rate": 5.131226691705285e-07, "loss": 0.016807083040475845, "step": 7448 }, { "epoch": 1.9338006230529596, "grad_norm": 4.100872039794922, "learning_rate": 5.130208443323829e-07, "loss": 0.039037566632032394, "step": 7449 }, { "epoch": 1.934060228452752, "grad_norm": 1.2697126865386963, "learning_rate": 5.129194149409321e-07, "loss": 0.003462970955297351, "step": 7450 }, { "epoch": 1.9343198338525442, "grad_norm": 3.0939013957977295, "learning_rate": 5.128183810141066e-07, "loss": 0.03131995350122452, "step": 7451 }, { "epoch": 1.9345794392523366, "grad_norm": 2.3114442825317383, "learning_rate": 5.127177425697662e-07, "loss": 0.0028629684820771217, "step": 7452 }, { "epoch": 1.9348390446521289, "grad_norm": 3.307260751724243, "learning_rate": 5.126174996257024e-07, "loss": 0.02233584225177765, "step": 7453 }, { "epoch": 1.9350986500519212, "grad_norm": 3.538508176803589, "learning_rate": 5.125176521996353e-07, "loss": 0.013551361858844757, "step": 7454 }, { "epoch": 1.9353582554517135, "grad_norm": 3.817626953125, "learning_rate": 5.124182003092153e-07, "loss": 0.02270021289587021, "step": 7455 }, { "epoch": 1.9356178608515058, "grad_norm": 2.4731557369232178, "learning_rate": 5.123191439720241e-07, "loss": 0.010163278318941593, "step": 7456 }, { "epoch": 1.935877466251298, "grad_norm": 0.9685937762260437, "learning_rate": 5.122204832055723e-07, "loss": 0.0025159872602671385, "step": 7457 }, { "epoch": 1.9361370716510904, "grad_norm": 3.74721097946167, "learning_rate": 5.121222180273004e-07, "loss": 0.04296819865703583, "step": 7458 }, { "epoch": 1.9363966770508827, "grad_norm": 2.186286687850952, "learning_rate": 5.120243484545798e-07, "loss": 0.005729283206164837, "step": 7459 }, { "epoch": 1.936656282450675, "grad_norm": 6.493672847747803, "learning_rate": 5.119268745047115e-07, "loss": 0.0193759985268116, "step": 7460 }, { "epoch": 1.9369158878504673, "grad_norm": 4.161322593688965, "learning_rate": 5.118297961949268e-07, "loss": 0.010053073987364769, "step": 7461 }, { "epoch": 1.9371754932502596, "grad_norm": 2.8838140964508057, "learning_rate": 5.117331135423866e-07, "loss": 0.0069350916892290115, "step": 7462 }, { "epoch": 1.937435098650052, "grad_norm": 4.032637119293213, "learning_rate": 5.116368265641828e-07, "loss": 0.033471472561359406, "step": 7463 }, { "epoch": 1.9376947040498442, "grad_norm": 5.432779312133789, "learning_rate": 5.115409352773361e-07, "loss": 0.036026712507009506, "step": 7464 }, { "epoch": 1.9379543094496365, "grad_norm": 0.19569683074951172, "learning_rate": 5.11445439698798e-07, "loss": 0.0004298343847040087, "step": 7465 }, { "epoch": 1.9382139148494288, "grad_norm": 3.7544896602630615, "learning_rate": 5.113503398454503e-07, "loss": 0.02710127830505371, "step": 7466 }, { "epoch": 1.9384735202492211, "grad_norm": 3.6477978229522705, "learning_rate": 5.11255635734104e-07, "loss": 0.028174329549074173, "step": 7467 }, { "epoch": 1.9387331256490135, "grad_norm": 0.6764193773269653, "learning_rate": 5.11161327381501e-07, "loss": 0.0011624848702922463, "step": 7468 }, { "epoch": 1.9389927310488058, "grad_norm": 4.30207633972168, "learning_rate": 5.110674148043129e-07, "loss": 0.0230291448533535, "step": 7469 }, { "epoch": 1.939252336448598, "grad_norm": 1.9694098234176636, "learning_rate": 5.109738980191408e-07, "loss": 0.005008261650800705, "step": 7470 }, { "epoch": 1.9395119418483904, "grad_norm": 3.7021143436431885, "learning_rate": 5.108807770425167e-07, "loss": 0.0149680245667696, "step": 7471 }, { "epoch": 1.9397715472481827, "grad_norm": 2.4371538162231445, "learning_rate": 5.107880518909023e-07, "loss": 0.004642263986170292, "step": 7472 }, { "epoch": 1.940031152647975, "grad_norm": 3.477860689163208, "learning_rate": 5.106957225806889e-07, "loss": 0.018910657614469528, "step": 7473 }, { "epoch": 1.9402907580477673, "grad_norm": 1.0863395929336548, "learning_rate": 5.106037891281988e-07, "loss": 0.001989062177017331, "step": 7474 }, { "epoch": 1.9405503634475596, "grad_norm": 1.8641644716262817, "learning_rate": 5.105122515496832e-07, "loss": 0.0018083445029333234, "step": 7475 }, { "epoch": 1.940809968847352, "grad_norm": 2.4393088817596436, "learning_rate": 5.10421109861324e-07, "loss": 0.004276149906218052, "step": 7476 }, { "epoch": 1.9410695742471442, "grad_norm": 4.9450154304504395, "learning_rate": 5.103303640792333e-07, "loss": 0.010108881630003452, "step": 7477 }, { "epoch": 1.9413291796469365, "grad_norm": 4.175941467285156, "learning_rate": 5.102400142194524e-07, "loss": 0.02467319369316101, "step": 7478 }, { "epoch": 1.9415887850467288, "grad_norm": 2.703181743621826, "learning_rate": 5.101500602979533e-07, "loss": 0.006167146377265453, "step": 7479 }, { "epoch": 1.9418483904465211, "grad_norm": 1.4988462924957275, "learning_rate": 5.100605023306377e-07, "loss": 0.003301743883639574, "step": 7480 }, { "epoch": 1.9421079958463134, "grad_norm": 0.4415879547595978, "learning_rate": 5.099713403333375e-07, "loss": 0.0008393257739953697, "step": 7481 }, { "epoch": 1.942367601246106, "grad_norm": 3.90922212600708, "learning_rate": 5.098825743218146e-07, "loss": 0.013390257954597473, "step": 7482 }, { "epoch": 1.9426272066458983, "grad_norm": 3.404989242553711, "learning_rate": 5.097942043117604e-07, "loss": 0.010456270538270473, "step": 7483 }, { "epoch": 1.9428868120456906, "grad_norm": 1.9113701581954956, "learning_rate": 5.097062303187973e-07, "loss": 0.004566126503050327, "step": 7484 }, { "epoch": 1.9431464174454829, "grad_norm": 2.1467182636260986, "learning_rate": 5.096186523584765e-07, "loss": 0.008058281615376472, "step": 7485 }, { "epoch": 1.9434060228452752, "grad_norm": 1.2738224267959595, "learning_rate": 5.095314704462803e-07, "loss": 0.0076960185542702675, "step": 7486 }, { "epoch": 1.9436656282450675, "grad_norm": 3.3924708366394043, "learning_rate": 5.0944468459762e-07, "loss": 0.029013823717832565, "step": 7487 }, { "epoch": 1.9439252336448598, "grad_norm": 3.47450852394104, "learning_rate": 5.093582948278375e-07, "loss": 0.010014649480581284, "step": 7488 }, { "epoch": 1.9441848390446521, "grad_norm": 3.902733325958252, "learning_rate": 5.092723011522047e-07, "loss": 0.029429053887724876, "step": 7489 }, { "epoch": 1.9444444444444444, "grad_norm": 1.3862674236297607, "learning_rate": 5.091867035859232e-07, "loss": 0.0031933877617120743, "step": 7490 }, { "epoch": 1.9447040498442367, "grad_norm": 2.4329211711883545, "learning_rate": 5.091015021441246e-07, "loss": 0.00553261348977685, "step": 7491 }, { "epoch": 1.944963655244029, "grad_norm": 2.6989638805389404, "learning_rate": 5.090166968418707e-07, "loss": 0.012526881881058216, "step": 7492 }, { "epoch": 1.9452232606438213, "grad_norm": 0.13738296926021576, "learning_rate": 5.08932287694153e-07, "loss": 0.0003897622227668762, "step": 7493 }, { "epoch": 1.9454828660436139, "grad_norm": 0.5511132478713989, "learning_rate": 5.088482747158933e-07, "loss": 0.0015558003215119243, "step": 7494 }, { "epoch": 1.9457424714434062, "grad_norm": 3.375520706176758, "learning_rate": 5.087646579219432e-07, "loss": 0.009588589891791344, "step": 7495 }, { "epoch": 1.9460020768431985, "grad_norm": 2.4299473762512207, "learning_rate": 5.086814373270836e-07, "loss": 0.001543646096251905, "step": 7496 }, { "epoch": 1.9462616822429908, "grad_norm": 1.402614712715149, "learning_rate": 5.085986129460268e-07, "loss": 0.004523990675806999, "step": 7497 }, { "epoch": 1.946521287642783, "grad_norm": 0.09082681685686111, "learning_rate": 5.085161847934141e-07, "loss": 0.00029743913910351694, "step": 7498 }, { "epoch": 1.9467808930425754, "grad_norm": 2.750553846359253, "learning_rate": 5.084341528838165e-07, "loss": 0.008623376488685608, "step": 7499 }, { "epoch": 1.9470404984423677, "grad_norm": 1.7290539741516113, "learning_rate": 5.083525172317356e-07, "loss": 0.006333081517368555, "step": 7500 }, { "epoch": 1.94730010384216, "grad_norm": 2.251948595046997, "learning_rate": 5.082712778516029e-07, "loss": 0.001580868847668171, "step": 7501 }, { "epoch": 1.9475597092419523, "grad_norm": 2.805119514465332, "learning_rate": 5.081904347577796e-07, "loss": 0.009531518444418907, "step": 7502 }, { "epoch": 1.9478193146417446, "grad_norm": 3.0425848960876465, "learning_rate": 5.081099879645564e-07, "loss": 0.004720083437860012, "step": 7503 }, { "epoch": 1.948078920041537, "grad_norm": 1.3473397493362427, "learning_rate": 5.080299374861551e-07, "loss": 0.003066414035856724, "step": 7504 }, { "epoch": 1.9483385254413292, "grad_norm": 3.3014490604400635, "learning_rate": 5.079502833367265e-07, "loss": 0.024640705436468124, "step": 7505 }, { "epoch": 1.9485981308411215, "grad_norm": 3.264451742172241, "learning_rate": 5.078710255303515e-07, "loss": 0.01986062154173851, "step": 7506 }, { "epoch": 1.9488577362409139, "grad_norm": 3.3184399604797363, "learning_rate": 5.077921640810416e-07, "loss": 0.022022683173418045, "step": 7507 }, { "epoch": 1.9491173416407062, "grad_norm": 3.508613109588623, "learning_rate": 5.077136990027372e-07, "loss": 0.007980894297361374, "step": 7508 }, { "epoch": 1.9493769470404985, "grad_norm": 0.7518192529678345, "learning_rate": 5.076356303093091e-07, "loss": 0.0016127959825098515, "step": 7509 }, { "epoch": 1.9496365524402908, "grad_norm": 1.8305715322494507, "learning_rate": 5.075579580145585e-07, "loss": 0.0023186595644801855, "step": 7510 }, { "epoch": 1.949896157840083, "grad_norm": 2.288707971572876, "learning_rate": 5.074806821322155e-07, "loss": 0.0031196819618344307, "step": 7511 }, { "epoch": 1.9501557632398754, "grad_norm": 4.514589786529541, "learning_rate": 5.074038026759412e-07, "loss": 0.04178778454661369, "step": 7512 }, { "epoch": 1.9504153686396677, "grad_norm": 2.094794988632202, "learning_rate": 5.07327319659326e-07, "loss": 0.0019659949466586113, "step": 7513 }, { "epoch": 1.95067497403946, "grad_norm": 1.8002856969833374, "learning_rate": 5.072512330958902e-07, "loss": 0.0019329142523929477, "step": 7514 }, { "epoch": 1.9509345794392523, "grad_norm": 3.84855318069458, "learning_rate": 5.071755429990844e-07, "loss": 0.031120218336582184, "step": 7515 }, { "epoch": 1.9511941848390446, "grad_norm": 2.231590747833252, "learning_rate": 5.071002493822885e-07, "loss": 0.006069133058190346, "step": 7516 }, { "epoch": 1.951453790238837, "grad_norm": 4.737893581390381, "learning_rate": 5.070253522588128e-07, "loss": 0.0069193653762340546, "step": 7517 }, { "epoch": 1.9517133956386292, "grad_norm": 1.5680745840072632, "learning_rate": 5.06950851641898e-07, "loss": 0.003552709473297, "step": 7518 }, { "epoch": 1.9519730010384215, "grad_norm": 2.093797445297241, "learning_rate": 5.068767475447131e-07, "loss": 0.005399708170443773, "step": 7519 }, { "epoch": 1.9522326064382138, "grad_norm": 5.292938709259033, "learning_rate": 5.068030399803587e-07, "loss": 0.01882251165807247, "step": 7520 }, { "epoch": 1.9524922118380061, "grad_norm": 3.397193431854248, "learning_rate": 5.067297289618643e-07, "loss": 0.007331852801144123, "step": 7521 }, { "epoch": 1.9527518172377984, "grad_norm": 3.8249762058258057, "learning_rate": 5.0665681450219e-07, "loss": 0.050135813653469086, "step": 7522 }, { "epoch": 1.9530114226375908, "grad_norm": 5.172621250152588, "learning_rate": 5.065842966142251e-07, "loss": 0.016378436237573624, "step": 7523 }, { "epoch": 1.953271028037383, "grad_norm": 1.5801786184310913, "learning_rate": 5.065121753107889e-07, "loss": 0.010786940343677998, "step": 7524 }, { "epoch": 1.9535306334371754, "grad_norm": 3.912216901779175, "learning_rate": 5.064404506046314e-07, "loss": 0.03159705922007561, "step": 7525 }, { "epoch": 1.9537902388369677, "grad_norm": 4.538243770599365, "learning_rate": 5.063691225084313e-07, "loss": 0.004434492439031601, "step": 7526 }, { "epoch": 1.95404984423676, "grad_norm": 0.9332162141799927, "learning_rate": 5.062981910347979e-07, "loss": 0.0056869760155677795, "step": 7527 }, { "epoch": 1.9543094496365523, "grad_norm": 0.711935818195343, "learning_rate": 5.062276561962705e-07, "loss": 0.001306129852309823, "step": 7528 }, { "epoch": 1.9545690550363446, "grad_norm": 5.159382343292236, "learning_rate": 5.06157518005318e-07, "loss": 0.017944209277629852, "step": 7529 }, { "epoch": 1.9548286604361371, "grad_norm": 3.675494909286499, "learning_rate": 5.060877764743391e-07, "loss": 0.0030859257094562054, "step": 7530 }, { "epoch": 1.9550882658359294, "grad_norm": 2.978135585784912, "learning_rate": 5.060184316156627e-07, "loss": 0.011935480870306492, "step": 7531 }, { "epoch": 1.9553478712357217, "grad_norm": 3.212930202484131, "learning_rate": 5.059494834415468e-07, "loss": 0.01008344255387783, "step": 7532 }, { "epoch": 1.955607476635514, "grad_norm": 1.2012085914611816, "learning_rate": 5.058809319641804e-07, "loss": 0.005651572719216347, "step": 7533 }, { "epoch": 1.9558670820353063, "grad_norm": 0.3340325653553009, "learning_rate": 5.05812777195682e-07, "loss": 0.0006922051543369889, "step": 7534 }, { "epoch": 1.9561266874350987, "grad_norm": 1.6596070528030396, "learning_rate": 5.057450191480994e-07, "loss": 0.0021932916715741158, "step": 7535 }, { "epoch": 1.956386292834891, "grad_norm": 1.774484395980835, "learning_rate": 5.056776578334107e-07, "loss": 0.009298348799347878, "step": 7536 }, { "epoch": 1.9566458982346833, "grad_norm": 3.714399814605713, "learning_rate": 5.056106932635236e-07, "loss": 0.011387830600142479, "step": 7537 }, { "epoch": 1.9569055036344756, "grad_norm": 3.9426965713500977, "learning_rate": 5.055441254502768e-07, "loss": 0.016678594052791595, "step": 7538 }, { "epoch": 1.9571651090342679, "grad_norm": 5.171436786651611, "learning_rate": 5.054779544054372e-07, "loss": 0.005473261699080467, "step": 7539 }, { "epoch": 1.9574247144340602, "grad_norm": 3.9053659439086914, "learning_rate": 5.054121801407024e-07, "loss": 0.014592276886105537, "step": 7540 }, { "epoch": 1.9576843198338525, "grad_norm": 3.9800186157226562, "learning_rate": 5.053468026677003e-07, "loss": 0.02109837532043457, "step": 7541 }, { "epoch": 1.957943925233645, "grad_norm": 4.648258209228516, "learning_rate": 5.052818219979871e-07, "loss": 0.007756874896585941, "step": 7542 }, { "epoch": 1.9582035306334373, "grad_norm": 4.399341106414795, "learning_rate": 5.05217238143051e-07, "loss": 0.027858642861247063, "step": 7543 }, { "epoch": 1.9584631360332296, "grad_norm": 1.5582990646362305, "learning_rate": 5.051530511143083e-07, "loss": 0.005045957397669554, "step": 7544 }, { "epoch": 1.958722741433022, "grad_norm": 3.5696518421173096, "learning_rate": 5.050892609231061e-07, "loss": 0.031577471643686295, "step": 7545 }, { "epoch": 1.9589823468328142, "grad_norm": 6.237905979156494, "learning_rate": 5.050258675807208e-07, "loss": 0.009293805807828903, "step": 7546 }, { "epoch": 1.9592419522326066, "grad_norm": 0.18280668556690216, "learning_rate": 5.049628710983592e-07, "loss": 0.0005257332231849432, "step": 7547 }, { "epoch": 1.9595015576323989, "grad_norm": 1.3096593618392944, "learning_rate": 5.049002714871572e-07, "loss": 0.002358240308240056, "step": 7548 }, { "epoch": 1.9597611630321912, "grad_norm": 3.9322543144226074, "learning_rate": 5.048380687581814e-07, "loss": 0.010748708620667458, "step": 7549 }, { "epoch": 1.9600207684319835, "grad_norm": 3.320430040359497, "learning_rate": 5.047762629224276e-07, "loss": 0.029635272920131683, "step": 7550 }, { "epoch": 1.9602803738317758, "grad_norm": 1.825967788696289, "learning_rate": 5.047148539908217e-07, "loss": 0.0014070577453821898, "step": 7551 }, { "epoch": 1.960539979231568, "grad_norm": 0.2012101411819458, "learning_rate": 5.046538419742194e-07, "loss": 0.0006197935435920954, "step": 7552 }, { "epoch": 1.9607995846313604, "grad_norm": 2.225989580154419, "learning_rate": 5.045932268834061e-07, "loss": 0.005449979100376368, "step": 7553 }, { "epoch": 1.9610591900311527, "grad_norm": 2.855358600616455, "learning_rate": 5.045330087290976e-07, "loss": 0.03578701242804527, "step": 7554 }, { "epoch": 1.961318795430945, "grad_norm": 3.7371633052825928, "learning_rate": 5.044731875219385e-07, "loss": 0.03129754960536957, "step": 7555 }, { "epoch": 1.9615784008307373, "grad_norm": 1.196786880493164, "learning_rate": 5.044137632725041e-07, "loss": 0.0033757067285478115, "step": 7556 }, { "epoch": 1.9618380062305296, "grad_norm": 1.0430840253829956, "learning_rate": 5.043547359912995e-07, "loss": 0.0010404516942799091, "step": 7557 }, { "epoch": 1.962097611630322, "grad_norm": 3.900380849838257, "learning_rate": 5.042961056887591e-07, "loss": 0.009945428930222988, "step": 7558 }, { "epoch": 1.9623572170301142, "grad_norm": 3.5356245040893555, "learning_rate": 5.042378723752473e-07, "loss": 0.0416438952088356, "step": 7559 }, { "epoch": 1.9626168224299065, "grad_norm": 2.7029225826263428, "learning_rate": 5.041800360610583e-07, "loss": 0.0040435814298689365, "step": 7560 }, { "epoch": 1.9628764278296988, "grad_norm": 1.30736243724823, "learning_rate": 5.041225967564167e-07, "loss": 0.0024007284082472324, "step": 7561 }, { "epoch": 1.9631360332294912, "grad_norm": 1.2119827270507812, "learning_rate": 5.040655544714762e-07, "loss": 0.0075993905775249004, "step": 7562 }, { "epoch": 1.9633956386292835, "grad_norm": 3.057420253753662, "learning_rate": 5.040089092163204e-07, "loss": 0.01047925092279911, "step": 7563 }, { "epoch": 1.9636552440290758, "grad_norm": 3.2130532264709473, "learning_rate": 5.039526610009634e-07, "loss": 0.014728670939803123, "step": 7564 }, { "epoch": 1.963914849428868, "grad_norm": 3.667527437210083, "learning_rate": 5.038968098353482e-07, "loss": 0.0262027308344841, "step": 7565 }, { "epoch": 1.9641744548286604, "grad_norm": 3.480243444442749, "learning_rate": 5.038413557293478e-07, "loss": 0.014923986978828907, "step": 7566 }, { "epoch": 1.9644340602284527, "grad_norm": 4.102890968322754, "learning_rate": 5.037862986927657e-07, "loss": 0.008079590275883675, "step": 7567 }, { "epoch": 1.964693665628245, "grad_norm": 4.701935768127441, "learning_rate": 5.037316387353347e-07, "loss": 0.034703947603702545, "step": 7568 }, { "epoch": 1.9649532710280373, "grad_norm": 2.6245033740997314, "learning_rate": 5.036773758667168e-07, "loss": 0.005351499188691378, "step": 7569 }, { "epoch": 1.9652128764278296, "grad_norm": 4.128364562988281, "learning_rate": 5.036235100965053e-07, "loss": 0.020958058536052704, "step": 7570 }, { "epoch": 1.965472481827622, "grad_norm": 3.7432456016540527, "learning_rate": 5.035700414342218e-07, "loss": 0.023857446387410164, "step": 7571 }, { "epoch": 1.9657320872274142, "grad_norm": 4.1635661125183105, "learning_rate": 5.035169698893185e-07, "loss": 0.02096475660800934, "step": 7572 }, { "epoch": 1.9659916926272065, "grad_norm": 0.44552168250083923, "learning_rate": 5.034642954711773e-07, "loss": 0.0008029385935515165, "step": 7573 }, { "epoch": 1.9662512980269988, "grad_norm": 3.173351287841797, "learning_rate": 5.034120181891097e-07, "loss": 0.019787147641181946, "step": 7574 }, { "epoch": 1.9665109034267911, "grad_norm": 3.5787346363067627, "learning_rate": 5.033601380523576e-07, "loss": 0.047493550926446915, "step": 7575 }, { "epoch": 1.9667705088265834, "grad_norm": 0.709382951259613, "learning_rate": 5.033086550700915e-07, "loss": 0.003354399697855115, "step": 7576 }, { "epoch": 1.9670301142263757, "grad_norm": 3.395005702972412, "learning_rate": 5.03257569251413e-07, "loss": 0.03486979752779007, "step": 7577 }, { "epoch": 1.9672897196261683, "grad_norm": 1.552744746208191, "learning_rate": 5.032068806053528e-07, "loss": 0.004033694975078106, "step": 7578 }, { "epoch": 1.9675493250259606, "grad_norm": 3.835766553878784, "learning_rate": 5.031565891408709e-07, "loss": 0.024488944560289383, "step": 7579 }, { "epoch": 1.967808930425753, "grad_norm": 1.6068445444107056, "learning_rate": 5.031066948668585e-07, "loss": 0.0014461857499554753, "step": 7580 }, { "epoch": 1.9680685358255452, "grad_norm": 0.10932987183332443, "learning_rate": 5.030571977921354e-07, "loss": 0.00035253167152404785, "step": 7581 }, { "epoch": 1.9683281412253375, "grad_norm": 1.0270766019821167, "learning_rate": 5.030080979254515e-07, "loss": 0.007226345129311085, "step": 7582 }, { "epoch": 1.9685877466251298, "grad_norm": 2.173823356628418, "learning_rate": 5.029593952754865e-07, "loss": 0.009699328802525997, "step": 7583 }, { "epoch": 1.9688473520249221, "grad_norm": 3.3418495655059814, "learning_rate": 5.029110898508502e-07, "loss": 0.011384665966033936, "step": 7584 }, { "epoch": 1.9691069574247144, "grad_norm": 3.362823247909546, "learning_rate": 5.028631816600815e-07, "loss": 0.012110330164432526, "step": 7585 }, { "epoch": 1.9693665628245067, "grad_norm": 0.6954183578491211, "learning_rate": 5.028156707116497e-07, "loss": 0.0009924222249537706, "step": 7586 }, { "epoch": 1.969626168224299, "grad_norm": 0.38621988892555237, "learning_rate": 5.027685570139535e-07, "loss": 0.0005904160207137465, "step": 7587 }, { "epoch": 1.9698857736240913, "grad_norm": 3.7614824771881104, "learning_rate": 5.02721840575322e-07, "loss": 0.009810728952288628, "step": 7588 }, { "epoch": 1.9701453790238836, "grad_norm": 2.0405538082122803, "learning_rate": 5.02675521404013e-07, "loss": 0.007003099657595158, "step": 7589 }, { "epoch": 1.9704049844236762, "grad_norm": 4.527587413787842, "learning_rate": 5.026295995082151e-07, "loss": 0.02971731126308441, "step": 7590 }, { "epoch": 1.9706645898234685, "grad_norm": 3.0548954010009766, "learning_rate": 5.025840748960458e-07, "loss": 0.009370802901685238, "step": 7591 }, { "epoch": 1.9709241952232608, "grad_norm": 1.705471396446228, "learning_rate": 5.025389475755532e-07, "loss": 0.005753234960138798, "step": 7592 }, { "epoch": 1.971183800623053, "grad_norm": 3.7709975242614746, "learning_rate": 5.024942175547146e-07, "loss": 0.026797305792570114, "step": 7593 }, { "epoch": 1.9714434060228454, "grad_norm": 1.6960023641586304, "learning_rate": 5.024498848414372e-07, "loss": 0.00251772440969944, "step": 7594 }, { "epoch": 1.9717030114226377, "grad_norm": 3.440354585647583, "learning_rate": 5.024059494435582e-07, "loss": 0.020435528829693794, "step": 7595 }, { "epoch": 1.97196261682243, "grad_norm": 2.5186705589294434, "learning_rate": 5.023624113688442e-07, "loss": 0.018469233065843582, "step": 7596 }, { "epoch": 1.9722222222222223, "grad_norm": 3.495666027069092, "learning_rate": 5.023192706249917e-07, "loss": 0.029697053134441376, "step": 7597 }, { "epoch": 1.9724818276220146, "grad_norm": 3.061187982559204, "learning_rate": 5.022765272196273e-07, "loss": 0.00781929586082697, "step": 7598 }, { "epoch": 1.972741433021807, "grad_norm": 3.3190793991088867, "learning_rate": 5.022341811603066e-07, "loss": 0.025393791496753693, "step": 7599 }, { "epoch": 1.9730010384215992, "grad_norm": 1.7397433519363403, "learning_rate": 5.021922324545157e-07, "loss": 0.0037625478580594063, "step": 7600 }, { "epoch": 1.9732606438213915, "grad_norm": 2.6358320713043213, "learning_rate": 5.021506811096703e-07, "loss": 0.009153921157121658, "step": 7601 }, { "epoch": 1.9735202492211839, "grad_norm": 3.157693386077881, "learning_rate": 5.021095271331153e-07, "loss": 0.00942960474640131, "step": 7602 }, { "epoch": 1.9737798546209762, "grad_norm": 3.9211323261260986, "learning_rate": 5.020687705321261e-07, "loss": 0.012818707153201103, "step": 7603 }, { "epoch": 1.9740394600207685, "grad_norm": 2.585702657699585, "learning_rate": 5.020284113139077e-07, "loss": 0.011937536299228668, "step": 7604 }, { "epoch": 1.9742990654205608, "grad_norm": 3.7092247009277344, "learning_rate": 5.019884494855942e-07, "loss": 0.007992163300514221, "step": 7605 }, { "epoch": 1.974558670820353, "grad_norm": 1.5286412239074707, "learning_rate": 5.019488850542504e-07, "loss": 0.006021425593644381, "step": 7606 }, { "epoch": 1.9748182762201454, "grad_norm": 3.1394095420837402, "learning_rate": 5.0190971802687e-07, "loss": 0.014288831502199173, "step": 7607 }, { "epoch": 1.9750778816199377, "grad_norm": 5.846381664276123, "learning_rate": 5.018709484103771e-07, "loss": 0.013359444215893745, "step": 7608 }, { "epoch": 1.97533748701973, "grad_norm": 4.008836269378662, "learning_rate": 5.018325762116251e-07, "loss": 0.017703650519251823, "step": 7609 }, { "epoch": 1.9755970924195223, "grad_norm": 0.39530661702156067, "learning_rate": 5.017946014373977e-07, "loss": 0.0007432764396071434, "step": 7610 }, { "epoch": 1.9758566978193146, "grad_norm": 1.1522704362869263, "learning_rate": 5.017570240944074e-07, "loss": 0.005643286742269993, "step": 7611 }, { "epoch": 1.976116303219107, "grad_norm": 2.8384928703308105, "learning_rate": 5.017198441892972e-07, "loss": 0.0071398126892745495, "step": 7612 }, { "epoch": 1.9763759086188992, "grad_norm": 2.0863802433013916, "learning_rate": 5.016830617286399e-07, "loss": 0.002548264106735587, "step": 7613 }, { "epoch": 1.9766355140186915, "grad_norm": 1.189605474472046, "learning_rate": 5.016466767189378e-07, "loss": 0.0012110212119296193, "step": 7614 }, { "epoch": 1.9768951194184838, "grad_norm": 1.8383897542953491, "learning_rate": 5.016106891666227e-07, "loss": 0.0038485703989863396, "step": 7615 }, { "epoch": 1.9771547248182761, "grad_norm": 0.7927461266517639, "learning_rate": 5.015750990780564e-07, "loss": 0.001254072180017829, "step": 7616 }, { "epoch": 1.9774143302180685, "grad_norm": 3.7161872386932373, "learning_rate": 5.015399064595305e-07, "loss": 0.009969467297196388, "step": 7617 }, { "epoch": 1.9776739356178608, "grad_norm": 3.380990743637085, "learning_rate": 5.01505111317266e-07, "loss": 0.01870480738580227, "step": 7618 }, { "epoch": 1.977933541017653, "grad_norm": 2.9164295196533203, "learning_rate": 5.014707136574146e-07, "loss": 0.007025807164609432, "step": 7619 }, { "epoch": 1.9781931464174454, "grad_norm": 3.9902851581573486, "learning_rate": 5.014367134860563e-07, "loss": 0.011643130332231522, "step": 7620 }, { "epoch": 1.9784527518172377, "grad_norm": 2.9930121898651123, "learning_rate": 5.014031108092017e-07, "loss": 0.009347246028482914, "step": 7621 }, { "epoch": 1.97871235721703, "grad_norm": 1.5672476291656494, "learning_rate": 5.01369905632791e-07, "loss": 0.007371821440756321, "step": 7622 }, { "epoch": 1.9789719626168223, "grad_norm": 2.6866440773010254, "learning_rate": 5.013370979626944e-07, "loss": 0.007175474427640438, "step": 7623 }, { "epoch": 1.9792315680166146, "grad_norm": 1.749691128730774, "learning_rate": 5.013046878047113e-07, "loss": 0.0019077277975156903, "step": 7624 }, { "epoch": 1.979491173416407, "grad_norm": 1.1325353384017944, "learning_rate": 5.01272675164571e-07, "loss": 0.002581992419436574, "step": 7625 }, { "epoch": 1.9797507788161994, "grad_norm": 2.730212450027466, "learning_rate": 5.012410600479326e-07, "loss": 0.008243432268500328, "step": 7626 }, { "epoch": 1.9800103842159917, "grad_norm": 3.3539445400238037, "learning_rate": 5.012098424603852e-07, "loss": 0.020964641124010086, "step": 7627 }, { "epoch": 1.980269989615784, "grad_norm": 1.399735450744629, "learning_rate": 5.011790224074471e-07, "loss": 0.0017896509962156415, "step": 7628 }, { "epoch": 1.9805295950155763, "grad_norm": 2.5457799434661865, "learning_rate": 5.011485998945668e-07, "loss": 0.004383097868412733, "step": 7629 }, { "epoch": 1.9807892004153687, "grad_norm": 0.5460506081581116, "learning_rate": 5.01118574927122e-07, "loss": 0.0008879487868398428, "step": 7630 }, { "epoch": 1.981048805815161, "grad_norm": 4.037732124328613, "learning_rate": 5.010889475104205e-07, "loss": 0.027322940528392792, "step": 7631 }, { "epoch": 1.9813084112149533, "grad_norm": 4.523144245147705, "learning_rate": 5.010597176497002e-07, "loss": 0.01151201780885458, "step": 7632 }, { "epoch": 1.9815680166147456, "grad_norm": 4.5775628089904785, "learning_rate": 5.010308853501275e-07, "loss": 0.01102073211222887, "step": 7633 }, { "epoch": 1.9818276220145379, "grad_norm": 3.675381898880005, "learning_rate": 5.010024506168001e-07, "loss": 0.029989521950483322, "step": 7634 }, { "epoch": 1.9820872274143302, "grad_norm": 2.0657148361206055, "learning_rate": 5.009744134547438e-07, "loss": 0.006802580319344997, "step": 7635 }, { "epoch": 1.9823468328141225, "grad_norm": 3.4362144470214844, "learning_rate": 5.009467738689157e-07, "loss": 0.03134489431977272, "step": 7636 }, { "epoch": 1.9826064382139148, "grad_norm": 3.112071990966797, "learning_rate": 5.009195318642012e-07, "loss": 0.02340259589254856, "step": 7637 }, { "epoch": 1.9828660436137073, "grad_norm": 3.5536410808563232, "learning_rate": 5.008926874454166e-07, "loss": 0.01947358250617981, "step": 7638 }, { "epoch": 1.9831256490134996, "grad_norm": 4.016260623931885, "learning_rate": 5.008662406173071e-07, "loss": 0.011868325062096119, "step": 7639 }, { "epoch": 1.983385254413292, "grad_norm": 2.3600735664367676, "learning_rate": 5.008401913845478e-07, "loss": 0.008978668600320816, "step": 7640 }, { "epoch": 1.9836448598130842, "grad_norm": 1.517071008682251, "learning_rate": 5.008145397517439e-07, "loss": 0.005665549077093601, "step": 7641 }, { "epoch": 1.9839044652128766, "grad_norm": 3.0330655574798584, "learning_rate": 5.007892857234296e-07, "loss": 0.004568387754261494, "step": 7642 }, { "epoch": 1.9841640706126689, "grad_norm": 4.404943466186523, "learning_rate": 5.007644293040695e-07, "loss": 0.014390437863767147, "step": 7643 }, { "epoch": 1.9844236760124612, "grad_norm": 0.15960684418678284, "learning_rate": 5.007399704980578e-07, "loss": 0.0003634539607446641, "step": 7644 }, { "epoch": 1.9846832814122535, "grad_norm": 0.19246022403240204, "learning_rate": 5.007159093097181e-07, "loss": 0.0003238226636312902, "step": 7645 }, { "epoch": 1.9849428868120458, "grad_norm": 0.547163724899292, "learning_rate": 5.006922457433039e-07, "loss": 0.00045053320354782045, "step": 7646 }, { "epoch": 1.985202492211838, "grad_norm": 0.9471994638442993, "learning_rate": 5.006689798029981e-07, "loss": 0.0034115698654204607, "step": 7647 }, { "epoch": 1.9854620976116304, "grad_norm": 1.244896411895752, "learning_rate": 5.00646111492914e-07, "loss": 0.002591707743704319, "step": 7648 }, { "epoch": 1.9857217030114227, "grad_norm": 2.5013320446014404, "learning_rate": 5.006236408170939e-07, "loss": 0.004482487682253122, "step": 7649 }, { "epoch": 1.985981308411215, "grad_norm": 1.4660072326660156, "learning_rate": 5.006015677795102e-07, "loss": 0.002318429993465543, "step": 7650 }, { "epoch": 1.9862409138110073, "grad_norm": 1.0233521461486816, "learning_rate": 5.005798923840652e-07, "loss": 0.0021583077032119036, "step": 7651 }, { "epoch": 1.9865005192107996, "grad_norm": 3.401463747024536, "learning_rate": 5.005586146345901e-07, "loss": 0.009752778336405754, "step": 7652 }, { "epoch": 1.986760124610592, "grad_norm": 1.1861634254455566, "learning_rate": 5.005377345348463e-07, "loss": 0.003092085011303425, "step": 7653 }, { "epoch": 1.9870197300103842, "grad_norm": 3.808107376098633, "learning_rate": 5.005172520885257e-07, "loss": 0.035186585038900375, "step": 7654 }, { "epoch": 1.9872793354101765, "grad_norm": 3.767850160598755, "learning_rate": 5.004971672992484e-07, "loss": 0.015429653227329254, "step": 7655 }, { "epoch": 1.9875389408099688, "grad_norm": 4.637451648712158, "learning_rate": 5.004774801705649e-07, "loss": 0.0033011212944984436, "step": 7656 }, { "epoch": 1.9877985462097612, "grad_norm": 0.8024053573608398, "learning_rate": 5.00458190705956e-07, "loss": 0.005031614564359188, "step": 7657 }, { "epoch": 1.9880581516095535, "grad_norm": 2.992755174636841, "learning_rate": 5.00439298908831e-07, "loss": 0.00348443491384387, "step": 7658 }, { "epoch": 1.9883177570093458, "grad_norm": 0.3676956295967102, "learning_rate": 5.004208047825299e-07, "loss": 0.000504242954775691, "step": 7659 }, { "epoch": 1.988577362409138, "grad_norm": 1.1026217937469482, "learning_rate": 5.004027083303222e-07, "loss": 0.00506350863724947, "step": 7660 }, { "epoch": 1.9888369678089304, "grad_norm": 1.693337082862854, "learning_rate": 5.003850095554065e-07, "loss": 0.0012046564370393753, "step": 7661 }, { "epoch": 1.9890965732087227, "grad_norm": 1.9167406558990479, "learning_rate": 5.00367708460912e-07, "loss": 0.007462998386472464, "step": 7662 }, { "epoch": 1.989356178608515, "grad_norm": 3.6415131092071533, "learning_rate": 5.003508050498967e-07, "loss": 0.026118900626897812, "step": 7663 }, { "epoch": 1.9896157840083073, "grad_norm": 4.920409202575684, "learning_rate": 5.00334299325349e-07, "loss": 0.009738795459270477, "step": 7664 }, { "epoch": 1.9898753894080996, "grad_norm": 3.7956905364990234, "learning_rate": 5.003181912901865e-07, "loss": 0.008790184743702412, "step": 7665 }, { "epoch": 1.990134994807892, "grad_norm": 1.9135185480117798, "learning_rate": 5.003024809472572e-07, "loss": 0.0011316947638988495, "step": 7666 }, { "epoch": 1.9903946002076842, "grad_norm": 2.823338508605957, "learning_rate": 5.002871682993378e-07, "loss": 0.00449011567980051, "step": 7667 }, { "epoch": 1.9906542056074765, "grad_norm": 0.8347780108451843, "learning_rate": 5.002722533491354e-07, "loss": 0.0005846216809004545, "step": 7668 }, { "epoch": 1.9909138110072688, "grad_norm": 2.274026393890381, "learning_rate": 5.002577360992869e-07, "loss": 0.005083481315523386, "step": 7669 }, { "epoch": 1.9911734164070611, "grad_norm": 0.4572412073612213, "learning_rate": 5.002436165523583e-07, "loss": 0.000955139403231442, "step": 7670 }, { "epoch": 1.9914330218068534, "grad_norm": 4.188867568969727, "learning_rate": 5.002298947108454e-07, "loss": 0.005985077004879713, "step": 7671 }, { "epoch": 1.9916926272066457, "grad_norm": 4.175419807434082, "learning_rate": 5.002165705771746e-07, "loss": 0.011608187109231949, "step": 7672 }, { "epoch": 1.991952232606438, "grad_norm": 4.151275157928467, "learning_rate": 5.002036441537007e-07, "loss": 0.024120721966028214, "step": 7673 }, { "epoch": 1.9922118380062306, "grad_norm": 0.828506350517273, "learning_rate": 5.00191115442709e-07, "loss": 0.0011153031373396516, "step": 7674 }, { "epoch": 1.992471443406023, "grad_norm": 2.5184788703918457, "learning_rate": 5.001789844464145e-07, "loss": 0.006473041605204344, "step": 7675 }, { "epoch": 1.9927310488058152, "grad_norm": 5.629329204559326, "learning_rate": 5.001672511669613e-07, "loss": 0.013063925318419933, "step": 7676 }, { "epoch": 1.9929906542056075, "grad_norm": 3.578644037246704, "learning_rate": 5.001559156064238e-07, "loss": 0.008149993605911732, "step": 7677 }, { "epoch": 1.9932502596053998, "grad_norm": 2.8938522338867188, "learning_rate": 5.00144977766806e-07, "loss": 0.0042179906740784645, "step": 7678 }, { "epoch": 1.9935098650051921, "grad_norm": 1.569982647895813, "learning_rate": 5.001344376500408e-07, "loss": 0.004936933051794767, "step": 7679 }, { "epoch": 1.9937694704049844, "grad_norm": 2.3357036113739014, "learning_rate": 5.001242952579924e-07, "loss": 0.004555203951895237, "step": 7680 }, { "epoch": 1.9940290758047767, "grad_norm": 3.8786919116973877, "learning_rate": 5.001145505924532e-07, "loss": 0.007913514040410519, "step": 7681 }, { "epoch": 1.994288681204569, "grad_norm": 3.746588945388794, "learning_rate": 5.001052036551458e-07, "loss": 0.036595311015844345, "step": 7682 }, { "epoch": 1.9945482866043613, "grad_norm": 2.5042786598205566, "learning_rate": 5.000962544477227e-07, "loss": 0.007362606003880501, "step": 7683 }, { "epoch": 1.9948078920041536, "grad_norm": 2.9486844539642334, "learning_rate": 5.000877029717655e-07, "loss": 0.00700925849378109, "step": 7684 }, { "epoch": 1.995067497403946, "grad_norm": 3.941046953201294, "learning_rate": 5.000795492287867e-07, "loss": 0.006746686529368162, "step": 7685 }, { "epoch": 1.9953271028037385, "grad_norm": 2.964338541030884, "learning_rate": 5.000717932202271e-07, "loss": 0.0025159253273159266, "step": 7686 }, { "epoch": 1.9955867082035308, "grad_norm": 4.006583213806152, "learning_rate": 5.000644349474577e-07, "loss": 0.01894819550216198, "step": 7687 }, { "epoch": 1.995846313603323, "grad_norm": 1.9354352951049805, "learning_rate": 5.000574744117799e-07, "loss": 0.003839645767584443, "step": 7688 }, { "epoch": 1.9961059190031154, "grad_norm": 2.2868752479553223, "learning_rate": 5.000509116144236e-07, "loss": 0.009545192122459412, "step": 7689 }, { "epoch": 1.9963655244029077, "grad_norm": 4.182133674621582, "learning_rate": 5.000447465565491e-07, "loss": 0.008546622470021248, "step": 7690 }, { "epoch": 1.9966251298027, "grad_norm": 3.4699573516845703, "learning_rate": 5.000389792392462e-07, "loss": 0.026379946619272232, "step": 7691 }, { "epoch": 1.9968847352024923, "grad_norm": 3.8159492015838623, "learning_rate": 5.000336096635346e-07, "loss": 0.004520926624536514, "step": 7692 }, { "epoch": 1.9971443406022846, "grad_norm": 1.3513662815093994, "learning_rate": 5.000286378303634e-07, "loss": 0.0030995351262390614, "step": 7693 }, { "epoch": 1.997403946002077, "grad_norm": 4.032857418060303, "learning_rate": 5.000240637406114e-07, "loss": 0.04472598060965538, "step": 7694 }, { "epoch": 1.9976635514018692, "grad_norm": 0.3730095326900482, "learning_rate": 5.000198873950875e-07, "loss": 0.0008403674000874162, "step": 7695 }, { "epoch": 1.9979231568016615, "grad_norm": 4.271195411682129, "learning_rate": 5.000161087945296e-07, "loss": 0.005435860250145197, "step": 7696 }, { "epoch": 1.9981827622014539, "grad_norm": 3.4121007919311523, "learning_rate": 5.000127279396061e-07, "loss": 0.04163118451833725, "step": 7697 }, { "epoch": 1.9984423676012462, "grad_norm": 3.078913927078247, "learning_rate": 5.000097448309143e-07, "loss": 0.006627815309911966, "step": 7698 }, { "epoch": 1.9987019730010385, "grad_norm": 1.2514686584472656, "learning_rate": 5.000071594689815e-07, "loss": 0.0034825641196221113, "step": 7699 }, { "epoch": 1.9989615784008308, "grad_norm": 4.265908718109131, "learning_rate": 5.000049718542651e-07, "loss": 0.0327887237071991, "step": 7700 }, { "epoch": 1.999221183800623, "grad_norm": 1.7714189291000366, "learning_rate": 5.000031819871517e-07, "loss": 0.003244661958888173, "step": 7701 }, { "epoch": 1.9994807892004154, "grad_norm": 4.108805179595947, "learning_rate": 5.000017898679574e-07, "loss": 0.007375925313681364, "step": 7702 }, { "epoch": 1.9997403946002077, "grad_norm": 2.698460340499878, "learning_rate": 5.000007954969287e-07, "loss": 0.011394234374165535, "step": 7703 }, { "epoch": 2.0, "grad_norm": 0.22801631689071655, "learning_rate": 5.000001988742409e-07, "loss": 0.0005169946234673262, "step": 7704 } ], "logging_steps": 1, "max_steps": 7704, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }