{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 3852, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00025960539979231567, "grad_norm": 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 } ], "logging_steps": 1, "max_steps": 7704, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }